CN101047869A - Method and device for correction gamma property of vedio communication - Google Patents

Abstract

The present invention provides a method and device for correcting gamma characteristics of video communication: obtain the brightness histogram of the output brightness signal, convert the brightness histogram of the output brightness signal into a probability density function of the brightness distribution of the output brightness signal, and determine the brightness of the output brightness signal The extreme point of the distribution probability density function; establish the mathematical relationship between the extreme point of the output brightness signal brightness distribution probability density function and the extreme point of the input brightness signal brightness distribution probability density function; use the mathematical relationship of the extreme point to input , the mathematical relationship between the brightness distribution probability density function and the Gamma characteristic function of the output signal is converted into: at the extreme point, the mathematical relationship between the output brightness signal brightness distribution probability density function and the Gamma characteristic function; Solve the mathematical relationship to determine the gamma characteristic parameters and perform gamma correction on the gamma link. The blind gamma characteristic parameter determination method of the present invention improves the usability of Gamma correction and broadens the application range of Gamma correction.

Description

Method and device for correcting gamma characteristics of video communication

technical field

The invention relates to the technical field of video communication, in particular to a method and device for correcting gamma characteristics of video communication.

Background technique

Video communication, especially multi-party video communication, is currently being widely used with the rapid development of broadband networks. Domestically and internationally, video conferencing and videophone services are becoming basic services on NGN (Next Generation Network, Next Generation Network). Telecom operators in various countries also attach great importance to this market opportunity. It is expected that in the next few years, the video communication service will become an important business growth point for operators.

A key issue in the development of video communication services is to improve end-to-end user experience (User Experience or Quality of Experience). In user experience, in addition to network QoS such as packet loss, delay, jitter, R factor and other parameters, for video, the distortion (Distortion) of the brightness signal caused by the Gamma (gamma) nonlinearity caused by each link also affects An important factor in the end user experience.

At present, methods and technologies for improving end-to-end user experience are mainly focused on ensuring network QoS and video compression and encoding related pre-processing (Pre-processing, post-processing). There is a lack of attention and systematic solutions to the problem of brightness distortion caused by Gamma characteristics. However, the importance of this issue has attracted the attention of some major international telecom operators.

The Gamma feature is briefly introduced below.

The process of video communication is: the optical signal of the scene to be transmitted, such as characters, background, files, etc., enters the video communication terminal (hereinafter referred to as the terminal) such as camera/camera, etc., converts it into a digital image signal through A/D, and then compresses it. It is coded, transmitted, and reaches the terminal of the other party. Then, after decompression and decoding, it is restored to a digital image signal, and then displayed on a display device, and finally becomes an optical signal to be perceived by human eyes. In the above process, the image brightness signal has gone through multiple links. According to the definition, the Gamma characteristic is that the input-output relationship of the image brightness signal of a link is not linear, but a nonlinear relationship. The image luminance signal (Luminance) here is a generalized luminance signal, that is, the initial optical signal, to the electrical signal, and then to the digitized image luminance/grayscale signal. The signal at each stage contains the information of the image luminance signal. , therefore, in a broad sense, the image brightness signal has gone through multiple links.

The general model of the Gamma characteristic of a single link is shown in Figure 1.

In Fig. 1, the nonlinear relationship between the input luminance signal and the output luminance signal can be expressed as: L _out = G(L _in ), wherein, L _out is the output luminance signal, L _in is the input luminance signal, and the function G(.) is a non-linear function.

A typical Gamma characteristic example is shown in Figure 2.

The number marked in each square in FIG. 2 is a brightness value, and the gray scale of the square represents the brightness of the brightness signal. In Figure 2(a), the brightness of the upper row of grayscale squares increases linearly, that is, from 0.1 to 1.0, and the brightness of the lower row of grayscale squares increases according to the power function, that is, the lower row of grayscale The brightness of the square is affected by the distortion of Gamma nonlinearity. Figure 2(b) gives the Gamma characteristic represented by the curve.

When multiple links are cascaded or connected in series, the total Gamma characteristic is equal to the composition of the Gamma functions of each link.

The general model of the Gamma characteristic of cascading multiple links is shown in Figure 3.

In Fig. 3, the nonlinear relationship between the input luminance signal and the output luminance signal of each link is respectively: Lout=G ⁽¹⁾ (Lin), Lout=G ⁽²⁾ (Lin), Lout=G ⁽³⁾ ( Lin).

From this, it can be known that the composition of the Gamma function of each link is shown in formula (1):

G _CT (.)＝G ⁽¹⁾ (.)oG ⁽²⁾ (.)oG ⁽³⁾ (.)......G ^(n-1) (.)oG ⁽ⁿ⁾ (. )

(1)

l _out ＝G _CT (l _in )＝G ⁽ⁿ⁾ (G ^(n-1) (G ^(n-2) (...G ⁽²⁾ (G ⁽¹⁾ (l _in )) )))

Among them, "." represents the compound operation of the function. CT stands for cascaded total, which means cascaded total Gamma.

In practice, Gamma nonlinearity is caused by different reasons. Gamma characteristics of display devices such as CRT monitors are ideally:

L _out =L _in ^2.2 (2)

The ideal Gamma characteristics of the corresponding camera/camera are:

L _out =L _in ^0.45 (3)

From the origin of the Gamma problem, the Gamma problem originated from the CRT monitor, because its Gamma value is 2.2. In order to compensate for this non-linearity, a Gamma value of 0.45 is introduced into the camera. In this example, the form of the Gamma characteristic is a power function (Power Function). It should be noted that the input and output luminance signals here are normalized (Normalized) in their respective coordinate spaces, that is, 0≤L _out ≤1, 0≤L _in ≤1. For other types of displays such as liquid crystals, the form of the Gamma function may be different, or although the form is also a power function, the parameters are different.

Ideally, there is a linear relationship between the input luminance signal and the output luminance signal, that is, L _out =L _in . To obtain a linear relationship, Gamma Correction (Gamma Correction) must be performed on links with nonlinear Gamma characteristics.

The principle diagram of Gamma correction for a Gamma link is shown in Figure 4.

In Fig. 4, for a link, its Gamma characteristic is given, that is, L _out = Gg(L _in ), so that another correction link L _out = Gc(L _in ) can be cascaded with it to make the total The Gamma characteristic of Gamma becomes a true linear relationship, so as to achieve the purpose of correcting the nonlinearity of a given link.

Obviously, Gg(.) and Gc(.) are inverse functions of each other. In general, for a function, there may not be a solution to obtain its inverse function, and even if there is an inverse function, it cannot be obtained by calculation.

In practical applications, there are more cases where there are multiple Gamma links, and the principle diagram of Gamma correction for multiple Gamma links is shown in Fig. 5 .

In Fig. 5, the correction link needs to be inserted between the two given Gamma links before and after. The Gamma characteristic of the previous given link is L _out =Ga(L _in ), and the Gamma characteristic of the later given link is L _out =Gp(L _in ). At this time, Gc(.) in the calibration link is very complicated, and the relationship between Gc(.) and Ga(.) or Gp(.) is no longer a simple inverse function relationship.

To realize the above-mentioned Gamma correction method, the premise is that the Gamma characteristic parameter can be determined for a given Gamma link or a cascade of multiple given Gamma links. The Gamma characteristic parameter here is the parameter of the Gamma characteristic function curve.

In the general case of communications, corrections need to involve more than two communicating terminals. For example, in a two-party video communication, if the video of terminal A is transmitted to terminal B, then the correction of this video involves both the Gamma link on terminal A and the Gamma link on terminal B.

At present, there are two main methods for determining the Gamma characteristic parameters:

Method 1: Instrument measurement method. That is, some points on the Gamma characteristic function curve are measured by special instruments, and then the data fitting method is used to perform curve fitting to determine the Gamma characteristic parameters.

Method 2: The method of inputting a luminance signal and outputting a luminance signal is adopted. That is, for a single given Gamma link, as long as Gg(.) satisfies certain conditions, Gc(.) can be found to perform Gamma correction on Gg(.); for multiple given Gamma links, as long as Ga(.), Gp(.) satisfies certain conditions, and Gc(.) that performs Gamma correction on Ga(.) and Gp(.) can be found.

From the description of the above two methods, it can be seen that the current implementation methods for determining the Gamma characteristic parameters have a prerequisite, that is, the specific values of the input luminance signal and the output luminance signal of the Gamma link that needs to be determined for the Gamma characteristic parameter. It is to clearly know all the knowledge of the input luminance signal and the output luminance signal of the Gamma link. Therefore, the above two implementation methods for determining the Gamma characteristic parameters belong to non-blind measurement methods.

The realization principle of the non-blind measurement method is shown in Fig. 6 .

In Fig. 6, the gamma characteristic parameter measurement system measures the gamma characteristic parameters by using all the knowledge of the input luminance signal and the output luminance signal of the gamma link. The gamma link here can be a single gamma link or a cascaded gamma link.

However, the scope of application of the non-blind measurement method in practical applications is very limited, that is to say, the above preconditions cannot be satisfied in many cases. The following are examples of three common applications that cannot meet the above preconditions.

Application Scenario 1: For streaming media services and applications such as IPTV (Internet Protocol Television), due to the influence of the Gamma characteristics of the video input device during the program production process, when the program is broadcast, especially in the case of on-demand, it has already The Gamma characteristic of the video input device used to collect the video signal during the original program production cannot be obtained, and it is impossible to measure the Gamma characteristic parameter of the video input device.

Application Scenario 2: The above problems also exist for applications such as data conferencing. At present, the development of video conferencing is synchronized with the development of data conferencing, and the perfect combination of the two is of great significance to collaborative applications. In environments such as enterprises, there is a strong market demand for the aforementioned collaborative application services. However, in data conferencing applications, the source of many multimedia materials such as pictures is unreliable, it is difficult to obtain the Gamma characteristics of the video input device that generated these data at that time, and it is also impossible to carry out the Gamma characteristic parameters of the video input device Measured.

Application Scenario 3: For public video communication services for tens of millions of household users, in order to reduce costs and use thresholds of video communication services, a large number of cheap cameras are often used, especially those very cheap USB interface cameras. The Gamma characteristic curve of these cheap video input devices is far from the standard L _out =L _in ^0.45 , and is not even in the form of a power function at all. Moreover, it is generally impossible to obtain the Gamma characteristic parameters of these cheap cameras from the factory technical data. Even some cheap cameras do not have factory technical data at all. When the user uses these cameras at home, it is also impossible to obtain the Gamma characteristic parameter through the above-mentioned method for determining the Gamma characteristic parameter.

The above three applications are very important, and all of the above three applications have great market potential, especially the IPTV and collaborative data conferencing markets are developing very fast. If video communication is really used in a huge market, it must rely on the road of public operation to attract thousands of households. This requires that the entry requirements must be very low, and the price of video input equipment must be very low. The current non-blind measurement method for determining Gamma characteristic parameters makes Gamma correction difficult to apply, and the method of measuring with special instruments increases the implementation cost of video communication services.

Contents of the invention

The object of the present invention is to provide a method and device for correcting gamma characteristics of video communication, which uses the histogram of the output brightness signal to determine gamma characteristic parameters, so as to reduce the cost of video communication and improve the usability of gamma correction.

In order to achieve the above object, a method for correcting gamma characteristics of video communication provided by the present invention includes:

a. Acquiring a luminance histogram of the output luminance signal;

b. converting the luminance histogram of the output luminance signal into a luminance distribution probability density function of the output luminance signal, and determining the extreme point of the luminance distribution probability density function of the output luminance signal;

c. Establishing the mathematical relationship between the extreme point of the brightness distribution probability density function of the output brightness signal and the extreme value point of the brightness distribution probability density function of the input brightness signal;

d. Using the extreme point and the mathematical relationship between the extreme points to convert the mathematical relationship between the respective brightness distribution probability density functions and Gamma characteristic functions of the input and output signals to: at the extreme point, the output brightness The mathematical relationship between the signal brightness distribution probability density function and the Gamma characteristic function;

e. Solving the converted mathematical relationship to determine the gamma characteristic parameter;

f. Performing gamma correction on the gamma link according to the gamma characteristic parameter.

The gamma link is: a single given gamma link, or a cascade combination of multiple given gamma links.

Described step b comprises:

Convert the luminance histogram of the output luminance signal into a polynomial form of the output luminance signal brightness distribution probability density function:

f _r (r) = c _d r ^d +c _d-1 r ^d-1 +c _d-2 r ^d-2 +.....+c ₁ r+c ₀ ;

Among them: c _d , c _d-1 , c _d-2 , ..., c ₀ are d+1 polynomial coefficients, and r is the amplitude of the input brightness signal.

Said step d comprises:

The mathematical relationship between the brightness distribution probability density functions f _e (x, t), f _r (x, t) and the Gamma characteristic function of the input and output brightness signals is:

d(e; p) f _r (r) = f _e (e);

Where: r=g(e,p), e∈[0,1];

The number of extreme points of the brightness distribution probability density function of the input and output brightness signals is the same, and the quantitative mathematical relationship is:

r _k = g(e _k ; p), p = [p ₁ , p ₂ , p ₃ , . . . , p _M ] ^T

Wherein: k=1,2,3,...,J; J is the number of extreme points of the brightness distribution probability density function;

At the extreme point, the mathematical relationship between the converted output brightness signal brightness distribution probability density function and its derivative function and the Gamma characteristic function and its inverse function and derivative function is:

$\mathrm{<span\; class="notranslate">\; z</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; g</span>}}^{<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; ;</span>\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ;</span>\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; )</span>{\mathrm{<span\; class="notranslate">\; f</span>}}_{\mathrm{<span\; class="notranslate">\; r</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; d</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; g</span>}}^{<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; ;</span>\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ;</span>\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; )</span>\frac{{\mathrm{<span\; class="notranslate">\; df</span>}}_{\mathrm{<span\; class="notranslate">\; r</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; )</span>}{\mathrm{<span\; class="notranslate">\; dr</span>}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1,2,3</span>}<span\; class="notranslate">\; ,</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; J</span>}<span\; class="notranslate">\; ;</span>$

为：Wherein: k=1,2,3,..., J, J is the extreme value point quantity of brightness distribution probability density function, r _k is the extreme value point of output brightness signal brightness distribution probability function, and, derivation function

for:

$\frac{{\mathrm{<span\; class="notranslate">\; df</span>}}_{\mathrm{<span\; class="notranslate">\; r</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; r</span>}<span\; class="notranslate">\; )</span>}{\mathrm{<span\; class="notranslate">\; dr</span>}}<span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; dc</span>}}_{\mathrm{<span\; class="notranslate">\; d</span>}}{\mathrm{<span\; class="notranslate">\; r</span>}}^{\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; +</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; )</span>{\mathrm{<span\; class="notranslate">\; c</span>}}_{\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; r</span>}}^{\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; +</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; )</span>{\mathrm{<span\; class="notranslate">\; c</span>}}_{\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 2</span>}}{\mathrm{<span\; class="notranslate">\; r</span>}}^{\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 3</span>}}<span\; class="notranslate">\; +</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; <figure-callout\; id="1"\; label="c"\; filenames="A20061008286100301.png,A20061008286100333.png"\; state="\{\{state\}\}">c</figure-callout></span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; .</span>$

Described step e comprises:

Mathematical relationships using direct equation-solving methods, or nonlinear function optimization methods: $z(g^{-1}(r_k;p);p)f_r\left(r_k\right)+d^2\left(g^{-1}(r_k;p)p\right)\frac{{\mathrm{df}}_r\left(r_k\right)}{\mathrm{dr}}=0,k=\mathrm{1,2,3},....,J$ A solution is performed to determine the gamma characteristic parameters.

Nonlinear function optimization methods include:

According to the mathematical relationship between the output brightness signal brightness distribution probability density function and the Gamma characteristic function, the cost function is constructed:

$\mathrm{<span\; class="notranslate">\; J</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; J</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; z</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; g</span>}}^{<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; ;</span>\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ;</span>\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; )</span>{\mathrm{<span\; class="notranslate">\; f</span>}}_{\mathrm{<span\; class="notranslate">\; r</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; d</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; g</span>}}^{<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; ;</span>\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ;</span>\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; )</span>\frac{{\mathrm{<span\; class="notranslate">\; df</span>}}_{\mathrm{<span\; class="notranslate">\; r</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; )</span>}{\mathrm{<span\; class="notranslate">\; dr</span>}}<span\; class="notranslate">\; )</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; ;</span>$

Randomly select M-1 parameters among the M parameters, and reduce the dimension of the parameter space to M-1 dimensions;

The method of determining M-1 parameters is: determine the parameter vector p _true , so that for any p∈R ^M-1 , the relationship J(p)>=J(p _ture ) holds true, that is, find the global minimum point of the cost function, The parameter vector p _true is the value determined for the M-1 parameters;

Using the relation g(1; p)=1 in combination with p _true , the remaining gamma characteristic parameter is determined.

Methods for determining p _true include: combinatorial optimization methods, neural network methods, and brute force search methods.

The brute force search method comprises the steps of:

Divide the one-dimensional gamma characteristic parameter space into multiple hypercubes;

Select an initial search point, and perform a traversal search from the hypercube where the initial search point is located according to a predetermined order;

Calculate the geometric center coordinate Q of each hypercube entering in the search process, and calculate J (Q) according to the expression of the cost function and the hypercube entering in the search process;

If J(Q) is less than or equal to the predetermined threshold, or the hypercube searched meets the predetermined condition, then the geometric center coordinate Q of the hypercube entered in this search is used as p _true , and the search process ends;

Otherwise, the search process continues.

The initial search point is set according to the empirical value of the gamma characteristic parameter in practical application.

The traversal search includes: taking the hypercube where the initial search point is located as the initial hypercube, dividing the hypercube surrounding the initial hypercube into multi-layer hypercube arrays wrapping the previous layer in turn according to the distance from the initial hypercube, and layer to search.

In the brute force search method, the gamma characteristic parameter space whose dimension is reduced by one dimension is divided into a plurality of coarse-grained hypercubes, and the traversal search includes:

With the hypercube where the initial search point is located as the initial hypercube, the hypercube surrounding the initial hypercube is divided into multi-layer hypercube arrays that wrap the previous layer in turn according to the distance from the initial hypercube, and search layer by layer;

The searched hypercube that satisfies the conditions is used as a new gamma characteristic parameter space, and it is divided into finer-grained hypercubes, and so on, and a layer-by-layer search is performed from coarse to fine.

The present invention also provides a correction device for gamma characteristics, said device comprising: a histogram acquisition module, a first conversion module, an extreme point calculation module, a storage module, a second conversion module, a gamma characteristic parameter solving module and a gamma Calibration module;

Obtaining a histogram module: used to obtain the brightness histogram of the output brightness signal, and output it to the first conversion module;

The first conversion module: used to convert the brightness histogram of the output brightness signal received by it into a probability density function of the brightness distribution of the output brightness signal, and output it to the extreme point calculation module and the second conversion module;

Extreme point calculation module: used to calculate each extreme point of the brightness distribution probability density function of the output brightness signal received by it, and output to the second conversion module;

Storage module: used to store the mathematical relationship between the extreme point of the brightness distribution probability density function of the output brightness signal and the extreme point of the brightness distribution probability density function of the input brightness signal, and store the respective brightness distribution probability densities of the input and output brightness signals The mathematical relationship between the function and the Gamma characteristic function;

The second conversion module: used for converting the respective luminance distribution probability density functions of input and output luminance signals and The mathematical relationship between the Gamma characteristic functions is transformed into: at the extreme point, the mathematical relationship between the output brightness signal brightness distribution probability density function and the Gamma characteristic function;

Gamma characteristic parameter solving module: used for solving and calculating the converted mathematical relationship to determine the gamma characteristic parameter;

Gamma correction module: for performing gamma correction on the gamma link according to the gamma characteristic parameters.

The device is located in a video data source device, and/or in an intermediate device of a video communication network, and/or in a video data destination device. The present invention also provides a correction device for gamma characteristics, comprising:

Known by the description of above-mentioned technical scheme, the Gamma correction method that the present invention provides only needs to output the histogram of luminance signal, like this, the Gamma parameter determination method of the present invention can be referred to as full-blind measuring method; Because the present invention does not need to input luminance signal Therefore, the gamma correction method of the present invention has high application feasibility; the gamma correction method of the present invention is particularly suitable for IPTV, collaborative data conference, public video communication that widely uses low-end video input equipment; the present invention When adopting the brute force search method to determine the parameter of the gamma characteristic function, adopted layer by layer search, search from coarse to fine, select the methods such as initial search point according to actual situation, improved search efficiency; Thereby the technology provided by the present invention The solution achieves the purpose of improving the usability of Gamma correction, broadening the application range of Gamma correction, and improving user experience and service quality.

Description of drawings

Figure 1 is a schematic diagram of the model of the link Gamma characteristic;

Figure 2(a) is a schematic diagram of the Gamma characteristic;

Figure 2(b) is a schematic diagram of the Gamma characteristic II;

Fig. 3 is a schematic diagram of the model of the Gamma characteristic of cascading multiple links;

Fig. 4 is a schematic diagram of the Gamma correction principle of a Gamma link;

Fig. 5 is a schematic diagram of the Gamma correction principle for multiple Gamma links;

Fig. 6 is a schematic diagram of the realization principle of the non-blind measurement method in the prior art;

Fig. 7 is a schematic diagram of the implementation principle of the method for determining the full-blind Gamma characteristic parameter according to the embodiment of the present invention;

Figure 8 is an example diagram of a brightness histogram of a video signal;

9 is a schematic diagram of the relationship between the input signal and output brightness signal brightness distribution probability density function according to an embodiment of the present invention;

Figure 10(a) is a frame of input image;

Figure 10(b) is a frame of output image;

Fig. 10(c) is a schematic diagram of the corresponding relationship between the input signal and the output brightness signal brightness distribution probability density function extreme points;

11 is a schematic diagram of a probability density function of brightness distribution of an output brightness signal obtained through interpolation and data fitting according to an embodiment of the present invention;

FIG. 12 is a schematic diagram of the principle of a brute force search method according to an embodiment of the present invention;

Figure 13 is a schematic diagram of the initial hypercube and its peripheral multilayer hypercube in two dimensions according to the embodiment of the present invention;

Fig. 14 is a schematic diagram of a brute force search method according to an embodiment of the present invention.

Detailed ways

In many application scenarios, the specific value of the output luminance signal, that is, all knowledge, is known, but the specific value of the input luminance signal is not known, or even any knowledge of the input luminance signal is unknown. The present invention only utilizes the knowledge of the output luminance signal to determine the Gamma characteristic parameter for Gamma correction. Since no knowledge of the input luminance signal is used in the technical solution of the present invention, the method for determining the Gamma characteristic parameter in the technical solution of the present invention can be called a method for determining the Gamma characteristic parameter of a full blindness.

The implementation principle of the blind Gamma characteristic parameter determination method of the present invention is shown in FIG. 7 .

In FIG. 7 , for the link that needs to determine the Gamma characteristic, the present invention determines the Gamma characteristic parameter of the link according to the known knowledge of the output brightness signal. In the full-blind Gamma characteristic parameter determination method of the present invention, all the knowledge of the output luminance signal is known, but the present invention does not necessarily use all the knowledge of the output luminance signal.

After the Gamma characteristic parameter is determined according to the knowledge of the output luminance signal, the Gamma correction can be performed on the Gamma link according to the Gamma characteristic parameter. In the present invention, the link that needs to determine the Gamma characteristic can be a single given Gamma link, or a cascade combination of multiple given Gamma links.

The technical solution provided by the present invention will be described in detail below in conjunction with the accompanying drawings.

First, the present invention needs to obtain the brightness histogram of the output brightness signal, and the brightness histogram is shown in Fig. 8 . In FIG. 8 , the brightness of the image is set from 0 to 255 levels, and different brightness levels correspond to a brightness distribution probability.

Histogram is a technical term in the field of image processing technology. In fact, histogram is a discrete form of distribution probability density function.

The video signal is composed of continuous images frame by frame, and the histogram of the output brightness signal can be obtained from a certain frame image. The method of obtaining the histogram of the luminance signal from the image belongs to conventional technology, and will not be described in detail here.

Obtaining the histogram of the output luminance signal can also be performed in other stages, for example, when the output luminance signal is still a one-dimensional signal, the histogram of the output luminance signal is obtained. At this time, the output luminance signal has not been converted into an image.

There is a close relationship between the histogram information and the probability density function of the continuous distribution. Generally speaking, the brightness histogram can be directly obtained from the continuous distribution probability density function; conversely, the continuous distribution probability density function can also be obtained by data interpolation or fitting methods from the brightness histogram. In fact, there is a strict proportional-quantity relationship between the histogram information and the probability density function of the continuous distribution. The above relationship is explained as follows.

For a single given Gamma link or a cascaded combination of multiple given Gamma links, the entire set of luminance signals is {s(t)|t∈R, 0≤s(t)≤1}, where R represents The set of all real numbers. That is to say, the entire set of luminance signals is a set of non-negative time signals whose overall signal amplitude (amplitude) is less than or equal to 1. The value of the luminance signal here is non-negative, and the value of the luminance signal is determined according to the physical meaning, because the negative luminance has no physical meaning. After any signal is normalized, it must meet the condition that the signal amplitude is less than or equal to 1. The luminance signal here is a luminance signal in a general sense, so the following description is applicable to both the input luminance signal and the output luminance signal. For simplicity of description, the following takes the output brightness signal as an example to illustrate the relationship between the histogram information and the continuous distribution probability density function.

Due to the existence of random interference, these output brightness signals can be regarded as a random process. The statistical characteristics of these output luminance signals may be different, but the output signals can be classified according to the statistical characteristics of the signals, especially according to the distribution probability characteristics. As a random process, any signal has a distribution probability density function corresponding to it. If the random process is stable (here, the stability is strictly stable), then the distribution probability density function has nothing to do with time; if the random process is not stable , then the probability density function of this distribution may be time-dependent. Therefore, in general, for a random process s(t)(t∈R, 0≤s(t)≤1), the probability density function of its distribution can be represented by f _s (x, t), t∈R . If it is a stationary random process in the strict sense, f _s (x, t), t∈R has nothing to do with t, that is, the distribution probability density function does not change with time. At this time, f _s (x, t)=f _s (x).

The normalization processing method of the signal is as follows:

If a signal s(t) does not satisfy the condition t∈R, 0≤s(t)≤1, then the signal needs to be normalized so that it satisfies t∈R, 0≤s(t)≤1 . For example: if the actual value range of the signal is [0, S _max ], the normalized signal s _n (t) is:

s _n (t) = s (t) / S _max (4)

The subscript n in formula (4) means normalized in English, which means normalization.

Correspondingly, if the signal is restored from the normalized value to the actual value, that is, the signal is denormalized, and the calculation formula is as follows:

s(t)=S _max s _n (t) (5)

According to the definition of the distribution probability density function, the distribution probability density function has the following properties:

${\&Integral;}_{-\&infin;}^{+\&infin;}{f}_{\mathrm{the\; s}}(x,t)\mathrm{dx}=1,$ for any t

and (6)

f _s (x,t)≥0, for any t

Moreover, for a non-negative signal with a signal amplitude less than or equal to 1, satisfy:

f _s (x, t)=0, x<0 or x>1 (7)

That is, it is impossible for a signal value to be greater than 1 or less than 0, and the probability is zero.

As a natural corollary it is:

${\&Integral;}_0^1{f}_{\mathrm{the\; s}}(x,t)\mathrm{dx}=1,$ for any t(8)

According to the definition of the probability density function, for a very small interval length δ and a point x ₀ on the interval [0, 1], f _s (x ₀ , t)δ≈Prob{x ₀ ≤s(t)≤x ₀ +δ} (9)

or equivalently

${\mathrm{<span\; class="notranslate">\; f</span>}}_{\mathrm{<span\; class="notranslate">\; the\; s</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; 0</span>}}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; t</span>}<span\; class="notranslate">\; )</span>\mathrm{<span\; class="notranslate">\; \&delta;</span>}<span\; class="notranslate">\; \&ap;</span>\mathrm{<span\; class="notranslate">\; Prob</span>}<span\; class="notranslate">\; \{</span>{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; 0</span>}}<span\; class="notranslate">\; -</span>\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; 2</span>}}\mathrm{<span\; class="notranslate">\; \&delta;</span>}<span\; class="notranslate">\; \&le;</span>\mathrm{<span\; class="notranslate">\; the\; s</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; t</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; \&le;</span>{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; 0</span>}}<span\; class="notranslate">\; +</span>\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; 2</span>}}\mathrm{<span\; class="notranslate">\; \&delta;</span>}<span\; class="notranslate">\; \}</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 10</span>}<span\; class="notranslate">\; )</span>$

Among them: the symbol Prob represents the probability (Probability).

Its intuitive meaning is that at time t, the luminance signal falls in the interval [x ₀ , x ₀ +δ] or The probability of is approximately equal to f _s (x ₀ ,t)δ. This is actually a way to turn the probability density function of a continuous distribution into a discrete probability density. It can be seen that, from the continuous probability density function, the brightness histogram of the signal can be obtained through such discretization.

For a normalized luminance signal, the interval [0, 1] can be equally divided into N subintervals, and the length of each subinterval is 1/N. The kth (k=0, 1, 2, ..., N-1) subinterval is [k/N, (k+1)/N]. If N is large enough and 1/N is small enough, then it can be considered that:

$\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; N</span>}}{\mathrm{<span\; class="notranslate">\; f</span>}}_{\mathrm{<span\; class="notranslate">\; the\; s</span>}}<span\; class="notranslate">\; (</span>\frac{\mathrm{<span\; class="notranslate">\; 2</span>}\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; 2</span>}\mathrm{<span\; class="notranslate">\; N</span>}}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; t</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; Prob</span>}<span\; class="notranslate">\; \{</span>\frac{\mathrm{<span\; class="notranslate">\; k</span>}}{\mathrm{<span\; class="notranslate">\; N</span>}}<span\; class="notranslate">\; \&le;</span>\mathrm{<span\; class="notranslate">\; the\; s</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; t</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; \&le;</span>\frac{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; N</span>}}<span\; class="notranslate">\; \}</span><span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0,1,2</span>}<span\; class="notranslate">\; ,</span><span\; class="notranslate">\; \&Center\; Dot;</span><span\; class="notranslate">\; \&CenterDot;</span><span\; class="notranslate">\; \&CenterDot;</span><span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; N</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 11</span>}<span\; class="notranslate">\; )</span>$

Thus, a probability sequence (sequence) can be formed:

$<span\; class="notranslate">\; \{</span>\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; N</span>}}{\mathrm{<span\; class="notranslate">\; f</span>}}_{\mathrm{<span\; class="notranslate">\; the\; s</span>}}<span\; class="notranslate">\; (</span>\frac{\mathrm{<span\; class="notranslate">\; 2</span>}\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; 2</span>}\mathrm{<span\; class="notranslate">\; N</span>}}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; t</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; |</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0,1,2</span>}<span\; class="notranslate">\; .</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; N</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; \}</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 12</span>}<span\; class="notranslate">\; )</span>$

If the signal is restored to its non-normalized signal space, as in video communication, the luminance signal usually takes an integer from 0 to 255, with a total of 256 levels of luminance. Of course, the luminance signal can also be generalized to ^2D level luminance. , at this time, it is necessary to linearly map the unit interval [0, 1] into a set {0, 1, 2, 3, ..., 2 ^D -2, 2 ^D -1}, and each sub-interval is correspondingly enlarged by 2 ^D times , becomes (1/N)2 ^D . The corresponding probability sequence then becomes a continuous probability density function:

$<span\; class="notranslate">\; \{</span>\mathrm{<span\; class="notranslate">\; h</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; |</span>\mathrm{<span\; class="notranslate">\; h</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; N</span>}}{\mathrm{<span\; class="notranslate">\; f</span>}}_{\mathrm{<span\; class="notranslate">\; the\; s</span>}}<span\; class="notranslate">\; (</span>\frac{\mathrm{<span\; class="notranslate">\; 2</span>}\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; 2</span>}\mathrm{<span\; class="notranslate">\; N</span>}}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; t</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0,1,2</span>}<span\; class="notranslate">\; .</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; N</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; \}</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 13</span>}<span\; class="notranslate">\; )</span>$

According to formula (8) and formula (10), it is obvious that:

$\underset{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0</span>}}{\overset{{\mathrm{<span\; class="notranslate">\; 2</span>}}^{\mathrm{<span\; class="notranslate">\; D.</span>}}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}\mathrm{<span\; class="notranslate">\; h</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 14</span>}<span\; class="notranslate">\; )</span>$

This probability sequence in formula (13) is called the histogram of the luminance signal s(t).

From the above derivation, it can be clearly seen that the histogram can be directly obtained from the probability density function of the continuous distribution of the signal brightness, and conversely, the probability density function of the continuous distribution of the brightness signal can also be processed by the histogram through data interpolation and fitting. after getting.

A specific example of a histogram is given below. When the brightness of the signal includes 256 brightness levels, the corresponding relationship between the probability sequence and the specific values in the histogram is:

h(0)=0

h(1)=0

....

h(64)=0.005

h(65)=0.006

....

h(190)=0.006

h(191)=0.005

h(192)=0.001

h(193)=0

...

h(255)=0

In the present invention, the specific expression form of setting the gamma characteristic function is clearly known, and one of the two forms of the present invention as an embodiment can be selected for use. Of course, it can also be a function of other forms, as long as it is continuous smooth and at least second-order differentiable.

Below, the function y=g(x; p), p=[p ₁ , p ₂ ,..., p _M ] ^T is used to represent the Gamma characteristic of the Gamma link whose Gamma characteristic parameter is unknown, where the Gamma link includes a single Gamma link Or the cascade combination of multiple Gamma links. In the above expression of the Gamma characteristic function, p=[p ₁ , p ₂ , . . . , P _M ] ^T is a parameter vector, and generally, the parameter vector is composed of M parameters. All or some of these parameters need to be determined. Therefore, according to this very general form, the Gamma characteristic function can be almost any form of function, as long as the function is continuous, and, generally speaking, the Gamma characteristic function is smooth and derivable, at least piecewise smooth Differentiable, therefore, it is reasonable to assume that there are first and second derivatives of the Gamma characteristic function with respect to the variable x. The first derivative of the Gamma characteristic function can be represented by the following symbols:

$\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ;</span>\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; d\; g</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ;</span>\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; )</span>}{\mathrm{<span\; class="notranslate">\; dx</span>}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 15</span>}<span\; class="notranslate">\; )</span>$

And, the Gamma characteristic function should also satisfy:

g(1;p)=1 (16)

Generally speaking, the Gamma characteristic function can be expressed in the following two common ways:

Method 1, power function:

$\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; g</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ;</span>\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; p</span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; <figure-callout\; id="2"\; label="x\; p"\; filenames="A20061008286100301.png,A20061008286100333.png"\; state="\{\{state\}\}">x</figure-callout></span><figure-callout\; id="2"\; label="x\; p"\; filenames="A20061008286100301.png,A20061008286100333.png"\; state="\{\{state\}\}">\; </figure-callout>}}^{{\mathrm{<span\; class="notranslate">\; </span>}}_{}}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; <figure-callout\; id="3"\; label="p"\; filenames="A20061008286100301.png"\; state="\{\{state\}\}">p</figure-callout></span>}}_{\mathrm{<span\; class="notranslate">\; 3</span>}}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; =</span>{<span\; class="notranslate">\; [</span>{\mathrm{<span\; class="notranslate">\; <figure-callout\; id="1"\; label="p"\; filenames="A20061008286100301.png,A20061008286100333.png"\; state="\{\{state\}\}">p</figure-callout></span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; <figure-callout\; id="2"\; label="p"\; filenames="A20061008286100301.png,A20061008286100333.png"\; state="\{\{state\}\}">p</figure-callout></span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; p</span>}}_{\mathrm{<span\; class="notranslate">\; 3</span>}}<span\; class="notranslate">\; ]</span>}^{\mathrm{<span\; class="notranslate">\; T</span>}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 17</span>}<span\; class="notranslate">\; )</span>$

Method 2, polynomial function:

y=g(x;p)=p ₁ x ^K +p ₂ x ^K-1 +...+p _K x+p _K+1 , where p=[p ₁ , p ₂ , p ₃ ,. .., p _K+1 ] ^T (18)

The above formula (18) can also be transformed into the expression form of formula (19):

y=g(x;p)=p ₁ (xx ₀ ) ^K +p ₂ (xx ₀ ) ^K-1 +....+p _K (xx ₀ )+p _K+1 (19)

Wherein, p=[p ₁ , p ₂ , p ₃ , . . . , p _K+1 , x ₀ ] ^T .

If e(t) and r(t) are used to denote the input luminance signal and output luminance signal respectively, then the respective distribution probability density functions of e(t) and r(t) are: f _e (x, t) and f _r (x, t), and there is the following relationship between e(t), r(t) and the Gamma characteristic function: r(t)=g(e(t); p), p=[p ₁ , p ₂ , p ₃ , ..., p _M ] ^T .

According to probability theory, it can be deduced that:

d(e; p) f _r (r) = f _e (e), where

  (20)

r=g(e;p), e∈[0, 1]

The specific process of deriving formula (20) can refer to commonly used probability books, and will not be described in detail in this embodiment.

It can be seen from formula (20) that formula (20) has nothing to do with the time variable t. In fact, the Gamma characteristic function itself has nothing to do with the time variable. Therefore, as long as the Gamma characteristic parameters are measured once in a relatively long period of time, this set of Gamma characteristic parameters can be used throughout the communication process. For example, in IPTV, a program's The Gamma characteristic parameters can be considered to be the same, so it is enough to measure the Gamma characteristic parameters at most once at the beginning of each program.

After obtaining the brightness distribution probability density function of the output brightness signal, it is necessary to calculate each extreme point of the brightness distribution probability density function of the output brightness signal. Calculating each extreme point of the function belongs to conventional technology, and will not be described in detail in the embodiments of the present invention.

An important basis of the present invention is that there is a strict one-to-one correspondence between the extreme points of the probability density function of the brightness distribution of the input brightness signal and the extreme points of the probability density function of the brightness distribution of the output brightness signal. This corresponding relationship is shown in FIG. 10 .

In Fig. 10, graph (a) is a frame of input video signal, graph (b) is a frame of output video signal corresponding to graph (a), and the two curves in graph (c) are input luminance signal respectively The brightness distribution probability density function and the output brightness signal brightness distribution probability density function.

It can be seen from Figure (c) that the extreme points of the two curves are strictly one-to-one correspondence. This correspondence is:

e ₁ ->r ₁ (maximum point), e ₂ ->r ₂ (minimum point), e ₃ ->r ₃ (maximum point), e ₄ ->r ₄ (minimum point), e ₅ ->r ₅ (maximum point).

Without loss of generality, it is assumed that both the probability density function of the brightness distribution of the input brightness signal and the probability density function of the brightness distribution of the output brightness signal have J extreme points, which are e ₁ , e ₂ ,..., e _J and r ₁ , r ₂ ,...,r _J .

On the basis of this extremum point correspondence, the present invention is for the extremum point e _k of the luminance distribution probability density function of the input luminance signal and the extremum point r _k of the corresponding output luminance signal luminance distribution probability density function (k=1 , 2,..., J) has performed a reasonable approximation, that is, setting the formula (21) to hold:

r _k = g(e _k ; p), p = [p ₁ , p ₂ , p ₃ , . . . , p _M ] ^T k = 1, 2, 3, . . . , J (21)

Where: k=1, 2, 3, . . . , J; J is the number of extreme points of the probability density function of the brightness distribution.

Since the number of extreme points of the probability density function of the brightness distribution of the output brightness signal is the same as the number of extreme points of the probability density function of the brightness distribution of the input brightness signal, J here can be the extreme point of the probability density function of the brightness distribution of the output brightness signal The quantity can also be the number of extreme points of the probability density function of the brightness distribution of the input brightness signal.

Strictly speaking, the relationship r _k =g(e _k ; p), p=[p ₁ , p ₂ , p ₃ , . . . , p _M ] ^T is not valid, which can be proved mathematically. However, in practical applications, after analyzing a large number of actual video signals, it can be found that r _k and g(e _k ; p) are very approximately equal, therefore, the present invention can approximately consider that formula (21) is established. Of course, the formula (21) can also be slightly changed, such as adding an additive constant term or multiplying a proportional factor to one side of the equation. In the following description, the mathematical relationship between the extreme point of the probability density function of the brightness distribution of the output brightness signal and the extreme point of the probability density function of the brightness distribution of the input brightness signal is used as an example to describe the formula (21). After the formula (21) is slightly changed, the process of determining the gamma parameters is basically the same as the following description, and will not be described in detail here.

It can be seen from the above description that the extreme points of the probability density function of the brightness distribution of the input brightness signal and the extreme points of the probability density function of the brightness distribution of the output brightness signal have the following two relationships: Relationship 1, qualitative geometric topological relationship, that is, two There is a one-to-one correspondence between the extreme points of the two brightness distribution probability density functions, that is, the number of extreme points of the two brightness distribution probability density functions is the same. Relation 2. Quantitative mathematical relationship: r _k =g(e _k ; p), p=[p ₁ , p ₂ , p ₃ , . . . , p _M ] ^T . Of course, the mathematical relationship here allows for slight transformations.

Suppose f _e (x) and f _r (x) are continuously differentiable, and d(x; p) is continuously differentiable, that is, g(x; p) is second-order continuously differentiable. Like this, the present invention derivates both sides of formula (20), can obtain:

$z(e;p)f_r\left(r\right)+d^2(e;p)\frac{{\mathrm{df}}_r\left(r\right)}{\mathrm{dr}}=\frac{{\mathrm{df}}_e\left(e\right)}{\mathrm{de}},$ in

r=g(e;p), (22)

$\mathrm{<span\; class="notranslate">\; z</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; e</span>}<span\; class="notranslate">\; ;</span>\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; e</span>}<span\; class="notranslate">\; ;</span>\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; )</span>}{\mathrm{<span\; class="notranslate">\; de</span>}}<span\; class="notranslate">\; \&ForAll;</span>\mathrm{<span\; class="notranslate">\; e</span>}<span\; class="notranslate">\; \&Element;</span><span\; class="notranslate">\; [</span>\mathrm{<span\; class="notranslate">\; 0,1</span>}<span\; class="notranslate">\; ]</span>$

Since e ₁ , e ₂ , . . . , e _J are extreme points of f _e (x), therefore, the derivatives of f _e (x) at these extreme points are zero. Thus, the following formula exists:

$\frac{{\mathrm{<span\; class="notranslate">\; df</span>}}_{\mathrm{<span\; class="notranslate">\; e</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; e</span>}<span\; class="notranslate">\; )</span>}{\mathrm{<span\; class="notranslate">\; de</span>}}{<span\; class="notranslate">\; |</span>}_{\mathrm{<span\; class="notranslate">\; e</span>}<span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; e</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1,2,3</span>}<span\; class="notranslate">\; ,</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; J</span>}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; twenty\; three</span>}<span\; class="notranslate">\; )</span>$

Combining formula (22) and formula (23), the following relationship is established:

$\mathrm{<span\; class="notranslate">\; z</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; e</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; ;</span>\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; )</span>{\mathrm{<span\; class="notranslate">\; f</span>}}_{\mathrm{<span\; class="notranslate">\; r</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; r</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; d</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; e</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; ;</span>\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; )</span>\frac{{\mathrm{<span\; class="notranslate">\; df</span>}}_{\mathrm{<span\; class="notranslate">\; r</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; r</span>}<span\; class="notranslate">\; )</span>}{\mathrm{<span\; class="notranslate">\; dr</span>}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0</span>}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; twenty\; four</span>}<span\; class="notranslate">\; )</span>$

r=g(e _k ; p), k=1, 2, 3, ..., J

Since formula (21) is approximately true, formula (24) can be transformed into the following relationship:

$\mathrm{<span\; class="notranslate">\; z</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; e</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; ;</span>\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; )</span>{\mathrm{<span\; class="notranslate">\; f</span>}}_{\mathrm{<span\; class="notranslate">\; r</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; d</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; e</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; ;</span>\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; )</span>\frac{{\mathrm{<span\; class="notranslate">\; df</span>}}_{\mathrm{<span\; class="notranslate">\; r</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; )</span>}{\mathrm{<span\; class="notranslate">\; dr</span>}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1,2,3</span>}<span\; class="notranslate">\; ,</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; J</span>}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 25</span>}<span\; class="notranslate">\; )</span>$

$\mathrm{<span\; class="notranslate">\; z</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; g</span>}}^{<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; ;</span>\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ;</span>\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; )</span>{\mathrm{<span\; class="notranslate">\; f</span>}}_{\mathrm{<span\; class="notranslate">\; r</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; d</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; g</span>}}^{<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; ;</span>\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ;</span>\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; )</span>\frac{{\mathrm{<span\; class="notranslate">\; df</span>}}_{\mathrm{<span\; class="notranslate">\; r</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; )</span>}{\mathrm{<span\; class="notranslate">\; dr</span>}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; 3</span>}<span\; class="notranslate">\; ,</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; J</span>}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 26</span>}<span\; class="notranslate">\; )</span>$

为：In the formula (26), k=1,2,3,..., J, J is the number of extremum points of the brightness distribution probability density function, r _k is the extremum point of the output brightness signal brightness distribution probability function , and, the derivative function

for:

$\frac{{\mathrm{<span\; class="notranslate">\; df</span>}}_{\mathrm{<span\; class="notranslate">\; r</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; r</span>}<span\; class="notranslate">\; )</span>}{\mathrm{<span\; class="notranslate">\; dr</span>}}<span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; dc</span>}}_{\mathrm{<span\; class="notranslate">\; d</span>}}{\mathrm{<span\; class="notranslate">\; r</span>}}^{\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; +</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; )</span>{\mathrm{<span\; class="notranslate">\; c</span>}}_{\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; r</span>}}^{\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; +</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; )</span>{\mathrm{<span\; class="notranslate">\; c</span>}}_{\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 2</span>}}{\mathrm{<span\; class="notranslate">\; r</span>}}^{\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 3</span>}}<span\; class="notranslate">\; +</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; <figure-callout\; id="1"\; label="c"\; filenames="A20061008286100301.png,A20061008286100333.png"\; state="\{\{state\}\}">c</figure-callout></span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; .</span>$

In practical applications, the Gamma characteristic functions are all monotone, therefore, there is an inverse function of the Gamma characteristic function. This inverse function can be recorded as g ^-1 (x; p), obviously, the inverse function of the Gamma characteristic function also depends on the gamma parameter vector p.

It can be seen from formula (26) that formula (26) does not contain any information about the input signal, and formula (26) completely depends on the output brightness signal brightness distribution probability density function and the polarity of the output brightness signal brightness distribution probability density function value points. Therefore, the present invention can determine the Gamma characteristic parameter only according to the output signal and its brightness distribution probability density function.

The histogram of the output brightness signal obtained by the present invention is set as: {h _r (k)|k=0, 1, 2, . . . , N-1}. That is to say, each histogram contains N items, and each item is called a "column" (bin) in histogram terminology.

According to formula (12), the output brightness signal brightness distribution probability density function can be obtained from the output image brightness histogram:

$\left\{h_r\left(k\right)\right|h_r\left(k\right)=\frac{1}{N}{f}_r\left(\frac{2k+1}{2N}\right),k=\mathrm{0,1,2},...,N-1\},$ then:

${\mathrm{<span\; class="notranslate">\; f</span>}}_{\mathrm{<span\; class="notranslate">\; r</span>}}<span\; class="notranslate">\; (</span>\frac{\mathrm{<span\; class="notranslate">\; 2</span>}\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; 2</span>}\mathrm{<span\; class="notranslate">\; N</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; Nh</span>}}_{\mathrm{<span\; class="notranslate">\; r</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0,1,2</span>}<span\; class="notranslate">\; .</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; N</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 27</span>}<span\; class="notranslate">\; )</span>$

For N points evenly distributed on the interval [0, 1] $a_k=\frac{2k-1}{2N},k=\mathrm{1,2},...,N,$ The values of the function f _r (x) at these discrete points can be obtained, namely $f_r\left(\frac{2k+1}{2N}\right)k=\mathrm{0,1,2}...,N-1.$ The positions of these discrete points on the r-axis of the coordinate system are shown in Figure 11.

If N is large enough, then the expression of the output brightness distribution probability density can be obtained by interpolation or data fitting:

f _r (r)＝c _d r ^d +c _d-1 r ^d-1 +c _d-2 r ^d-2 +.....+c ₁ r+c ₀ (28)

Among them: c _d , c _d-1 , c _d-2 , ..., c ₀ are d+1 polynomial coefficients, and r is the amplitude of the input brightness signal.

The expression is a polynomial including a polynomial spline.

In general, for an image with 256 levels of brightness, N=256, at this time, N is already large enough. There are many ways to implement the interpolation or data fitting technology, and the present invention does not limit the specific implementation scope of the interpolation or data fitting.

The derivative of formula (28) is:

$\frac{{\mathrm{<span\; class="notranslate">\; df</span>}}_{\mathrm{<span\; class="notranslate">\; r</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; r</span>}<span\; class="notranslate">\; )</span>}{\mathrm{<span\; class="notranslate">\; dr</span>}}<span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; dc</span>}}_{\mathrm{<span\; class="notranslate">\; d</span>}}{\mathrm{<span\; class="notranslate">\; r</span>}}^{\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; +</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; )</span>{\mathrm{<span\; class="notranslate">\; c</span>}}_{\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; r</span>}}^{\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; +</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; )</span>{\mathrm{<span\; class="notranslate">\; c</span>}}_{\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 2</span>}}{\mathrm{<span\; class="notranslate">\; r</span>}}^{\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 3</span>}}<span\; class="notranslate">\; +</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; c</span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 29</span>}<span\; class="notranslate">\; )</span>$

The present invention can select (17), (18), the expression of the gamma characteristic function of any form of expression that provides in (19), at this moment, obtained the concrete form of Gamma characteristic function, only parameter vector p needs Sure. Any item in the formula (26) is calculated for each given value of the parameter vector p, that is, each multiplication factor and addition item in the formula (26) is calculated.

If there are J equations, as long as J≥M-1, where M is the number of gamma characteristic parameters that need to be determined, then the unique solution must be solved through this set of equations as the M-1 gamma parameters Determine the value, and then use formula (16) as a constraint condition to determine the constant term, that is, determine the remaining gamma parameter.

That is to say, due to the existence of formula (16), all gamma parameters satisfy a constraint condition. If there are M gamma parameters, then only M-1 parameters are independent, as long as it is determined that For any M-1 parameters among the M gamma parameters, the remaining gamma parameter can be solved by formula (16), so as to determine all the parameters of the Gamma characteristic function.

In general, the condition J≥M-1 can be satisfied.

There are many ways to obtain the parameter vector p according to the formula (26). The following mainly introduces two methods of determining the parameter vector p:

The first method is to solve the equation directly.

J equations are obtained from formula (26), from which M-1 equations are arbitrarily selected to form a system of equations to be solved simultaneously. Generally speaking, this set of equations is non-linear and Transcendental, such as the Gamma characteristic function in the form of a power function. Therefore there is no closed-form solution. A numerical solution method is required. The numerical solution of the equations belongs to conventional technology, and will not be described in detail in this embodiment.

Method 2, nonlinear function optimization method.

Construct a cost function according to formula (26):

$\mathrm{<span\; class="notranslate">\; J</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; J</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; z</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; g</span>}}^{<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; ;</span>\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ;</span>\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; )</span>{\mathrm{<span\; class="notranslate">\; f</span>}}_{\mathrm{<span\; class="notranslate">\; r</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; d</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; g</span>}}^{<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; ;</span>\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ;</span>\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; )</span>\frac{{\mathrm{<span\; class="notranslate">\; df</span>}}_{\mathrm{<span\; class="notranslate">\; r</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; )</span>}{\mathrm{<span\; class="notranslate">\; dr</span>}}<span\; class="notranslate">\; )</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 30</span>}<span\; class="notranslate">\; )</span>$

Obviously, for the real parameter vector p _true of the Gamma characteristic function, J(p _true )=0 should be made in theory, because the parameter vector p _true satisfies each equation in the formula (30), therefore, in the formula (30) Each summation term of is zero. But in practical applications, because there are errors and approximations such as the approximation in the process of obtaining the histogram, so each summation item in formula (30) will not be equal to zero, but should be a very small value, and The following condition should be satisfied: for any ^p∈RM , there is J(p)≥J(p _true ). That is, p _true is the global minimum point of the function J(p).

从而就可以计算出公式(30)中的每个求和项，也就可以计算出总的求和结果J。It can be known from the above description that for any of the Gamma characteristic functions given by formulas (17), (18) and (19), g can be calculated according to each given value of the parameter vector p ^{- 1} (r _k ; p), z(g ^-1 (r _k ; p); p), d ² (g ^-1 (r _k ; p); p), at the same time, the output brightness signal brightness distribution probability density function is known, so that f _r (r _k ) and

Thus, each summation item in formula (30) can be calculated, and the total summation result J can also be calculated.

Therefore, in the nonlinear function optimization method, the problem of determining the parameter vector p is transformed into a mathematical problem of finding the global minimum point of the cost function J(p).

In the nonlinear function optimization method, the parameter vector p can be determined in the following three ways.

Mode (1), conventional mathematical optimization method.

Since there is a derivative in J(p), classical mathematical optimization methods such as gradient method and conjugate gradient method can be used to determine the parameter vector p.

The specific process of determining the parameter vector p through a conventional mathematical optimization method belongs to conventional technology, and will not be described in detail in this embodiment.

Mode (2), neural network method.

The specific process of determining the parameter vector p through the neural network method belongs to conventional technology, and will not be described in detail in this embodiment.

Method (3), Brutal Force Search Method. The so-called brute force search, as the name suggests, is to exhaust all possibilities.

For the case where the parameter takes a discrete value, the set of all possible values of the parameter is a finite set, then, by searching each point in the set one by one, the point that makes J(p) the smallest can be found, so that the global minimum Point p _true . However, such cases are rare. In most cases, parameters take continuous values. Therefore, the set of all possible values of parameters is an infinite set, and exhaustive search cannot be truly performed.

For the case where the parameters take continuous values, the brute force search method is to divide the parameter space into multiple small hypercubes (Hypercube), and then take a point in each hypercube as a sampling point, such as the geometric center point of the hypercube, etc. , and finally, calculate the function value of the cost function at each sampling point of the hypercube, find the sampling point of the hypercube that minimizes the cost function, and use this sampling point as the global minimum point.

The implementation process of using the brute-force search method to determine the parameters of the Gamma characteristic function will be described in detail below for the case where the parameters take continuous values and in conjunction with the accompanying drawings.

Although the brute force search method is a bit "stupid", the brute force search method has a wide range of applications in engineering practice, such as password deciphering, etc. The most effective method is the brute force search method. Moreover, for mathematical optimization problems, except for brute force search, the existing technologies all have the problem of falling into a local minimum (local minimal point), but the brute force search method does not have this problem.

In the present invention, the prior knowledge about the parameters can be used to find a suitable initial search point, thus greatly reducing the number of searches required, thereby improving the efficiency of the brute force search method.

The principle of the brute force search method is shown in FIG. 12 .

In Fig. 12, the set of all possible values of the M-1 parameters constitutes the parameter space (ParameterSpace, referred to as PS), and the parameter space is a subset of the M-1-dimensional Euclidean space R ^M-1 . After the parameter space is determined, the implementation method of brute force search includes the following steps:

Step 1, hypercube division.

Divide the PS into a plurality of M-1 hypercubes (Hypercube), in Figure 12, ABCDEFGH, 8 points form a hypercube. Since the value range of each parameter is different, the length of each side of the hypercube is also different.

Set the value range of the kth (k=1, 2, ..., M-1) parameter p _k to [min _k , max _k ], and perform P _k equal division on the kth dimension, so that the dimension The edge length of each hypercube is

${\mathrm{<span\; class="notranslate">\; \&Delta;</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; =</span>\frac{{\mathrm{<span\; class="notranslate">\; max</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; min</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}}{{\mathrm{<span\; class="notranslate">\; P</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 31</span>}<span\; class="notranslate">\; )</span>$

The volume of each hypercube is thus:

$\mathrm{<span\; class="notranslate">\; V</span>}<span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; \&Pi;</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}^{\mathrm{<span\; class="notranslate">\; m</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; \&Delta;</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; \&Pi;</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}^{\mathrm{<span\; class="notranslate">\; m</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; (</span>\frac{{\mathrm{<span\; class="notranslate">\; max</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; min</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}}{{\mathrm{<span\; class="notranslate">\; P</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 32</span>}<span\; class="notranslate">\; )</span>$

Therefore, the total number of hypercubes is greater than $T={\mathrm{\&Pi;}}_{k=1}^{m-1}{P}_k.$ This value is the maximum possible value. If the shape of PS is not a hypercube, then the number of long cubes it contains may be much smaller than $T={\mathrm{\&Pi;}}_{k=1}^{m-1}{P}_k.$

For each hypercube, it is represented by an index vector I=[i ₁ , i ₂ , ..., i _M-1 ] ^T , where i _k (k=1, 2, ..., M-1, i _k =1, 2, 3, ..., P _k ) means that on the kth dimension, the hypercube is the i _kth .

Therefore, for the I=[i ₁ , i ₂ ,...,i _M-1 ] ^T hypercube, its coordinate range in each dimension is:

[min _k +(i _k-1 )Δ _k ，min _k +i _k Δ _k ] (33)

The coordinates of the geometric center Q _I of this hypercube are:

[min ₁ +(i ₁ -1/2)Δ ₁ , min ₂ +(i ₂ -1/2)Δ ₂ ,......, min _M-1 +( i _M-1 -1/2)Δ _M-1 ] ^T (34)

Step 2. Select the initial search point.

Generally speaking, for a certain parameter p _k , there is a relatively reasonable value, which can be used as an initial value. For example, when the Gamma characteristic function adopts the form of a power function, for video input devices, the value of the Gamma parameter is generally around 2.2, because the industry standard requires 2.2, and due to manufacturing technology and product quality, the Gamma parameter may deviate positively or negatively. 2.2, but in most cases, closer to 2.2. In this way, if 2.2 is used as the initial search point to start searching, since 2.2 is closer to the real value, the number of times to find the real value is relatively small. In the same way, it is possible to find an appropriate initial value p _{int k} for each parameter p _k (k=1, 2, ..., M-1), then these initial values form a vector, which is the initial value of the parameter vector p Value p _int = [p _{int k} , p _{int k} , p _{int k} , . . . , p _int(M-1) ] ^T .

Step 3, start searching according to the selected initial search point.

First judge which hypercube the p _int falls in, and determine which hypercube the p _int falls in by comparing coordinates and other methods. The specific implementation process of the comparison coordinate method is as follows:

The coordinates of the hypercube are set to be I _int =[i _int1 , i _int2 ,...., i _int(M-1) ] ^T , and the judgment condition is: for k=0, 1, 2.....M- When 1, if and only if the following formula (35) is established, it is determined that the coordinates of p _int are: I _int = [i _int1 , i _int2 , ..., i _int(M-1) ] ^T super in the cube.

min _k +(i _intk -1)Δ _k ≤p _intk < min _k +i _intk Δ _k (35)

After determining the hypercube where the initial search point is located, start searching from this hypercube. Calculate J(p _int ) according to formula (30), if J(p _int ) can make formula (36) hold, or the searched hypercube meets the predetermined conditions, such as the number of searched hypercubes reaches the predetermined value, etc., then , the whole search process ends, go to step five. At this time, p _true = p _int ; wherein, p _true is the global optimal point, that is, the parameter vector of the finally obtained Gamma characteristic function.

J(p _int )≤J _threshold (36)

Wherein, J _threshold is a predetermined threshold value.

If J(p _int ) cannot make formula (36) valid, go to step four.

Step 4. Continue to search.

The continued search in step 4 may be performed hierarchically. The hypercube surrounding the initial hypercube can be one or more layers.

In the two-dimensional case, the initial hypercube and its surrounding multilayer hypercube are shown in Fig. 13 .

In Fig. 13, the square in the middle gray is the initial hypercube, the cube adjacent to the edge of the initial hypercube is the first layer hypercube, and the cube adjacent to the edge of the first layer hypercube is the second layer hypercube, and the second layer hypercube is the cube adjacent to the edge of the first layer hypercube. The edge cheap cube of the layer hypercube is the third layer hypercube.

Hierarchical search method of the present invention is: successively search each hypercube in each layer except initial hypercube, in the search of each layer hypercube, should traverse each hypercube in this layer according to predetermined order . The predetermined order can be various, and the present invention does not limit the form of the predetermined order, as long as the hypercube in one layer can be traversed.

During the search process of a certain layer of hypercube, when a certain hypercube is searched according to the predetermined order, it is necessary to calculate the coordinates of the geometric center Q of the hypercube according to the formula (34), and then calculate the function value J(Q), if J(Q) can make the formula (37) valid, or the number of searched hypercubes reaches a predetermined value, then the whole search process is over, go to step five. At this time, p _true = Q;

J(Q) _{≤Jthreshold} (37)

If J(Q) cannot make the formula (37) valid, and the number of searched hypercubes has not reached the predetermined value, continue to search in the hypercubes in this layer according to the predetermined order. If the hypercube search of this layer is completed, and the J(Q) of each hypercube in this layer cannot make the formula (37) established, then continue to search the hypercube of the next layer.

When the L-th layer is searched, no matter whether the geometric center Q of the hypercube satisfying the condition (37) is found, the search process will end. At this time, the Q in the smallest J(Q) should be searched as p _true . Go to step five. Here, L is a predetermined threshold value, indicating the maximum number of layers to be searched.

Step five, the search process ends.

The above search method can also be applied in the search process from coarse to fine to achieve the fastest and best search effect.

The search process from coarse to fine is shown in Figure 14.

In Fig. 14, firstly, the first search is performed according to the above steps 1 to 5. The first search can be regarded as a rough search. In this way, the side length of the hypercube in the parameter space can be set larger, so that the number of hypercubes in the parameter space is less. In the coarse-to-fine search process, the predetermined condition satisfied by the hypercube searched in step 3 may be whether the division granularity is coarser than the predetermined division granularity. If a hypercube satisfying condition (36) or (37) is found in the first search, the search process ends. The first search process can adopt a hierarchical search method.

If no hypercube satisfying the condition (36) or (37) is found in the first search process, and the granularity division of the hypercube has not yet reached the predetermined granularity, the smallest J(Q ) to perform a second finer search in the corresponding hypercube. At this time, the hypercube corresponding to the smallest J(Q) found in the first search should be regarded as the new entire parameter space. At this time, the side length of each hypercube in the new parameter space becomes smaller, and the search process from step 1 to step 5 above is repeated. Similarly, in the second finer search process, if a hypercube satisfying condition (36) or (37) is found, the second search process ends. The second search process may adopt a layered search method.

If no hypercube satisfying the condition (36) or (37) is found in the second finer search process, and the granularity division of the hypercube has not yet reached the predetermined granularity, then the smallest hypercube found in the second search Carry out a third finer search in the hypercube corresponding to J(Q), and so on, take the geometric center of the hypercube that meets the condition (36) or (37) found in the last fine search as the global optimal point p _true , which is the parameter vector of the Gamma characteristic function.

If the granularity division of the hypercube reaches the predetermined granularity, but no hypercube satisfying the condition (36) or (37) is still found, the search process ends. At this time, the set center of the hypercube corresponding to the found smallest J(Q) should be taken as the global optimal point p _true , that is, the parameter vector of the Gamma characteristic function.

A layered search method can be used in each search process from coarse to fine.

After the parameter vector of the Gamma characteristic function is determined by the above method, the Gamma correction can be performed on the Gamma link. Here, the link requiring Gamma correction may be a video data source device, an intermediate device in a video communication network, or a video data destination device.

As can be seen from the description of the above technical solution, the Gamma correction method provided by the present invention only needs to know the histogram of the output luminance signal and a group of loose assumptions to determine the Gamma characteristic parameter of the given Gamma link, thereby providing a new solution for multimedia communication. The system provides an easy-to-implement Gamma correction method. The Gamma correction method provided by the present invention has high application feasibility, thereby greatly broadening the application range of Gamma correction, especially for IPTV, collaborative data conferences, and widely used low-end The public video communication of the video input device provides a very good Gamma correction function, which greatly improves the user experience and service quality, further enhances the competitiveness of the above services, and brings huge economic benefits to telecom operators, service providers and equipment manufacturers.

The device for correcting gamma characteristics of video communication provided by the present invention mainly includes: a histogram acquisition module, a first conversion module, an extremum point calculation module, a storage module, a second conversion module, a gamma characteristic parameter solving module and a gamma correction module.

The obtaining histogram module is mainly used to obtain the brightness histogram of the output brightness signal, and output the obtained brightness histogram to the first conversion module. The histogram obtaining module can obtain the brightness histogram of the output brightness signal before converting the output brightness signal into an output image frame, or can obtain the brightness histogram of the output brightness signal from the output image frame. Specifically as described in the above method.

The first conversion module is mainly used to convert the luminance histogram of the output luminance signal received by it into the luminance distribution probability density function of the output luminance signal, and output the luminance distribution probability density function of the output luminance signal to the extreme point calculation module and the second conversion respectively module. Here the output brightness signal brightness distribution probability density function can be: f _r (r) = c _d r ^d + c _d-1 r ^d-1 + c _d-2 r ^d-2 +.....+c ₁ r+c ₀ .

The extreme point calculation module is mainly used to calculate each extreme point of the brightness distribution probability density function of the output brightness signal after receiving the brightness distribution probability density function of the output brightness signal, and transmit the calculated extreme points to the second conversion module. The method of calculating the extreme point of the probability density function of the brightness distribution of the output brightness signal belongs to conventional technology, and will not be described in detail here.

The storage module is mainly used to store the mathematical relationship between the extreme point of the brightness distribution probability density function of the output brightness signal and the extreme point of the brightness distribution probability density function of the input brightness signal, and store the respective brightness distribution probability densities of the input and output brightness signals The mathematical relationship between the function and the Gamma characteristic function d(e; p) f _r (r) = f _e (e).

Here, the mathematical relationship between the extreme point of the output brightness signal brightness distribution probability density function and the extreme value point of the input brightness signal brightness distribution probability density function can be:

r _k = g(e _k ; p), p = [p ₁ , p ₂ , p ₃ , . . . , p _M ] ^T . Of course, the mathematical relationship here allows a slight transformation, specifically as described in the above method.

The second conversion module is mainly used to convert the respective luminance distributions of the input and output luminance signals in the storage module according to the extreme points it receives, the probability density function of the output luminance signal brightness distribution, and the mathematical relationship of the extreme points stored in the storage module The mathematical relationship between the probability density function and the Gamma characteristic function d(e; p) f _r (r) = f _e (e) is transformed into: at the extreme point, the output brightness signal brightness distribution probability density function and its derivative function The mathematical relationship between the Gamma characteristic function and its inverse function, first-order derivative function, and second-order derivative function:

$\mathrm{<span\; class="notranslate">\; z</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; g</span>}}^{<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; ;</span>\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ;</span>\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; )</span>{\mathrm{<span\; class="notranslate">\; f</span>}}_{\mathrm{<span\; class="notranslate">\; r</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; d</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; g</span>}}^{<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; ;</span>\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ;</span>\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; )</span>\frac{{\mathrm{<span\; class="notranslate">\; df</span>}}_{\mathrm{<span\; class="notranslate">\; r</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; )</span>}{\mathrm{<span\; class="notranslate">\; dr</span>}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1,2,3</span>}<span\; class="notranslate">\; ,</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; J</span>}<span\; class="notranslate">\; ;</span>$

为：Wherein: k=1,2,3,..., J, J is the extreme value point quantity of brightness distribution probability density function, r _k is the extreme value point of output brightness signal brightness distribution probability function, and, derivation function

for:

$\frac{{\mathrm{<span\; class="notranslate">\; df</span>}}_{\mathrm{<span\; class="notranslate">\; r</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; r</span>}<span\; class="notranslate">\; )</span>}{\mathrm{<span\; class="notranslate">\; dr</span>}}<span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; dc</span>}}_{\mathrm{<span\; class="notranslate">\; d</span>}}{\mathrm{<span\; class="notranslate">\; r</span>}}^{\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; +</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; )</span>{\mathrm{<span\; class="notranslate">\; c</span>}}_{\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; r</span>}}^{\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; +</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; )</span>{\mathrm{<span\; class="notranslate">\; c</span>}}_{\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 2</span>}}{\mathrm{<span\; class="notranslate">\; r</span>}}^{\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 3</span>}}<span\; class="notranslate">\; +</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; <figure-callout\; id="1"\; label="c"\; filenames="A20061008286100301.png,A20061008286100333.png"\; state="\{\{state\}\}">c</figure-callout></span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; .</span>$

The Gamma characteristic parameter solving module is mainly used to solve and calculate the mathematical relationship converted by the second conversion module, so as to determine the Gamma characteristic parameter. The Gamma characteristic parameter solving module can use the direct solution equation method, the nonlinear function optimization method, etc. to determine the Gamma characteristic parameter. The nonlinear function optimization methods here include conventional mathematical optimization methods, neural network methods, brute force search, etc. When the Gamma characteristic parameter solving module adopts the brute-force search method, it can adopt a layered search method, or a layered search method from coarse to fine, etc., as described in the above method specifically.

The gamma correction module is mainly used to perform gamma correction on the gamma link according to the gamma characteristic parameters obtained by the gamma characteristic parameter solving module. The gamma link here includes: a single given Gamma link or a cascade combination of multiple given Gamma links.

The device provided by the present invention is located in a video device, such as in a video data source device, in an intermediate device of a video communication network, or in a video data destination device.

Although the present invention has been described by way of example, those of ordinary skill in the art know that there are many variations and changes in the present invention without departing from the spirit of the invention, and the claims of the application document of the present invention include these variations and changes.

Claims (13)

1. A method for correcting video communication gamma characteristics, comprising: a. Acquiring a luminance histogram of the output luminance signal; b. converting the luminance histogram of the output luminance signal into a luminance distribution probability density function of the output luminance signal, and determining the extreme point of the luminance distribution probability density function of the output luminance signal; c. Establishing the mathematical relationship between the extreme point of the brightness distribution probability density function of the output brightness signal and the extreme value point of the brightness distribution probability density function of the input brightness signal; d. Using the extreme point and the mathematical relationship between the extreme points to convert the mathematical relationship between the respective brightness distribution probability density functions and Gamma characteristic functions of the input and output signals to: at the extreme point, the output brightness The mathematical relationship between the signal brightness distribution probability density function and the Gamma characteristic function; e. Solving the converted mathematical relationship to determine the gamma characteristic parameter; f. Performing gamma correction on the gamma link according to the gamma characteristic parameter. 2. The method according to claim 1, wherein the gamma link is: a single given gamma link, or a cascade combination of multiple given gamma links. 3. The method according to claim 1 or 2, characterized in that said step b comprises: Convert the luminance histogram of the output luminance signal into a polynomial form of the output luminance signal brightness distribution probability density function: f _r (r) = c _d r ^d +c _d-1 r ^d-1 +c _d-2 r ^d-2 +.....+c ₁ r+c ₀ ; Among them: c _d , c _d-1 , c _d-2 ,..., c ₀ are d+1 polynomial coefficients, r is the amplitude of the input brightness signal. 4. The method according to claim 3, wherein said step d comprises: The mathematical relationship between the brightness distribution probability density functions f _e (x, t), f _r (x, t) and the Gamma characteristic function of the input and output brightness signals is: d(e; p) f _r (r) = f _e (e); Where: r=g(e,p), e∈[0,1]; The number of extreme points of the brightness distribution probability density function of the input and output brightness signals is the same, and the quantitative mathematical relationship is: r _k = g(e _k ; p), p = [p ₁ , p ₂ , p ₃ , . . . , p _M ] ^T Among them: k=1, 2, 3, ..., J; J is the number of extreme points of the probability density function of the brightness distribution; At the extreme point, the mathematical relationship between the converted output brightness signal brightness distribution probability density function and its derivative function and the Gamma characteristic function and its inverse function and derivative function is: $\mathrm{<span\; class="notranslate">\; z</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; g</span>}}^{<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; ;</span>\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ;</span>\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; )</span>{\mathrm{<span\; class="notranslate">\; f</span>}}_{\mathrm{<span\; class="notranslate">\; r</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; d</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; g</span>}}^{<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; ;</span>\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ;</span>\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; )</span>\frac{{\mathrm{<span\; class="notranslate">\; df</span>}}_{\mathrm{<span\; class="notranslate">\; r</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; )</span>}{\mathrm{<span\; class="notranslate">\; dr</span>}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; 2,3</span>}<span\; class="notranslate">\; ,</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; J</span>}<span\; class="notranslate">\; ;</span>$ Wherein: k=1,2,3,..., J, J is the number of extremum points of the brightness distribution probability density function, r _k is the extreme point of the probability function of the brightness distribution of the output brightness signal, and the derivative function for: $\frac{{\mathrm{<span\; class="notranslate">\; df</span>}}_{\mathrm{<span\; class="notranslate">\; r</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; r</span>}<span\; class="notranslate">\; )</span>}{\mathrm{<span\; class="notranslate">\; dr</span>}}<span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; dc</span>}}_{\mathrm{<span\; class="notranslate">\; d</span>}}{\mathrm{<span\; class="notranslate">\; r</span>}}^{\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; +</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; )</span>{\mathrm{<span\; class="notranslate">\; c</span>}}_{\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; r</span>}}^{\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; +</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; )</span>{\mathrm{<span\; class="notranslate">\; c</span>}}_{\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 2</span>}}{\mathrm{<span\; class="notranslate">\; r</span>}}^{\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 3</span>}}<span\; class="notranslate">\; +</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; .</span><span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; c</span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; .</span>$ 5. The method according to claim 4, wherein said step e comprises: Mathematical relationships using direct equation-solving methods, or nonlinear function optimization methods: $z(g^{-1}(r_k;p);p)f_r\left(r_k\right)+d^2(g^{-1}(r_k;p);p)\frac{{\mathrm{df}}_r\left(r_k\right)}{\mathrm{dr}}=0,k=1,\mathrm{2,3},....,J$ A solution is performed to determine the gamma characteristic parameters. 6. The method according to claim 5, wherein the nonlinear function optimization method comprises: According to the mathematical relationship between the output brightness signal brightness distribution probability density function and the Gamma characteristic function, the cost function is constructed: $\mathrm{<span\; class="notranslate">\; J</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; J</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; z</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; g</span>}}^{<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; ;</span>\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ;</span>\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; )</span>{\mathrm{<span\; class="notranslate">\; f</span>}}_{\mathrm{<span\; class="notranslate">\; r</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; d</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; g</span>}}^{<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; ;</span>\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ;</span>\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; )</span>\frac{{\mathrm{<span\; class="notranslate">\; df</span>}}_{\mathrm{<span\; class="notranslate">\; r</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; )</span>}{\mathrm{<span\; class="notranslate">\; dr</span>}}<span\; class="notranslate">\; )</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; ;</span>$ Randomly select M-1 parameters among the M parameters, and reduce the dimension of the parameter space to M-1 dimensions; The method to determine the M-1 parameters is: determine the parameter vector p _true , so that for any p∈R ^M-1 , the relationship J(p)>=J(p _ture ) holds true, that is, find the global minimum point of the cost function, The parameter vector p _true is the value determined for the M-1 parameters; Using the relation g(1; p)=1 in combination with p _true , the remaining gamma characteristic parameter is determined. 7. The method according to claim 6, wherein the method for determining p _true comprises: a combinatorial optimization method, a neural network method, and a brute force search method. 8. The method according to claim 7, wherein the brute force search method comprises the steps of: Divide the one-dimensional gamma characteristic parameter space into multiple hypercubes; Select an initial search point, and perform a traversal search from the hypercube where the initial search point is located according to a predetermined order; Calculate the geometric center coordinate Q of each hypercube entering in the search process, and calculate J (Q) according to the expression of the cost function and the hypercube entering in the search process; If J(Q) is less than or equal to the predetermined threshold, or the hypercube searched meets the predetermined condition, then the geometric center coordinate Q of the hypercube entered in this search is used as p _true , and the search process ends; Otherwise, the search process continues. 9. The method according to claim 8, wherein the initial search point is set according to the empirical value of the gamma characteristic parameter in actual application. 10. The method according to claim 8, wherein the traversal search comprises: taking the hypercube where the initial search point is located as the initial hypercube, and dividing the hypercube surrounding the initial hypercube according to the distance from the initial hypercube is a multi-layer hypercube array that wraps the previous layer in turn, and searches layer by layer. 11. The method of claim 8, wherein: In the brute force search method, the gamma characteristic parameter space whose dimension is reduced by one dimension is divided into a plurality of coarse-grained hypercubes, and the traversal search includes: With the hypercube where the initial search point is located as the initial hypercube, the hypercube surrounding the initial hypercube is divided into multi-layer hypercube arrays that wrap the previous layer in turn according to the distance from the initial hypercube, and search layer by layer; The searched hypercube that satisfies the conditions is used as a new gamma characteristic parameter space, and it is divided into finer-grained hypercubes, and so on, and a layer-by-layer search is performed from coarse to fine. 12. A device for correcting gamma characteristics of video communication, characterized in that the device includes: a histogram acquisition module, a first conversion module, an extreme point calculation module, a storage module, a second conversion module, and a Gamma characteristic parameter solution module and gamma correction module; Obtaining a histogram module: used to obtain the brightness histogram of the output brightness signal, and output it to the first conversion module; The first conversion module: used to convert the brightness histogram of the output brightness signal received by it into a probability density function of the brightness distribution of the output brightness signal, and output it to the extreme point calculation module and the second conversion module; Extreme point calculation module: used to calculate each extreme point of the brightness distribution probability density function of the output brightness signal received by it, and output to the second conversion module; Storage module: used to store the mathematical relationship between the extreme point of the brightness distribution probability density function of the output brightness signal and the extreme point of the brightness distribution probability density function of the input brightness signal, and store the respective brightness distribution probability densities of the input and output brightness signals The mathematical relationship between the function and the Gamma characteristic function; The second conversion module: used for converting the respective luminance distribution probability density functions of input and output luminance signals and The mathematical relationship between the Gamma characteristic functions is transformed into: at the extreme point, the mathematical relationship between the output brightness signal brightness distribution probability density function and the Gamma characteristic function; Gamma characteristic parameter solving module: used for solving and calculating the converted mathematical relationship to determine the gamma characteristic parameter; Gamma correction module: for performing gamma correction on the gamma link according to the gamma characteristic parameters. 13. The device according to claim 12, characterized in that the device is located in a video data source device, and/or is located in an intermediate device of a video communication network, and/or is located in a video data destination device.

Images