Backlight brightness control method and system based on reinforcement learning

Abstract

The application discloses a backlight brightness control method and system based on reinforcement learning, and relates to the technical field of intelligent control. The method comprises the following steps: state sensing is performed on a target display area to obtain a plurality of state sensing data; a plurality of display content images are extracted, image semantic conversion is performed, and a plurality of display content image semantic description texts are obtained; a plurality of fused semantic description features are obtained; the plurality of state sensing data are updated, backlight brightness control parameter analysis is performed, and a plurality of backlight brightness control parameters are obtained; the plurality of backlight brightness control parameters are centrally screened and analyzed, a target backlight brightness control parameter is obtained, and backlight brightness control is performed. The technical problem that, in the prior art, backlight brightness adjustment relies on fixed rules, lacks semantic feature sensing and adaptive control of display content, and display quality and energy-saving effect are difficult to balance is solved, and the technical effect that intelligent brightness adjustment according to display content is realized through reinforcement learning is achieved.

Description

Technical Field

[0001] This invention relates to the field of intelligent control technology, specifically to a backlight brightness control method and system based on reinforcement learning. Background Technology

[0002] With the widespread application of LCD displays, automotive displays, and mobile terminals, backlight brightness control has become a crucial aspect of improving display quality and reducing energy consumption. However, most existing backlight brightness adjustment methods still rely on fixed rules or single sensor feedback. Their adjustments typically only correct for the overall ambient brightness, lacking a deep understanding of the differences in displayed content. In practical applications, displays often contain complex situations such as high-brightness and low-brightness areas, and static and dynamic scenes. Relying solely on fixed rules for brightness adjustment makes it difficult to simultaneously meet the comprehensive requirements of image clarity, visual comfort, and energy efficiency. For example, when there are details in dark areas, traditional methods can easily lead to detail loss, while in high-brightness scenes, excessive brightness and increased energy consumption may occur. Furthermore, existing methods generally lack deep perception of the semantic features and state data of the displayed content, resulting in a lack of intelligence and adaptability in brightness control, leading to insufficient control precision and failing to meet the dual requirements of high-quality display and energy-saving optimization. Summary of the Invention

[0003] This application provides a backlight brightness control method and system based on reinforcement learning, which solves the technical problems in the prior art where backlight brightness adjustment relies on fixed rules and lacks deep perception of the semantic features and state data of the displayed content, resulting in insufficient brightness control accuracy and difficulty in balancing display quality and energy saving effect.

[0004] The first aspect of this application provides a backlight brightness control method based on reinforcement learning, the method comprising:

[0005] A multi-agent reinforcement learning technique is employed to invoke multiple agents to perceive the state of a target display area according to preset state indicators, thereby obtaining multiple state perception data. Multiple display content images are extracted from the multiple state perception data, and image semantic transformation is performed on each of these images to obtain multiple semantic description texts. Directional features are fused based on these semantic description texts to obtain multiple fused semantic description features. The multiple state perception data are updated according to these fused semantic description features, and multiple agents are invoked to analyze backlight brightness control parameters on the updated state perception data to obtain multiple backlight brightness control parameters. These backlight brightness control parameters are then centrally filtered and analyzed to obtain the target backlight brightness control parameters, which are then input into the backlight controller of the target display area for backlight brightness control.

[0006] A second aspect of this application provides a backlight brightness control system based on reinforcement learning, the system comprising:

[0007] The system comprises the following modules: a State Awareness Module, a Semantic Conversion Module, and a Brightness Control Module. The former employs multi-agent reinforcement learning to invoke multiple agents to perceive the state of the target display area according to preset state indicators, thereby obtaining multiple state awareness data. The latter extracts multiple display content images from the state awareness data and performs image semantic conversion on each image to obtain multiple semantic description texts. The former fuses directional features based on the semantic description texts of the display content images, obtaining multiple fused semantic description features. The latter updates the state awareness data according to the fused semantic description features and invokes multiple agents to parse backlight brightness control parameters from the updated state awareness data, obtaining multiple backlight brightness control parameters. The latter performs centralized filtering and analysis on the multiple backlight brightness control parameters to obtain target backlight brightness control parameters, which are then input into the backlight controller of the target display area for backlight brightness control.

[0008] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0009] First, a multi-agent reinforcement learning technique is employed to invoke multiple agents to perceive the state of the target display area according to preset state indicators, obtaining multiple state perception data. Next, multiple display content images are extracted from the multiple state perception data, and image semantic transformation is performed on each of these images to obtain multiple semantic description texts. Further, directional features are fused based on the multiple semantic description texts to obtain multiple fused semantic description features. Then, the multiple state perception data are updated according to the multiple fused semantic description features, and multiple agents are invoked to analyze the backlight brightness control parameters of the updated state perception data, obtaining multiple backlight brightness control parameters. Finally, the multiple backlight brightness control parameters are centrally filtered and analyzed to obtain the target backlight brightness control parameters, which are then input into the backlight controller of the target display area for backlight brightness control. This invention solves the technical problem in existing technologies where backlight brightness adjustment relies on fixed rules and lacks deep perception of the semantic features and state data of the displayed content, resulting in insufficient brightness control precision and difficulty in balancing display quality and energy saving. It achieves adaptive brightness control through multi-agent reinforcement learning and semantic feature fusion, thereby effectively reducing energy consumption while improving display clarity and visual comfort. Attached Figure Description

[0010] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0011] Figure 1 A schematic flowchart of a backlight brightness control method based on reinforcement learning provided in an embodiment of this application;

[0012] Figure 2 This is a schematic diagram of the backlight brightness control system based on reinforcement learning provided in an embodiment of this application.

[0013] Figure labeling: 11 State perception module, 12 Semantic conversion module, 13 Feature fusion module, 14 Parameter parsing module, 15 Brightness control module. Detailed Implementation

[0014] This application provides a backlight brightness control method and system based on reinforcement learning, which solves the technical problems in the prior art where backlight brightness adjustment relies on fixed rules and lacks deep perception of the semantic features and state data of the displayed content, resulting in insufficient brightness control accuracy and difficulty in balancing display quality and energy saving.

[0015] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0016] It should be noted that the terms "comprising" and "having" are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or server that includes a series of steps or units is not necessarily limited to those steps or units that are explicitly listed, but may include other steps or modules that are not explicitly listed or that are inherent to these processes, methods, products, or devices.

[0017] Example 1, as Figure 1 As shown, this application provides a backlight brightness control method based on reinforcement learning, wherein the method includes:

[0018] By employing multi-agent reinforcement learning technology, multiple agents are invoked to perceive the state of the target display area according to preset state indicators, thereby obtaining multiple state perception data.

[0019] Furthermore, the preset status indicators include the brightness of the target display area, the ambient light intensity, the displayed content image, and the backlight energy efficiency.

[0020] Specifically, multiple state acquisition channels are arranged or virtually modeled in the target display area. Each agent corresponds to an independent state acquisition channel. These channels can collect real-time data on the brightness distribution, ambient light intensity, displayed content image frames, and backlight energy efficiency parameters of the target display area. The acquisition results are then subjected to feature quantization and normalization to form a standardized state vector. Subsequently, multiple agents in a multi-agent reinforcement learning framework are invoked to perceive and respond to the aforementioned state vector. Different agents perform policy exploration and policy utilization tasks respectively, generating multiple state perception results. Each state perception result includes perceived values ​​of the real-time brightness level of the target display area, local area illumination differences, display content feature strength, and backlight energy efficiency consumption. By aggregating the perception results of multiple agents, multiple state perception data are obtained.

[0021] Multiple display content images are extracted from the multiple state-aware data, and image semantic transformation is performed on the multiple display content images to obtain semantic description text of the multiple display content images.

[0022] Furthermore, multiple display content images are extracted from the multiple state-aware data, and image semantic transformation is performed on each of the multiple display content images to obtain semantic description text for the multiple display content images, including:

[0023] A pre-built image semantic converter is used based on a convolutional neural network; the image semantic converter is used to perform image semantic conversion on the multiple display content images to obtain semantic description text of the multiple display content images.

[0024] An image semantic converter is pre-built based on a convolutional neural network. This converter extracts features through convolutional, pooling, and fully connected layers, enabling multi-level analysis of local and global features in the input image. During training, it utilizes a pre-labeled set of display content images for supervised learning, thus acquiring the ability to recognize text, icons, image subjects, background lighting characteristics, and moving edges. Multiple display content images are sequentially input into the image semantic converter, where feature extraction and semantic mapping operations are performed. Low-level pixel information is progressively converted into high-level semantic feature vectors. Combined with a semantic dictionary generated during training, these feature vectors are converted into corresponding semantic descriptive text. This semantic descriptive text can represent the core semantic information of the display content in a natural language symbolic form, such as "the image subject is text content, the background brightness is dark," "the image subject is a high dynamic range video, containing fast-moving objects," or "the image subject is a static icon, the edge area has high color saturation," etc.

[0025] Based on the semantic description text of the multiple displayed content images, directional features are fused to obtain multiple fused semantic description features.

[0026] Furthermore, based on the semantic description text of the multiple displayed content images, directional feature fusion is performed to obtain multiple fused semantic description features, including:

[0027] Semantic description features are extracted by traversing the semantic description text of the multiple displayed content images to obtain multiple semantic description features; pairwise enumeration similarity calculation is performed on the multiple semantic description features to obtain an enumerated similarity set; the multiple semantic description features are filtered based on the enumerated similarity set to determine the leading semantic description feature and multiple following semantic description features; the multiple following semantic description features are directionally fused based on the leading semantic description feature to obtain the multiple fused semantic description features.

[0028] Specifically, the process iterates through multiple semantic description texts of displayed content images, performs feature decomposition on each semantic description text, extracts the corresponding semantic description feature vectors, forming multiple semantic description features. These multiple semantic description features are then enumerated in pairwise combinations, and similarity calculations are performed on each pair of semantic description features. Similarity can be calculated using methods such as cosine similarity or Euclidean distance, ultimately resulting in an enumerated similarity set covering all feature pairs. Based on the enumerated similarity set, the multiple semantic description features are filtered, and the average similarity of each feature with other features is calculated. The semantic description feature with the highest average similarity is identified as the leading semantic description feature, and the remaining semantic description features are marked as following semantic description features. A directional fusion matrix is ​​constructed centered on the leading semantic description feature, and a weighted directional fusion process is performed on the multiple following semantic description features. During the fusion process, the semantic dominance of the leading feature is maintained, and the fused multidimensional semantic vector representation is output through iterative superposition, ultimately obtaining multiple fused semantic description features.

[0029] Furthermore, based on the enumerated similarity set, the multiple semantic description features are filtered to determine the leading semantic description feature and multiple following semantic description features, including:

[0030] Using the multiple semantic description features as indices, the enumerated similarity set is retrieved to obtain multiple semantic description feature association enumerated similarity sets; the mean of the multiple semantic description feature association enumerated similarity sets is calculated to obtain the mean of multiple semantic description feature association enumerated similarity; the semantic description feature corresponding to the maximum value of the multiple semantic description feature association enumerated similarity sets is taken as the leading semantic description feature, and the remaining semantic description features are taken as multiple following semantic description features.

[0031] First, the enumerated similarity sets are indexed according to semantic descriptive features. Using each semantic descriptive feature as a search key, the similarity scores between that feature and all other semantic descriptive features are extracted, forming multiple enumerated similarity sets associated with different semantic descriptive features. Next, the mean of each enumerated similarity set associated with a semantic descriptive feature is calculated, yielding the corresponding mean enumerated similarity score, reflecting the centrality and representativeness of that semantic descriptive feature within the overall set. Then, the mean enumerated similarities of multiple semantic descriptive feature associations are compared, and the one with the largest value is selected as the leading semantic descriptive feature, representing its most representative and directional nature in the global semantic space. Finally, the remaining semantic descriptive features are labeled as following semantic descriptive features in order of their mean scores.

[0032] Furthermore, based on the leading semantic description feature, the plurality of following semantic description features are directionally fused to obtain the plurality of fused semantic description features, including:

[0033] Based on the leading semantic description features, fine-grained sub-feature similarity recognition is performed on the multiple following semantic description features to obtain multiple sets of following fine-grained sub-feature similarities; the multiple sets of following fine-grained sub-feature similarities are traversed and preprocessed to construct multiple directional leading matrices; the multiple directional leading matrices are used to perform directional fusion on the multiple following semantic description features, and combined with the leading semantic description features, the multiple fused semantic description features are obtained.

[0034] Furthermore, the multiple sets of similarity of following fine-grained sub-features are normalized respectively, and the results of the multiple normalization processes are filled into the initially empty matrix to obtain the multiple directional guidance matrices.

[0035] Specifically, using the leading semantic description feature as a reference, each following semantic description feature is divided into fine-grained sub-features, and its semantic vector is decomposed into multiple sub-feature units according to dimensions. Then, the similarity between each sub-feature unit and the corresponding sub-feature of the leading semantic description feature is calculated, resulting in multiple sets of similarity for the following fine-grained sub-features. Similarity can be measured using methods such as cosine similarity or Euclidean distance. Next, these sets of similarity are traversed and preprocessed, with the similarity values ​​in each set normalized. The normalized results are then filled into an initially empty matrix to construct multiple directional leading matrices. Finally, using these multiple directional leading matrices as constraints, the multiple following semantic description features are weighted and directionally fused. During the fusion process, the dominant direction of the leading semantic description feature is maintained, while the following features are weighted and superimposed based on similarity, forming a fused high-level semantic feature representation. This fused representation is then combined with the leading semantic description feature for output, ultimately yielding multiple fused semantic description features.

[0036] The multiple state-aware data are updated based on the multiple fused semantic description features, and multiple intelligent agents are invoked to parse the backlight brightness control parameters of the multiple updated state-aware data to obtain multiple backlight brightness control parameters.

[0037] Specifically, multiple fused semantic description features obtained by fusing directional features are mapped to multiple state-aware data. The brightness parameters, ambient light parameters, display content feature parameters, and backlight energy efficiency parameters in each state-aware data are then updated with weights. The weighting factors are adaptively adjusted based on the similarity between the fused semantic description features and the original features, thereby generating multiple updated state-aware data. Subsequently, these updated state-aware data are input into a multi-agent reinforcement learning framework. Each agent parses the input data based on its own policy network and outputs corresponding candidate values ​​for backlight brightness control parameters. Specifically, the agent generates control parameters, including brightness adjustment range, backlight power allocation ratio, and energy efficiency constraint factors, by combining the display content type and ambient light intensity through a state-action-reward loop mechanism. Finally, the parsing results of all agents are collected to obtain multiple backlight brightness control parameters.

[0038] The multiple backlight brightness control parameters are centrally screened and analyzed to obtain the target backlight brightness control parameters, which are then input into the backlight controller of the target display area for backlight brightness control.

[0039] By centrally screening and analyzing multiple backlight brightness control parameters, the target backlight brightness control parameters are obtained. The target backlight brightness control parameters are then input to the backlight controller of the target display area. The backlight controller performs adjustment operations on the drive current and voltage according to the target backlight brightness control parameters, and controls the backlight brightness output in real time, thereby realizing dynamic adaptive adjustment of backlight brightness under different display content and ambient lighting conditions.

[0040] Furthermore, the target backlight brightness control parameters are obtained by performing centralized screening and analysis on the multiple backlight brightness control parameters, including:

[0041] The average value of the multiple backlight brightness control parameters is calculated to obtain the average value of the backlight brightness control parameters. Starting from the average value of the backlight brightness control parameters, the multiple backlight brightness control parameters are centrally screened according to a preset central screening step size until the preset screening conditions are met. The result obtained from the last screening is used as the target backlight brightness control parameter.

[0042] Furthermore, the preset filtering condition is that the number of times the data is filtered is less than or equal to a preset threshold.

[0043] Specifically, statistical processing is performed on all backlight brightness control parameters output by multiple agents to calculate their arithmetic mean, obtaining the mean value of the backlight brightness control parameters. This mean value serves as the initial reference point for centralized screening. Starting from the mean value of the backlight brightness control parameters, the screening range is iteratively updated in the parameter set according to a preset centralized screening step size. In each iteration, parameters whose deviation from the current reference value exceeds a preset tolerance threshold are removed, and the mean value of the remaining parameters is recalculated as the new reference point. The parameter set is gradually narrowed in this way until the screening process meets the preset screening conditions. That is, when the number of centralized screenings reaches the preset threshold, further screening operations are immediately terminated, and the parameter set obtained from the last iteration is taken as the final result, outputting the target backlight brightness control parameters.

[0044] In summary, the embodiments of this application have at least the following technical effects:

[0045] First, a multi-agent reinforcement learning technique is employed to invoke multiple agents to perceive the state of the target display area according to preset state indicators, obtaining multiple state perception data. Next, multiple display content images are extracted from the multiple state perception data, and image semantic transformation is performed on each of these images to obtain multiple semantic description texts. Further, directional features are fused based on the multiple semantic description texts to obtain multiple fused semantic description features. Then, the multiple state perception data are updated according to the multiple fused semantic description features, and multiple agents are invoked to analyze the backlight brightness control parameters of the updated state perception data, obtaining multiple backlight brightness control parameters. Finally, the multiple backlight brightness control parameters are centrally filtered and analyzed to obtain the target backlight brightness control parameters, which are then input into the backlight controller of the target display area for backlight brightness control. This invention solves the technical problem in existing technologies where backlight brightness adjustment relies on fixed rules and lacks deep perception of the semantic features and state data of the displayed content, resulting in insufficient brightness control precision and difficulty in balancing display quality and energy saving. It achieves adaptive brightness control through multi-agent reinforcement learning and semantic feature fusion, thereby effectively reducing energy consumption while improving display clarity and visual comfort.

[0046] Example 2, based on the same inventive concept as the reinforcement learning-based backlight brightness control method in the previous examples, such as... Figure 2 As shown, this application provides a backlight brightness control system based on reinforcement learning, wherein the system includes:

[0047] State perception module 11: Employs multi-agent reinforcement learning technology, calling multiple agents to perceive the state of the target display area according to preset state indicators, obtaining multiple state perception data; Semantic conversion module 12: Extracts multiple display content images from the multiple state perception data, performs image semantic conversion on the multiple display content images respectively, and obtains multiple display content image semantic description texts; Feature fusion module 13: Performs directional feature fusion based on the multiple display content image semantic description texts, obtaining multiple fused semantic description features; Parameter parsing module 14: Updates the multiple state perception data according to the multiple fused semantic description features, and calls multiple agents to parse the backlight brightness control parameters of the multiple updated state perception data, obtaining multiple backlight brightness control parameters; Brightness control module 15: Performs centralized screening and analysis on the multiple backlight brightness control parameters, obtains target backlight brightness control parameters, and inputs the target backlight brightness control parameters into the backlight controller of the target display area for backlight brightness control.

[0048] Furthermore, the state-aware module 11 is used to perform the following methods:

[0049] The preset status indicators include the brightness of the target display area, the ambient light intensity, the displayed content image, and the backlight energy efficiency.

[0050] Furthermore, the semantic conversion module 12 is used to perform the following method:

[0051] A pre-built image semantic converter is used based on a convolutional neural network; the image semantic converter is used to perform image semantic conversion on the multiple display content images to obtain semantic description text of the multiple display content images.

[0052] Furthermore, the feature fusion module 13 is used to perform the following method:

[0053] Semantic description features are extracted by traversing the semantic description text of the multiple displayed content images to obtain multiple semantic description features; pairwise enumeration similarity calculation is performed on the multiple semantic description features to obtain an enumerated similarity set; the multiple semantic description features are filtered based on the enumerated similarity set to determine the leading semantic description feature and multiple following semantic description features; the multiple following semantic description features are directionally fused based on the leading semantic description feature to obtain the multiple fused semantic description features.

[0054] Furthermore, the feature fusion module 13 is used to perform the following method:

[0055] Using the multiple semantic description features as indices, the enumerated similarity set is retrieved to obtain multiple semantic description feature association enumerated similarity sets; the mean of the multiple semantic description feature association enumerated similarity sets is calculated to obtain the mean of multiple semantic description feature association enumerated similarity; the semantic description feature corresponding to the maximum value of the multiple semantic description feature association enumerated similarity sets is taken as the leading semantic description feature, and the remaining semantic description features are taken as multiple following semantic description features.

[0056] Furthermore, the feature fusion module 13 is used to perform the following method:

[0057] Based on the leading semantic description features, fine-grained sub-feature similarity recognition is performed on the multiple following semantic description features to obtain multiple sets of following fine-grained sub-feature similarities; the multiple sets of following fine-grained sub-feature similarities are traversed and preprocessed to construct multiple directional leading matrices; the multiple directional leading matrices are used to perform directional fusion on the multiple following semantic description features, and combined with the leading semantic description features, the multiple fused semantic description features are obtained.

[0058] Furthermore, the feature fusion module 13 is used to perform the following method:

[0059] The multiple sets of similarity of the following fine-grained sub-features are normalized respectively, and the results of the multiple normalization processes are filled into the initially empty matrix to obtain the multiple directional guidance matrices.

[0060] Furthermore, the brightness control module 15 is used to perform the following method:

[0061] The average value of the multiple backlight brightness control parameters is calculated to obtain the average value of the backlight brightness control parameters. Starting from the average value of the backlight brightness control parameters, the multiple backlight brightness control parameters are centrally screened according to a preset central screening step size until the preset screening conditions are met. The result obtained from the last screening is used as the target backlight brightness control parameter.

[0062] Furthermore, the brightness control module 15 is used to perform the following method:

[0063] The preset filtering condition is that the number of times the data is filtered is less than or equal to the preset number of times threshold.

[0064] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, the above description focuses on specific embodiments of this specification. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some implementations, multitasking and parallel processing are possible or may be advantageous.

[0065] The above description is only a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

[0066] This specification and accompanying drawings are merely illustrative examples of this application and are intended to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Clearly, those skilled in the art can make various alterations and modifications to this application without departing from its scope. Therefore, if such modifications and modifications fall within the scope of this application and its equivalents, this application intends to include such modifications and modifications.

Claims

1. A backlight brightness control method based on reinforcement learning, characterized in that, The method includes: By employing multi-agent reinforcement learning technology, multiple agents are invoked to perceive the state of the target display area according to preset state indicators, thereby obtaining multiple state perception data. Extract multiple display content images from the multiple state-aware data, and perform image semantic transformation on the multiple display content images respectively to obtain semantic description text of multiple display content images; Based on the semantic description text of the multiple displayed content images, directional features are fused to obtain multiple fused semantic description features; The multiple state-aware data are updated based on the multiple fused semantic description features, and multiple intelligent agents are invoked to parse the backlight brightness control parameters of the multiple updated state-aware data to obtain multiple backlight brightness control parameters. The multiple backlight brightness control parameters are centrally screened and analyzed to obtain the target backlight brightness control parameters, and the target backlight brightness control parameters are input into the backlight controller of the target display area to control the backlight brightness. Based on the semantic description text of the multiple displayed content images, directional features are fused to obtain multiple fused semantic description features, including: The semantic description features are extracted by traversing the semantic description text of the multiple displayed content images to obtain multiple semantic description features; The pairwise similarity calculation is performed on the multiple semantic description features to obtain an enumerated similarity set; Based on the enumerated similarity set, the multiple semantic description features are filtered to determine the leading semantic description feature and multiple following semantic description features; Based on the leading semantic description feature, the multiple following semantic description features are directionally fused to obtain the multiple fused semantic description features; Based on the leading semantic description feature, the multiple following semantic description features are directionally fused to obtain the multiple fused semantic description features, including: Based on the leading semantic description features, fine-grained sub-feature similarity recognition is performed on the multiple following semantic description features to obtain multiple sets of following fine-grained sub-feature similarities; The multiple sets of similarity to follow-fine-grained sub-features are traversed for preprocessing to construct multiple directional guiding matrices; Based on the multiple directional guiding matrices, the multiple following semantic description features are directionally fused, and combined with the guiding semantic description features, the multiple fused semantic description features are obtained.

2. The backlight brightness control method based on reinforcement learning as described in claim 1, characterized in that, The preset status indicators include the brightness of the target display area, the ambient light intensity, the displayed content image, and the backlight energy efficiency.

3. The backlight brightness control method based on reinforcement learning as described in claim 1, characterized in that, Extract multiple display content images from the multiple state-aware data, and perform image semantic transformation on each of the multiple display content images to obtain semantic description text for the multiple display content images, including: Pre-built image semantic converter based on convolutional neural network; The image semantic converter is used to perform image semantic conversion on the plurality of display content images to obtain semantic description text for the plurality of display content images.

4. The backlight brightness control method based on reinforcement learning as described in claim 1, characterized in that, Based on the enumerated similarity set, the multiple semantic description features are filtered to determine the leading semantic description feature and multiple following semantic description features, including: Using the multiple semantic description features as indexes, the enumerated similarity set is retrieved to obtain the enumerated similarity set associated with multiple semantic description features; Calculate the mean of the multiple semantic description feature association enumeration similarity sets to obtain the mean of the multiple semantic description feature association enumeration similarity; The maximum value among the average similarity values ​​of the multiple semantic description features is taken as the leading semantic description feature, and the remaining semantic description features are taken as multiple following semantic description features.

5. The backlight brightness control method based on reinforcement learning as described in claim 1, characterized in that, The multiple sets of similarity of the following fine-grained sub-features are normalized respectively, and the results of the multiple normalization processes are filled into the initially empty matrix to obtain the multiple directional guidance matrices.

6. The backlight brightness control method based on reinforcement learning as described in claim 1, characterized in that, The target backlight brightness control parameters are obtained by performing centralized screening and analysis on the multiple backlight brightness control parameters, including: The average value of the backlight brightness control parameters is calculated by averaging the multiple backlight brightness control parameters. Starting from the average value of the backlight brightness control parameters, multiple backlight brightness control parameters are centrally screened according to a preset central screening step size until the preset screening conditions are met. The result obtained from the last screening is then used as the target backlight brightness control parameter.

7. The backlight brightness control method based on reinforcement learning as described in claim 6, characterized in that, The preset filtering condition is that the number of times the data is filtered is less than or equal to the preset number of times threshold.

8. A backlight brightness control system based on reinforcement learning, characterized in that, The system is used to implement the backlight brightness control method based on reinforcement learning as described in any one of claims 1-7, the system comprising: State perception module: Employs multi-agent reinforcement learning technology, calling multiple agents to perceive the state of the target display area according to preset state indicators, and obtaining multiple state perception data; Semantic conversion module: Extracts multiple display content images from the multiple state-aware data, performs image semantic conversion on the multiple display content images respectively, and obtains semantic description text of multiple display content images; Feature fusion module: Performs directional feature fusion based on the semantic description text of the multiple displayed content images to obtain multiple fused semantic description features; Parameter parsing module: Updates the multiple state perception data according to the multiple fused semantic description features, and calls multiple intelligent agents to parse the backlight brightness control parameters of the multiple updated state perception data to obtain multiple backlight brightness control parameters; Brightness control module: performs centralized screening and analysis on the multiple backlight brightness control parameters to obtain target backlight brightness control parameters, and inputs the target backlight brightness control parameters into the backlight controller of the target display area for backlight brightness control.

Images