阅读说明：本技术 图像处理用的基于改进烟花算法的区域背光动态调光方法 (Regional backlight dynamic dimming method based on improved firework algorithm for image processing ) 是由 张涛 曾琴 王伊飞 赵鑫 于 2019-09-21 设计创作，主要内容包括：一种图像处理用的基于改进烟花算法的区域背光动态调光方法,包括：对图像进行区域划分,基于参数方法确定每一个区域的初始背光亮度,将每一个区域的初始背光亮度作为该区域烟花的初始位置；将烟花初的始位置作为中心,生成搜索区间；在搜索区间中随机选择烟花初始位置以外的n-1个位置；在n个位置设置n个烟花；计算每个烟花生成的火花数量；计算每个烟花的爆炸幅度；生成通用火花的位置；生成指导火花的位置；判断是否达到设定的N次迭代；将当前最佳位置作为输出解,所述的输出解为最佳区域背光亮度分配方案。本发明有利于满足背光亮度与图像之间的关系,加快搜索速度,提高输出解决方案的质量；可以使区域调光后的图像获得更高的显示质量。(An area backlight dynamic dimming method based on an improved firework algorithm for image processing comprises the following steps: dividing the image into areas, determining the initial backlight brightness of each area based on a parameter method, and taking the initial backlight brightness of each area as the initial position of fireworks in the area; generating a search interval by taking the initial position of the firework as a center; randomly selecting n-1 positions except the initial position of the fireworks in the search interval; arranging n fireworks at n positions; calculating the number of sparks generated by each firework; calculating the explosion amplitude of each firework; a location at which a generic spark is generated; generating a location of a pilot spark; judging whether the set N times of iteration is reached; and taking the current optimal position as an output solution, wherein the output solution is an optimal area backlight brightness distribution scheme. The invention is beneficial to meeting the relation between the backlight brightness and the image, quickening the searching speed and improving the quality of the output solution; the image after the regional dimming can obtain higher display quality.)

1. An area backlight dynamic dimming method based on an improved firework algorithm for image processing is characterized by comprising the following steps:

1) dividing the image into areas, determining the initial backlight brightness of each area based on a parameter method, and taking the initial backlight brightness of each area as the initial position of fireworks in the area;

2) generating a search interval by taking the initial position of the firework as a center;

3) randomly selecting n-1 positions except the initial position of the fireworks in the search interval;

4) arranging n fireworks at n positions;

5) calculating the number of sparks generated by each firework;

6) calculating the explosion amplitude of each firework;

7) a location at which a generic spark is generated;

8) generating a location of a pilot spark;

9) judging whether set N iterations are reached, if not, calculating the probability of fireworks selection by taking the current optimal position in a mixed set formed by mixing the positions of the fireworks and the positions of the general sparks as the initial position of fireworks in the next iteration, selecting other N-1 positions according to the calculated probability, returning to the step 4), and if so, entering the next step;

10) and taking the current optimal position as an output solution, wherein the output solution is an optimal area backlight brightness distribution scheme.

2. The method for dynamically dimming regional backlight based on an improved firework algorithm in image processing according to claim 1, wherein the parameter method in step 1) is one of a maximum value method, an average value method, an error correction method and a root mean square method.

3. The method for dynamically dimming the regional backlight based on the improved firework algorithm for image processing according to claim 1, wherein the step 2) comprises: the brightness level of each region is 0 to 255, the initial position of the firework is I, the radius value of the set region is R, R takes one value from 0 to 255, and the search region of each region is obtained as follows: I-R to I + R.

4. The regional backlight dynamic dimming method based on the improved firework algorithm for image processing as claimed in claim 1, wherein the step 5) is to calculate the number of sparks generated by each firework by using the following formula:

wherein the content of the first and second substances,

5. The regional backlight dynamic dimming method based on the improved firework algorithm for image processing as claimed in claim 1, wherein step 6) is to calculate the explosion amplitude of each firework by using the following formula:

wherein A is_iIndicating the ith Firework

6. The regional backlight dynamic dimming method for image processing based on an improved firework algorithm as claimed in claim 1, wherein the positions for generating the guidance sparks in step 8) are obtained by mixing the positions of the fireworks and the positions of the general sparks to form a mixed set, calculating the quality (SSIM) of each position in the mixed set, sorting according to the quality from high to low, wherein the position with the lowest quality is a bad position, the position with the highest quality is an optimal position, and the guidance of the bad position in the optimal position searches for a better position, so that the possibility of finding a better solution is increased, and if the better position is found, the quality of the solution is improved; if the bad position does not find a better position to replace itself, the good position in the mixture set generates a new position based on the Gaussian function, and the bad position is replaced by the new position.

7. The regional backlight dynamic dimming method based on the improved firework algorithm for image processing as claimed in claim 1, wherein step 9) is to calculate the probability of firework being selected by using the following formula:

where G is the set of all fireworks and generic spark positions, R'_iRepresenting the position, R ', of the ith Firework in the set G'_jRepresents the position of the j-th firework in the set G, P (R'_i) Represents R'_iThe probability of being selected.

Technical Field

The invention relates to a dynamic dimming method for regional backlight. In particular to a regional backlight dynamic dimming method based on an improved firework algorithm for image processing.

Background

Due to continuous maturity of production technology, continuous perfection of industrial chain and continuous fusion of new technology, the liquid crystal display has obvious advantages in the aspects of low power consumption, high resolution, good color performance, cost performance and the like, and becomes the best-selling equipment in the flat panel display market. With the continuous development of national economy, the requirements of people on display equipment are continuously improved. Display devices with high color gamut, high contrast, and low power consumption are increasingly favored by more and more people. In addition, the development of mobile devices has also placed more demand on display devices. The local backlight dynamic dimming technology has the characteristics of high contrast and low power consumption, so that the local backlight dynamic dimming technology becomes a great research hotspot for the reality of the LCD.

The regional dynamic backlight divides the backlight plate into a plurality of regions which are independent of each other, and each region is independently controlled. The backlight plate under the 2D dynamic backlight is divided into m multiplied by n areas, and the LED backlight is arranged in an array mode, so that the division with high degree of freedom can be realized. Theoretically, the more partitions are arranged, the fewer pixel points of each partition are arranged, the better the finally realized energy-saving effect is, the higher the contrast is improved, and due to the limitation of a manufacturing process, the backlight partition cannot be infinitely small. Meanwhile, the more backlight partitions are, the larger the area occupied by the driving circuit for controlling the backlight plate is. There is generally a corresponding limit to the number of partitions. The 2D backlight has better energy-saving and contrast-improving effects, so the 2D backlight is widely applied to the field of dynamic backlight. The invention is mainly directed to 2D-LED area backlighting.

Regional backlight dynamic dimming technology: the method comprises the steps of firstly carrying out partition processing on an input image, enabling partition sizes to correspond to backlight partition, then extracting brightness information of each partition, calculating backlight of each partition according to an extraction result, and carrying out corresponding compensation on liquid crystal pixels according to a smoothed backlight signal.

The backlight module adopts an LCD system of a 2D-area backlight mode, the size of a liquid crystal panel of the LCD system is mxn, and for a pixel i, the brightness perceived by human eyes from a liquid crystal screen is y_iThen it is expressed as:

y_i＝a_i·b_i(1)

written in the form of a Hadamard product matrix:

Y＝a°b (2)

in the formula, a_iAnd b_iRespectively, the liquid crystal panel projection factor and the backlight brightness corresponding to pixel i. a is in the form of R^m×nIs a matrix formed by the light projection coefficients of each liquid crystal pixel in the liquid crystal panel, b ∈ R^m×nAnd backlight brightness matrix corresponding to each pixel point. If b is_iWhen the value is 0, the backlight brightness at the pixel i is 0, and no light irradiates the pixel; if b is_iWhen the backlight corresponding to the pixel i is fully on, the display device is described as 1. Due to the light leakage characteristics of the liquid crystal itself, a_iAnd cannot be 0.

Theoretically, each pixel point of the backlight module can correspond to one backlight partition, namely each pixel corresponds to one backlight source LED lamp, and considering factors in multiple aspects such as cost, heat dissipation performance and the like, in practical application, one backlight partition comprises hundreds of pixel points, so that the problem of light crosstalk between the backlight partitions exists, namely the actual backlight brightness of the pixel point i comes from the backlight brightness of the current partition and the backlight brightness of other partitions. Therefore, after the brightness of the backlight of each subarea is determined, in order to ensure the brightness and the effect of the display image after the backlight is reduced, the diffusion effect of light rays in the backlight module is considered, and then the liquid crystal pixels are accurately compensated. In the process of simulating light diffusion, the contribution of all the partition backlight brightness to the backlight brightness of the pixel i can be expressed as a Point Spread Function (PSF) coefficient matrix H, and then b_iThat is, the sum of the product of the corresponding point spread function vector H in the matrix H and the corresponding LED lamp luminous intensity, so there are:

written in the form of matrix multiplication:

b＝Hr (4)

wherein N represents the number of the partitioned backlights, h_i,jIs the point spread function coefficient of the jth sub-area corresponding to pixel i, representing the contribution of the jth backlight sub-area to pixel i; r is_jIs the backlight brightness of the jth partition.

Assuming that the display quality of the image is denoted as Q and the power consumption is denoted as P, the objective function and the constraint in the optimization process can be expressed as follows:

wherein P is_limitIs an upper limit of the power consumption constraint.

Structural Similarity Index (SSIM) is an Index for measuring the Similarity of two images, and the Structural Similarity Index (SSIM) takes human visual characteristics into consideration, and is comprehensively evaluated from three aspects of image structure, brightness and contrast. According to experience, the local SSIM index in the quality assessment field is better than global.

Wherein mu_xAnd mu_yRepresenting the mean, σ, of the images x and y, respectively_xAnd σ_yRepresenting the standard deviation, σ, of the images x and y, respectively_xyTable covariance of images x and y. C₁、C₂And C₃The constant value is set to avoid the denominator being 0 and to maintain stability. Taking generally:

where MSSIM represents the average structural similarity between the reference image X and the evaluation image Y, X_jAnd y_jRespectively, and M is the total number of the local image blocks. The MSSIM is used as an evaluation index of the display image,is an objective function, where R represents the backlight partition luminance solution; r ═ R1, R2, …, R_i,…r_k]K is the number of backlight partitions; r is not less than 0_i255, the luminance of the ith backlight block. Y (R) represents a display image obtained by the back photolysis R.

As a new type of group intelligent optimization algorithm, the optimization process of firework algorithm (FWA) is inspired by the firework explosion process. In the process of optimizing the firework algorithm, sparks generated by each firework and explosion are considered as a feasible solution. The number of sparks generated by the explosion of the fireworks with good quality is large, and the sparks are distributed more densely; fireworks with poor quality have fewer sparks after explosion, and the spark distribution is relatively dispersed. In the original firework algorithm, two sparks, namely an explosion spark and a Gaussian spark, are generated by firework explosion, the explosion spark is used for carrying out neighborhood search, the Gaussian spark is used for ensuring the diversity of solutions by utilizing Gaussian random numbers, and then fireworks and sparks which are good in quality and distributed dispersedly are selected to enter the next generation.

Disclosure of Invention

The invention aims to provide a regional backlight dynamic dimming method based on an improved firework algorithm for processing an image, which enables the image after regional dimming to have the highest display quality.

The technical scheme adopted by the invention is as follows: an area backlight dynamic dimming method based on an improved firework algorithm for image processing comprises the following steps:

1) dividing the image into areas, determining the initial backlight brightness of each area based on a parameter method, and taking the initial backlight brightness of each area as the initial position of fireworks in the area;

2) generating a search interval by taking the initial position of the firework as a center;

3) randomly selecting n-1 positions except the initial position of the fireworks in the search interval;

4) arranging n fireworks at n positions;

5) calculating the number of sparks generated by each firework;

6) calculating the explosion amplitude of each firework;

7) a location at which a generic spark is generated;

8) generating a location of a pilot spark;

9) judging whether set N iterations are reached, if not, calculating the probability of fireworks selection by taking the current optimal position in a mixed set formed by mixing the positions of the fireworks and the positions of the general sparks as the initial position of fireworks in the next iteration, selecting other N-1 positions according to the calculated probability, returning to the step 4), and if so, entering the next step;

10) and taking the current optimal position as an output solution, wherein the output solution is an optimal area backlight brightness distribution scheme.

The parameter method in the step 1) is one of a maximum value method, an average value method, an error correction method and a root mean square method.

The step 2) comprises the following steps: the brightness level of each region is 0 to 255, the initial position of the firework is I, the radius value of the set region is R, R takes one value from 0 to 255, and the search region of each region is obtained as follows: I-R to I + R.

And step 5) calculating the number of sparks generated by each firework by adopting the following formula:

wherein the content of the first and second substances,

indicating the position of the ith firework,is composed of

The target function of (a) is determined,

x represents a reference image and X represents a reference image,

the smaller the size of the tube is,

the higher the corresponding display image quality;

indicating the ith fireworks

The number of sparks generated, the parameter Q is used to control the number of sparks generated by the fireworks, n is the number of fireworks,

maximum value of the objective function

ξ is a constant, and 0 < b < a < 1 is set in order to avoid setting the denominator to 0.

And step 6) calculating the explosion amplitude of each firework by adopting the following formula:

wherein A is_iIndicating the ith Firework

The amplitude of the explosion of the gas turbine,is the maximum value of the explosion amplitude and,

minimum of the objective function

ξ is a constant and n is the number of fireworks.

The position for generating the guide sparks in the step 8) is to mix the positions of the fireworks and the positions of the general sparks together to form a mixed set, calculate the quality (SSIM) of each position in the mixed set, sort the positions according to the quality from high to low, wherein the position with the lowest quality is a bad position, the position with the highest quality is an optimal position, and the bad position guides and searches a better position in the optimal position, so that the possibility of finding a better solution is increased, and if the better position is found, the quality of the solution is improved; if the bad position does not find a better position to replace itself, the good position in the mixture set generates a new position based on the Gaussian function, and the bad position is replaced by the new position.

Step 9) calculating the probability of fireworks selection by adopting the following formula:

where G is the set of all fireworks and generic spark positions, R'_iRepresenting the position, R ', of the ith Firework in the set G'_jRepresents the position of the j-th firework in the set G, P (R'_i) Represents R'_iThe probability of being selected.

The regional backlight brightness extraction problem is regarded as an optimization problem, a group of initial backlight brightness is determined through an algorithm based on parameters of image brightness characteristics, the initial value is used as the initial position of fireworks, a plurality of new solutions are generated around the fireworks, the optimal backlight brightness of each partition is solved through the improved firework algorithm, the process of improving the firework algorithm is a search process of the optimal solution, and the optimal solution after the algorithm is ended is a final regional backlight brightness adjustment scheme. The initial position of the firework generated based on the parameter algorithm has higher quality, which is beneficial to meeting the relation between the backlight brightness and the image, accelerating the searching speed and improving the quality of the output solution; the fireworks for the next iteration are selected based on the distance between the solutions, when the sum of the distances between the current position and other positions is larger, the position has higher selection probability, and the selection strategy effectively ensures the diversity of the solutions; the improved firework algorithm can enable the image after the local dimming to obtain higher display quality under the condition of not increasing the power consumption of a local dimming system.

Drawings

FIG. 1 is a flow chart of the present invention image processing method for dynamic dimming of area backlight based on the improved firework algorithm.

Detailed Description

The following describes the method for dynamic dimming of regional backlight based on the improved firework algorithm for image processing according to the present invention in detail with reference to the following embodiments and the accompanying drawings.

The regional backlight dynamic dimming method based on the improved firework algorithm for image processing is characterized in that a regional backlight dynamic dimming problem is regarded as an optimization problem on the basis of image brightness characteristics, a guidance strategy is introduced into an original firework algorithm, structural similarity is used as an evaluation index, and a regional backlight optimization problem model is established. Firstly, determining a group of initial backlight brightness by a method based on image brightness characteristics, taking the initial value as the initial position of a firework, generating a plurality of new solutions around the firework, and solving the optimal backlight brightness of each partition by an improved firework algorithm so as to obtain an optimal partition backlight brightness distribution scheme. To find a method for obtaining the highest display quality of the local dimming image without increasing the power consumption of the local dimming system in all the partitioned brightness distribution schemes.

As shown in fig. 1, the method for dynamically dimming regional backlight based on the improved firework algorithm for image processing of the present invention includes the following steps:

1) the method includes the steps of dividing an image into regions, and determining initial backlight brightness of each region based on a parameter method, wherein the parameter method is one of a maximum value method, an average value method, an error correction method (LUT) and a root mean square method. Taking the initial backlight brightness of each area as the initial position of fireworks in the area; compared with a random position, the position generated by the parameter-based algorithm has higher quality, which is beneficial to meeting the relation between the backlight brightness and the image, accelerating the searching speed and improving the quality of the output solution.

2) Generating a search interval by taking the initial position of the firework as a center; to ensure the diversity of the solution, the interval radius is set to a relatively large value, including: the brightness level of each region is 0 to 255, the initial position of the firework is I, the radius value of the set region is R, R takes one value from 0 to 255, and the search region of each region is obtained as follows: I-R to I + R.

3) Randomly selecting n-1 positions except the initial position of the fireworks in the search interval;

4) arranging n fireworks at n positions;

5) calculating the number of sparks generated by each firework;

the number of sparks generated by each firework is calculated by adopting the following formula:

wherein the content of the first and second substances,

indicating the position of the ith firework,is composed of

The target function of (a) is determined,

x represents a reference image and X represents a reference image,the smaller the size of the tube is,

the higher the corresponding display image quality;

indicating the ith fireworks

The number of sparks generated, the parameter Q is used to control the number of sparks generated by the fireworks, n is the number of fireworks,maximum value of the objective function

ξ is a constant, and 0 < b < a < 1 is set in order to avoid setting the denominator to 0.

6) Calculating the explosion amplitude of each firework;

the explosion amplitude of each firework is calculated by adopting the following formula:

wherein A is_iIndicating the ith Firework

The amplitude of the explosion of the gas turbine,

is the maximum value of the explosion amplitude and,

minimum of the objective function

ξ is a constant and n is the number of fireworks.

7) A location at which a generic spark is generated;

8) generating a location of a pilot spark;

the positions for generating the guide sparks are that the positions of the fireworks and the positions of the general sparks are mixed together to form a mixed set, the quality (SSIM) of each position is calculated in the mixed set, the positions are sorted from high to low according to the quality, the position with the lowest quality is a bad position, the position with the highest quality is an optimal position, the bad position guides and searches a better position in the optimal position, the possibility of finding a better solution is increased, and the solution quality is improved if the better position is found; if the bad position does not find a better position to replace itself, the good position in the mixture set generates a new position based on the Gaussian function, and the bad position is replaced by the new position.

9) Judging whether set N iterations are reached, if not, calculating the probability of fireworks selection by taking the current optimal position in a mixed set formed by mixing the positions of the fireworks and the positions of the general sparks as the initial position of fireworks in the next iteration, selecting other N-1 positions according to the calculated probability, returning to the step 4), and if so, entering the next step;

the probability of fireworks being selected is calculated by the following formula:

where G is the set of all fireworks and generic spark positions, R'_iRepresenting the position, R ', of the ith Firework in the set G'_jRepresents the position of the j-th firework in the set G, P (R'_i) Represents R'_iThe probability of being selected. Because the selection of fireworks for the next iteration is based on the distance between the solutions, the current location will have a higher probability of selection when the sum of the distances between the location and other locations is larger. This selection strategy effectively ensures solution diversity.

10) And taking the current optimal position as an output solution, wherein the output solution is an optimal area backlight brightness distribution scheme.

In order to test the performance of the regional backlight dynamic dimming method based on the improved firework algorithm for image processing, 4 images with wide brightness coverage and different scenes are selected, and performance comparison simulation test is carried out on the images and three algorithms of LUT, FWA and ISFLA. The simulation experiments were performed in a MATLAB R2016a environment, with a resolution of 1920 × 1080 for all test images. For 4 test images, the display quality of the image after dimming obtained by the algorithm and other algorithms under the condition of not increasing the power consumption of the local dimming system is represented by the structure similarity MSSIM, the higher the MSSIM is, the better the image structure similarity is, the better the display quality is, and the comparison result is shown in table 1. The experimental result shows that compared with the LUT algorithm, the FWA algorithm and the ISFLA algorithm, the algorithm can enable the image after the local dimming to obtain higher display quality under the condition of not increasing the power consumption of a local dimming system.

TABLE 1 improved Firework Algorithm vs. other Algorithm Performance comparison

Specific examples are given below:

when the backlight module is divided into 35 regions, and the backlight value of each region is an integer between 0 and 255, the example is as follows:

(1) the solution generated based on the parameter method is used as the initial position.

(2) The section is generated with the initial position as the center. The section radius is set to a relatively large value, the luminance levels of the regions are 0 to 255, and the radius value is set to 50, for example, if the luminance level of a region is calculated as 100 by a parameter-based algorithm, the search section of the region is (50,150).

(3) Randomly selecting other 19 positions in the interval;

(4) 20 fireworks are arranged at the 20 positions;

(5) the number of universal sparks generated per firework is calculated based on the formula (1) of the present invention, which calculates the number of sparks generated per firework. Wherein

The position of each of the fireworks is indicated,

is an objective function of

X represents a reference image and X represents a reference image,

is shown to pass through

And (5) displaying an image obtained by the scheme.

The smaller the size of the tube is,

the higher the corresponding display image quality.

(6) The explosion amplitude of each firework is calculated based on formula (2) of the present invention, which calculates the explosion amplitude of each firework.

(7) A location at which a generic spark is generated;

(8) generating a location of a pilot spark;

(9) judging whether a set termination condition is reached, terminating after 20000 times of evaluation, if the termination condition is not reached, taking the current optimal position as the position of the next iteration firework, and selecting other 19 positions based on the selected probability calculated by the formula (3) for calculating the explosion amplitude of each firework.

(10) If the termination condition is reached, the current best position is taken as the output solution. I.e. the optimal area backlight brightness distribution scheme.