A method for reducing energy consumption of LED display backlight

Abstract

The invention relates to the technical field of LED display backlight, and discloses a method for reducing energy consumption of LED display backlight, comprising: after a display device receives an instruction, fuzzy recognition is performed on an obtained frame image to obtain a main information part, a secondary information part and a fuzzy information part of the frame image; a backlight panel is fixedly partitioned according to a preset rule to obtain a plurality of candidate areas, each partition including the same number of backlight panels; the main information part, the secondary information part and the fuzzy information part of the frame image are matched to the candidate areas, each candidate area responds to a corresponding frame image part; a first response, a second response and a third response are respectively performed in the candidate areas responding to the main information part, the fuzzy information part and the secondary information part of the frame image; a first relationship between the working time and the temperature of the backlight panel, and a second relationship between the stopping time and the temperature of the backlight panel are determined, and the backlight panel of the candidate area is dynamically scheduled to work according to the first relationship and the second relationship.

Description

Technical Field

[0001] The invention relates to the field of LED display backlight, and more specifically, to a method for reducing energy consumption of LED display backlight. Background Art

[0002] Mini-LED display backlight, as the core technology of display devices, is widely used in small devices such as mobile phones and watches. However, in these portable devices, the energy consumption of Mini-LED backlight has a significant impact on the battery life of the device. The energy consumption of Mini-LED display backlight is affected by many factors, mainly including temperature and content information density. First, temperature has a direct impact on the luminous efficiency of Mini-LED backlight - as the temperature rises, the luminous efficiency of LED will decrease, resulting in increased energy consumption. Secondly, the information density of the displayed content also affects energy consumption, but the existing Mini-LED backlight design does not intelligently adjust for differences in information density, so that content with low information density and content with high information density use the same light intensity output, resulting in unnecessary energy waste. Summary of the invention

[0003] The invention provides a method for reducing energy consumption of LED display backlight, which solves the technical problem raised in the background technology.

[0004] The present invention provides a method for reducing energy consumption of LED display backlight, comprising:

[0005] Step 1, performing fixed partitioning on the Mini-LED backlight panel according to a preset rule to obtain a plurality of candidate areas, each candidate area including the same number of Mini-LED backlight panels;

[0006] Step 2, after the display device receives the instruction, it performs fuzzy recognition on the obtained frame image to obtain the main information part, the secondary information part and the fuzzy information part of the frame image;

[0007] Step 3, matching the main information part, the secondary information part and the blurred information part of the frame image to the candidate regions, each candidate region responding to the corresponding frame image part;

[0008] Step 4, performing a first response, a second response and a third response respectively in the candidate areas in response to the main information part, the blurred information part and the secondary information part of the frame picture;

[0009] Step 5, determine the first relationship between the working time and temperature of the Mini-LED backlight panel, and the second relationship between the stopping time and temperature of the Mini-LED backlight panel, and dynamically schedule the backlight panel of the alternative area to work according to the first relationship and the second relationship.

[0010] Furthermore, the obtained frame image is subjected to fuzzy recognition, including:

[0011] Obtain all historical frame images within a first preset time period before the current moment;

[0012] Input the historical frame pictures into the preset recognition model for recognition, and obtain the information recognition results of the historical frame pictures respectively;

[0013] Perform association matching on the information recognition results of the historical frame images to obtain the main information of the historical frame images;

[0014] The current frame images are divided according to the main information of the historical frame images, as follows:

[0015] Step 6, initialize several occlusion blocks of the same size as the candidate area;

[0016] Step 7, randomly blocking the frame image at the current moment with the blocking block, and obtaining the information recognition result of the frame image at the current moment after each blocking through the preset recognition model;

[0017] Step 8, repeating step 7 for a preset number of times to obtain the occlusion arrangement with the largest number of occlusion blocks and the largest overlap between the information recognition result of the current frame image and the main information of the historical frame image;

[0018] Step 9, in the occlusion arrangement, the occluded part of the frame image at the current moment is used as the secondary information part, and the unoccluded part is used as the primary information part;

[0019] Step 10: Select a preset progressive area in the secondary information part as the fuzzy information part according to the part of the secondary information part adjacent to the primary information area.

[0020] Furthermore, the Mini-LED backlight panel is partitioned according to preset rules to obtain multiple candidate areas, including:

[0021] According to the size of the display device, the matrix constructed by the Mini-LED backlight panel is divided into rows and columns to obtain several candidate areas, each of which includes at least 4 Mini-LED backlight panels.

[0022] Further, matching the main information part, the secondary information part and the blurred information part of the frame image to the candidate area includes:

[0023] Divide the frame image in the same manner as the division method of the Mini-LED backlight panel matrix to obtain a plurality of sub-regions of the frame image, and determine whether the sub-regions of the frame image belong to the main information part, the secondary information part, or the fuzzy information part of the frame image;

[0024] Overlap the frame image with the Mini-LED backlight panel matrix according to a preset display arrangement to obtain a matching relationship between a sub-region of each frame image and each candidate region;

[0025] Matching the candidate area of ​​the main information part of the frame picture, and making a first response;

[0026] Matching the candidate area of ​​the blurred information part of the frame image, and performing a second response;

[0027] A third response is made to the candidate area matching the secondary information portion of the frame picture.

[0028] Furthermore, the first response, the second response and the third response include:

[0029] Determine that the number of Mini-LED backlight panels in the candidate area is K;

[0030] The first response is to set X Mini-LED backlight panels in the candidate area to a working state, K>X;

[0031] The second response is to set the Y Mini-LED backlight panels in the candidate area to the working state, X>Y;

[0032] The third response is to set the Z Mini-LED backlight panels in the candidate area to the working state, Y>Z, Z≠0.

[0033] Furthermore, dynamically scheduling the backlight panel of the candidate area to work according to the first relationship and the second relationship includes:

[0034] Step 11, obtaining the cumulative working time and the cumulative non-working time of the K Mini-LED backlight panels in the ith candidate area at the tth moment in the second preset time period, and the response state of the ith candidate area at the tth moment;

[0035] Step 12, determining the number of features of the Mini-LED backlight panel that needs to be set to a working state in the i-th candidate area at the t-th moment according to the response state of the i-th candidate area at the t-th moment;

[0036] Step 13, weighting the accumulated working time and accumulated non-working time of the j-th Mini-LED backlight panel in the i-th candidate area at the t-th moment respectively according to the first relationship and the second relationship, and taking the difference between the weighted accumulated working time and the accumulated non-working time as the approximate temperature of the j-th Mini-LED backlight panel in the i-th candidate area at the t-th moment;

[0037] Step 14, sorting the approximate temperatures of the K Mini-LED backlight panels in the i-th candidate area at the t-th moment from small to large to obtain a feature sorting;

[0038] Step 15, sort the features from front to back to intercept the approximate temperature of the number of features, and set the corresponding Mini-LED backlight panel to the working state at time t.

[0039] Furthermore, the preset recognition model includes:

[0040] The recognition model includes a text extraction layer and an image feature extraction layer;

[0041] A text extraction layer, used to extract text information from the frame image and use the text information as the first recognition result;

[0042] The image feature extraction layer is used to identify the frame image in multiple dimensions, obtain the description words of the frame image in multiple dimensions, and use the description words as the second recognition result;

[0043] The recognition model combines the first recognition result and the second recognition result into an information recognition result for output.

[0044] Furthermore, the information recognition results of the historical frame images are associated and matched, including:

[0045] Determine a maximum intersection of first recognition results of the preset recognition model on the historical frame image as a first association set;

[0046] Determine a maximum intersection of second recognition results of the preset recognition model on the historical frame image as a second association set;

[0047] The first association set and the second association set are combined to obtain main information of the historical frame image.

[0048] The beneficial effects of the present invention are: by dynamically zoning and intelligently adjusting the LED backlight, the energy waste of the Mini-LED display device under different information densities is reduced; by controlling the working and stopping time of the Mini-LED backlight panel, the negative impact of temperature increase on the LED luminous efficiency is reduced; and the energy efficiency of the overall display is improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 It is a flow chart of a method for reducing energy consumption of LED display backlight of the present invention;

[0050] Figure 2 It is a flow chart of the present invention for performing fuzzy recognition on frame images. DETAILED DESCRIPTION

[0051] The subject matter described herein will now be discussed with reference to example implementations. It should be understood that the discussion of these implementations is only to enable those skilled in the art to better understand and implement the subject matter described herein, and the functions and arrangements of the elements discussed may be changed without departing from the scope of protection of the contents of this specification. Various examples may omit, replace, or add various processes or components as needed. In addition, the features described in some examples may also be combined in other examples.

[0052] like Figure 1-2 As shown, a method for reducing energy consumption of LED display backlight includes:

[0053] Step 1, performing fixed partitioning on the Mini-LED backlight panel according to a preset rule to obtain a plurality of candidate areas, each candidate area including the same number of Mini-LED backlight panels;

[0054] Step 2, after the display device receives the instruction, it performs fuzzy recognition on the obtained frame image to obtain the main information part, the secondary information part and the fuzzy information part of the frame image;

[0055] Step 3, matching the main information part, the secondary information part and the blurred information part of the frame image to the candidate regions, each candidate region responding to the corresponding frame image part;

[0056] Step 4, performing a first response, a second response and a third response respectively in the candidate areas in response to the main information part, the fuzzy information part and the secondary information part of the frame image;

[0057] Step 5, determine the first relationship between the working time and temperature of the Mini-LED backlight panel, and the second relationship between the stopping time and temperature of the Mini-LED backlight panel, and dynamically schedule the backlight panel of the alternative area to work according to the first relationship and the second relationship.

[0058] It should be noted that the display device obtains the instruction, which may be that the advertising device obtains the content preloading instruction, that is, loads and processes the content that needs to be advertised in advance.

[0059] In one embodiment of the present invention, performing fuzzy recognition on the obtained frame image includes:

[0060] Obtain all historical frame images within a first preset time period before the current moment;

[0061] Input the historical frame pictures into the preset recognition model for recognition, and obtain the information recognition results of the historical frame pictures respectively;

[0062] Perform association matching on the information recognition results of the historical frame images to obtain the main information of the historical frame images;

[0063] The current frame images are divided according to the main information of the historical frame images, as follows:

[0064] Step 6, initialize several occlusion blocks of the same size as the candidate area;

[0065] Step 7, randomly blocking the frame image at the current moment with the blocking block, and obtaining the information recognition result of the frame image at the current moment after each blocking through the preset recognition model;

[0066] Step 8, repeating step 7 for a preset number of times to obtain the occlusion arrangement with the largest number of occlusion blocks and the largest overlap between the information recognition result of the current frame image and the main information of the historical frame image;

[0067] Step 9, in the occlusion arrangement, the occluded part of the frame image at the current moment is used as the secondary information part, and the unoccluded part is used as the primary information part;

[0068] Step 10: Select a preset progressive area in the secondary information part as the fuzzy information part according to the part of the secondary information part adjacent to the primary information area.

[0069] In one embodiment of the present invention, the occlusion block implements steps 7 and 8, including:

[0070] Spread the occlusion blocks across the frame image to completely occlude it;

[0071] Randomly remove an occluded block. If the information recognition result of the frame image overlaps with the main information of the historical frame image, the block is retained, otherwise the block is cancelled.

[0072] Repeat a preset number of times, for example 50 times, and then return to the occlusion arrangement.

[0073] Specifically, first, all historical frame images displayed by the display device are collected within a preset time period before the current moment. The purpose is to find the most frequently appearing or more important area in the display by analyzing the past display content. These historical frame images are input into a preset recognition model, and the model will respectively identify and extract the key information in these images. This recognition process can generate the main information features of each frame image, laying the foundation for subsequent matching. The main information in the historical frame image is found by associating and matching the recognition results. In other words, this step determines the most important or most frequently appearing area in the display content, which helps to determine the key part of the current frame image. Region division of the current frame image: set a number of occlusion blocks of the same size as the candidate backlight area to simulate the occlusion effect of the image in different areas. Randomly occlude the current frame image, and obtain the recognition result after occlusion through the recognition model. In this way, it can be observed whether the information of the occluded part contains important information of the current frame. Repeat this operation multiple times, and find an occlusion arrangement by randomly arranging the occlusion blocks, in which the overlap between the frame image information after occlusion and the main information of the historical frame image is the highest. In this way, the important area in the current frame can be effectively located. Information area classification: After determining the best occlusion arrangement, the occluded part of the current frame image is marked as the secondary information part, and the unoccluded part is marked as the main information part. This division is intended to distinguish the key areas and secondary areas of the display. According to the junction between the secondary information part and the main information area, an additional layer of occlusion blocks is set inside and outside the main information area. This layer of occlusion blocks is located in the secondary information part, and this part is used as the blurred information part. The main information part is the brightest, and the secondary information part is the darkest. The blurred information part is used to provide a transitional brightness change between the main information part and the secondary information part, thereby optimizing the visual experience.

[0074] In one embodiment of the present invention, the Mini-LED backlight panel is partitioned according to a preset rule to obtain a plurality of candidate areas, including:

[0075] According to the size of the display device, the matrix constructed by the Mini-LED backlight panel is divided into rows and columns to obtain several candidate areas, each of which includes at least 4 Mini-LED backlight panels.

[0076] In one embodiment of the present invention, matching the main information part, the secondary information part and the fuzzy information part of the frame picture to the candidate area includes:

[0077] Divide the frame image in the same manner as the division method of the Mini-LED backlight panel matrix to obtain a plurality of sub-regions of the frame image, and determine whether the sub-regions of the frame image belong to the main information part, the secondary information part, or the fuzzy information part of the frame image;

[0078] Overlap the frame image with the Mini-LED backlight panel matrix according to a preset display arrangement to obtain a matching relationship between a sub-region of each frame image and each candidate region;

[0079] Matching the candidate area of ​​the main information part of the frame picture, and making a first response;

[0080] Matching the candidate area of ​​the blurred information part of the frame image, and performing a second response;

[0081] A third response is made to the candidate area matching the secondary information portion of the frame picture.

[0082] Specifically, the partition matching of the frame image and the backlight panel:

[0083] First, the frame image is partitioned in the same way according to the partitioning method of the Mini-LED backlight panel. For example, assuming that the backlight panel is divided into a matrix partition of several rows and columns, the frame image is also divided into sub-areas of the same size according to the corresponding rows and columns.

[0084] This matching method can make each sub-area of ​​each frame image directly correspond to a specific backlight panel partition. Through such matching, the system can independently control the brightness requirements of different sub-areas.

[0085] Identification and division of information areas:

[0086] In the partitioned frame image, the system will classify each sub-area into the main information part, the secondary information part and the fuzzy information part according to the preset recognition model.

[0087] The main information part is usually the core area of ​​the displayed content, containing the main content that users are concerned about, while the secondary information part is the secondary background or edge content, and the blurred information part is an area with low information density and little impact on viewing.

[0088] Response patterns in different regions:

[0089] After completing the matching between the frame image area and the backlight panel area, the system sets the backlight response mode of different partitions according to the priority of each area information:

[0090] First response: The backlight areas that match the main information part enter the "first response" mode, that is, the Mini-LED backlight panels in these areas work at a higher brightness to ensure the clarity and brightness of the main content.

[0091] Second response: The backlight areas that partially match the ambiguous information enter the "second response" mode, and the backlight panels of these areas work at a medium brightness state.

[0092] Third response: The backlight areas that partially match the secondary information enter the "third response" mode, that is, the Mini-LED backlight panels in these areas operate at a dimmer brightness.

[0093] In one embodiment of the present invention, the first response, the second response and the third response include:

[0094] Determine that the number of Mini-LED backlight panels in the candidate area is K;

[0095] The first response is to set X Mini-LED backlight panels in the candidate area to a working state, K>X;

[0096] The second response is to set the Y Mini-LED backlight panels in the candidate area to the working state, X>Y;

[0097] The third response is to set the Z Mini-LED backlight panels in the candidate area to the working state, Y>Z, Z≠0.

[0098] In one embodiment of the present invention, dynamically scheduling the backlight panel of the candidate area to work according to the first relationship and the second relationship includes:

[0099] Step 11, obtaining the cumulative working time and the cumulative non-working time of the K Mini-LED backlight panels in the ith candidate area at the tth moment in the second preset time period, and the response state of the ith candidate area at the tth moment;

[0100] Step 12, determining the number of features of the Mini-LED backlight panel that needs to be set to a working state in the i-th candidate area at the t-th moment according to the response state of the i-th candidate area at the t-th moment;

[0101] Step 13, weighting the accumulated working time and accumulated non-working time of the j-th Mini-LED backlight panel in the i-th candidate area at the t-th moment respectively according to the first relationship and the second relationship, and taking the difference between the weighted accumulated working time and the accumulated non-working time as the approximate temperature of the j-th Mini-LED backlight panel in the i-th candidate area at the t-th moment;

[0102] Step 14, sorting the approximate temperatures of the K Mini-LED backlight panels in the i-th candidate area at the t-th moment from small to large to obtain a feature sorting;

[0103] Step 15, sort the features from front to back to intercept the approximate temperature of the number of features, and set the corresponding Mini-LED backlight panel to the working state at time t.

[0104] Specifically, the backlight panel working data is collected, and the working status of each candidate area at each moment (the tth moment) is recorded within a preset time period (the "second preset time period"). This includes the accumulated working time and non-working time of the Mini-LED backlight panel in each area, so as to determine the response state of the area at a specific moment (i.e., high brightness response, low brightness response, or non-working state).

[0105] The record of accumulated working and non-working time is helpful for subsequent temperature estimation, because the temperature of the backlight panel will increase with the increase of working time.

[0106] Determine the number of working backlight panels, and determine the number of Mini-LED backlight panels that need to work in this area at the current moment according to the response status of each candidate area (that is, the brightness requirement of the area at this time). For example, for the main information area that requires high brightness, a larger number of backlight panels are required, while for the secondary or blurred information area, a smaller number of backlight panels are required.

[0107] This method of dynamically determining the number of working backlight panels allows the system to flexibly adjust the backlight working intensity of different areas according to changes in display content, thereby reducing unnecessary energy consumption.

[0108] Estimate the approximate temperature of the backlight panel, and make a weighted estimate of the temperature of each Mini-LED backlight panel based on the accumulated working time and non-working time, combined with the first relationship (the relationship between the working time of the backlight panel and the temperature) and the second relationship (the relationship between the non-working time and the temperature).

[0109] The core of this step is to calculate the temperature difference, that is, the difference between the weighted cumulative working time and non-working time, to approximate the current temperature of the backlight. A higher cumulative working time will result in a higher backlight temperature, while an increase in non-working time will help cool the backlight.

[0110] Temperature sorting: sort the Mini-LED backlight panels in each candidate area from low to high by temperature. The purpose of this step is to give priority to backlight panels with lower temperatures to reduce the overall temperature burden and extend the life of working parts under high temperature conditions.

[0111] This sequencing process allows the system to select a relatively cool backlight panel at each moment, helping to maintain high luminous efficacy and reducing cooling requirements.

[0112] Select the working backlight panel, and according to the characteristic quantity requirement (that is, the number of backlight panels required to work according to the response state at the t moment), select the corresponding number of Mini-LED backlight panels from the sorted low-temperature backlight panels and set them to the working state.

[0113] By giving priority to the use of backlight panels with lower temperatures, the system can not only meet the brightness requirements of the displayed content, but also effectively control the temperature rise and avoid the impact of high temperature on LED efficiency, thereby achieving the purpose of energy saving and improving luminous efficiency.

[0114] In one embodiment of the present invention, the preset recognition model includes:

[0115] The recognition model includes a text extraction layer and an image feature extraction layer;

[0116] A text extraction layer, used to extract text information from the frame image and use the text information as the first recognition result;

[0117] The image feature extraction layer is used to identify the frame image in multiple dimensions, obtain the description words of the frame image in multiple dimensions, and use the description words as the second recognition result;

[0118] The recognition model combines the first recognition result and the second recognition result into an information recognition result for output.

[0119] Specifically, the text extraction layer has the task of extracting text information from the frame image and using these texts as the “first recognition result” of the image.

[0120] Image feature extraction layer: The image feature extraction layer is responsible for multi-dimensional analysis of the visual features of the frame image, thereby generating feature words that describe the image content. These feature words may describe the objects, background, color, shape, etc. in the image, providing data support for image understanding and classification.

[0121] It should be noted that the text extraction layer is implemented based on the ABBYY FineReader program; the image feature extraction layer is implemented based on the Amazon Rekognition program.

[0122] In one embodiment of the present invention, the information recognition results of the historical frame images are associated and matched, including:

[0123] Determine a maximum intersection of first recognition results of the preset recognition model on the historical frame image as a first association set;

[0124] Determine a maximum intersection of second recognition results of the preset recognition model on the historical frame image as a second association set;

[0125] The first association set and the second association set are combined to obtain main information of the historical frame image.

[0126] Specifically, the first association set first performs association matching on the "first recognition result" of each historical frame image. The first recognition result is usually obtained from the text extraction layer, such as subtitles in a video, annotated text in an image, and the like.

[0127] During the association matching process, the system will traverse the first recognition results in the historical frame and look for the part with the "largest intersection".

[0128] The second association set, the second recognition result is usually obtained from the image feature extraction layer, including visual feature description words in the image, such as object category, scene description, etc.

[0129] Similarly, by traversing the historical frames, the system will find the parts with the highest overlap rate among these image features to form a "second association set".

[0130] The above describes an embodiment of the present embodiment, but the present embodiment is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the guidance of the present embodiment, ordinary technicians in this field can also make many forms, all of which are within the protection of the present embodiment.

Claims

1. A method for reducing energy consumption of LED display backlight, characterized in that: include: Step 1, performing fixed partitioning on the Mini-LED backlight panel according to a preset rule to obtain a plurality of candidate areas, each candidate area including the same number of Mini-LED backlight panels; Step 2, after the display device receives the instruction, it performs fuzzy recognition on the obtained frame image to obtain the main information part, the secondary information part and the fuzzy information part of the frame image; Step 3, matching the main information part, the secondary information part and the blurred information part of the frame image to the candidate regions, each candidate region responding to the corresponding frame image part; Step 4, performing a first response, a second response and a third response respectively in the candidate areas in response to the main information part, the blurred information part and the secondary information part of the frame picture; Step 5, determine the first relationship between the working time and temperature of the Mini-LED backlight panel, and the second relationship between the stopping time and temperature of the Mini-LED backlight panel, and dynamically schedule the backlight panel of the alternative area to work according to the first relationship and the second relationship.

2. The method for reducing energy consumption of LED display backlight according to claim 1, characterized in that: Perform fuzzy recognition on the obtained frame image, including: Obtain all historical frame images within a first preset time period before the current moment; Input the historical frame pictures into the preset recognition model for recognition, and obtain the information recognition results of the historical frame pictures respectively; Perform association matching on the information recognition results of the historical frame images to obtain the main information of the historical frame images; The current frame images are divided according to the main information of the historical frame images, as follows: Step 6, initialize several occlusion blocks of the same size as the candidate area; Step 7, randomly blocking the frame image at the current moment with the blocking block, and obtaining the information recognition result of the frame image at the current moment after each blocking through the preset recognition model; Step 8, repeating step 7 for a preset number of times to obtain the occlusion arrangement with the largest number of occlusion blocks and the largest overlap between the information recognition result of the current frame image and the main information of the historical frame image; Step 9, in the occlusion arrangement, the occluded part of the frame image at the current moment is used as the secondary information part, and the unoccluded part is used as the primary information part; Step 10: Select a preset progressive area in the secondary information part as the fuzzy information part according to the part of the secondary information part adjacent to the primary information area.

3. The method for reducing energy consumption of LED display backlight according to claim 1, characterized in that: The Mini-LED backlight panel is partitioned according to preset rules to obtain multiple candidate areas, including: According to the size of the display device, the matrix constructed by the Mini-LED backlight panel is divided into rows and columns to obtain several candidate areas, each of which includes at least 4 Mini-LED backlight panels.

4. The method for reducing energy consumption of LED display backlight according to claim 3, characterized in that: Matching the main information part, the secondary information part and the blurred information part of the frame image to the candidate area includes: Divide the frame image in the same manner as the division method of the Mini-LED backlight panel matrix to obtain a plurality of sub-regions of the frame image, and determine whether the sub-regions of the frame image belong to the main information part, the secondary information part, or the fuzzy information part of the frame image; Overlap the frame image with the Mini-LED backlight panel matrix according to a preset display arrangement to obtain a matching relationship between a sub-region of each frame image and each candidate region; Matching the candidate area of ​​the main information part of the frame picture, and making a first response; Matching the candidate area of ​​the blurred information part of the frame image, and performing a second response; A third response is made to the candidate area matching the secondary information portion of the frame picture.

5. A method for reducing energy consumption of LED display backlight according to claim 4, characterized in that: First response, second response and third response, including: Determine that the number of Mini-LED backlight panels in the candidate area is K; First response: set X Mini-LED backlight panels in the candidate area to work, K>X; The second response is to set the Y Mini-LED backlight panels in the candidate area to the working state, X>Y; The third response is to set the Z Mini-LED backlight panels in the candidate area to the working state, Y>Z, Z≠0.

6. A method for reducing energy consumption of LED display backlight according to claim 5, characterized in that: Dynamically scheduling the backlight panel of the candidate area to work according to the first relationship and the second relationship, including: Step 11, obtaining the cumulative working time and the cumulative non-working time of the K Mini-LED backlight panels in the ith candidate area at the tth moment in the second preset time period, and the response state of the ith candidate area at the tth moment; Step 12, determining the number of features of the Mini-LED backlight panel that needs to be set to a working state in the i-th candidate area at the t-th moment according to the state of the response of the i-th candidate area at the t-th moment; Step 13, weighting the accumulated working time and accumulated non-working time of the j-th Mini-LED backlight panel in the i-th candidate area at the t-th moment respectively according to the first relationship and the second relationship, and taking the difference between the weighted accumulated working time and the accumulated non-working time as the approximate temperature of the j-th Mini-LED backlight panel in the i-th candidate area at the t-th moment; Step 14, sorting the approximate temperatures of the K Mini-LED backlight panels in the i-th candidate area at the t-th moment from small to large to obtain a feature sorting; Step 15, sort the features from front to back to intercept the approximate temperature of the number of features, and set the corresponding Mini-LED backlight panel to the working state at time t.

7. The method for reducing energy consumption of LED display backlight according to claim 2, characterized in that: Preset recognition models include: The recognition model includes a text extraction layer and an image feature extraction layer; A text extraction layer, used to extract text information from the frame image and use the text information as the first recognition result; The image feature extraction layer is used to identify the frame image in multiple dimensions, obtain the description words of the frame image in multiple dimensions, and use the description words as the second recognition result; The recognition model combines the first recognition result and the second recognition result into an information recognition result for output.

8. A method for reducing energy consumption of LED display backlight according to claim 7, characterized in that: The information recognition results of the historical frame images are associated and matched, including: Determine a maximum intersection of first recognition results of the preset recognition model on the historical frame image as a first association set; Determine a maximum intersection of second recognition results of the preset recognition model on the historical frame image as a second association set; The first association set and the second association set are combined to obtain main information of the historical frame image.

Images