JP4695015B2 - Code amount estimation method, frame rate estimation method, code amount estimation device, frame rate estimation device, code amount estimation program, frame rate estimation program, and computer-readable recording medium recording those programs - Google Patents

Description

  The present invention relates to a code amount estimation method and apparatus used in video coding with inter-frame prediction to estimate the amount of code generated when a video is coded at a given frame rate, and its code amount estimation technique , A frame rate estimation method and apparatus for estimating the maximum frame rate achievable under a given code amount constraint condition, a code amount estimation program used for realizing the code amount estimation method, and the program And a frame rate estimation program used for realizing the frame rate estimation method and a computer readable recording medium recording the program.

  In recent years, there are growing expectations for super-high-quality images such as large-screen sports images and digital cinema that are full of realism. In response to this, research on high-quality video has been vigorously conducted.

  The following four elements are necessary to realize super high-quality video. That is, spatial resolution, pixel value depth, color reproducibility, and temporal resolution. In response, the former three elements are being studied in applications such as digital cinema and natural vision projects.

  However, sufficient studies have not been made on the improvement of time resolution, that is, the increase of the video frame rate, which is indispensable for expressing the natural movement of the subject.

  According to Spillmann et al., Physiological knowledge is shown that the upper limit of the number of pulses output by ganglion cells, which are output cells of the retina, is about 300 to 400 pulses per second. For this reason, it is presumed that the human visual system is the shortest and can perceive the difference in light emission up to 1/150 to 1/200 second. This means that the perceivable frame rate detection limit is 150 to 200 [frames / second]. The current video frame rate of 30, 60 [frames / second] is determined from the flicker detection limit and is not a sufficient value to express natural motion.

  On the other hand, since the super high image quality of the video leads to an explosive increase in the amount of data, an efficient code amount control technique is required when the code amount is limited for reasons such as network bandwidth limitation. .

  The purpose of the code amount control is to achieve the highest image quality under the constraint conditions related to the code amount. Elements of code amount control include a quantization parameter, the number of pixels per frame, a frame rate, and the like. In encoding a high frame rate video, the effect of reducing the code amount by down-sampling the frame rate is large, so it is important to grasp the relationship between the frame rate and the code amount. Intuitively, it is expected that the amount of information per frame included in the inter-frame prediction error signal decreases as the frame rate increases. However, until now, its quantitative nature has not been grasped.

Also, as a conventional study, time scalability with variable frame rate focusing on the relationship between frame rate and information amount has been studied. However, the content is mainly related to low frame rate video below 60 [frame / sec]. The knowledge about the high frame rate video signal has not been obtained (for example, see Non-Patent Documents 1 and 2).
H.Song and C.-CJKuo.Rate control for low-bit-rate video via variable-encoding frame rates.IEEE trans.CSVT, Vol.11, No.4, pp.512-521, 2001. Yasuhiro Inazumi, Toshiyuki Yoshida, Yoshinori Sakai, Hirohiro Hotta. Estimation of optimal frame rate for video communication under bit rate limitation. IEICE Theory, Vol.J85-B, No.7, pp.1130-1142, 2002.

  As described above, in encoding of a high frame rate video, the effect of reducing the code amount by down-sampling the frame rate is great, so it is important to grasp the relationship between the frame rate and the code amount. Intuitively, it is expected that the amount of information per frame included in the inter-frame prediction error signal decreases as the frame rate increases. However, until now, its quantitative nature has not been grasped.

  Thus, according to the prior art, there is a problem that an efficient code amount control method design method has not been established in encoding processing for a high frame rate video signal.

  In view of such circumstances, the present inventors, in the previously filed Japanese Patent Application No. 2006-20082, assume frame rate down-sampling based on frame decimation and the frame rate and inter-frame prediction error signal. An invention has been disclosed in which a theoretical model representing the relationship with the information amount is constructed, the generated code amount for each frame rate is estimated based on the model, and a design method for an efficient code amount control method is established.

  However, the invention disclosed in Japanese Patent Application No. 2006-20082 assumes frame rate down-sampling based on frame decimation and does not take into account the difference in the shutter opening time of the imaging system.

  The present invention has been made in view of such circumstances, and is advanced one step from the invention disclosed in Japanese Patent Application No. 2006-20082, and is generated in accordance with the open time of the imaging system in video encoding with inter-frame prediction. Establishing a theoretical model that represents the relationship between the frame rate and the amount of information in the inter-frame prediction error signal, taking into account the effects of integration effects, and estimating the amount of generated code for each frame rate based on that model It aims at establishing the design method of a code amount control method.

[1] First Invention In order to achieve the above object, the code amount estimation apparatus of the present invention occurs when video is encoded at a given frame rate in video encoding with inter-frame prediction. In order to realize the estimation of the code amount, (b) the input means for inputting four model parameters and (b) the four model parameters input by the input means are used as coefficients, and the estimated prediction error power (C) a logarithm of the cost function determined by the determining means; and (c) a determining function for determining a cost function as a linear sum of a reciprocal of the frame rate, a reciprocal of the square root of the frame rate, a square value of the frame rate, and a constant term. The calculated value is obtained by substituting the value of the given frame rate for the calculation formula obtained by multiplying the value by the frame rate. And calculating means for calculating the amount of code generated when the video is encoded at the specified frame rate.

  The code amount estimation method of the present invention realized by the operation of each of the above processing means can also be realized by a computer program, and this computer program is provided by being recorded on an appropriate computer-readable recording medium. The present invention is realized by being provided via a network, installed when the present invention is implemented, and operating on a control means such as a CPU.

[2] Second invention In order to achieve the above object, the code amount estimation apparatus of the present invention occurs when video is encoded at a given frame rate in video encoding with inter-frame prediction. In order to realize the estimation of the code amount, (b) the input means for inputting three model parameters and (b) the three model parameters input by the input means are used as coefficients, and the estimated prediction error power A determining means for determining a cost function as a reciprocal of the square root of the frame rate used for calculation, a square value of the frame rate and a linear sum of constant terms ; and (c) a logarithm of the cost function determined by the determining means and the frame rate. A calculated value is obtained by substituting the value of the given frame rate into the calculation formula obtained by multiplying, and based on the calculated value, the image is obtained at the given frame rate. And a calculating means for calculating the amount of code generated when the image is encoded.

  The code amount estimation method of the present invention realized by the operation of each of the above processing means can also be realized by a computer program, and this computer program is provided by being recorded on an appropriate computer-readable recording medium. The present invention is realized by being provided via a network, installed when the present invention is implemented, and operating on a control means such as a CPU.

[3] Third invention In order to achieve the above object, the frame rate estimation apparatus according to the present invention is the maximum achievable under the constraint condition of a given code amount in video coding with inter-frame prediction. In order to realize the estimation of the frame rate, (a) four model parameters are input, and the four model parameters thus input are used as coefficients, and the reciprocal of the frame rate used to calculate the estimated prediction error power. A cost function that is a reciprocal of the square root of the frame rate, a square value of the frame rate, and a linear sum of the constant terms is determined, and a calculation formula obtained by multiplying the logarithm of the determined cost function and the frame rate is derived. Deriving means and (b) Substituting multiple frame rate values into the calculation formula derived by the deriving means to obtain calculated values and calculate them Based on the value, the code amount generated when the video is encoded at those frame rates is calculated, and the calculated code amount is compared with the code amount that is the constraint condition, so that the code amount And estimating means for estimating the maximum frame rate that can be achieved under the constraint condition.

  When adopting this configuration, the estimation means sets the value of the frame rate to be evaluated, and obtains the calculated value by substituting the value of the frame rate for the calculation formula derived by the derivation means. Based on the calculated value, the code amount generated when the video is encoded at the frame rate is calculated, the calculated code amount is compared with the code amount as a constraint condition, and the comparison result is obtained. Thus, there are cases where processing is performed so as to estimate the maximum achievable frame rate by repeatedly increasing or decreasing the value of the frame rate to be evaluated.

  The frame rate estimation method of the present invention realized by the operation of each of the above processing means can also be realized by a computer program, and this computer program is provided by being recorded on an appropriate computer-readable recording medium. The present invention is realized by being provided via a network, installed when the present invention is implemented, and operating on a control means such as a CPU.

[4] Fourth invention In order to achieve the above object, the frame rate estimation apparatus of the present invention provides the maximum achievable under the constraint condition of a given code amount in video coding with inter-frame prediction. In order to realize the estimation of the frame rate, (a) three model parameters are input, and the input three model parameters are used as coefficients to calculate the square root of the frame rate used for calculating the estimated prediction error power. Derivation means for deriving a calculation formula obtained by determining a cost function as a linear sum of a reciprocal of the frame rate, a square value of the frame rate and a constant term , and multiplying the logarithm of the determined cost function and the frame rate; B) Substituting multiple frame rate values into the calculation formula derived by the deriving means to obtain calculated values, and based on those calculated values, Achievable under the constraint condition of the code amount by calculating the code amount generated when encoding the video at the frame rate and comparing the calculated code amount with the code amount as the constraint condition And an estimation means for estimating the maximum frame rate.

  When adopting this configuration, the estimation means sets the value of the frame rate to be evaluated, and obtains the calculated value by substituting the value of the frame rate for the calculation formula derived by the derivation means. Based on the calculated value, the code amount generated when the video is encoded at the frame rate is calculated, the calculated code amount is compared with the code amount as a constraint condition, and the comparison result is obtained. Thus, there are cases where processing is performed so as to estimate the maximum achievable frame rate by repeatedly increasing or decreasing the value of the frame rate to be evaluated.

  The frame rate estimation method of the present invention realized by the operation of each of the above processing means can also be realized by a computer program, and this computer program is provided by being recorded on an appropriate computer-readable recording medium. The present invention is realized by being provided via a network, installed when the present invention is implemented, and operating on a control means such as a CPU.

  According to the present invention, it is possible to estimate in advance the amount of code required for encoding at a desired frame rate in consideration of the influence of the integration effect generated according to the open time of the imaging system. This makes it possible to appropriately set the network transmission band.

  According to the present invention, it is possible to estimate the maximum frame rate that can be realized using a given code amount in consideration of the influence of the integration effect generated according to the open time of the imaging system. . As a result, it is possible to set an appropriate frame rate at the time of encoding, and to improve the image quality of the decoded image.

  Furthermore, according to the second and fourth inventions, when realizing this, it is possible to reduce the degree of freedom (number of parameters) of the model representing the relationship between the code amount and the frame rate, and the parameters relating to the model. Estimation can be simplified. As a result, the model estimation can be speeded up.

  Next, the present invention will be described in detail according to embodiments.

  In the present invention, in video coding with inter-frame prediction, the relationship between the frame rate and the amount of information of the inter-frame prediction error signal is calculated in consideration of the influence of the integration effect that occurs according to the open time of the imaging system. It is possible to construct a theoretical model to represent and estimate the generated code amount for each frame rate based on the model, and to estimate the maximum frame rate achievable under the constraints of the given code amount. Realize.

[1] Relationship between Frame Rate and Information Amount Next, the relationship between the frame rate and the information amount of the inter-frame prediction error signal is analytically derived. Here, for simplicity, a one-dimensional signal will be described as an example.

At a position x, a signal obtained by opening the shutter from time t to t + δ and taking a picture is defined as _ft ′ (x, δ). At this time, a signal f _mt (x, mδ) obtained when the shutter opening time is extended to mδ (m is a natural number) can be expressed as follows using the signal f _t ′ (x, δ).

This equation (1) indicates that the signal f _mt (x, mδ) is defined by an average value of m signals f _t ′ (x, δ) that have been taken continuously.

When motion compensation (estimated displacement d _m ^* ) is performed on the signal f _mt (x, mδ) using the section L as a unit, the prediction error after motion compensation in the section can be expressed as follows. In the following, the true displacement amount at the position x is d _m (x).

  Here, φ (x) is the second and subsequent terms of Taylor expansion. N (m) is a term due to a noise component.

If the movement between frames is regarded as a constant velocity motion, the amount of displacement is proportional to the frame interval (the reciprocal of the frame rate), and thus has an inversely proportional relationship with the frame rate. Therefore, if F = (m · δ) ⁻¹ , d _m (x) −d _m ^* can be expressed as the following equation.

d _m (x) −d _m ^* = (v _m (x) −v _m ^* ) · m · δ
Here, v _m (x) is a proportionality constant determined for each pixel at the position x, and v _m ^* is a proportionality constant determined for the section L. The physical meaning of both proportionality constants is speed.

Note that for simplicity in the _{following, f t '(x, δ} ) a f _t' is abbreviated as (x).

  By substituting the above equation into equation (2) and using the independence of noise components and the first order approximation of Taylor expansion, the following equation is obtained.

  Here, A (m), B (m), and N (m) are as follows.

A (m) and B (m) can be expanded as follows, and the noise component n (m) constituting N (m) can be expanded as follows. In the following,
μ _mt (x) = (1 / m) Σf _{m (t-1) + i} '(x)
Where Σ is the sum of i = 0 to m−1
ε (x) = (v _m (x) −v _m ^* )
far.

  A (m) can be expanded as follows.

Here, it is assumed that β = Σ _x ε (x) ² . In the above approximation, it is assumed that ε (x) and μ _mt (x) are statistically independent. The following homogeneous model was used.

  Furthermore, the following approximation was used.

Here, <d _i > and <d _j > are average values of d _i (x) and d _j (x) (x∈L), respectively. Α _{i, j} is a correction parameter for approximation using the average displacement (<d _i >, <d _j >).

ρ is 1 or less because of autocorrelation of the image signal, and can be regarded as a value close to 1. For this reason,
(1-ρ) / ρ << 1
The inequality is established.

  By using this inequality, equation (5) can be approximated as follows.

Since m is the ratio of the downsampled frame rate F to the maximum frame rate F ₀ , we obtain:

  Similarly, B (m) is approximated as follows:

  Here, γ is 1 or −1.

Next, the noise component n (m) constituting N (m) will be considered. Since it is assumed that the noise component n (m) is statistically independent from the image signal, the averaging process of Equation (1) reduces the noise component. That is, if the noise component included in the sequence at the frame rate F ₀ is n ₀ , n (m) is expressed as the following equation.

  Here, “L” is the number of elements in the segment L.

  Assuming that the prediction error follows a Laplace distribution and estimating the information amount I (F) [bits / pel] of the prediction error using the differential entropy of the distribution, the following relationship is obtained. In the following, the base of the logarithm is 2.

  Here, A ″, B ″, C ″ and D ″ are as follows.

In Expression (13), when the frame rate is large, F ⁻¹ becomes a small value, so that the first term is considered negligible, and I (F) is approximated as I ^# (F) below.

When using equation (13),
σ _e ² = A ″ F ⁻¹ + B ″ F ^−1/2 + C ″ F + D ″
In contrast to calculating the estimated prediction error power according to the calculation formula, when using this formula (15),
σ _e ² = B ″ F ^−1/2 + C ″ F + D ″
The estimated prediction error power is calculated according to the calculation formula.

[2] Code Amount Estimation In the first and second inventions, a process for estimating the required code amount is performed when a constraint condition that the frame rate is encoded as F is given.

  This is a process of estimating the generated code amount corresponding to the designated frame rate F as shown in FIG. For example, the frame rate is given as a restriction condition at the time of encoding. (I) For the purpose of quality assurance in the video transmission service, the video display terminal on the decoding side is required to display at a frame rate higher than a certain value. (Ii) for the purpose of maximizing image quality (frame rate) in the time axis direction, encoding is performed at the maximum frame rate that can be displayed on the video display terminal on the decoding side.

[2-1] First Invention In the first invention, the generated code amount estimation using Expression (13) is performed. That is, when encoding a video with a given frame rate F, the estimated value of the generated code amount is given by the following equation.

  Note that A ″, B ″, C ″, and D ″ are separately given from the outside. A ″, B ″, C ″, D ″ are referred to as model parameters in the first invention.

[2-2] Second Invention In the second invention, the generated code amount estimation using Expression (15) is performed. That is, when encoding a video with a given frame rate F, the estimated value of the generated code amount is given by the following equation.

  Note that B ″, C ″, and D ″ are separately given from the outside. B ″, C ″, and D ″ are referred to as model parameters in the second invention.

[3] Frame Rate Estimation In the third and fourth aspects of the invention, when a constraint condition that the code amount is suppressed to R or less is given, a process for estimating the maximum frame rate for realizing it is performed.

  As shown in FIG. 2, this is a process of selecting the maximum frame rate that satisfies the code amount constraint from the selectable frame rate candidates. Such a restriction on the amount of codes occurs due to the network bandwidth or the like. Hereinafter, the allowable value of the code amount given as the constraint condition is referred to as an allowable code amount.

[3-1] Third Invention In the third invention, when the image quality (frame rate) in the time axis direction is maximized under the condition that the code amount is suppressed to R or less, Expression (13) is used. A frame rate F ₁ satisfying the following equation is obtained.

Here, argmax _F returns F that maximizes the next function. A ″, B ″, C ″, and D ″ in Equation (13) are separately given from the outside. A ″, B ″, C ″, D ″ are called model parameters in the third invention.

[3-2] Fourth Invention In the fourth invention, when maximizing the image quality (frame rate) in the time axis direction under the condition that the code amount is suppressed to R or less, the equation (15) is used. Then, a frame rate F ₂ satisfying the following equation is obtained.

Here, argmax _F returns F that maximizes the next function. B ″, C ″, and D ″ in Expression (15) are separately given from the outside. B ″, C ″, and D ″ are referred to as model parameters in the fourth invention.

  Next, the present invention will be described in detail according to examples.

[1] Configuration of Generated Code Amount Estimating Device of the Present Invention FIG. 3 shows an embodiment of the device configuration of the generated code amount estimating device 10 of the present invention.

  As shown in this figure, the generated code amount estimation apparatus 10 of the present invention includes an input unit 100, a model parameter storage unit 101, a frame rate storage unit 102, a frame rate calculation calculation unit 103, and a frame rate calculation result storage. Unit 104, estimated prediction error power calculation unit 105, estimated prediction error power storage unit 106, logarithmic value calculation unit 107, logarithmic value storage unit 108, generated code amount estimated value calculation unit 109, and generated code amount estimation A value storage unit 110 and a generated code amount estimated value output unit 111 are provided.

  The input unit 100 inputs a frame rate F that is an evaluation target of the generated code amount estimation. When the first invention is realized, the input unit 100 receives the above-described model parameters A ″, B ″, C ″, D ″. When the second invention is input, the model parameters B ″, C ″, D ″ described above are input.

  The model parameter storage unit 101 stores model parameters input from the input unit 100. The frame rate storage unit 102 stores the frame rate F input by the input unit 100.

When realizing the first invention, the frame rate calculation / calculation unit 103 obtains the reciprocal F ⁻¹ , the reciprocal square root F ^−1/2 , and the same power F of the frame rate F stored in the frame rate storage unit 102. When calculating and realizing the second invention, the inverse square root F ^−1/2 of the frame rate F stored in the frame rate storage unit 102 and the same power value F are calculated.

  The frame rate calculation result storage unit 104 stores the calculation result for the frame rate F calculated by the frame rate calculation calculation unit 103.

The estimated prediction error power calculation unit 105 uses the model parameter stored in the model parameter storage unit 101 and the calculation result for the frame rate F stored in the frame rate calculation result storage unit 104 to realize the first invention. In
σ _e ² = A ″ F ⁻¹ + B ″ F ^−1/2 + C ″ F + D ″
When the estimated prediction error power is calculated according to the calculation formula, and the second invention is realized,
σ _e ² = B ″ F ^−1/2 + C ″ F + D ″
The estimated prediction error power is calculated according to the following calculation formula.

  The estimated prediction error power storage unit 106 stores the estimated prediction error power calculated by the estimated prediction error power calculation unit 105.

  The logarithmic value calculation unit 107 calculates the logarithmic value of the estimated prediction error power stored in the estimated prediction error power storage unit 106. The logarithmic value storage unit 108 stores the logarithmic value of the estimated prediction error power calculated by the logarithmic value calculation unit 107.

  The generated code amount estimated value calculation unit 109 calculates a multiplication value of the logarithmic value of the estimated prediction error power stored in the logarithmic value storage unit 108 and the frame rate F stored in the frame rate storage unit 102, and further calculates the multiplication value. Is multiplied by 0.5 to calculate an estimated value of the amount of code generated when encoding is performed at the frame rate F to be evaluated.

  The generated code amount estimated value storage unit 110 stores the generated code amount estimated value calculated by the generated code amount estimated value calculating unit 109. The generated code amount estimated value output unit 111 outputs the generated code amount estimated value stored in the generated code amount estimated value storage unit 110.

  FIG. 4 shows an example of a processing flow executed by the generated code amount estimation apparatus 10 of the present invention configured as shown in FIG.

  Next, processing executed by the generated code amount estimation apparatus 10 according to the present invention will be described in detail according to the processing flow of FIG.

  When a code amount estimation request is issued, the generated code amount estimation apparatus 10 of the present invention first inputs a frame rate F to be evaluated in step S10 as shown in the processing flow of FIG. .

  Subsequently, in step S11, when the first invention is realized, the above-described model parameters A ″, B ″, C ″, D ″ are input, and when the second invention is realized, the above-described model parameters A ″, B ″, C ″, D ″ are input. Enter model parameters B ", C", D ".

Subsequently, in step S12, when the first invention is realized, the reciprocal F ⁻¹ , the reciprocal square root F ^−1/2 , and the same power value F of the input frame rate F are calculated, and the second invention is calculated. Is realized, the square root reciprocal F ^−1/2 and the same power value F of the input frame rate F are calculated.

Subsequently, when realizing the first invention in step S13,
σ _e ² = A ″ F ⁻¹ + B ″ F ^−1/2 + C ″ F + D ″
When the estimated prediction error power is calculated according to the calculation formula, and the second invention is realized,
σ _e ² = B ″ F ^−1/2 + C ″ F + D ″
The estimated prediction error power is calculated according to the following calculation formula.

Subsequently, in step S14, a logarithmic value of the calculated estimated prediction error power is calculated. That is, when the first invention is realized, a logarithmic value of the calculated estimated prediction error power “A ″ F ⁻¹ + B ″ F ^−1/2 + C ″ F + D ″” is calculated to realize the second invention. In this case, the logarithmic value of the calculated estimated prediction error power “B ″ F ^−1/2 + C ″ F + D ″” is calculated.

  Subsequently, in step S15, a multiplication value of the calculated logarithmic value of the estimated prediction error power and the input frame rate F is calculated.

  Subsequently, in step S16, by multiplying the value of the multiplication value calculated in step S15 by 0.5, an estimated value of the amount of code generated when encoding at the frame rate F to be evaluated is calculated. In other words, when the first invention is realized, an estimated value of a code amount generated when encoding is performed at the frame rate F to be evaluated according to the equation (16), and the second invention is realized. In this case, an estimated value of the amount of code generated when encoding is performed at the frame rate F to be evaluated according to the equation (17).

  Finally, in step S17, the calculated generated code amount estimation value is output, and the process is terminated.

  In this way, the generated code amount estimation apparatus 10 according to the present invention performs a process of estimating the necessary code amount when the constraint condition that the frame rate is F is given.

[2] Configuration of Frame Rate Estimation Device of the Present Invention FIG. 5 shows an embodiment of the device configuration of the frame rate estimation device 20 of the present invention.

  As shown in this figure, the frame rate estimation apparatus 20 of the present invention includes an input unit 200, an allowable code amount storage unit 201, a model parameter storage unit 202, a frame rate storage unit 203, and a frame rate increment value storage unit. 204, a frame rate update unit 205, a generated code amount estimation unit 206, a generated code amount estimated value storage unit 207, a generated code amount estimated value determination unit 208, and a frame rate output unit 209.

  The input unit 200 inputs an allowable code amount, an initial frame rate, and a frame rate increment value. When the third invention is realized, the above-described model parameters A ″, B ″, C ″, D ″ are input. When the fourth invention is realized, the above-described model parameters B ″, C ″, D ″ are input.

  Here, what is given as the allowable code amount is, for example, a target code amount determined by band limitation or the like. Also, a sufficiently small value is used as the initial frame rate.

  The allowable code amount storage unit 201 stores the allowable code amount input by the input unit 200. The model parameter storage unit 202 stores the model parameters input from the input unit 200. The frame rate storage unit 203 stores the current frame rate, and stores the initial frame rate input by the input unit 200 as its initial value. The frame rate increment value storage unit 204 stores the frame rate increment value input from the input unit 200.

  When there is a frame rate update instruction from the generated code amount estimated value determination unit 208, the frame rate update unit 205 stores the frame rate stored in the frame rate storage unit 203 in the frame rate increment stored in the frame rate increment value storage unit 204. Increment by the value and update it by overwriting.

  The generated code amount estimation unit 206 is configured with the same apparatus configuration as the generated code amount estimation device 10 of the present invention shown in FIG. 3, and stores the model parameters stored in the model parameter storage unit 202 and the frame rate storage unit 203. Using the frame rate as an input, an estimated value of the amount of code generated when encoding at that frame rate is calculated. The generated code amount estimated value storage unit 207 stores the generated code amount estimated value calculated by the generated code amount estimated unit 206.

  The generated code amount estimated value determination unit 208 compares the generated code amount estimated value stored in the generated code amount estimated value storage unit 207 with the allowable code amount stored in the allowable code amount storage unit 201, and generates the generated code amount estimated value. When the allowable code amount is larger than the value, the frame rate update unit 205 is instructed to update the frame rate. When the generated code amount estimated value is larger than the allowable code amount, the frame rate output unit 209 is instructed to output the value of the frame rate stored in the frame rate storage unit 203.

  The frame rate output unit 209 outputs the frame rate instructed to be output by the generated code amount estimated value determination unit 208.

  FIG. 6 shows an example of a processing flow executed by the frame rate estimation apparatus 20 of the present invention configured as shown in FIG.

  Next, processing executed by the frame rate estimation apparatus 20 of the present invention will be described in detail according to the processing flow of FIG.

  When a frame rate estimation request is issued, the frame rate estimation apparatus 20 of the present invention first inputs an allowable code amount in step S20 as shown in the processing flow of FIG.

  Subsequently, in step S21, when the third invention is realized, the above-described model parameters A ″, B ″, C ″, D ″ are input, and when the fourth invention is realized, the above-described model parameters A ″, B ″, C ″, D ″ are input. Enter model parameters B ", C", D ".

  Subsequently, in step S22, an initial value of the frame rate is input, the current frame rate is set to the initial value, and the value is stored in the register S. In step S23, the frame rate increment value is input to set the frame rate increment value.

  Subsequently, in step S24, the current frame rate (stored in the frame rate storage unit 203) according to the processing of the generated code amount estimation unit 206 having the same apparatus configuration as the generated code amount estimation apparatus 10 of the present invention shown in FIG. The estimated amount of code generated when encoding is performed at a frame rate).

  That is, when realizing the third invention, by executing the processing flow of FIG. 4 using the input model parameters A ″, B ″, C ″, D ″ and the current frame rate, When an estimated value of a code amount generated when encoding is performed at the current frame rate according to the equation (16) and the fourth invention is realized, the input model parameters B ″, C ″, D ″ and By executing the processing flow of FIG. 4 using the current frame rate, an estimated value of the code amount generated when encoding is performed at the current frame rate according to the equation (17) is calculated.

  Subsequently, in step S25, it is determined whether or not the calculated estimated amount of generated code is smaller than the input allowable code amount, and it is determined that the estimated generated code amount is smaller than the allowable code amount. If so, the process proceeds to step S26, and the current frame rate is stored in the register S.

  Subsequently, in step S27, the current frame rate is incremented according to the increment value of the frame rate, and then the process returns to step S24.

  In this way, by repeating the processing of step S24 to step S27, when it is determined in step S25 that the calculated estimated code amount is larger than the input allowable code amount, Proceeding to step S28, the frame rate (frame rate updated immediately before) stored in the register S is output, and the process is terminated.

  In this way, the frame rate estimation apparatus 20 according to the present invention performs a process of estimating the maximum frame rate that realizes the restriction when the code amount is limited to the allowable code amount or less. .

It is explanatory drawing of the generated code amount estimation process which the 1st and 2nd invention performs. It is explanatory drawing of the frame rate estimation process which the 3rd and 4th invention performs. It is one Example of the apparatus structure of the generated code amount estimation apparatus of this invention. It is a processing flow which the generated code amount estimation apparatus of this invention performs. It is one Example of the apparatus structure of the frame rate estimation apparatus of this invention. It is a processing flow which the frame rate estimation apparatus of this invention performs.

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 Generated code amount estimation apparatus 100 Input part 101 Model parameter memory | storage part 102 Frame rate memory | storage part 103 Frame rate calculation calculation part 104 Frame rate calculation result memory | storage part 105 Estimated prediction error electric power calculation part 106 Estimated prediction error electric power memory | storage part 107 Logarithmic value calculation Unit 108 logarithmic value storage unit 109 generated code amount estimated value calculation unit 110 generated code amount estimated value storage unit 111 generated code amount estimated value output unit

Claims (13)

A code amount estimation method that is used in video encoding with inter-frame prediction and estimates a code amount generated when video is encoded at a given frame rate,
Input four model parameters,
Using the four model parameters input as coefficients , a cost function that is a reciprocal of the frame rate used to calculate the estimated prediction error power, a reciprocal of the square root of the frame rate, a square value of the frame rate, and a linear sum of constant terms The decision process,
A calculated value is obtained by substituting the value of the given frame rate for the calculation formula obtained by multiplying the logarithm of the cost function determined above and the frame rate, and is given based on the calculated value. And a process of calculating a code amount generated when video is encoded at a different frame rate,
A characteristic code amount estimation method. A code amount estimation method that is used in video encoding with inter-frame prediction and estimates a code amount generated when video is encoded at a given frame rate,
The process of entering three model parameters;
A process of determining a cost function that uses the three input model parameters as coefficients and that is a reciprocal of a square root of a frame rate, a square value of a frame rate, and a linear sum of constant terms used to calculate an estimated prediction error power;
A calculated value is obtained by substituting the value of the given frame rate for the calculation formula obtained by multiplying the logarithm of the cost function determined above and the frame rate, and is given based on the calculated value. And a process of calculating a code amount generated when video is encoded at a different frame rate,
A characteristic code amount estimation method. A frame rate estimation method for estimating a maximum frame rate that is used in video encoding with inter-frame prediction and can be achieved under a given code amount constraint condition,
Four model parameters are input, and the input four model parameters are used as coefficients. The reciprocal of the frame rate, the reciprocal of the square root of the frame rate, the square value of the frame rate, and a constant are used to calculate the estimated prediction error power. Determining a cost function as a linear sum of terms and deriving a calculation formula obtained by multiplying the logarithm of the determined cost function and the frame rate;
The amount of code generated when a calculated value is obtained by substituting a plurality of frame rate values into the derived calculation formula, and video is encoded at the frame rate based on the calculated value. And calculating a maximum frame rate achievable under the constraint condition of the code amount by comparing the calculated code amount and the code amount that is the constraint condition,
A characteristic frame rate estimation method. A frame rate estimation method for estimating a maximum frame rate that is used in video encoding with inter-frame prediction and can be achieved under a given code amount constraint condition,
Three model parameters are input, and the input three model parameters are used as coefficients. The reciprocal of the square root of the frame rate used to calculate the estimated prediction error power, the square value of the frame rate, and the linear sum of the constant terms Determining a cost function to be calculated, and deriving a calculation formula obtained by multiplying the logarithm of the determined cost function and the frame rate,
The amount of code generated when a calculated value is obtained by substituting a plurality of frame rate values into the derived calculation formula, and video is encoded at the frame rate based on the calculated value. And calculating a maximum frame rate achievable under the constraint condition of the code amount by comparing the calculated code amount and the code amount that is the constraint condition,
A characteristic frame rate estimation method. In the frame rate estimation method according to claim 3 or 4,
In the estimation process, a frame rate value to be evaluated is set, and a calculated value is obtained by substituting the frame rate value for the derived calculation formula. Based on the calculated value, The amount of code generated when video is encoded at the frame rate is calculated, the calculated amount of code is compared with the amount of code that is a constraint condition, and the frame to be evaluated based on the comparison result Repeating increasing or decreasing rate values sequentially,
A characteristic frame rate estimation method. A code amount estimation device that is used in video encoding with inter-frame prediction and estimates a code amount generated when video is encoded at a given frame rate,
Means for inputting four model parameters;
Using the four model parameters input as coefficients , a cost function that is a reciprocal of the frame rate used to calculate the estimated prediction error power, a reciprocal of the square root of the frame rate, a square value of the frame rate, and a linear sum of constant terms Means to determine,
A calculated value is obtained by substituting the value of the given frame rate for the calculation formula obtained by multiplying the logarithm of the cost function determined above and the frame rate, and is given based on the calculated value. Means for calculating the amount of code generated when video is encoded at a different frame rate,
Characteristic code amount estimation apparatus. A code amount estimation device that is used in video encoding with inter-frame prediction and estimates a code amount generated when video is encoded at a given frame rate,
Means for inputting three model parameters;
Means for determining a cost function that uses the three input model parameters as coefficients and uses a reciprocal of a square root of a frame rate, a square value of a frame rate, and a linear sum of a constant term used to calculate an estimated prediction error power;
A calculated value is obtained by substituting the value of the given frame rate for the calculation formula obtained by multiplying the logarithm of the cost function determined above and the frame rate, and is given based on the calculated value. Means for calculating the amount of code generated when video is encoded at a different frame rate,
Characteristic code amount estimation apparatus. A frame rate estimation device that is used in video encoding with inter-frame prediction and estimates the maximum frame rate that can be achieved under a given code amount constraint condition,
Four model parameters are input, and the input four model parameters are used as coefficients. The reciprocal of the frame rate, the reciprocal of the square root of the frame rate, the square value of the frame rate, and a constant are used to calculate the estimated prediction error power. Means for deriving a calculation formula obtained by determining a cost function as a linear sum of terms and multiplying the logarithm of the determined cost function and a frame rate;
The amount of code generated when a calculated value is obtained by substituting a plurality of frame rate values into the derived calculation formula, and video is encoded at the frame rate based on the calculated value. And a means for estimating the maximum frame rate achievable under the constraint condition of the code amount by comparing the calculated code amount and the code amount that is the constraint condition,
A frame rate estimation device as a feature. A frame rate estimation device that is used in video encoding with inter-frame prediction and estimates the maximum frame rate that can be achieved under a given code amount constraint condition,
Three model parameters are input, and the input three model parameters are used as coefficients. The reciprocal of the square root of the frame rate used to calculate the estimated prediction error power, the square value of the frame rate, and the linear sum of the constant terms Determining a cost function to be calculated, and deriving a calculation formula obtained by multiplying the logarithm of the determined cost function and the frame rate;
The amount of code generated when a calculated value is obtained by substituting a plurality of frame rate values into the derived calculation formula, and video is encoded at the frame rate based on the calculated value. And a means for estimating the maximum frame rate achievable under the constraint condition of the code amount by comparing the calculated code amount and the code amount that is the constraint condition,
A frame rate estimation device as a feature.   A code amount estimation program for causing a computer to execute processing used to realize the code amount estimation method according to claim 1.   A frame rate estimation program for causing a computer to execute processing used to realize the frame rate estimation method according to any one of claims 3 to 5.   A computer-readable recording medium recording a code amount estimation program for causing a computer to execute processing used to realize the code amount estimation method according to claim 1.   A computer-readable recording medium recording a frame rate estimation program for causing a computer to execute processing used to realize the frame rate estimation method according to any one of claims 3 to 5.

Images