CN100372362C - A Real-time Image Processor for Reconstructing the Gray Scale of the Dark Region of Plasma Display - Google Patents

Abstract

The invention discloses a real-time image processor for reconstructing the gray level of the dark area of a plasma display, which is composed of an image sharpening module and an error diffusion module, the image sharpening module is in front, and the error diffusion module is arranged in the image sharpening module after. It can change the gray level of the plasma dark area to improve the image quality. The gray level input by the real-time image processor is converted into a floating-point number according to the set gamma value, and then divided into an integer part and a decimal part. Errors caused by rounding can be compensated by subsequent error diffusion processing. Therefore, the average gray level of the entire display area will be closer to ideal inverse gamma correction. Experimental results show that the output processed by error diffusion is very consistent with the output calculated according to the ideal inverse gamma correction curve, the complexity of the hardware structure for realizing the present invention is small, and the image quality of the plasma display is obviously improved.

Description

A kind of real time image processor of reestablishing plasma display dark zone grey grade

Technical field

The present invention relates to a kind of plasma display storage control module, be particularly related to a kind of real time image processor of reestablishing plasma display dark zone grey grade, this processor utilizes FPGA realization plasma display dark space to carry out image sharpening and error diffusion, can improve the dark zone grey grade minimizing that inverse gamma correction causes, alleviate the false contouring phenomenon, improve image quality.

Background technology

Compare frivolous liquid crystal and the employing more and more widely of plasma display quilt with CRT monitor.But liquid crystal has major defects such as visible angle is little, size is limited.By contrast, plasma display has many good qualities, as bigger size, fast the response and very wide visual angle.Therefore, along with plasma occupies the increasing market share, more and more higher requirement has just been proposed for its image quality.

In order to adapt to the visual characteristic of human eye, CRT is non-linear to the response of the brightness of input.But the luminosity response characteristics of plasma in normal working range is linear.Therefore, in order to produce the image the same with CRT on plasma display, must proofread and correct the digital signal of input, this process is called as inverse gamma correction (referring to accompanying drawing 1).Adopt after the inverse gamma correction, the gray scale in the dark space can significantly reduce, and the false contouring that produces in the dark space can significantly be perceived by human eye thus.Because the light characteristic of blue colour fluorescent powder is than redness and green poor, redness and the gray scale of green in usable range reduce manyly (referring to accompanying drawing 2).So, set white balance point and inverse gamma correction and can make the minimizing of plasma display gray scale so that false contouring occur in the dark space.

There have been pixel arrangement method, varitron field method etc. can improve the image quality of plasma display, but improved not obvious for the dark zone grey grade minimizing that inverse gamma correction causes.

Summary of the invention

The objective of the invention is to, a kind of real time image processor of ion display dark zone grey grade is provided, this processor can improve the plasma dark zone grey grade to improve image quality.

In order to realize above-mentioned task, the technical solution that the present invention takes is, a kind of real time image processor of reestablishing plasma display dark zone grey grade, it is characterized in that, this real time image processor is made up of image sharpening module and error diffusion module, the image sharpening module is preceding, and the error diffusion module is arranged on after the image sharpening module.

The real time image processor of ion display dark zone grey grade of the present invention, can change the plasma dark zone grey grade and improve image quality, show through experimental result, very consistent through the output that error diffusion is handled with the output that draws according to desirable inverse gamma correction curve calculation, if hardware can be handled decimal, the error between them can be ignored.

Realize that hardware configuration complexity of the present invention is very little, and the picture quality of plasma display improves significantly.

Description of drawings

Fig. 1 is the gamma curve of CRT and PDP; Wherein (a) is CRT (1.8) gamma curve, (b) is PDP (1.8) gamma curve;

Fig. 2 is RGB three look white balance point curves; Wherein the red balance point of green balance point 2 expressions of 1 expression is represented blue balance point;

Fig. 3 is an overview flow chart;

Fig. 4 is the error diffusion flow chart;

Fig. 5 is image sharpening modular structure figure;

Fig. 6 is error diffusion modular structure figure.

The present invention is described in further detail below in conjunction with accompanying drawing and principle and flow process thereof.

Embodiment

The real time image processor of reestablishing plasma display dark zone grey grade of the present invention is made up of image sharpening module and error diffusion module, and the image sharpening module is preceding, and the error diffusion module is arranged on after the image sharpening module.The gray scale of input is converted into floating number according to the gamma value of setting, and is divided into integer part and fractional part then.Fractional part branch is rounded to 0 or 1.The error that produces that rounds off can be handled compensation with the error diffusion of back.Therefore, the average gray level of whole viewing area can more approaching desirable inverse gamma correction.Because error-diffusion method is similar to low pass filter, the border of image can thicken, and therefore needs an extra image sharpening to handle.Two algorithms are all realized in FPGA.

When adopting error diffusion and not adopting the error diffusion dual mode to show the image of a width of cloth gray scale gradual change, the obvious gray scale of the latter is unsmooth, and false contouring is arranged, and the former has then alleviated this phenomenon.

1. image sharpening

The purpose of image sharpening is to cut down low frequency part, increases the weight of other parts.Its Mathematical Modeling can be with following formulate:

Rout＝s*Rin-(s-1)**avgneighbor(R)

Gout＝s*Gin-(s-1)**avgneighbor(G)

Bout＝s*Bin-(s-1)**avgneighbor(B) (1)

In the formula, S is a luminance factor, and Rin, Gin, Bin are the 8bit inputs.The appropriate S that is provided with just can the sharpening image.Avgneighbor (R), avgneighbor (G), avgneighbor (B) are the 3*3 windows that neighbor is averaged.

The 3*3 window is as shown in table 1:

Table 1

1/9	1/9	1/9
			1/9	1/9	1/9
1/9	1/9	1/9

2. overall procedure

Referring to shown in Figure 3, the signal of three kinds of colors of RGB of 8bit, exercise can, enable, global clock input imagery sharpening module, place, image border fuzzy carried out sharpening after, output to the error diffusion module, reduce the false contouring of plasma display dark space.(Rout, Gout Bout) are exactly the data that are presented on the plasma display in final output.Must be placed on error diffusion module front to the image sharpening module, if conversely, after the error diffusion processing, the gamma correction coefficient will change, and the image sharpening module just can't correctly have been handled.

3. error diffusion

The flow process of error diffusion is referring to shown in Figure 4, and this is the process of a feedback, and each pixel need receive from 3 neighbors of lastrow with the previous pixel of delegation and spread the data of coming, and also will go out the error diffusion of itself simultaneously.Constantly the decimal error of accumulation whenever is incremented to an integer value and just rests on corresponding pixel, and fractional part then continues diffusion and goes down.Do not improve precision if do not adopt error diffusion just to need to increase data bits, cost is too big; After adopting error diffusion, because the visual characteristic of human eye has reached similar effect.

Each input and output value relation is shown below:

U (x, y)=the inverse gamma correction look-up table [input (x, y)]+e (x, y)

O (x, y)=round [u (x, y)]

a(x，y)＝u(x，y)-o(x，y)

$e(x+1,y)=a(x,y)*\frac{1}{2}$

$e(x-1,y+1)=a(x,y)*\frac{1}{8}$

$e(x,y+1)=a(x,y)*\frac{1}{4}$

$e(x+1,y+1)=a(x,y)*\frac{1}{8}---\left(2\right)$

Error e (x+1 y) is diffused into next pixel, and e (x-1, y+1), e (x, y+1) (x+1 y+1) is diffused into next line to e; The information storage of previous row is at row cache.Weight coefficient: 1/2,1/8,1/4 and 1/8 is Freud and the Staenberg error diffusion coefficient of revising.

4. hardware is realized

Hardware is realized as Fig. 5, shown in Figure 6.The image sharpening module is mainly formed by three.First comprises a Memory Controller Hub, and two first in first out row caches and a pixel shift register are to realize real-time 3*3 image sharpening operation.Data are constantly imported, and are stored in two row caches.In order to handle the data of triplex row input, preceding two row are stored in row cache, and the third line is stored in one of two row caches and shift register.With delay unit the data of nine pixels are handled for second.The mean value computation device is obtained the average of 9 pixel datas.Form by floating number multiplier and subtracter, obtain end product for the 3rd.Data shift register latchs Rin, and sync section generates Venable, Henable, clk with synchronous with dateout.Memory Controller Hub is controlled two row caches and shift register based on finite state machine.When Henable when 0 becomes 1, handle beginning.First and second mean value of handling initial data and neighbor respectively, the output of two modules are given the 3rd and be multiply by respectively and measure coefficient s and s-1, subtract each other again the data after the sharpening.

The error diffusion module is formed by five.At first, the image sharpening module is output as 8 integers and 8 floating numbers, carries out anti-Gamma correction through two look-up tables respectively; At second, the error from lastrow (being the error of fifth line output) that each pixel in the whole delegation need receive is carried out row cache; At the 3rd, calculate the error that one's own profession need be diffused into next line; At the 4th, the integer part shift LD, decimal and error are partly finished the diffusion summation; At the 5th, integer and fractional part sum are rounded to nearest integer, and error is delivered to second as the output of one's own profession.This is the process of a feedback cycle, and data flow is:

Article one,------first (integer part)--------the 4th--------the 5th-------output of circuit: input.

Second circuit: the 5th------second------the 3rd (first fractional part is incorporated the 4th into together at this)------the 4th--five of-----Di (circulation).

Claims (2)

1. A real-time image processor for rebuilding the gray scale of the dark area of a plasma display, this real-time image processor is made up of an image sharpening module and an error diffusion module, the image sharpening module is in front, and the error diffusion module is arranged on the image After the sharpening module; it is characterized in that, the described image sharpening module is made up of memory controller, two line buffers, pixel shift register, delay unit, mean calculator, floating-point number multiplier and subtractor; The memory controller, two line buffers and pixel shift registers are used together for real-time 3*3 image sharpening operations for processing raw data; The delay unit and the mean value calculator are respectively used to process the data of nine pixels and calculate the mean value of the nine pixel data; Floating-point multipliers and subtractors are used to find the final result; Part of the original data enters the original data shift register, and then multiplies the brightness coefficient s by the floating point multiplier and enters the subtractor; The other part of the original data is first sent to the memory controller, the data flowing out from the memory controller is sent to the two FIFO buffers and the pixel shift register, and the data from the two FIFO buffers and the pixel shift register are sent to To the delay unit, the data from the delay unit enters the mean value calculator, and the data obtained from the mean value calculator enters the floating-point number multiplier and multiplies the brightness coefficient s-1 and then enters the subtractor; The subtractor subtracts the data entering the subtractor to obtain the final sharpened data. 2. the real-time image processor of reconstruction plasma display dark area gray scale as claimed in claim 1, is characterized in that, error diffusion module is made up of five, and the first piece, the output of image sharpening module is 8 integers and 8-bit floating-point numbers, respectively go through two lookup tables for inverse gamma correction. In the second block, the error from the previous line that each pixel in the entire line needs to receive is line-buffered, and the third block calculates the current line The error that needs to be diffused to the next line, in the fourth block, the integer part is shifted and registered, and the decimal and error parts are diffused and summed. In the fifth block, the sum of the integer and decimal parts is rounded to the nearest integer, and the error is as The output of this line is sent to the second block. This is a feedback loop process. The data flow direction is the first line, that is, the input --- the first integer part --- the fourth block --- the fifth block --- Output, the second line is the fifth block---the second block---the third block, and the fractional part of the first block is merged into the fourth block---the fourth block---the fifth block, So cycle.

Images