CN114880914A - Regional dimming model based on multi-objective optimization - Google Patents
Regional dimming model based on multi-objective optimization Download PDFInfo
- Publication number
- CN114880914A CN114880914A CN202110993249.7A CN202110993249A CN114880914A CN 114880914 A CN114880914 A CN 114880914A CN 202110993249 A CN202110993249 A CN 202110993249A CN 114880914 A CN114880914 A CN 114880914A
- Authority
- CN
- China
- Prior art keywords
- individual
- value
- backlight
- individuals
- image
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000005457 optimization Methods 0.000 title claims abstract description 27
- 238000000034 method Methods 0.000 claims abstract description 48
- 239000011159 matrix material Substances 0.000 claims abstract description 14
- 238000004220 aggregation Methods 0.000 claims description 14
- 230000002776 aggregation Effects 0.000 claims description 14
- 239000013598 vector Substances 0.000 claims description 14
- 238000000605 extraction Methods 0.000 claims description 7
- 230000002068 genetic effect Effects 0.000 claims description 5
- 238000012545 processing Methods 0.000 claims description 3
- 238000011156 evaluation Methods 0.000 claims description 2
- 238000012854 evaluation process Methods 0.000 claims description 2
- 238000009499 grossing Methods 0.000 claims description 2
- 238000010606 normalization Methods 0.000 claims 1
- 238000000354 decomposition reaction Methods 0.000 abstract description 16
- 238000005516 engineering process Methods 0.000 abstract description 4
- 238000004364 calculation method Methods 0.000 abstract 1
- 238000005192 partition Methods 0.000 description 5
- 230000000694 effects Effects 0.000 description 3
- 102100025912 Melanopsin Human genes 0.000 description 2
- 238000012937 correction Methods 0.000 description 2
- 239000004973 liquid crystal related substance Substances 0.000 description 2
- 239000013213 metal-organic polyhedra Substances 0.000 description 2
- 238000012011 method of payment Methods 0.000 description 2
- 238000009825 accumulation Methods 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 230000014509 gene expression Effects 0.000 description 1
- 238000012804 iterative process Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 238000012163 sequencing technique Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/12—Computing arrangements based on biological models using genetic models
- G06N3/126—Evolutionary algorithms, e.g. genetic algorithms or genetic programming
-
- G—PHYSICS
- G09—EDUCATION; CRYPTOGRAPHY; DISPLAY; ADVERTISING; SEALS
- G09G—ARRANGEMENTS OR CIRCUITS FOR CONTROL OF INDICATING DEVICES USING STATIC MEANS TO PRESENT VARIABLE INFORMATION
- G09G3/00—Control arrangements or circuits, of interest only in connection with visual indicators other than cathode-ray tubes
- G09G3/20—Control arrangements or circuits, of interest only in connection with visual indicators other than cathode-ray tubes for presentation of an assembly of a number of characters, e.g. a page, by composing the assembly by combination of individual elements arranged in a matrix no fixed position being assigned to or needed to be assigned to the individual characters or partial characters
- G09G3/34—Control arrangements or circuits, of interest only in connection with visual indicators other than cathode-ray tubes for presentation of an assembly of a number of characters, e.g. a page, by composing the assembly by combination of individual elements arranged in a matrix no fixed position being assigned to or needed to be assigned to the individual characters or partial characters by control of light from an independent source
- G09G3/3406—Control of illumination source
- G09G3/342—Control of illumination source using several illumination sources separately controlled corresponding to different display panel areas, e.g. along one dimension such as lines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2111/00—Details relating to CAD techniques
- G06F2111/06—Multi-objective optimisation, e.g. Pareto optimisation using simulated annealing [SA], ant colony algorithms or genetic algorithms [GA]
Abstract
The local dimming technology can improve the display quality of the display system and reduce the power consumption of the system. Setting the optimal backlight brightness matrix by the local dimming method is key to enable the local dimming system to achieve the optimal performance. The regional dimming problem model is established by regarding regional dimming as a multi-objective optimization problem, taking improvement of image distortion, improvement of image contrast and reduction of system power consumption as optimization targets. In order to effectively solve the multi-target optimization problem, the invention improves a multi-target evolutionary algorithm (MOEA/D) based on decomposition, fuses the MOEA/D and the SFLA, and provides an improved multi-target evolutionary calculation method which is named as MOEA/D-SFLA. Experimental results prove that compared with a regional dimming method based on parameters, the MOEA/D-SFLA can simultaneously improve image distortion, improve image contrast and reduce system power consumption. MOEA/D-SFLAs have higher performance than some improved MOEA/D.
Description
One, the technical field
The method is applied to the HDR display system, and based on multi-objective optimization, the regional dimming technology in the display system is improved and optimized. The method aims to find a balance point of image quality, power consumption and contrast of the display system.
Second, background Art
1. Decomposition-based multi-objective optimization problem
The idea of multi-objective decomposition mainly means that a multi-objective optimization problem is decomposed into a certain number of sub-problems, and then each sub-problem is optimized simultaneously. In a classical decomposition-based multi-objective optimization evolutionary algorithm (MOEA/D), the MOEA/D does not treat the MOP as a whole, but decomposes the multi-objective optimization problem into several single-objective sub-problems by weight vectors, optimizes all sub-problems simultaneously by using neighborhood relations between sub-problems determined by euclidean distances between weight vectors, wherein the decomposition is achieved by an aggregation method, and the common decomposition method is: weighted summation, Chebyshev method and boundary crossing method. The mathematical expressions of the three decomposition methods are respectively:
(1) weighted Summation (WS). The single-target optimization problem obtained after the decomposition by the method is as follows:
subject x∈Ω
the weighted sum decomposition method carries out accumulation processing on all objective functions and then solves the overall subproblems.
(2) Chebyshev method. The single-target optimization problem obtained after the decomposition by the method is as follows:
subject to x∈Ω
wherein
As a reference point, for each
By changing the weight vector, more different Pareto optimal solutions can be obtained by using MOEA/D of TCH decomposition method.
(3) Penalty based boundary intersection method (PBI). The single-target optimization problem obtained after the decomposition by the method is as follows:
min imize g pbi (x|λ，z * )＝d 1 +θd 2
subject to x∈Ω
where θ is a penalty factor. When the number of the solving targets is more than 2 MOPs, if the weight vectors with the same distribution are used, the MOEA/D using the PBI decomposition method has better effect than the MOEA/D using the TCH decomposition method, but the cost is required to be paid for the MOEA/D, and therefore an effective penalty factor theta must be set.
2. Local dimming
In an LCD display system, since liquid crystals cannot emit light autonomously, it is necessary to provide brightness by means of backlight LEDs to display video or images. The local dimming means that an input image is partitioned, and the LED of each partition is independently controlled to emit light according to the image characteristics of the partition, so that the image quality can be effectively improved and the power consumption can be reduced by the local dimming technology. The existing LED area dimming technology can be generally divided into two parts of area backlight extraction and liquid crystal pixel compensation. The backlight extraction means that when an image is input into a display system, the image is partitioned, backlight brightness of each partition is obtained according to different algorithms, and finally a backlight matrix of the whole image is obtained. After the backlight extraction is performed on the input image, since the size of the backlight value is determined by the corresponding image content characteristics, in order to ensure that the display brightness of the image in actual display is kept as constant as possible when the backlight is fully bright, the pixel value of the display image needs to be further adjusted according to the backlight information, so that the light transmission amount is effectively controlled according to the size of the backlight value, and the process is called pixel compensation.
The most basic and classical methods in area backlight extraction are the maximum and average methods. The maximum value method takes the maximum value of the pixel gray level in the image area as the partition backlight brightness, although the image brightness can be improved, the energy-saving effect is poor, and the method is very sensitive to noise; the average value method is to take the average value of the pixel gray levels in the image area as the partition backlight brightness, and compared with the maximum value method, the energy-saving effect is greatly improved, but the display brightness of the image highlight area is easily reduced, the display brightness of the image dark area is improved, and the image after dimming is still easily distorted. In addition, the backlight extraction algorithm also includes a table look-up method, a CDF method, a gaussian function method, a dynamic threshold method, and the like. The table look-up method is that on the basis of the average value method, a backlight correction value is added according to the brightness information of the subareas, and the obtained result is the backlight brightness of the area; the CDF method utilizes the gray level histogram of each subarea of the image to obtain the probability density curve of each subarea, the probability density curves of the subareas are accumulated to obtain the CDF curve, a preset threshold value K is mapped to a CDF curve gray level axis, the subarea backlight brightness value is obtained, a backlight correction value is added according to the brightness information of the subarea, and the obtained result is the backlight brightness of the area. The method comprises the steps of extracting backlight by using an average value and a variance of brightness of an input image, obtaining a gray value for separating a foreground from a background by using a maximum inter-class variance method through a dynamic threshold method, carrying out binarization processing on the image to calculate a subarea backlight adjusting coefficient, further determining a subarea backlight dimming gray level and a subarea backlight dimming ratio, and finally obtaining a subarea backlight value by using the subarea backlight dimming ratio in combination with a maximum value and the average value.
Third, the invention
The technical scheme adopted by the invention is as follows: optimizing a local dimming algorithm by utilizing decomposition-based multi-objective optimization, comprising the following steps of:
1. a regional dimming model based on multi-objective optimization is characterized by comprising the following steps:
(1) initialization: and initializing the multi-objective optimization parameters and the image parameters.
(2) Iterative evolution: and generating a filial generation individual by using a genetic operator and evaluating.
(3) Updating: updating the neighborhood by using a Chebyshev aggregation method, comparing the optimal conditions of the filial generation individuals and the neighbor individuals according to the aggregation value, and then updating the neighborhood. Finally, the outer population EP is updated, the non-dominant solution is added to the EP, and the dominant solution is removed from the EP.
2. The regional dimming model based on multi-objective optimization according to 1, is characterized in that the step 1) comprises:
(1) after the image is read in, the image is down-sampled to improve the operation speed.
(2) Setting the maximum iteration times maxgen, the population size N and the neighbor scale size T, generating a group of uniformly distributed weight vectors, calculating Euclidean distances of the weight vectors, and storing the index values of the nearest T neighbors of each weight vector in B.
(3) And carrying out backlight extraction on the sampled image to obtain a backlight matrix, changing the two-dimensional backlight matrix into a one-dimensional individual as an initial individual, setting the optimizing range within plus or minus 20 of the brightness value of the individual, and if the brightness is higher than 255 or lower than 0, setting the backlight value to be 255 or 0.
(4) N individuals are randomly generated by the initial individual, each individual is evaluated, and the evaluation values are normalized to improve the algorithm performance.
(5) Setting the EP of the external population to be empty, and storing the individual in the iterative process
3. The multi-objective optimization-based regional dimming model initialization process according to the method 2, wherein in the step 4), an individual objective function value is obtained from each individual in the initialization population, and the evaluation process comprises the following steps:
(1) and changing the one-dimensional individuals into a two-dimensional backlight matrix, performing backlight smoothing operation on the backlight matrix to obtain the backlight matrix with the same size as the input image, and performing pixel compensation on the matrix.
(2) Calculating the mean square error mse of the compensated image, wherein the formula is as follows:
(3) calculating power consumption pc and contrast cr, wherein the formula is as follows:
4. the multi-objective optimization-based local dimming model according to claim 1, wherein, in step 2), two references are randomly selected from b (i), then a new solution is generated for two parent individuals and xl by using a genetic operator, if a certain dimension of a child individual exceeds a set upper or lower brightness value, the brightness value of the dimension is set as the upper or lower brightness value, and finally the function value of the child individual is evaluated.
5. The regional dimming model based on multi-objective optimization according to 1 is characterized in that in step 3), the neighborhood is updated by using a frog leap thought, and the method comprises the following steps:
(1) comparing the aggregation function value of the filial generation individual with the aggregation function values of the T neighbors, if the filial generation individual is better than a certain neighbor, replacing the neighbor with the filial generation individual, otherwise, performing the second step
(2) And (3) sequencing according to the frog population fitness value to obtain the optimal frog and the worst frog in the neighbors, generating an individual by using the information of the optimal frog, comparing the aggregation function value of the individual with the aggregation function values of the T neighbors, replacing the neighbor by the individual if the individual is better than the neighbor, and otherwise, carrying out the third step.
(3) Two individuals were randomly generated to replace the worst two individuals in the neighborhood.
Description of the drawings
The attached drawing is an improved process for updating a neighborhood solution by multi-objective optimization based on decomposition adopted by the invention.
Fifth, detailed description of the invention
1. Initialization
1.1 setting experiment parameters including maximum iteration times gen, algorithm population scale N, neighbor scale T, cross probability pc and variation probability pm, and determining an upper limit xmax and a lower limit xmin of a decision variable.
1.2 obtaining N evenly distributed weight vectors
1.3 down-sampling the image gray-scale map, obtaining a group of backlight values based on LUT algorithm, and randomly generating a plurality of groups of backlight values by the group of backlight values
1.4 evaluate this set of solutions, get the original objective function values, store the minimum of the target values for each sub-problem into z, which is the best value found so far, and add the non-dominated solution to the outer population EP.
1.5, calculating Euclidean distance between the weight vectors, and searching indexes of T weight vectors nearest to each weight vector.
2. Iterative evolution
2.1 randomly taking two indexes k and l in B, generating a filial generation unit for two parents and using a genetic operator, and if a certain dimension of the filial generation unit exceeds a set upper limit value or lower limit value, setting the dimension as the upper limit value or lower limit value.
2.2 evaluating the offspring individuals to obtain three objective function values.
3 updating
And 3.1 updating the neighbor B (i) according to the selected aggregation method, selecting a Chebyshev method in the text, comparing the individuals in the neighbor according to a formula of the Chebyshev method to update the value of the corresponding subproblem, and generating a new solution to replace the worst individual in the neighbor solution through a frog-leap algorithm if the update does not occur.
3.2 update the outer population EP, update according to calculating the value of each subproblem, add non-dominant solutions to the EP, remove dominant solutions from the EP.
Claims (5)
1. A regional dimming model based on multi-objective optimization is characterized by comprising the following steps:
1) initialization: initializing multi-objective optimization parameters and image parameters;
2) iterative evolution: generating a filial generation individual by using a genetic operator and evaluating;
3) updating: updating the neighborhood by using a Chebyshev aggregation method, comparing the optimal conditions of the filial individuals and the neighbor individuals according to the aggregation value, then updating the neighborhood, finally updating the external population EP, adding the non-dominant solution into the EP, and removing the dominant solution from the EP.
2. The regional dimming model based on multi-objective optimization according to 1, is characterized in that the step 1) comprises:
1) after an image is read in, the image is down-sampled to improve the operation speed;
2) setting the maximum iteration times maxgen, the population size N and the neighbor scale size T, generating a group of uniformly distributed weight vectors, calculating Euclidean distances of the weight vectors, and storing index values of the nearest T neighbors of each weight vector in B;
3) carrying out backlight extraction on the sampled image to obtain a backlight matrix, changing a two-dimensional backlight matrix into a one-dimensional individual as an initial individual, setting the optimizing range within plus or minus 20 of the brightness value of the individual, and if the brightness is higher than 255 or lower than 0, setting the backlight value to be 255 or 0;
4) randomly generating N individuals by the initial individual, evaluating each individual, and carrying out normalization processing on the evaluation value to improve the algorithm performance;
5) and setting the external population EP to be empty, and storing the individuals in the iteration process.
3. The multi-objective optimization-based regional dimming model initialization process according to the method 2, wherein in the step 4), an individual objective function value is obtained from each individual in the initialization population, and the evaluation process comprises the following steps:
1) changing the one-dimensional individuals into a two-dimensional backlight matrix, performing backlight smoothing operation on the backlight matrix to obtain the backlight matrix with the same size as the input image, and performing pixel compensation on the matrix;
2) calculating the mean square error mse of the compensated image;
3) calculating power consumption pc and contrast cr, wherein the formula is as follows:
4. the multi-objective optimization-based local dimming model according to claim 1, in step 2), two indexes are randomly selected from b (i), then a genetic operator is used for two parent individuals to generate a new solution, if a certain dimension of a child individual exceeds a set upper limit value or a set lower limit value of brightness, the brightness value of the dimension is set as the upper limit value or the lower limit value, and finally the function value of the child individual is evaluated.
5. The regional dimming model based on multi-objective optimization according to 1 is characterized in that in step 3), the neighborhood is updated by using a frog leap thought, and the method comprises the following steps:
1) comparing the aggregation function value of the filial generation individual with the aggregation function values of the T neighbors, if the filial generation individual is better than a certain neighbor, replacing the neighbor with the filial generation individual, otherwise, performing the second step;
2) sorting according to the frog population fitness value to obtain the optimal frog and the worst frog in the neighbors, generating an individual by using the information of the optimal frog, comparing the aggregation function value of the individual with the aggregation function values of the T neighbors, replacing the neighbor by the individual if the individual is better than the neighbor, and otherwise, performing the third step;
3) two individuals were randomly generated to replace the worst two individuals in the neighborhood.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110993249.7A CN114880914A (en) | 2021-08-27 | 2021-08-27 | Regional dimming model based on multi-objective optimization |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110993249.7A CN114880914A (en) | 2021-08-27 | 2021-08-27 | Regional dimming model based on multi-objective optimization |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114880914A true CN114880914A (en) | 2022-08-09 |
Family
ID=82667190
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110993249.7A Pending CN114880914A (en) | 2021-08-27 | 2021-08-27 | Regional dimming model based on multi-objective optimization |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114880914A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117690388A (en) * | 2024-02-04 | 2024-03-12 | 深圳康荣电子有限公司 | Picture optimization method and system based on display module backlight |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20110134358A1 (en) * | 2009-06-03 | 2011-06-09 | Manufacturing Resources International, Inc. | Dynamic dimming led backlight |
| CN107342056A (en) * | 2017-07-31 | 2017-11-10 | 天津大学 | A kind of region backlight dynamic light adjustment method that the algorithm that leapfrogs is shuffled based on improvement |
| CN107507580A (en) * | 2017-09-06 | 2017-12-22 | 天津大学 | Local dimming method applied to Novel backlight |
| CN109887468A (en) * | 2019-04-02 | 2019-06-14 | 合肥工业大学 | A kind of liquid crystal display dynamic light adjustment method based on image local feature |
| US20210005162A1 (en) * | 2018-06-13 | 2021-01-07 | Shenzhen Skyworth-Rgb Electronic Co., Ltd. | Image quality optimization method based on local dimming, apparatus, and computer readable storage medium |
-
2021
- 2021-08-27 CN CN202110993249.7A patent/CN114880914A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20110134358A1 (en) * | 2009-06-03 | 2011-06-09 | Manufacturing Resources International, Inc. | Dynamic dimming led backlight |
| CN107342056A (en) * | 2017-07-31 | 2017-11-10 | 天津大学 | A kind of region backlight dynamic light adjustment method that the algorithm that leapfrogs is shuffled based on improvement |
| CN107507580A (en) * | 2017-09-06 | 2017-12-22 | 天津大学 | Local dimming method applied to Novel backlight |
| US20210005162A1 (en) * | 2018-06-13 | 2021-01-07 | Shenzhen Skyworth-Rgb Electronic Co., Ltd. | Image quality optimization method based on local dimming, apparatus, and computer readable storage medium |
| CN109887468A (en) * | 2019-04-02 | 2019-06-14 | 合肥工业大学 | A kind of liquid crystal display dynamic light adjustment method based on image local feature |
Non-Patent Citations (3)
| Title |
|---|
| ZHENGCAI CAO, ET AL: "An Improved MOEA/D for Order Scheduling Problem in Automated Warehouse", 《2018 IEEE 14TH INTERNATIONAL CONFERENCE ON AUTOMATION SCIENCE AND ENGINEERING (CASE)》, 24 August 2018 (2018-08-24), pages 797 - 802, XP033463245, DOI: 10.1109/COASE.2018.8560504 * |
| 刘恺文: "基于分解的多目标自动化立体仓库调度方法研究", 《中国优秀硕士学位论文全文数据库》, 15 June 2020 (2020-06-15), pages 10 - 12 * |
| 巫红英: "基于HDR显示的区域背光算法研究", 《中国优秀硕士学位论文全文数据库》, 15 April 2019 (2019-04-15), pages 35 - 56 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117690388A (en) * | 2024-02-04 | 2024-03-12 | 深圳康荣电子有限公司 | Picture optimization method and system based on display module backlight |
| CN117690388B (en) * | 2024-02-04 | 2024-04-19 | 深圳康荣电子有限公司 | Picture optimization method and system based on display module backlight |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Abd Elaziz et al. | Many-objectives multilevel thresholding image segmentation using knee evolutionary algorithm | |
| Zhang et al. | A dynamic three-way decision model based on the updating of attribute values | |
| Khaire et al. | High-dimensional microarray dataset classification using an improved adam optimizer (iAdam) | |
| CN108846526A (en) | A kind of CO2 emissions prediction technique | |
| CN109344994A (en) | A kind of prediction model method based on improvement moth optimization algorithm | |
| CN110245783B (en) | Short-term load prediction method based on C-means clustering fuzzy rough set | |
| CN111583031A (en) | Application scoring card model building method based on ensemble learning | |
| CN114880914A (en) | Regional dimming model based on multi-objective optimization | |
| CN116091455A (en) | Steel mesh surface defect judging method based on machine vision | |
| CN114463211A (en) | Underwater image enhancement method based on turbidity classification | |
| Dhanalakshmy et al. | A survey on adaptation strategies for mutation and crossover rates of differential evolution algorithm | |
| CN112381157A (en) | Fuzzy C-means clustering method based on goblet sea squirt group algorithm optimization | |
| Huang et al. | Segmentation on ripe Fuji apple with fuzzy 2D entropy based on 2D histogram and GA optimization | |
| Li et al. | A nondominated sorting genetic algorithm III with three crossover strategies for the combined heat and power dynamic economic emission dispatch with or without prohibited operating zones | |
| CN111210347A (en) | Transaction risk early warning method, device, equipment and storage medium | |
| CN114841872A (en) | Digital halftone processing method based on multi-agent deep reinforcement learning | |
| Zhang et al. | A local dimming method based on improved multi-objective evolutionary algorithm | |
| Wang et al. | A novel imperialist competitive algorithm for multithreshold image segmentation | |
| Hu et al. | Multiobjective image color quantization algorithm based on self-adaptive hybrid differential evolution | |
| CN116935054A (en) | Semi-supervised medical image segmentation method based on hybrid-decoupling training | |
| Xia et al. | A novel key influencing factors selection approach of P2P lending investment risk | |
| CN107220978B (en) | Multi-target threshold image segmentation method fusing interval fuzzy information and statistical information | |
| CN114140645B (en) | Photographic image aesthetic style classification method based on improved self-supervision feature learning | |
| CN107274357B (en) | Gray level image enhancement processing system with optimal parameters | |
| CN110084376B (en) | Method and device for automatically separating data into boxes |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |