A method and system for controlling energy consumption of an LED display screen

Abstract

The application discloses an LED display screen energy consumption control method and system, and relates to the technical field of display screen energy consumption control; the method comprises a first data acquisition module, a data source processing module, a second data acquisition module, a real-time processing module and a correction processing module; the first data acquisition module collects video uploaded by a user and a video label, and stores the video and the video label in a distributed file system; the technical key points are as follows: the data source is processed and analyzed, the calculation complexity is reduced, the video is processed in advance, the video does not need to be further adjusted in the playing link, the processing amount of runtime data is reduced, the display screen parameters are dynamically adjusted according to the environment and user demand, energy waste is avoided, the LED display screen energy consumption is significantly reduced, the video playing quality is guaranteed, and the user experience is improved; compared with a traditional scheme, the scheme not only realizes intelligent video playing, but also greatly reduces the operation cost, has good overall effect, and has good use prospect.

Description

Technical Field

[0001] The present invention relates to the technical field of display screen energy consumption control, and in particular to an LED display screen energy consumption control method and system. Background Art

[0002] With the rapid development of digital technology, LED displays, with their advantages of high brightness, wide viewing angles, and long lifespan, have become a vital medium for displaying information in public places like shopping malls. In shopping malls, large LED displays are often used to display brand advertisements, promotions, and navigation guides. They not only effectively attract consumers' attention and enhance brand image, but also provide customers with convenient service information, becoming an indispensable part of mall operations.

[0003] However, the widespread use of LED displays has also brought with it increasingly severe energy consumption issues. For example, large screens in shopping malls are often large, often covering dozens or even hundreds of square meters, and require continuous operation for extended periods of time. In daily operations, traditional LED displays employ a relatively crude display mode, maintaining a fixed brightness, content, and display mode regardless of ambient light levels or foot traffic. This not only wastes a significant amount of energy, contributing to high electricity costs for shopping malls, but also accelerates the aging of display equipment due to prolonged high-load operation, increasing maintenance costs. Furthermore, in the context of promoting green and low-carbon development, high-energy-consuming displays are contrary to the concept of sustainable development. Therefore, effectively controlling the energy consumption of LED displays has become a pressing need for shopping malls to reduce operating costs, implement environmental protection concepts, and enhance their overall competitiveness. To this end, several energy consumption control solutions for LED displays have been developed.

[0004] The existing patent authorization announcement number is "CN116863868B", and the patent name is "Display control method, device and display system for large-size LED screens for curtain walls". It records that "the method includes the following steps: S1: obtaining the distribution of people around the LED screen; S2: determining at least a portion of the LED screen as a target display area based on the distribution of people around the LED screen; S3: determining the target resolution for displaying each target display area based on the distribution of people around the screen; S4: controlling each target display area of ​​the LED screen to display at the corresponding target resolution. The present invention can reduce the energy consumption of the LED screen and extend the service life of the LED screen."

[0005] However, in actual use, it has certain defects:

[0006] First of all, it concentrates all data processing processes on the moment after identifying the crowd. At this time, the amount of data that needs to be processed is large, so the requirements for the equipment are very high. It cannot be applied in the scenario of the existing LED display screen that has been deployed. In addition, the solution it adopts is partition display, that is, displaying multiple pictures. Although part of the display screen is not working in this way, it is equivalent to realizing split-screen display of the video. There will be obvious interference during display, which reduces the display effect. Moreover, the multi-point centralized display method has smaller partitions, which will increase the current load of the centralized display area. If the centralized display area is in a high-brightness state for a long time, local heat accumulation will accelerate, shortening the life of the lamp beads. At the same time, the small screen display makes it difficult to display video effects. Split-screen control requires data transmission to multiple centralized display areas (multiple transmissions at the same time), which increases the energy consumption of this part. Therefore, the energy-saving effect is limited and does not meet market requirements. For this reason, we propose an energy consumption control method and system for LED display screens. Summary of the Invention

[0007] To achieve the above objectives, the present invention is implemented through the following technical solutions:

[0008] A method and system for controlling energy consumption of an LED display screen includes a first data acquisition module, a data source processing module, a second data acquisition module, a real-time processing module, and a correction processing module:

[0009] The first data collection module collects videos and video tags uploaded by users and stores them in a distributed file system;

[0010] Data source processing module: This module performs noise reduction and frame segmentation on video data stored in the distributed file system to form an image set. It then uses video tags to classify the image set, executes the corresponding image processing solution based on the classification results, and stores the processed energy-saving video data in the application database.

[0011] The second data acquisition module: collects relevant data of the LED display screen on site and verifies and analyzes the collected data;

[0012] Real-time processing module: further processes the verified and analyzed on-site data and formulates energy-saving playback plans;

[0013] Correction processing module: plays the video stored in the application database according to the energy-saving playback plan, collects user feedback on the video, and executes the corresponding video adjustment strategy based on the feedback.

[0014] Furthermore, the first data acquisition module includes:

[0015] Video acquisition submodule: used to obtain video data and video labels from various data sources;

[0016] Format conversion submodule: converts the collected raw video data into the target format required for subsequent processing and playback;

[0017] Data parsing submodule: Analyzes the data structure of common video formats, writes regular expression rules, and extracts resolution and frame rate information;

[0018] Data storage submodule: configure storage strategies, write data storage interfaces to connect with other submodules, and store processed data in a distributed file system.

[0019] Furthermore, when performing noise reduction on video data stored in a distributed file system, the video data stored in the distributed file system is first decoded into a lossless sequence of images, parameter information is extracted, and grayscale histogram statistics and pixel variance calculation are performed on the decoded images to determine whether there is noise;

[0020] If there is no noise, no processing is required;

[0021] If there is noise, the noise type is determined based on the grayscale histogram and pixel variance, and the corresponding noise processing method is used to denoise the image. The denoised image is enhanced and re-encoded to generate a video, which is saved in the denoising dataset.

[0022] Furthermore, the steps for classifying the image set using video tags are as follows:

[0023] The image set corresponds to the video tag of the video and is divided into first-level tags, second-level tags and third-level tags;

[0024] The third-level labels include average resolution and average brightness;

[0025] Match the first-level labels of the image set with the first-level labels in the knowledge graph, calculate the text similarity of the label names, obtain the first-level label corresponding to the maximum similarity, and thus obtain the second-level label list of the knowledge graph;

[0026] Match the second-level tags of the image set with the second-level tag list, calculate the text similarity of the tag names, and obtain the second-level tag corresponding to the maximum similarity, thereby obtaining the third-level tag list of the knowledge graph;

[0027] Match the three-level labels of the image set with the three-level label list, calculate the difference ratio, obtain the minimum value of each difference ratio, and obtain the classification of the image set based on the minimum value.

[0028] Furthermore, the implementation steps of the image processing solution are as follows:

[0029] Identify the pictures in the picture set and select the key areas;

[0030] Weaken the non-selected area by adjusting the brightness and clarity;

[0031] When weakening, the farther away from the selected area, the lower the brightness and clarity;

[0032] Obtain the original video duration, adjust the display duration of the image after image concentration weakening processing according to the original video duration, and generate a new video;

[0033] Extract the original video and audio stream, align the extracted original video and audio stream with the new video's timestamp, obtain the adjusted video data, and store the video data in the application database.

[0034] Furthermore, relevant data of the LED display screen on site are collected, including collecting lighting data of the four corners and the center position of the LED display screen and pictures of the LED display screen playback area. The collected data are verified and analyzed to verify the lighting data and the pictures of the LED display screen playback area, and the light intensity is analyzed based on the lighting data of the four corners and the center position of the LED display screen.

[0035] Furthermore, the objects of further processing of the on-site relevant data after verification and analysis include light intensity and pictures of the LED display screen playback area. The light intensity is processed and analyzed to formulate a playback display intensity plan. The pictures of the LED display screen playback area are processed and analyzed to formulate a display plan. The energy-saving playback plan includes a playback display intensity plan and a display plan. The steps for formulating a playback display intensity plan are as follows:

[0036] Compare the light intensity LXp with two preset light intensity thresholds LX1 and LX2, LX1 < LX2;

[0037] If LXp≤LX1, a fixed lower limit display intensity is executed;

[0038] If LXp>LX2, a fixed upper limit display intensity is executed;

[0039] If LX1<LXp≤LX2, the display intensity is adjusted according to a linear relationship, and the adjustment range is between the lower limit display intensity and the upper limit display intensity.

[0040] Furthermore, the steps for developing a display plan are as follows:

[0041] Audience number recognition: Audience detection is performed based on the target detection model, and the number of viewers in the viewing area is analyzed in real time;

[0042] Audience distribution identification: The image in the LED display screen playback area is evenly divided into N grid sub-areas, the number of audiences in each sub-area is counted, and the crowd density is calculated. The crowd density is the number of people divided by the area of ​​the sub-area, based on the audience density of each sub-area;

[0043] Display screen size calculation: Calculate the display size based on the number of viewers and the distribution of the crowd;

[0044] Display center setting: Set the display position corresponding to the area with the highest crowd density as the center area of ​​the video to be played. If part of the video cannot be displayed after setting, adjust the video position to make the video fully displayed.

[0045] Furthermore, executing corresponding video adjustment strategies based on the feedback includes:

[0046] If the number of positive reviews for the same video exceeds the preset value of one, no adjustment will be made;

[0047] If the number of positive reviews for the same video does not exceed the preset value of one, the video in the denoising dataset is extracted and reprocessed;

[0048] If the total number of positive reviews in the feedback of the same category does not exceed the preset value of two, the parameters of the image processing solution are adjusted. After the adjustment, all the image sets of the category are processed, and the video is regenerated and replaces the original video stored in the application database.

[0049] Furthermore, a method for controlling energy consumption of an LED display screen includes the following steps:

[0050] Collect videos and video tags uploaded by users and store them in a distributed file system;

[0051] Perform noise reduction and frame segmentation on video data stored in a distributed file system to form an image set. The image set is then classified using video tags. Based on the classification results, the corresponding image processing solution is executed. The processed energy-saving video data is then stored in the application database.

[0052] Collect relevant data from the LED display screen on site and verify and analyze the collected data;

[0053] Further process the relevant on-site data after verification and analysis to develop an energy-saving broadcasting plan;

[0054] The video stored in the application database is played according to the energy-saving playback scheme, and user feedback on the video is collected, and the corresponding video adjustment strategy is executed based on the feedback.

[0055] The present invention provides a method and system for controlling energy consumption of an LED display screen, which has the following beneficial effects:

[0056] The present invention realizes precise control of the energy consumption of the LED display screen, processes and analyzes the data source, reduces the computational complexity, processes the video in advance, and then does not need to further adjust the video in the playback stage, thereby reducing the processing amount of data at runtime, and dynamically adjusts the display screen parameters according to the environment and user needs to avoid energy waste. It not only significantly reduces the energy consumption of the LED display screen, but also ensures the video playback quality and improves the user experience. Compared with traditional solutions, this solution greatly reduces the operating costs while realizing intelligent video playback, has good overall effect, and has good prospects for use.

[0057] The present invention processes the video before playing and identifies key areas. For non-key areas, the module adopts an intelligent weakening strategy to reduce the brightness and clarity and reduce the complexity of pixel processing. This not only significantly reduces the energy consumption of video file storage and transmission, but also, during playback, the LED display does not need to consume too much electricity for non-critical images, and the overall playback energy consumption is greatly reduced. Without affecting the user's viewing of core content, a balance is achieved between energy consumption and viewing experience, with good use effect and good usage prospects.

[0058] The present invention intelligently controls the energy consumption of LED displays based on environmental and crowd data, collects and processes light intensity, crowd density, and distribution data in real time, dynamically generates and adjusts playback strategies, and automatically adjusts display brightness according to light intensity, reducing brightness in low-light environments to reduce power consumption. The invention also dynamically adjusts the size and position of the playback screen based on crowd density and distribution to avoid energy consumption in ineffective display areas, allowing the energy consumption of the LED display to be dynamically optimized as the environment and crowd change, achieving significant energy savings while ensuring viewing quality.

[0059] The present invention obtains user feedback and executes corresponding video adjustment strategies based on the user feedback. At the processing level of a single video, if the praise rate does not meet the standard, the advanced algorithm can be called to optimize the details in a targeted manner. For videos of the same category, when the overall praise rate is insufficient, the parameters of the image processing solution can be adjusted in batches to uniformly optimize the data of all videos in the category, thereby greatly improving the adaptability of video playback and user satisfaction.

[0060] The present invention constitutes a complete intelligent, energy-saving LED display video playback and processing system, which effectively solves the problems of high energy consumption of video playback and inability to intelligently adjust according to the environment and user needs in the existing technology. Through multi-module collaboration, it realizes the whole process optimization from data collection, processing to playback, reduces the energy consumption of LED display, and lays the foundation for the energy-saving application of LED display in commercial, public service and other fields. It not only has significant economic benefits, but also promotes the sustainable development of related industries and has good application prospects. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] Figure 1This is a system block diagram of an LED display energy consumption control system of the present invention;

[0062] Figure 2 This is a flow chart of a method for controlling energy consumption of an LED display screen according to the present invention;

[0063] Figure 3 This is a schematic diagram of a weakening processing area in an LED display energy consumption control system of the present invention. DETAILED DESCRIPTION

[0064] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0065] Example 1:

[0066] See also Figure 1 ,This embodiment provides an LED display energy consumption control system, which is aimed at large-scale display screens, such as large LED display screens in shopping malls. The system includes a first data acquisition module, a data source processing module, a second data acquisition module, a real-time processing module and a correction processing module, which operate in sequence;

[0067] The present invention realizes precise control of the energy consumption of the LED display screen, processes and analyzes the data source, reduces the computational complexity, processes the video in advance, and then does not need to further adjust the video in the playback stage, thereby reducing the processing amount of data at runtime, and dynamically adjusts the display screen parameters according to the environment and user needs to avoid energy waste. It not only significantly reduces the energy consumption of the LED display screen, but also ensures the video playback quality and improves the user experience. Compared with traditional solutions, this solution greatly reduces the operating costs while realizing intelligent video playback, has good overall effect, and has good prospects for use.

[0068] The specific plan is described as follows:

[0069] 1. The first data acquisition module:

[0070] The first data collection module collects various promotional videos and video tags uploaded by users, such as brand advertisements and promotional videos, and stores the data in a distributed file system;

[0071] The first data acquisition module is the starting point of the intelligent video playback and processing solution. It undertakes the important tasks of acquiring and basic processing of original video data. Its processing effect directly affects the work quality of subsequent modules.

[0072] The first data acquisition module includes a video acquisition submodule, a format conversion submodule, a data analysis submodule and a data storage submodule;

[0073] 1.1. The video acquisition submodule is based on the FFmpeg framework. Through its powerful multimedia processing capabilities, it can obtain video data and video tags from various data sources such as the local file system, network video stream RTMP, HTTP, etc. Each data source has different upload paths. FFmpeg supports multiple video acquisition protocols and device interfaces, and can adapt to different types of video input sources. Whether it is the original video captured by the camera or the advertising video files stored on the server, it can achieve efficient acquisition.

[0074] For local file acquisition, the video acquisition submodule uses the file input interface provided by FFmpeg to specify the video file path and related parameters, such as video stream retrieval and audio stream retrieval; for network video stream acquisition, the corresponding input parameters are set according to the stream protocol type, and at the same time, the acquisition parameters are set to ensure that the acquired video data meets the subsequent processing requirements.

[0075] The acquisition parameters include sampling rate and number of channels. The sampling rate is set to 44100Hz or 48000Hz, and the number of channels is two-channel stereo.

[0076] When in use, the video acquisition submodule can achieve compatibility with different data sources, and can realize parallel acquisition of multiple videos to improve acquisition efficiency. The video acquisition submodule adopts an error retry mechanism. When encountering abnormal situations such as network interruption and file reading errors, it automatically retries multiple times and records error logs to facilitate subsequent problem investigation.

[0077] 1.2. The format conversion submodule calls the built-in functions of FFmpeg to implement format adaptation, establish the mapping relationship between input and output streams, and ensure that the video and audio streams are correctly converted and merged;

[0078] The format conversion submodule calls the built-in format conversion function of FFmpeg to convert the collected original video data into the target format required for subsequent processing and playback of the system.

[0079] FFmpeg supports many video formats and can convert between MP4, AVI, MKV and other video formats. It can adapt the formats through the process of encoding, decoding and re-encoding the videos, ensuring the uniformity of the stored video formats, thus facilitating the subsequent processing of the videos and reducing the workload of subsequent processing.

[0080] The format conversion submodule sets corresponding encoding parameters for different combinations of source format and target format. The encoding parameters include video encoding format, resolution, frame rate, audio encoding format, sampling rate, bit rate, etc.

[0081] The format conversion submodule is based on H.265 / HEVC encoding. Compared with the H.264 encoding technology commonly used in the existing technology, it can reduce the bit rate by 50% under the same conditions, and can further improve the compression effect, thereby facilitating the storage of more data.

[0082] 1.3. The data parsing submodule analyzes the data structure of common video formats, writes regular expression rules, and extracts information such as resolution and frame rate.

[0083] Regular expression rules are grammatical rules used to match string patterns. By writing specific regular expressions, key parameters such as resolution and frame rate can be accurately extracted from the metadata of video files, ensuring a unified format for subsequent processing.

[0084] The data parsing submodule utilizes a regular expression matching model to extract and parse data from video files. Video data contains a wealth of information, including resolution, frame rate, encoding format, duration, and audio and video track information. This information is crucial for subsequent video processing and playback strategy development. Regular expressions define specific pattern rules to perform matching searches within the data fields of video files, extracting the required information and facilitating further analysis and processing. Regular expression patterns are designed to improve the accuracy of data extraction. For complex data structures, advanced regular expression features such as grouping and capturing are used to precisely match required fields. Furthermore, the module handles the diversity and irregularities of data formats and improves the robustness of parsing by adding fault-tolerance mechanisms.

[0085] The data parsing submodule adopts multi-threading or parallel computing technology when in use, thereby improving the parsing speed; and caches commonly used data parsing rules to avoid repeated calculations.

[0086] The data parsing submodule determines the key area where the data is located according to the characteristics of different video formats, and directly performs regular expression matching in this area. At the same time, the regular expression is compiled and optimized to reduce the computational overhead at runtime.

[0087] 1.4. The data storage submodule configures the storage strategy, writes the data storage interface to connect with other submodules, and stores the processed data in the distributed file system.

[0088] The data storage submodule stores the video data after format conversion and metadata parsing in a distributed file system. By distributing the data across multiple storage nodes, it achieves data redundancy and load balancing, ensuring data security and availability.

[0089] The data storage submodule ensures the security of data during storage through the multi-copy mechanism and data redundancy strategy of the distributed file system. When one of the storage nodes fails, the data can be restored from other replica nodes to ensure data integrity and availability.

[0090] A distributed file system is used to achieve redundant storage and efficient access to video data. It is a file system that stores data in a dispersed manner across multiple independent physical nodes and works collaboratively over a network. Its core advantages are high fault tolerance, high scalability, and efficient parallel access capabilities, enabling the storage and rapid reading of large amounts of video data.

[0091] The first data acquisition module, through the collaborative work of the above submodules, achieves efficient video data acquisition, format conversion, data analysis, and reliable storage, providing a solid data foundation for subsequent data source processing, video playback, and other modules. In actual applications, the module can be further optimized and adjusted according to the specific needs and operating conditions of the shopping mall to improve the performance and stability of the entire intelligent video playback and processing solution.

[0092] 2. Data source processing module:

[0093] Process the video data stored in the distributed file system and store the processed data to build an application database;

[0094] As the core data processing unit for video playback and processing, the data source processing module is responsible for in-depth processing of video data.

[0095] The data source processing module processes the data stored in the distributed file system in the following steps.

[0096] 2.1, noise reduction processing;

[0097] Preprocessing: Decode the video data stored in the distributed file system into lossless sequential images, extract information such as the video frame rate and resolution, perform grayscale histogram statistics and pixel variance calculation on the decoded images, and determine whether noise exists and the type of noise if it exists.

[0098] At this time, decoding is used to obtain the image instead of framing, mainly to retain the characteristics of the video. If the subsequent analysis and processing is not effective, the video after noise reduction can be directly retrieved and processed again.

[0099] For example, if the grayscale histogram shows discrete burrs and the pixel variance is higher than the normal range, it is judged that impulse noise exists; if the overall distribution of the grayscale histogram is discrete and there are high-frequency fluctuations, it is judged that Gaussian noise exists.

[0100] A grayscale histogram is a graph that describes the distribution of the number of pixels at each grayscale level in an image. The grayscale histogram of a normal image is relatively continuous. If salt and pepper noise or Gaussian noise is present, the histogram will show discrete spikes or abnormal fluctuations. The presence and type of noise can be quickly determined by combining the pixel variance.

[0101] Data denoising: Use the fast non-local mean denoising algorithm in the non-local mean denoising method to remove static noise in the image;

[0102] The non-local mean denoising method is a denoising algorithm based on image redundancy. The core idea is to find multiple pixel blocks similar to the target pixel block in the image, and calculate the denoising value of the target pixel by weighted average. Compared with traditional denoising algorithms, it can better preserve image details and is suitable for denoising processing of video images.

[0103] The inter-frame weighted averaging method is used with weights of 0.2, 0.6, and 0.2 to eliminate the noise caused by inter-frame jitter, reduce motion blur, and ensure the smoothness of dynamic elements in the video.

[0104] Post-processing: Use OpenCV's contrast-limited adaptive histogram equalization algorithm to enhance the denoised image, increasing the color saturation and contrast of the advertisement image to make the video visual effect more prominent. Re-encode the processed image sequence and restore it to the H.265 format. Adjust the video frame rate and resolution to the original settings, and output the final denoised advertisement video.

[0105] Processing videos uploaded by users can effectively reduce noise, and the use of H.265 / HEVC encoding algorithm can improve efficiency by 30%-50% compared to existing solutions. At the same time, it enhances color and contrast while reducing noise, highlighting the core information of the video, further facilitating subsequent core identification, and laying the foundation for weakening the edge areas of the image.

[0106] In actual use, the video can also be processed based on the non-local mean denoising algorithm of deep learning to achieve video denoising. However, this method requires corresponding data as support for calculation. The overall calculation requires a lot of data and consumes a lot of energy, so its adaptability for LED display screens is relatively low.

[0107] The video data after denoising is saved in a separate dataset to form a denoising dataset.

[0108] 2.2, Frame processing;

[0109] Based on video coding and decoding technology, the continuous video after noise reduction is decomposed into independent picture frames to form a picture set. The frame rate information of the video is read, and the video frames are extracted at fixed time intervals or frame intervals to retain all the content information of the video.

[0110] The number of images after frame processing is lower than the number of decoded images during the noise reduction process, usually between one-third and one-half of the number of decoded images. By reducing the number of frames in a single video, the amount of data processing can be effectively reduced.

[0111] Video processing technology is used for frame processing. The frame extraction interval is set according to the video frame rate and processing requirements. Usually 2-8 frames are extracted per second to avoid large differences after video processing. The extracted images are obtained and aggregated to form an image set. One image set is constructed for each video to avoid image confusion.

[0112] 2.3 Data classification;

[0113] Data classification is based on knowledge graph technology. The video tags of the corresponding videos in the picture set are compared with the tags in the knowledge graph to achieve the classification of the picture set.

[0114] Use historical knowledge data related to videos and images to build a knowledge graph.

[0115] A knowledge graph is a database that graphically represents entities and their relationships. The knowledge graph pre-stores the hierarchical relationship of video tags and automatically classifies image sets through tag matching, providing a basis for subsequent targeted processing.

[0116] Video tags include topic tags, scene tags, and video tags, which constitute a three-level tag system. The three-level tag system is hierarchical and progressive.

[0117] Topic tags, scene tags, and video tags are first-level tags, second-level tags, and third-level tags, respectively.

[0118] Thematic tags are used to describe the functions of videos, such as product promotion, atmosphere creation, customer interaction, brand promotion, technology display, dynamic special effects, etc.

[0119] For example, if the video is a promotional video for a mobile phone, the topic tag is product promotion-electronic products, and the topic tag data is the relevant data filled in by the user when uploading the video.

[0120] When in use, the labels can be further refined, but the amount of data compared in the subsequent comparison is large, which is not suitable for large LED display screens such as shopping malls.

[0121] Scene labels are labels formed by combining buildings and weather, such as indoors on a sunny day, outdoors on a sunny day, and streets at night.

[0122] Video tags include labels of video characteristics such as average resolution and average brightness.

[0123] The average resolution is calculated by obtaining the resolutions of the images in the image set and then calculating their average.

[0124] The average brightness is to convert the picture set into a gray picture, and then calculate the average value of the brightness values of all pixels.

[0125] The picture set classification obtains a classification result, and a corresponding picture processing scheme is selected according to the classification result.

[0126] The steps of the picture set classification are as follows:

[0127] Primary label matching: match the theme label of the picture set with the theme label in the knowledge graph.

[0128] All primary theme label nodes in the knowledge graph are traversed, the cosine similarity algorithm is used to calculate the text similarity of the extracted label and the name of each theme label node, and then the calculated text similarity is compared. The theme label corresponding to the maximum similarity is the matching successful label. Through the matching of the primary theme label, the belonging theme direction of the picture set is preliminarily determined, and a secondary label list is obtained.

[0129] Secondary label matching: after determining the primary theme label, the scene label of the picture set is matched with the scene label belonging to the same primary label in the knowledge graph.

[0130] All secondary scene label nodes in the knowledge graph are traversed, the cosine similarity algorithm is used to calculate the text similarity of the extracted label and the name of each scene label node, and then the calculated text similarity is compared. The scene label corresponding to the maximum similarity is the matching successful label. Through the matching of the secondary theme label, the belonging scene direction of the picture set is determined, that is, the scene label of the picture set is compared with the secondary label list to obtain a tertiary label list.

[0131] Tertiary label matching: obtain the data in the tertiary video label, compare each item in the data in the tertiary video label with each item of the tertiary video label belonging to the same primary label and secondary label in the knowledge graph, calculate the difference ratio, obtain the minimum value of each difference ratio, and the type corresponding to the minimum value realizes the classification of the picture set.

[0132] Each item of the tertiary video label in the knowledge graph contains 3 data with large differences. It is obtained by analyzing all historical data under the historical secondary theme label category. The setting method is to subtract the minimum value Amin from the maximum value Amax to obtain the difference, and then divide the difference by 3 to obtain the grade difference Ac. According to the grade difference, three data intervals are set , the average value of the data located in the three data intervals is calculated to obtain 3 average values, that is, 3 data with large differences.

[0133] For example, the average resolution of the three-level video tags is Ap, and the three data with relatively large differences in the average resolution of the three-level video tags in the corresponding knowledge graph are Ap1, Ap2 and Ap3. The difference ratios between Ap and Ap1, Ap2 and Ap3 are calculated respectively. The calculation formulas are (Ap-Ap1) / Ap1, (Ap-Ap2) / Ap2 and (Ap-Ap3) / Ap3, and two groups of difference ratios are obtained, from which the difference ratio of this item is obtained.

[0134] For example, the third-level video tag has average resolution and average brightness. Each of the average resolution and average brightness outputs a difference ratio. Each item may output three groups of results. For example, the average resolution may output A11, A12, and A13, and the average brightness may output B11, B12, and B13. After combining them, there are a total of nine results, as follows: A11B11, A11B12, A11B13, A12B11, A12B12, A12B13, A13B11, A13B12, A13B13. Each result corresponds to a set image processing solution, and only some parameters in the image processing solution differ.

[0135] The above is a fine division. In actual application, the second-level label can be removed, and each item in the third-level label can be modified from three data with large differences to two data with large differences. This method can greatly reduce the amount of data processing.

[0136] 2.4 Data processing;

[0137] Data processing is to process the pictures in the picture set according to the set picture processing plan.

[0138] Key area identification: The YOLOv7 target detection model is used to identify images in the image set and select key areas.

[0139] The YOLOv7 target detection model is built based on historical data.

[0140] The YOLOv7 target detection model is a real-time target detection algorithm based on deep learning. Its core advantages are fast detection speed and high accuracy. It can quickly identify targets in images and output the bounding box coordinates of the target.

[0141] In order to further reduce the amount of data processing, a fixed frame selection method can be used, that is, a part of the image is set as the key area in the image processing solution, such as the center one-third of the image as the key area. Different image processing solutions use different frame selection areas.

[0142] A fixed frame selection method is used. The size of the frame selection is mainly based on the first-level tag. The specific size is set according to the type of subject in the first-level tag. For example, if the subjects are mobile phones and washing machines, the sizes of the fixed frame selections are different. The area of ​​the washing machine frame selection is larger than that of the mobile phone frame selection.

[0143] like Figure 3 As shown, weakening processing: weakening processing is performed on the non-framed area. The brightness and clarity of the screen display picture are positively correlated with power consumption. For every 10% increase in brightness, power consumption increases by about 8%; changes in clarity will affect the luminous intensity and display complexity of the pixels, and thus affect energy consumption. Therefore, processing non-key content can effectively reduce the energy consumption of the display during subsequent playback.

[0144] When weakening the image, the closer it is to the selected area, the lower the degree of weakening, thereby effectively ensuring the display effect of the image and avoiding abrupt display of the image.

[0145] The image processing solution includes the brightness and clarity of the edge area of ​​the image, that is, the lowest clarity and brightness after weakening processing.

[0146] Based on the clarity and brightness of the selected area, a solution is developed in which the brightness and clarity decrease as the distance from the selected area increases.

[0147] In specific applications, the Sigmoid function can be used to achieve the mapping from distance to weakening coefficient.

[0148] The specific weakening coefficient calculation formula is: , where is the attenuation coefficient, k is the slope parameter, the larger the k value is, the faster the attenuation coefficient increases with distance, and the k in different image processing schemes is different, mainly because the requirements of different label videos are different, that is, the minimum clarity and brightness requirements are different, and the clarity and brightness of the pictures in the picture set are different. Therefore, the k value in different categories of image processing schemes is different, generally set to k < 0.2. When setting the image processing scheme, the lower the requirement for the video and the higher the clarity and brightness of the pictures in the corresponding picture set, the larger the k value is, d is the shortest distance from the pixel to the bounding box of the key area, d0 is the inflection point distance, that is, the critical point of the attenuation degree. If it is before the critical point, that is, d ≥ d0, the degree of attenuation processing is low, and if it is after the critical point, that is, d < d0, the degree of attenuation processing is low.

[0149] During the weakening process, the clarity and brightness are processed separately. The formula for the weakened brightness is Lc=L×(1-Wear), where Lc is the pixel brightness value in the image after weakening, and L is the original pixel brightness value of the image before weakening.

[0150] For example, the original brightness of the pixel in the non-key area of ​​one image is L=0.8, the weakening coefficient of this position is 0.9, and the brightness of this position after weakening processing is 0.08, which means that the brightness is greatly reduced.

[0151] The clarity of the weakened processing is controlled by adjusting the sharpness and detail retention of the image. The specific formula is Ic=I-Wear×(I-IG), where Ic is the pixel matrix of the image after weakening processing, I is the pixel matrix of the original image before weakening processing, Bt is the clarity adjustment factor, -1<Bt<0, and IG is the pixel matrix after Gaussian blur processing of the images in the image set.

[0152] For example, if the value of a pixel in the original image is 200, and the corresponding pixel value in the Gaussian blurred image is 150, the weakening coefficient at this position is 0.9. After the weakening process, the pixel value at this position is 155, which greatly reduces the clarity.

[0153] 2.5, video generation;

[0154] Video conversion: Obtain the duration of the video in the denoising dataset corresponding to the image set, adjust the display duration of the weakened images in the image set according to the duration of the video in the denoising dataset, and generate a new video so that the generated new video is synchronized with the original video duration.

[0155] For example, a video in the denoising dataset is 10 seconds long and contains 500 images. The display time of each photo is 0.02 seconds. After the de-noising process, the number of images in the image set is 250. After calculation, the display time of each photo is 0.04 seconds.

[0156] Audio and video processing: Use FFmpeg to extract the original video and audio streams, retain them as separate audio files, and then align the timestamps to ensure audio and video synchronization and achieve audio and video integration. This avoids repeated data processing and increased energy consumption caused by audio and video asynchrony, and obtains adjusted video data.

[0157] The adjusted video data is stored in the application database. The video in the application database occupies a small amount of memory, has a low data volume, is convenient for subsequent playback, and is an energy-saving video.

[0158] The present invention processes the video before playing and identifies key areas. For non-key areas, the module adopts an intelligent weakening strategy to reduce the brightness and clarity and reduce the complexity of pixel processing. This not only significantly reduces the energy consumption of video file storage and transmission, but also, during playback, the LED display does not need to consume too much electricity for non-critical images, and the overall playback energy consumption is greatly reduced. Without affecting the user's viewing of core content, a balance is achieved between energy consumption and viewing experience, with good use effect and good usage prospects.

[0159] 3. Second data acquisition module;

[0160] The second data acquisition module is used to collect relevant data on the LED display screen, such as light intensity, crowd density and distribution, to lay the foundation for further adjustments.

[0161] 3.1, Light data collection;

[0162] Light data collection uses BH1750 digital light intensity sensors, one installed at each of the four corners and the center of the LED display to avoid data deviation caused by uneven lighting;

[0163] During collection, light data is collected every 10 minutes. Each light intensity sensor transmits data once every 10 seconds, and transmits data 5 times. The data is timestamped for each collection. After the data is collected, the data is checked using the cyclic redundancy check algorithm to achieve data verification, prevent loss or errors during data transmission, and ensure the accuracy of the collected data.

[0164] The cyclic redundancy check algorithm is a data transmission error detection algorithm. It generates a fixed-length check value by performing polynomial operations on the original data. The receiver recalculates the check value and compares it with the sender's check value. If they are inconsistent, the data transmission is determined to be incorrect, thereby ensuring the integrity of the collected lighting data and audience images.

[0165] The analysis is to calculate the light intensity, which is the average of 25 sets of data collected by 5 sensors. If data loss and errors occur, all data collected by the light intensity sensor at that time will be excluded, that is, 5 sets of data will be eliminated at one time.

[0166] 3.2、Collection of audience data;

[0167] High-resolution network cameras are used for data acquisition, specifically model DS-2CD3T86FWDV2-IZS. The high-resolution network cameras are installed directly above and at appropriate positions on both sides of the LED display to ensure that the entire playback area is covered. The high-resolution network cameras are connected to the data processing server via Gigabit Ethernet to ensure that video data can be transmitted quickly and stably. The high-resolution network cameras are equipped with infrared fill light function to ensure clear image acquisition even at night or in low-light environments.

[0168] The verification of the image is done in the same way as the light data collection verification.

[0169] During acquisition, the camera continuously collects image data at a frame rate of 1 frame per second, and does not use the method of shooting video, which can effectively reduce the amount of data storage.

[0170] 4. Real-time processing module;

[0171] The real-time processing module is used to process and analyze the LED display screen field-related data collected by the second data acquisition module, so as to formulate a suitable energy-saving playback plan.

[0172] 4.1, Light data processing;

[0173] According to the corresponding relationship between light intensity and display intensity, a segmented adjustment strategy is formulated;

[0174] The scheme of the segmented adjustment strategy is to set the display intensity into three levels, among which two light intensity thresholds LX1 and LX2 for comparison are preset, LX1<LX2. The light intensity thresholds LX1 and LX2 are set according to the scene. For example, in a shopping mall, LX1 is generally set to 50lx and LX2 is generally set to 250lx.

[0175] Compare the detected light intensity LXp with LX1. If LXp≤LX1, a fixed lower limit display intensity is executed. Generally, the display intensity of the LED display is set to 15-25%.

[0176] For example: the lower limit display intensity is set to 20%. If LXp≤LX1, the display intensity of the LED display is set to 20%.

[0177] If LXp>LX2, a fixed upper limit display intensity is implemented, generally setting the display intensity of the LED display to 45-55%;

[0178] For example: the upper limit display intensity is set to 50%. If LXp>LX2, the display intensity of the LED display will be set to 50%.

[0179] If LX1<LXp≤LX2, the display intensity is adjusted according to a linear relationship, and the adjustment range is between the lower limit display intensity and the upper limit display intensity.

[0180] For example: the lower limit display intensity is set to 20%, the upper limit display intensity is set to 50%, LX1 is 50lx, LX2 is 250lx, the difference between LX2 and LX1 is 200lx, if LXp is 110lx, then the display intensity of the LED display is set to 20% + (150-50) × (50%-20%) / 200 = 35%.

[0181] This method overcomes the existing defects of using fixed display brightness to play, which causes high power consumption and poor display effects due to the display being too bright or too dark when in use.

[0182] 4.2. Audience data analysis;

[0183] Before analysis, the images collected by the second data acquisition module are uniformly scaled to the same size and the same pixels.

[0184] Audience number identification: The YOLOv7-Tiny target detection model is used for audience detection, and the number of viewers in the viewing area is analyzed in real time. Existing solutions can also be used to identify the number of viewers.

[0185] For example, real-time monitoring based on AI cameras uses AI cameras combined with deep learning algorithms, and can achieve millimeter-level precision statistics through multispectral imaging and three-dimensional point cloud reconstruction technology.

[0186] Audience distribution identification: The images collected by the second data acquisition module are evenly divided into N grid sub-areas. The number of spectators in each sub-area is counted, and the crowd density is calculated. The crowd density is the number of people divided by the area of ​​the sub-area. Based on the audience density of each sub-area, an audience density heat map is generated, so that the distribution of people can be clearly understood.

[0187] For example, a 1920×1080 image is divided into a 3×3 grid, with each sub-region being 640×360 pixels. If 8 people are detected in the fifth sub-region, the calculated crowd density is 3.47×10 -5 People / Pixels 2 .

[0188] 4.3、Display plan formulation;

[0189] Display screen size calculation: Calculate the appropriate display size based on the number of viewers and the distribution of the crowd.

[0190] The specific solution is to first calculate the weight of each sub-area, that is, divide the number of people in the area by the total number of people, and use the weighted average method to calculate the screen size. Calculate the display width, and set the display height according to the original width and height ratio of the video;

[0191] For example: N is 9, and the number of people in the 9 areas are 2, 3, 1, 5, 8, 4, 2, 1, and 3, respectively, totaling 29 people. The fifth area has the largest number of people, and its weight is about 0.276. Calculate the width size data contributed by this area, that is, the width of the LED display × 0.276 × 1 / 3. Calculate the width size data contributed by the 9 areas in turn, superimpose the calculated data to obtain the display width, and then combine it with the width and height ratio of the video in the application database to obtain the size of the playback screen.

[0192] Display center setting: Set the display position corresponding to the area with the highest crowd density as the center area of ​​the video to be played. If part of the video cannot be displayed after setting, adjust the video position to make the video fully displayed.

[0193] For example, if the 5th area has the highest population density, then the position corresponding to the 5th area is the center of the video. At this time, the 5th area is exactly the center of the LED display, so there is no need to set it separately (9-grid).

[0194] If the crowd density in area 1 is the highest, the position corresponding to area 1 will be the center of the video. If the calculated size of the playback screen is smaller than the size corresponding to area 1 (both length and width), the video will be played at the position corresponding to area 1.

[0195] If the crowd density in area 1 is the highest, then the position corresponding to area 1 is the center position of the video. If the calculated size of the playback screen is not smaller than the size corresponding to area 1 (either length or width), the video position is adjusted so that the boundary (upper left corner) of the playback screen coincides with the boundary (upper left corner) of area 1, thereby achieving complete playback of the video.

[0196] During video playback, in order to improve the playback effect and avoid repeated adjustments, when there are fewer people, you can use a 20-minute interval calculation and analysis to adjust the playback position and playback area once. If there are more people, use full-screen display. This method can effectively avoid the situation where the screen changes frequently and affect the video viewing effect.

[0197] The energy-saving playback solution is to set the display intensity of the LED display, set the playback position and playback size, and play the video after the settings are synchronized.

[0198] The present invention intelligently controls the energy consumption of LED displays based on environmental and crowd data, collects and processes light intensity, crowd density, and distribution data in real time, dynamically generates and adjusts playback strategies, and automatically adjusts display brightness according to light intensity, reducing brightness in low-light environments to reduce power consumption. The invention also dynamically adjusts the size and position of the playback screen based on crowd density and distribution to avoid energy consumption in ineffective display areas, allowing the energy consumption of the LED display to be dynamically optimized as the environment and crowd change, achieving significant energy savings while ensuring viewing quality.

[0199] 5. Correction processing module;

[0200] Correction Processing Module: Plays the video stored in the application database according to the energy-saving playback plan, collects user feedback on the video, and executes the corresponding video adjustment strategy based on the feedback. The content of the video adjustment strategy is as follows:

[0201] If the number of positive reviews for the same video exceeds 70%, no adjustment will be made and 70% will be the default value.

[0202] If the number of good reviews in the same video feedback is not more than 70%, the video in the noise reduction data set is extracted and reprocessed.

[0203] For example, the video picture has noise and is not clear enough. The video is extracted from the distributed file system for reprocessing. During reprocessing, the noise reduction processing link of the video is strengthened, a more advanced noise reduction algorithm is used, and after processing, the newly generated video replaces the video in the application database.

[0204] If the feedback video picture is not coordinated and the picture transition is abrupt, the video in the noise reduction data set is obtained, and more pictures are retained during the frame processing. After processing, the newly generated video replaces the video in the application database.

[0205] If the total number of good reviews in the same category (the same picture set classification) is not more than 60%, the parameters of the picture processing scheme are adjusted, such as increasing the minimum brightness and clarity in the picture processing scheme, adjusting the size of the key area frame, etc. After adjustment, all picture sets in the category are processed, a new video is generated, and the original video stored in the application database is replaced. 60% is the second preset value.

[0206] Feedback data is obtained through various user feedback channels, including scoring and commenting functions in mobile applications, web-based feedback forms, and offline questionnaires, making it easy for viewers to submit evaluations of videos from different channels.

[0207] The feedback data includes a scoring mechanism, such as 1-5 points for poor reviews and 6-10 points for good reviews, making it easy to determine whether it is a good review.

[0208] The feedback data is divided into multiple dimensions, such as picture quality, content presentation, and playback smoothness. Picture quality covers color deviation, geometric distortion, and clarity. Content presentation focuses on the prominence of key areas and the weakening effect of non-key areas. Playback smoothness involves frame rate matching and issues such as lag.

[0209] The present application obtains user feedback and executes corresponding video adjustment strategies based on user feedback. On the level of individual video processing, if the good review rate does not meet the standard, advanced algorithms can be called to optimize details. For videos of the same category, when the overall good review rate is insufficient, the parameters of the picture processing scheme can be adjusted in bulk, and all videos in the category can be optimized uniformly, greatly improving the adaptability of video playback and user satisfaction.

[0210] Verification and optimization: After executing the video adjustment strategy, the user feedback of the new video is continuously tracked, and by comparing the good review rate and user comment content before and after adjustment, the effectiveness of the adjustment strategy is evaluated. If the effect after adjustment does not meet the expectation, further analysis is conducted to optimize the adjustment strategy.

[0211] Through the above detailed correction processing flow and feedback-driven adjustment strategy, the correction processing module can dynamically optimize the video playback effect according to user needs, while ensuring video quality, and realize intelligent and personalized video playback and processing.

[0212] The present invention constitutes a complete intelligent, energy-saving LED display video playback and processing system, which effectively solves the problems of high energy consumption of video playback and inability to intelligently adjust according to the environment and user needs in the existing technology. Through multi-module collaboration, it realizes the whole process optimization from data collection, processing to playback, reduces the energy consumption of LED display, and lays the foundation for the energy-saving application of LED display in commercial, public service and other fields. It not only has significant economic benefits, but also promotes the sustainable development of related industries and has good application prospects.

[0213] Example 2:

[0214] Based on Example 1, Figure 2 As shown, this embodiment also provides a method for controlling energy consumption of an LED display screen, comprising the following specific steps:

[0215] Collect videos and video tags uploaded by users and store them in a distributed file system;

[0216] Perform noise reduction and frame segmentation on video data stored in a distributed file system to form a picture set. The picture set is then classified using video tags. Based on the classification results, the corresponding image processing solution is executed. The processed energy-saving video data is then stored in the application database.

[0217] Collect relevant data from the LED display screen on site and verify and analyze the collected data;

[0218] Further process the relevant on-site data after verification and analysis to develop an energy-saving broadcasting plan;

[0219] The video stored in the application database is played according to the energy-saving playback scheme, and user feedback on the video is collected, and the corresponding video adjustment strategy is executed based on the feedback.

[0220] The above embodiments can be implemented in whole or in part by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. Those skilled in the art will appreciate that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution.

[0221] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, and may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment as needed.

[0222] The above is only a specific implementation method of the present application, but the scope of protection of the present application is not limited thereto. Any technician familiar with this technical field can easily think of changes or replacements within the technical scope disclosed in this application, which should be covered by the scope of protection of the present application.

Claims

1. An LED display energy consumption control system, characterized by: include: The first data collection module collects videos and video tags uploaded by users and stores them in a distributed file system; Data source processing module: This module performs noise reduction and frame segmentation on video data stored in the distributed file system to form an image set. It then uses video tags to classify the image set, executes the corresponding image processing solution based on the classification results, and stores the processed energy-saving video data in the application database. The steps to classify the image set using video tags are as follows: The image set corresponds to the video tag of the video and is divided into first-level tags, second-level tags and third-level tags; The third-level labels include average resolution and average brightness; Match the first-level labels of the image set with the first-level labels in the knowledge graph, calculate the text similarity of the label names, obtain the first-level label corresponding to the maximum similarity, and thus obtain the second-level label list of the knowledge graph; Match the second-level tags of the image set with the second-level tag list, calculate the text similarity of the tag names, and obtain the second-level tag corresponding to the maximum similarity, thereby obtaining the third-level tag list of the knowledge graph; Match the three-level labels of the image set with the three-level label list, calculate the difference ratio, obtain the minimum value of each difference ratio, and obtain the classification of the image set based on the minimum value; The second data acquisition module: collects relevant data of the LED display screen on site and verifies and analyzes the collected data; Real-time processing module: further processes the verified and analyzed on-site data and formulates energy-saving playback plans; Correction processing module: plays the video stored in the application database according to the energy-saving playback plan, collects user feedback on the video, and executes the corresponding video adjustment strategy based on the feedback.

2. The LED display energy consumption control system according to claim 1, characterized in that: The first data acquisition module includes: Video acquisition submodule: used to obtain video data and video labels from various data sources; Format conversion submodule: converts the collected raw video data into the target format required for subsequent processing and playback; Data parsing submodule: Analyzes the data structure of common video formats, writes regular expression rules, and extracts resolution and frame rate information; Data storage submodule: configure storage strategies, write data storage interfaces to connect with other submodules, and store processed data in a distributed file system.

3. The LED display energy consumption control system according to claim 2, characterized in that: When performing noise reduction on video data stored in a distributed file system, the video data stored in the distributed file system is first decoded into a lossless sequence of images, parameter information is extracted, and grayscale histogram statistics and pixel variance calculation are performed on the decoded images to determine whether there is noise; If there is no noise, no processing is required; If there is noise, the noise type is determined based on the grayscale histogram and pixel variance, and the corresponding noise processing method is used to denoise the image. The denoised image is enhanced and re-encoded to generate a video, which is saved in the denoising dataset.

4. The LED display energy consumption control system according to claim 1, characterized in that: The implementation steps of the image processing solution are as follows: Identify the pictures in the picture set and select the key areas; Weaken the non-selected area by adjusting the brightness and clarity; When weakening, the farther away from the selected area, the lower the brightness and clarity; Obtain the original video duration, adjust the display duration of the image after image concentration weakening processing according to the original video duration, and generate a new video; Extract the original video and audio stream, align the extracted original video and audio stream with the new video's timestamp, obtain the adjusted video data, and store the video data in the application database.

5. The LED display energy consumption control system according to claim 1, characterized in that: Collecting relevant data on the LED display screen at the scene includes collecting the lighting data of the four corners and the center of the LED display screen and the picture of the LED display screen playback area. The collected data is verified and analyzed to verify the lighting data and the picture of the LED display screen playback area, and the light intensity is analyzed based on the lighting data of the four corners and the center of the LED display screen.

6. The LED display energy consumption control system according to claim 5, characterized in that: The objects of further processing of the relevant on-site data after verification and analysis include light intensity and pictures of the LED display screen playback area. The light intensity is processed and analyzed to formulate a playback display intensity plan. The pictures of the LED display screen playback area are processed and analyzed to formulate a display plan. The energy-saving playback plan includes a playback display intensity plan and a display plan. The steps for formulating a playback display intensity plan are as follows: Compare the light intensity LXp with two preset light intensity thresholds LX1 and LX2, LX1 < LX2; If LXp≤LX1, a fixed lower limit display intensity is executed; If LXp>LX2, a fixed upper limit display intensity is executed; If LX1<LXp≤LX2, the display intensity is adjusted according to a linear relationship, and the adjustment range is between the lower limit display intensity and the upper limit display intensity.

7. The LED display energy consumption control system according to claim 6, characterized in that: The steps to develop a display plan are as follows: Audience number recognition: Audience detection is performed based on the target detection model, and the number of viewers in the viewing area is analyzed in real time; Audience distribution identification: The image in the LED display screen playback area is evenly divided into N grid sub-areas, the number of audiences in each sub-area is counted, and the crowd density is calculated. The crowd density is the number of people divided by the area of ​​the sub-area, based on the audience density of each sub-area; Display screen size calculation: Calculate the display size based on the number of viewers and the distribution of the crowd; Display center setting: Set the display position corresponding to the area with the highest crowd density as the center area of ​​the video to be played. If part of the video cannot be displayed after setting, adjust the video position to make the video fully displayed.

8. The LED display energy consumption control system according to claim 1, characterized in that: The corresponding video adjustment strategies based on the feedback include: If the number of positive reviews for the same video exceeds the preset value of one, no adjustment will be made; If the number of positive reviews for the same video does not exceed the preset value of one, the video in the denoising dataset is extracted and reprocessed; If the total number of positive reviews in the feedback of the same category does not exceed the preset value of two, the parameters of the image processing solution are adjusted. After the adjustment, all the image sets of the category are processed, and the video is regenerated and replaces the original video stored in the application database.

9. A method for controlling energy consumption of an LED display screen, using the system according to any one of claims 1 to 8, characterized in that: The following steps are involved: Collect videos and video tags uploaded by users and store them in a distributed file system; Perform noise reduction and frame segmentation on video data stored in a distributed file system to form a picture set. The picture set is then classified using video tags. Based on the classification results, the corresponding image processing solution is executed. The processed energy-saving video data is then stored in the application database. Collect relevant data from the LED display screen on site and verify and analyze the collected data; Further process the relevant on-site data after verification and analysis to develop an energy-saving broadcasting plan; The video stored in the application database is played according to the energy-saving playback scheme, and user feedback on the video is collected, and the corresponding video adjustment strategy is executed based on the feedback.

Images