A High-Efficiency Video Data Compression Method and System Based on Artificial Intelligence

Abstract

This invention relates to the field of intelligent safety management technology, and specifically discloses a method and system for efficient video data compression based on artificial intelligence. The method includes: acquiring video data of a target area based on a preset sensor, processing the data to obtain a first processed video frame, and then performing frame-by-frame analysis based on a preset deep learning model to determine the video analysis results; determining an intelligent compression strategy based on the video analysis results, and dynamically adjusting it in conjunction with real-time device characteristics and real-time user needs to obtain an optimized compression strategy; selecting a matching compression algorithm based on the optimized compression strategy, optimizing compression parameters, and thus compressing the first processed video frame; evaluating the quality of the compressed video stream, and further optimizing and adjusting the optimized compression strategy based on user feedback to achieve efficient compression; thereby making video data compression in the industrial field more efficient and accurate.

Description

Technical Field

[0001] This invention relates to the field of intelligent compression technology, and in particular to a method and system for efficient compression of video data based on artificial intelligence. Background Technology

[0002] In recent years, the rise of industrial big data has brought unprecedented changes to various industries. In the industrial sector, big data is widely used to optimize production processes, improve product quality, and reduce energy consumption. Similarly, industrial big data also holds enormous potential in the field of video compression. By collecting and analyzing large amounts of video compression-related data, patterns and trends in the video compression process can be uncovered, providing strong support for developing intelligent compression strategies.

[0003] However, the application of industrial big data in video compression is still relatively limited. On the one hand, video compression is a complex process involving multiple factors and variables, making it difficult to directly apply traditional big data analysis methods. On the other hand, video compression generates massive amounts of data, placing high demands on storage and processing capabilities.

[0004] Therefore, this invention proposes a method and system for efficient compression of video data based on artificial intelligence. Summary of the Invention

[0005] This invention provides a method and system for efficient video data compression based on artificial intelligence. By using artificial intelligence to intelligently analyze video content and employing intelligent compression strategies and optimized compression algorithms, it can compress video while maintaining video quality, avoiding image quality loss. Furthermore, it can adjust the compression loading speed according to user needs, thereby making video data compression in the industrial field more efficient and accurate.

[0006] This invention provides an efficient video data compression method based on artificial intelligence, comprising:

[0007] S1: Acquire video data of the target area based on a preset sensor, process the data to obtain the first processed video frame, and then perform frame-by-frame analysis on the first processed video frame based on a preset deep learning model to determine the video analysis result;

[0008] S2: Determine the intelligent compression strategy based on the video analysis results, and dynamically adjust it in combination with real-time device characteristics and real-time user needs to obtain the optimized compression strategy;

[0009] S3: Select a matching compression algorithm based on the optimized compression strategy, and optimize the compression parameters based on the preset machine learning algorithm, thereby compressing the first video frame.

[0010] S4: Evaluate the quality of the compressed video stream after compression, and optimize the compression strategy based on the evaluation quality and user feedback to achieve efficient compression.

[0011] Preferably, S1: Video data of the target area is acquired based on a preset sensor, and the data is processed to obtain a first processed video frame. Then, the first processed video frame is analyzed frame-by-frame based on a preset deep learning model to determine the video analysis results, including:

[0012] The system acquires real-time video of the device's operating status in the target area based on preset sensors, and then decodes and generates continuous initial video frames.

[0013] The initial video frame is denoised and contrast-enhanced to obtain the first processed video frame.

[0014] Based on the video features extracted from the first processed video frame, a deep learning model is pre-built and trained in the video model database.

[0015] The first processed video frame is analyzed frame by frame based on a deep learning model to determine the video analysis results in the first processed video frame.

[0016] Preferably, the first processed video frame is analyzed frame by frame based on a deep learning model to determine the video analysis results in the first processed video frame, including:

[0017] Based on a deep learning model, video analysis is performed on the inter-frame differences between each adjacent video frame of the first processed video frame to determine key frame data.

[0018] The video compression accuracy is determined based on the real-time user demand in the target area, thereby obtaining the standard contrast of the first processed video frame.

[0019] Extract the region with a contrast higher than the standard contrast from each video frame of the first processed video frame as the detail region data of the current video frame;

[0020] Compare the inter-frame similarity between each adjacent video frame of the first processed video frame to identify and determine redundant information in the first processed video frame;

[0021] The video analysis results in the first processed video frame are obtained by combining the keyframe data, detail region data, and redundancy information of the first processed video frame.

[0022] Preferably, S2: Based on the video analysis results, a smart compression strategy is determined and dynamically adjusted in conjunction with real-time device characteristics and real-time user needs to obtain an optimized compression strategy, including:

[0023] Based on the video analysis results, a comprehensive compression target is determined, and in combination with real-time user needs, it is judged whether the comprehensive compression target can meet the video's watchability.

[0024] If the video compression of the first processed video frame achieves the comprehensive compression target and satisfies the video's watchability, then based on the comprehensive compression target and combined with keyframe data, detail area data, and redundant information, an intelligent compression strategy for the first processed video frame is obtained.

[0025] The intelligent compression strategy is dynamically adjusted based on the content type of the first processed video frame and the real-time device characteristics of the playback device to obtain an optimized compression strategy.

[0026] Preferably, the intelligent compression strategy is dynamically adjusted based on the content type of the first processed video frame and the real-time device characteristics of the playback device to obtain an optimized compression strategy, including:

[0027] The application scenario and real-time performance of the first processed video frame are determined based on the user's historical requirements.

[0028] Determine whether the intelligent compression strategy can meet the application scenario and real-time performance requirements of the first processed video frame.

[0029] If the intelligent compression strategy can meet the application scenario and real-time performance requirements of the first processed video frame, then the intelligent compression strategy will be used as the first compression strategy.

[0030] If the intelligent compression strategy cannot meet the application scenario or real-time performance requirements of the first processed video frame, then extract the first compression scheme that can meet the application scenario or real-time performance requirements.

[0031] Compare the corresponding compression parameters in the intelligent compression strategy with those in the first compression scheme, and replace the compression parameters in the intelligent compression strategy that are not greater than the corresponding compression parameters in the first compression scheme with the compression parameters of the first compression scheme to obtain the initial compression strategy;

[0032] Determine the feasibility of the initial compression strategy, and optimize the compression parameters that conflict with the strategy to obtain the first compression strategy;

[0033] Obtain the content type of the first processed video frame, obtain the historical compression degree in the effective historical compression process corresponding to the content type, and calculate the average value of the historical compression degree to obtain the average compression degree range corresponding to the current content type.

[0034] The first compression strategy and the first processed video frame are input into the virtual machine, and compression simulation is performed in combination with the real-time device characteristics of the playback device to obtain the first compression degree of the first compression strategy.

[0035] Compare and determine whether the first degree of compression is within the range of average compression.

[0036] If the first compression level is within the range of the average compression level, then the first compression strategy will be used as the optimized compression strategy.

[0037] If the first compression degree is less than the minimum compression threshold of the average compression degree range, it is determined that the first compression strategy is under-compressed. The compression parameter adjustment strategy that increases the compression parameter is obtained from the compression strategy database, the amount of compression parameter adjustment is determined, and the first compression strategy is optimized to obtain the optimized compression strategy.

[0038] If the first compression degree is greater than the maximum compression threshold of the average compression degree range, it is determined that the first compression strategy is in an over-compression state. The compression parameter adjustment strategy that reduces the compression parameter is obtained from the compression strategy database, the compression parameter adjustment amount is determined, and the first compression strategy is optimized to obtain an optimized compression strategy.

[0039] Preferably, S3: Selecting a matching compression algorithm based on an optimized compression strategy, and optimizing compression parameters based on a preset machine learning algorithm, thereby compressing the first processed video frame, including:

[0040] A comprehensive compression algorithm based on optimized compression strategy selection and matching of the first processed video frame;

[0041] Based on the types of historical compression parameters that affect the video compression effect during the historical video compression process, the key parameter types of the first processed video frame are determined.

[0042] The initial parameter values ​​for key parameter types are determined based on the video characteristics of the first processed video frame and the overall compression target.

[0043] Input the historical compressed video of the target area and the corresponding historical compression parameters into the machine learning algorithm for training, and determine the initial set of compression parameters;

[0044] Based on the real-time characteristics of the device, the compression parameters of the initial compression parameter set are optimized in real time to obtain the optimal compression parameter set.

[0045] The comprehensive compression algorithm is configured based on the optimal compression parameter set and deployed to the preset compression tool to compress the first video frame.

[0046] Preferably, S4: The compressed video stream after compression is evaluated for quality, and the evaluation quality is combined with user feedback to optimize and adjust the compression strategy, thereby achieving efficient compression, including:

[0047] The quality of the compressed video stream after compression is evaluated based on a preset subjective and objective quality evaluation scheme.

[0048] Real-time feedback ratings from target users on compressed video streams, and overall user experience based on these ratings;

[0049] Based on the quality assessment results and the overall user experience, the real-time compression defects of the optimized compression strategy are analyzed and optimized to obtain the second optimized compression strategy.

[0050] The second optimized compression strategy is applied to the first processed video frame to obtain the second compression result;

[0051] A second quality assessment is performed based on the second compression result, and the second quality assessment result is compared with the quality assessment result to obtain the optimal compression strategy for the first processed video frame, thereby achieving efficient compression of the first processed video frame.

[0052] Preferably, after achieving efficient compression, the method further includes: verifying the compression effect, specifically including:

[0053] Obtain the compression time for video compression based on the optimal compression strategy, and compare the compression time with the historical compression time of the video.

[0054] If the compression time is less than the historical compression time, the optimal compression scheme is preliminarily judged to be qualified.

[0055] Randomly extract any time segment of the compressed video from the compressed video stream that has been preliminarily judged to be qualified by the optimal compression scheme, and obtain the first compressed video;

[0056] Determine whether there is missing key video information in the first compressed video;

[0057] If the first compressed video does not contain any missing key video information, then the optimal compression strategy is deemed acceptable.

[0058] Conversely, if the optimal compression strategy is deemed unqualified, the first video frame needs to be reprocessed and compressed.

[0059] This invention provides an artificial intelligence-based high-efficiency video data compression system for executing any one of the artificial intelligence-based high-efficiency video data compression methods described in Examples 1 to 8, comprising:

[0060] The content analysis module is used to acquire video data of the target area based on a preset sensor, and to process the data to obtain the first processed video frame. Then, based on a preset deep learning model, the first processed video frame is analyzed frame by frame to determine the video analysis results.

[0061] Strategy formulation module: used to determine intelligent compression strategies based on video analysis results, and dynamically adjust them in combination with real-time device characteristics and real-time user needs to obtain optimized compression strategies;

[0062] Compression Implementation Module: Used to select a matching compression algorithm based on an optimized compression strategy and optimize compression parameters based on a preset machine learning algorithm, thereby compressing the first video frame.

[0063] Evaluation and optimization module: This module evaluates the quality of the compressed video stream after compression and combines the evaluation quality with user feedback to optimize and adjust the compression strategy, thereby achieving efficient compression.

[0064] The beneficial effects of this invention compared to existing technologies are as follows: by using artificial intelligence to intelligently analyze video content, and through intelligent compression strategies and optimized compression algorithms, it is possible to compress video while maintaining video quality, avoiding image quality loss. At the same time, it can also adjust the compression loading speed according to user needs, thereby making the compression of video data in the industrial field more efficient and accurate.

[0065] Other features and advantages of the invention will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in this application.

[0066] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0067] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0068] Figure 1 This is a schematic diagram of an efficient video data compression method based on artificial intelligence, as described in an embodiment of the present invention.

[0069] Figure 2 This is a structural diagram of an artificial intelligence-based high-efficiency video data compression system according to an embodiment of the present invention. Detailed Implementation

[0070] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0071] Example 1:

[0072] This invention provides an efficient video data compression method based on artificial intelligence, referencing... Figure 1 ,include:

[0073] S1: Acquire video data of the target area based on a preset sensor, process the data to obtain the first processed video frame, and then perform frame-by-frame analysis on the first processed video frame based on a preset deep learning model to determine the video analysis result;

[0074] S2: Determine the intelligent compression strategy based on the video analysis results, and dynamically adjust it in combination with real-time device characteristics and real-time user needs to obtain the optimized compression strategy;

[0075] S3: Select a matching compression algorithm based on the optimized compression strategy, and optimize the compression parameters based on the preset machine learning algorithm, thereby compressing the first video frame.

[0076] S4: Evaluate the quality of the compressed video stream after compression, and optimize the compression strategy based on the evaluation quality and user feedback to achieve efficient compression.

[0077] In this embodiment, the preset sensor refers to a pre-set device for capturing video data, such as a camera.

[0078] In this embodiment, the video data is a continuous sequence of images of the target area captured by a preset sensor, which typically includes audio and visual information.

[0079] In this embodiment, data processing refers to preprocessing the original video data, such as denoising, enhancement, and format conversion, to obtain a first processed video frame suitable for further analysis.

[0080] In this embodiment, the first processed video frame is a video frame that has undergone preliminary data processing.

[0081] In this embodiment, the preset deep learning model is a pre-trained deep learning network capable of identifying and analyzing key information in video frames. It is used to analyze video data frame by frame to determine the video analysis results, which include the identification and analysis of key frames (such as frames containing important events or changes), detailed regions (such as image parts requiring special attention), and redundant information.

[0082] In this embodiment, the video analysis results are information about video frames obtained through analysis using a deep learning model, including the location and content of keyframes and detail regions, as well as the analysis results of redundant information.

[0083] In this embodiment, the intelligent compression strategy is a video compression scheme formulated based on video analysis results, which aims to optimize video storage and transmission efficiency by optimizing video compression efficiency.

[0084] In this embodiment, real-time device characteristics refer to the current performance parameters of the device that performs compressed video playback output, such as computing power and storage space.

[0085] In this embodiment, real-time user demand refers to the user's current requirements for video quality, compression speed, and other aspects.

[0086] In this embodiment, the optimized compression strategy is a compression strategy that is dynamically adjusted based on real-time device characteristics and user needs, so that the compressed video after implementing the compression optimization strategy is more in line with the actual application scenario.

[0087] In this embodiment, the compression algorithm is an algorithm used to reduce the amount of video data. It achieves compression by removing redundant information and reducing resolution. It is a comprehensive compression algorithm that includes lossless compression and lossy compression.

[0088] In this embodiment, the preset machine learning algorithm is a machine learning model used to optimize compression parameters. It can automatically adjust the compression parameters according to the video content and user needs to achieve the best compression effect.

[0089] In this embodiment, the compressed video stream is a sequence of video data that has been compressed, typically used for storage or network transmission.

[0090] In this embodiment, quality assessment is a subjective and objective evaluation of the compressed video stream to determine whether the compressed video meets the quality requirements.

[0091] In this embodiment, user feedback is direct feedback from users on the quality of the compressed video stream, such as satisfaction ratings and opinions, which can be used to further optimize the compression strategy.

[0092] The beneficial effects of the above technologies are as follows: by using artificial intelligence to intelligently analyze video content, and through intelligent compression strategies and optimized compression algorithms, compression can be performed while maintaining video quality, avoiding image quality loss. At the same time, the compression loading speed can be adjusted according to user needs, thus making the compression of video data in the industrial field more efficient and accurate.

[0093] Example 2:

[0094] Based on Example 1, an efficient video data compression method based on artificial intelligence is provided, S1: acquiring video data of a target area based on a preset sensor, and processing the data to obtain a first processed video frame, thereby performing frame-by-frame analysis on the first processed video frame based on a preset deep learning model to determine the video analysis results, including:

[0095] The system acquires real-time video of the device's operating status in the target area based on preset sensors, and then decodes and generates continuous initial video frames.

[0096] The initial video frame is denoised and contrast-enhanced to obtain the first processed video frame.

[0097] Based on the video features extracted from the first processed video frame, a deep learning model is pre-built and trained in the video model database.

[0098] The first processed video frame is analyzed frame by frame based on a deep learning model to determine the video analysis results in the first processed video frame.

[0099] In this embodiment, the real-time device operation status video is a continuous video image captured by a preset sensor that reflects the current operating status of the device.

[0100] In this embodiment, decoding refers to the process of converting captured video data into original video frames.

[0101] In this embodiment, the initial video frame refers to the video frame obtained after decoding and without further processing.

[0102] In this embodiment, denoising is a video processing technique used to eliminate noise and interference in video frames. Denoising can improve the clarity and quality of video frames, providing more accurate information for subsequent analysis.

[0103] In this embodiment, contrast enhancement processing is a video processing technique used to adjust the contrast of video frames, making them sharper and easier to observe. Contrast enhancement processing helps highlight key information in the video, improving the accuracy of analysis.

[0104] In this embodiment, the first processed video frame is a video frame obtained after noise reduction and contrast enhancement. These video frames have undergone preliminary processing and optimization, making them suitable for subsequent deep learning analysis.

[0105] In this embodiment, video characteristics are image features reflected in video frames, such as color, texture, and shape.

[0106] In this embodiment, the video model database is a database used to store pre-built and trained deep learning models. These models are used to analyze and identify video frames and extract useful information.

[0107] In this embodiment, a deep learning model refers to an algorithm or network structure based on deep learning, used for complex analysis and processing of video frames. Deep learning models can automatically learn features in video frames and classify, identify, or predict them.

[0108] In this embodiment, frame-by-frame analysis refers to the process of analyzing video frames one by one. Frame-by-frame analysis can capture subtle changes and key information in the video.

[0109] In this embodiment, the video analysis result is obtained by performing frame-by-frame analysis of the first processed video frame using a deep learning model.

[0110] In this embodiment, the video analysis results include the analysis results that identify key frames (such as frames containing important events or changes), detailed regions (such as image parts that require special attention), and redundant information.

[0111] The beneficial effects of the above technologies are as follows: by processing video data and performing frame-by-frame analysis based on deep learning models, the analysis of video data can be made more intelligent, improving the efficiency and accuracy of video analysis, thereby improving video compression performance and optimizing video storage and transmission efficiency.

[0112] Example 3:

[0113] Based on Example 2, an efficient video data compression method based on artificial intelligence analyzes the first processed video frame frame by frame using a deep learning model to determine the video analysis results in the first processed video frame, including:

[0114] Based on a deep learning model, video analysis is performed on the inter-frame differences between each adjacent video frame of the first processed video frame to determine key frame data.

[0115] The video compression accuracy is determined based on the real-time user demand in the target area, thereby obtaining the standard contrast of the first processed video frame.

[0116] Extract the region with a contrast higher than the standard contrast from each video frame of the first processed video frame as the detail region data of the current video frame;

[0117] Compare the inter-frame similarity between each adjacent video frame of the first processed video frame to identify and determine redundant information in the first processed video frame;

[0118] The video analysis results in the first processed video frame are obtained by combining the keyframe data, detail region data, and redundancy information of the first processed video frame.

[0119] In this embodiment, inter-frame difference refers to the difference between two adjacent video frames. Inter-frame difference can reflect information such as the motion of objects and changes in the scene. By analyzing inter-frame difference, key change points in the video can be identified.

[0120] In this embodiment, video analysis refers to using a deep learning model to analyze the inter-frame differences of the first processed video frame in order to determine keyframe data.

[0121] In this embodiment, keyframe data refers to representative or significant frames in the video, typically containing significant changes or important events. Keyframe data is used for video processing, compression, etc.

[0122] In this embodiment, video compression accuracy refers to the amount of information or quality level retained during video compression. Higher video compression accuracy retains more information, but also requires more storage space or transmission bandwidth. Video compression accuracy is determined based on real-time user needs.

[0123] In this embodiment, the standard contrast ratio is a contrast threshold determined based on video compression accuracy and video content. Areas with a contrast ratio higher than the standard contrast ratio are considered to have higher contrast, i.e., areas of detail.

[0124] In this embodiment, detail region data refers to areas in a video frame with contrast higher than the standard contrast, typically containing important image details or texture information. Detail region data requires special attention during video compression and transmission to ensure video transmission quality.

[0125] In this embodiment, inter-frame similarity refers to the degree of similarity between two adjacent video frames. By analyzing inter-frame similarity, redundant information in the video can be identified, that is, content that does not change significantly or repeat.

[0126] In this embodiment, the inter-frame similarity between adjacent i-th frame video images and j-th frame video images is sim(F i F j );

[0127] sim(F i ,F j )=1-ham(Q (i) Q (j) )

[0128] Where, sim(F) i F j ) represents the inter-frame similarity between the current i-th frame and the adjacent j-th frame, ham(Q) (i) Q (j) ) represents the Hamming distance between the pixel value of the current i-th frame and the pixel value of the adjacent j-th frame, sim(F) i F j The value range of ) is [0,1], sim(F i F j The closer the value is to 1, the higher the similarity between two adjacent video image frames.

[0129] In this embodiment, redundant information refers to repetitive or redundant content in the video data, such as static backgrounds or repetitive actions. During video compression, redundant information can be reduced or deleted to save storage space or transmission bandwidth.

[0130] In this embodiment, the video analysis result is the result obtained after analyzing and processing the video data, including keyframe data, detail region data, and redundant information.

[0131] The beneficial effects of the above technologies are as follows: by performing frame-by-frame analysis of the first processed video frame based on a deep learning model, the analysis of video data can be made more intelligent, improving the efficiency and accuracy of video analysis, thereby improving video compression performance and optimizing video storage and transmission efficiency.

[0132] Example 4:

[0133] Based on Example 2, an efficient video data compression method based on artificial intelligence is provided, S2: An intelligent compression strategy is determined based on video analysis results, and dynamically adjusted in conjunction with real-time device characteristics and real-time user needs to obtain an optimized compression strategy, including:

[0134] Based on the video analysis results, a comprehensive compression target is determined, and in combination with real-time user needs, it is judged whether the comprehensive compression target can meet the video's watchability.

[0135] If the video compression of the first processed video frame achieves the comprehensive compression target and satisfies the video's watchability, then based on the comprehensive compression target and combined with keyframe data, detail area data, and redundant information, an intelligent compression strategy for the first processed video frame is obtained.

[0136] The intelligent compression strategy is dynamically adjusted based on the content type of the first processed video frame and the real-time device characteristics of the playback device to obtain an optimized compression strategy.

[0137] In this embodiment, the comprehensive compression objective is one or more compression metrics determined based on video analysis results and real-time user needs, aiming to achieve video compression while maintaining viewability. The comprehensive compression objective may include multiple aspects such as compression ratio, compression speed, and video quality.

[0138] In this embodiment, real-time user requirements refer to the user's specific demands for video compression at the current moment, such as the compressed video size, playback smoothness, and video clarity. Real-time user requirements are dynamic and need to be adjusted according to the actual situation.

[0139] In this embodiment, video viewability refers to the quality of the video during playback, including aspects such as clarity, smoothness, and color reproduction. Video viewability is an important indicator for evaluating video compression effectiveness.

[0140] In this embodiment, the intelligent compression strategy is a video compression scheme formulated based on comprehensive compression objectives and video analysis results. The intelligent compression strategy performs targeted compression on the video according to important features such as keyframe data, detailed region data, and redundant information, achieving efficient compression while ensuring video quality.

[0141] In this embodiment, content type refers to the variety of video content, such as video data of industrial equipment, video data of industrial personnel, and video data of industrial conditions. Different content types of videos may require different strategies and methods during compression to better preserve their characteristics and quality.

[0142] In this embodiment, the real-time device characteristics of the playback device refer to the performance parameters of the device playing the video at the current moment, such as screen resolution, processor performance, and memory size. These device characteristics affect the video playback effect and the formulation of compression strategies.

[0143] In this embodiment, the optimized compression strategy refers to a compression scheme that is dynamically adjusted based on the intelligent compression strategy, taking into account the video content type and the real-time device characteristics of the playback device. The optimized compression strategy can better adapt to the video compression needs of different scenarios, achieving higher compression efficiency and better playback performance.

[0144] The beneficial effects of the above technologies are as follows: by combining the characteristics of real-time devices and the needs of real-time users, the compression strategy can be adjusted so that the adjusted and optimized compression strategy can better meet the needs of real-time use, thereby improving video compression efficiency.

[0145] Example 5:

[0146] Based on Example 4, an efficient video data compression method based on artificial intelligence dynamically adjusts the intelligent compression strategy based on the content type of the first processed video frame and the real-time device characteristics of the playback device to obtain an optimized compression strategy, including:

[0147] The application scenario and real-time performance of the first processed video frame are determined based on the user's historical requirements.

[0148] Determine whether the intelligent compression strategy can meet the application scenario and real-time performance requirements of the first processed video frame.

[0149] If the intelligent compression strategy can meet the application scenario and real-time performance requirements of the first processed video frame, then the intelligent compression strategy will be used as the first compression strategy.

[0150] If the intelligent compression strategy cannot meet the application scenario or real-time performance requirements of the first processed video frame, then extract the first compression scheme that can meet the application scenario or real-time performance requirements.

[0151] Compare the corresponding compression parameters in the intelligent compression strategy with those in the first compression scheme, and replace the compression parameters in the intelligent compression strategy that are not greater than the corresponding compression parameters in the first compression scheme with the compression parameters of the first compression scheme to obtain the initial compression strategy;

[0152] Determine the feasibility of the initial compression strategy, and optimize the compression parameters that conflict with the strategy to obtain the first compression strategy;

[0153] Obtain the content type of the first processed video frame, obtain the historical compression degree in the effective historical compression process corresponding to the content type, and calculate the average value of the historical compression degree to obtain the average compression degree range corresponding to the current content type.

[0154] The first compression strategy and the first processed video frame are input into the virtual machine, and compression simulation is performed in combination with the real-time device characteristics of the playback device to obtain the first compression degree of the first compression strategy.

[0155] Compare and determine whether the first degree of compression is within the range of average compression.

[0156] If the first compression level is within the range of the average compression level, then the first compression strategy will be used as the optimized compression strategy.

[0157] If the first compression degree is less than the minimum compression threshold of the average compression degree range, it is determined that the first compression strategy is under-compressed. The compression parameter adjustment strategy that increases the compression parameter is obtained from the compression strategy database, the amount of compression parameter adjustment is determined, and the first compression strategy is optimized to obtain the optimized compression strategy.

[0158] If the first compression degree is greater than the maximum compression threshold of the average compression degree range, it is determined that the first compression strategy is in an over-compression state. The compression parameter adjustment strategy that reduces the compression parameter is obtained from the compression strategy database, the compression parameter adjustment amount is determined, and the first compression strategy is optimized to obtain an optimized compression strategy.

[0159] In this embodiment, historical user requirements refer to the specific requirements of users for video processing or compression over a past period, including application scenarios, compression levels, and real-time performance. Historical user requirements can be used to predict current or future user needs, thereby guiding the formulation of video compression strategies.

[0160] In this embodiment, the application scenario refers to the specific environment and scenario in which the video content is used or displayed, such as online playback, mobile device playback, and large-screen display. Different application scenarios have different requirements for video compression and playback, necessitating the development of corresponding compression strategies.

[0161] In this embodiment, application real-time performance refers to the real-time requirements that the video needs to meet during playback or processing, such as playback smoothness and loading speed. Real-time performance is one of the important indicators for measuring the effectiveness of video compression.

[0162] In this embodiment, the intelligent compression strategy is a video compression scheme formulated using intelligent algorithms or models based on video content, application scenarios, and real-time performance requirements. The intelligent compression strategy aims to achieve efficient compression and playback while ensuring video quality.

[0163] In this embodiment, the determination of whether the intelligent compression strategy can meet the application scenario and real-time performance requirements of the first processed video frame can be based on the satisfaction index S.

[0164] S=αM+βN

[0165] Where S is the satisfaction index, α and β are the weights of application scenario matching degree and real-time performance satisfaction degree, and the sum of the weights of application scenario matching degree and real-time performance satisfaction degree is 1, M is the matching degree between intelligent compression strategy and application scenario, and N is the satisfaction degree between intelligent optimization strategy and real-time performance. The values ​​of M and N are in the range of (0,1). When S is greater than 1.2, it is considered that the intelligent compression strategy meets the requirements of the first processed video frame.

[0166] In this embodiment, the first compression scheme is a pre-determined video compression scheme that meets the current application scenario and real-time performance requirements. When the intelligent compression strategy cannot meet the requirements, strategy optimization can be performed based on the first compression scheme.

[0167] In this embodiment, compression parameters refer to various parameters used to control compression quality and speed during video compression, such as bitrate, frame rate, and resolution. Different combinations of compression parameters will affect the compression effect and playback quality of the video.

[0168] In this embodiment, strategy feasibility is used to assess the likelihood that a compression strategy can be effectively implemented under the current environment or conditions. Strategy feasibility needs to consider various factors, such as device performance and network bandwidth.

[0169] In this embodiment, parameter optimization refers to adjusting and improving the parameters in the compression strategy to enhance compression performance or meet specific requirements. Parameter optimization can be based on experimental data, user feedback, or intelligent algorithms.

[0170] In this embodiment, the content type refers to the category or type of video content, such as industrial equipment data, industrial personnel data, industrial environment data, etc. Different content types have different requirements for compression strategies, necessitating the development of corresponding compression schemes.

[0171] In this embodiment, a valid historical compression process refers to a compression process that was successfully implemented in the past and was able to maintain video quality and meet user needs. A valid historical compression process can provide a reference for the formulation of current compression strategies.

[0172] In this embodiment, historical compression level refers to the degree or proportion of video compression during the effective historical compression process. Historical compression level can reflect the characteristics and patterns of different content types during the compression process.

[0173] In this embodiment, the average compression range is calculated based on historical compression levels, reflecting the range of average compression levels for a certain content type during the compression process. This range can be used to evaluate whether the current compression strategy is reasonable.

[0174] In this embodiment, a virtual machine is a software technology that simulates a real computer environment, in which various applications, including video compression and playback software, can be run and tested. Virtual machines can be used to simulate video compression effects under different device and network conditions.

[0175] In this embodiment, compression simulation refers to the process of compressing video using specific compression strategies and parameters in a virtual machine. Compression simulation can simulate the video compression effect in a real-world environment, providing a basis for optimizing compression strategies.

[0176] In this embodiment, the degree of compression refers to the amount or proportion of data reduced during the video compression process. The degree of compression is one of the important indicators for measuring the effectiveness of video compression, and the value range of the degree of compression is generally (0.05, 0.5).

[0177] In this embodiment, the compression strategy database is a database used to store various compression strategies and parameter adjustments. This database can provide a reference and basis for optimizing compression strategies.

[0178] In this embodiment, the compression parameter adjustment amount is Δp1 or Δp2;

[0179]

[0180] Where Δp1 is the parameter adjustment amount corresponding to the undercompression state of the first compression strategy, Δp2 is the parameter adjustment amount corresponding to the overcompression state of the first compression strategy, k is the parameter adjustment coefficient, and Min t Max t Com represents the minimum and maximum compression thresholds for the average compression range. t The compression level of the first compression strategy, u user E is a user preference factor. t D is the entropy of the video image in the current frame. tγ1 is the standard deviation of the video image in the current frame, e is the base of the natural logarithm, γ2 is the weight of the influence of image entropy on the adjustment of compression parameters, and γ3 is the weight of the influence of image standard deviation on the adjustment of compression parameters.

[0181] In this embodiment, the compression parameter adjustment strategy refers to the strategies and methods used in the compression strategy database to adjust compression parameters for specific situations or needs. It can be used to optimize the current compression strategy.

[0182] The beneficial effects of the above technologies are as follows: by combining the characteristics of real-time devices and the needs of real-time users, the compression strategy can be adjusted so that the adjusted and optimized compression strategy can better meet the needs of real-time use, thereby improving video compression efficiency.

[0183] Example 6:

[0184] Based on Example 4, an efficient video data compression method based on artificial intelligence, S3: Selecting a matching compression algorithm based on an optimized compression strategy and optimizing compression parameters based on a preset machine learning algorithm, thereby compressing the first processed video frame, including:

[0185] A comprehensive compression algorithm based on optimized compression strategy selection and matching of the first processed video frame;

[0186] Based on the types of historical compression parameters that affect the video compression effect during the historical video compression process, the key parameter types of the first processed video frame are determined.

[0187] The initial parameter values ​​for key parameter types are determined based on the video characteristics of the first processed video frame and the overall compression target.

[0188] Input the historical compressed video of the target area and the corresponding historical compression parameters into the machine learning algorithm for training, and determine the initial set of compression parameters;

[0189] Based on the real-time characteristics of the device, the compression parameters of the initial compression parameter set are optimized in real time to obtain the optimal compression parameter set.

[0190] The comprehensive compression algorithm is configured based on the optimal compression parameter set and deployed to the preset compression tool to compress the first video frame.

[0191] In this embodiment, the integrated compression algorithm is a video compression method that combines multiple compression technologies and algorithms. The integrated compression algorithm can select the most suitable compression strategy and parameters based on different video content and requirements to achieve the best compression effect.

[0192] In this embodiment, historical compression parameter types refer to the parameter types that have significantly affected the video compression effect during past video compression processes for the target region. These parameter types typically include bitrate, frame rate, resolution, encoding format, etc.

[0193] In this embodiment, the key parameter type refers to the parameter type that plays a decisive role in the compression effect and goal achievement in the current video compression task. These parameter types are typically determined based on the video content, application scenario, and compression target.

[0194] In this embodiment, video characteristics are inherent properties of the video content itself, such as color, brightness, contrast, and motion complexity. These characteristics affect the effectiveness of video compression and the required compression parameters.

[0195] In this embodiment, the overall compression objective refers to the general goals and requirements that are desired to be achieved during the video compression process, such as compression ratio, video quality, and real-time performance. The overall compression objective is an important basis for formulating compression strategies and selecting compression algorithms.

[0196] In this embodiment, the initial parameter values ​​refer to the initial values ​​set for the key parameter types before video compression begins. These values ​​are typically determined based on video characteristics and overall compression objectives.

[0197] In this embodiment, the machine learning algorithm is an algorithm that learns and improves its performance using training data. In video compression, machine learning algorithms can be used to analyze historical compressed videos and corresponding compression parameters to predict and optimize the compression effect of the current video.

[0198] In this embodiment, the initial compression parameter set is obtained through training with a machine learning algorithm and is used as a combination of parameters for initial video compression. The initial compression parameter set is typically determined based on historical compressed videos of the target area and the corresponding compression parameters.

[0199] In this embodiment, real-time device characteristics refer to the performance characteristics and limitations of the device performing the video compression task, such as processor speed, memory size, and storage speed. Real-time device characteristics influence the selection of compression algorithms and the optimization of parameters.

[0200] In this embodiment, the optimal compression parameter set is the best combination of parameters obtained by real-time optimization of the initial compression parameter set, taking into account the characteristics of the real-time device. These parameter combinations typically achieve the best compression effect and meet real-time performance requirements.

[0201] In this embodiment, parameter configuration refers to the process of setting the optimal set of compression parameters into the comprehensive compression algorithm. Parameter configuration is a crucial step in ensuring that the algorithm can compress as expected.

[0202] In this embodiment, the preset compression tool is a pre-installed and configured software or hardware tool used to perform video compression tasks. This tool typically includes a comprehensive compression algorithm and necessary parameter configuration options.

[0203] In this embodiment, the compression process involves compressing the first video frame using a preset compression tool. The compression process is performed according to configured parameters and algorithms to generate a compressed video that meets the requirements.

[0204] The beneficial effects of the above technologies are as follows: by determining a comprehensive compression algorithm based on an optimized compression strategy and combining it with machine learning algorithms to optimize compression parameters, the compression processing accuracy can be improved, making the video compression of the first processed video frame more efficient, thereby improving storage and transmission efficiency.

[0205] Example 7:

[0206] Based on Example 6, an AI-based high-efficiency video data compression method, S4: Evaluating the quality of the compressed video stream after compression, and optimizing the compression strategy based on the evaluation quality and user feedback to achieve high-efficiency compression, includes:

[0207] The quality of the compressed video stream after compression is evaluated based on a preset subjective and objective quality evaluation scheme.

[0208] Real-time feedback ratings from target users on compressed video streams, and overall user experience based on these ratings;

[0209] Based on the quality assessment results and the overall user experience, the real-time compression defects of the optimized compression strategy are analyzed and optimized to obtain the second optimized compression strategy.

[0210] The second optimized compression strategy is applied to the first processed video frame to obtain the second compression result;

[0211] A second quality assessment is performed based on the second compression result, and the second quality assessment result is compared with the quality assessment result to obtain the optimal compression strategy for the first processed video frame, thereby achieving efficient compression of the first processed video frame.

[0212] In this embodiment, the preset subjective and objective quality assessment scheme is a scheme used to evaluate the quality of compressed video streams, including two parts: subjective assessment and objective assessment. Subjective assessment usually judges video quality through the visual perception of human observers, while objective assessment uses mathematical algorithms or models to quantify various aspects of video quality, such as sharpness, distortion, and frame rate.

[0213] In this embodiment, compressed video stream refers to video data that has undergone compression processing, typically transmitted or stored as a stream. Compressed video streams have a smaller file size compared to the original video data, but the original video quality needs to be restored during decoding.

[0214] In this embodiment, the quality assessment result is a conclusion or data obtained after evaluating the compressed video stream using preset subjective and objective quality assessment schemes. It typically includes one or more quality indicators to quantify the video's quality level.

[0215] In this embodiment, the target users are a designated group of users who need to receive and watch the compressed video stream. Feedback from the target users is crucial for optimizing the compression strategy and improving the user experience. In industrial production processes, the target users are typically quality inspectors, security personnel, etc.

[0216] In this embodiment, the feedback rating refers to the target user's subjective evaluation of the compressed video stream, typically expressed as a score or rating. The feedback rating can reflect the user's satisfaction and acceptance of the video quality.

[0217] In this embodiment, the overall user experience is a user experience metric obtained by comprehensively evaluating user feedback ratings and other relevant data (such as viewing time, number of interactions, etc.). The overall user experience reflects the performance of the compressed video stream in practical applications and user satisfaction.

[0218] In this embodiment, the real-time compression defects of the optimized compression strategy refer to problems that occur when using the optimized compression strategy for real-time compression, leading to a decrease in video quality or a poor user experience. Examples include loss of detail due to over-compression and excessively large files due to under-compression.

[0219] In this embodiment, optimization refers to the analysis and improvement measures taken to address defects in real-time compression. Optimization may involve adjusting compression parameters, improving compression algorithms, or adopting new compression technologies.

[0220] In this embodiment, the second optimized compression strategy refers to the new compression strategy after optimization and adjustment.

[0221] In this embodiment, the second compression result is obtained by compressing the first processed video frame using the second optimized compression strategy. This result is used to perform a second quality assessment on the first processed video frame to verify the effectiveness of the optimization adjustment.

[0222] In this embodiment, the second quality assessment refers to a quality assessment of the second compression result. This assessment will use the same preset subjective and objective quality assessment scheme as the initial quality assessment.

[0223] In this embodiment, the optimal compression strategy refers to the compression strategy that achieves the best video quality and user experience after multiple quality assessments and optimizations. The optimal compression strategy is the best choice for the current video content and application scenario.

[0224] In this embodiment, efficient compression is the process of compressing video using an optimal compression strategy, aiming to achieve a balance between minimum file size and best visual quality. Efficient compression can improve the efficiency of video transmission and storage while maintaining a high-quality viewing experience.

[0225] The beneficial effects of the above technologies are: by evaluating the quality of the compressed video stream and optimizing the compression strategy based on user feedback, the compression efficiency and video transmission loading speed can be improved, thus better meeting the needs of users and industrial production.

[0226] Example 8:

[0227] Based on Example 7, an efficient video data compression method based on artificial intelligence, after achieving efficient compression, further includes: verifying the compression effect, specifically including:

[0228] Obtain the compression time for video compression based on the optimal compression strategy, and compare the compression time with the historical compression time of the video.

[0229] If the compression time is less than the historical compression time, the optimal compression scheme is preliminarily judged to be qualified.

[0230] Randomly extract any time segment of the compressed video from the compressed video stream that has been preliminarily judged to be qualified by the optimal compression scheme, and obtain the first compressed video;

[0231] Determine whether there is missing key video information in the first compressed video;

[0232] If the first compressed video does not contain any missing key video information, then the optimal compression strategy is deemed acceptable.

[0233] Conversely, if the optimal compression strategy is deemed unqualified, the first video frame needs to be reprocessed and compressed.

[0234] In this embodiment, compression time refers to the time required to complete the video compression process. For example, compression time typically depends on the size and complexity of the video, the compression strategy, and the hardware and software resources used.

[0235] In this embodiment, historical compression time refers to the time required to compress similar videos using the same or different compression strategies in the past. Historical compression time can be used to compare and evaluate the efficiency of the current compression strategy.

[0236] In this embodiment, the preliminary judgment that the optimal compression scheme is qualified means that, based on the comparison between the compression time and the historical compression time, it is preliminarily considered that the currently used optimal compression strategy is qualified in terms of compression efficiency.

[0237] In this embodiment, a compressed video stream refers to a sequence of compressed video data, which typically exists in the form of consecutive data packets or frames.

[0238] In this embodiment, the first compressed video is video data of any time period randomly extracted from the compressed video stream corresponding to the optimal compression scheme that is initially judged to be qualified. This video data is used for further inspection and verification.

[0239] In this embodiment, key video information refers to information crucial for understanding the video content, such as faces, text, important objects, and actions. This key video information should be preserved as completely as possible during compression to avoid information loss and subsequent video quality degradation.

[0240] In this embodiment, information loss refers to the loss or damage of key video information during the compression process due to various reasons (such as excessive compression ratio, algorithm defects, etc.). Information loss affects the readability and comprehensibility of the video.

[0241] In this embodiment, determining that the optimal compression strategy is qualified means that, after preliminary judgment and further inspection of key video information, it is confirmed that the currently used optimal compression strategy is qualified in terms of maintaining video quality and compression efficiency.

[0242] In this embodiment, compression processing refers to the need to reprocess and compress the video frames if the optimal compression strategy is deemed unsuitable. Compression processing involves adjusting compression parameters, changing the compression algorithm, or adopting other optimization measures.

[0243] The beneficial effects of the above technologies are: by timely testing of video compression effects, the compression strategy can be adjusted in a timely manner, making video data compression more efficient, thereby improving video transmission and loading speed and better meeting the needs of users and industrial production.

[0244] Example 9:

[0245] This invention provides an artificial intelligence-based high-efficiency video data compression system for executing any one of the artificial intelligence-based high-efficiency video data compression methods described in Examples 1 to 8, with reference to... Figure 2 ,include:

[0246] The content analysis module is used to acquire video data of the target area based on a preset sensor, and to process the data to obtain the first processed video frame. Then, based on a preset deep learning model, the first processed video frame is analyzed frame by frame to determine the video analysis results.

[0247] Strategy formulation module: used to determine intelligent compression strategies based on video analysis results, and dynamically adjust them in combination with real-time device characteristics and real-time user needs to obtain optimized compression strategies;

[0248] Compression Implementation Module: Used to select a matching compression algorithm based on an optimized compression strategy and optimize compression parameters based on a preset machine learning algorithm, thereby compressing the first video frame.

[0249] Evaluation and optimization module: This module evaluates the quality of the compressed video stream after compression and combines the evaluation quality with user feedback to optimize and adjust the compression strategy, thereby achieving efficient compression.

[0250] The beneficial effects of the above technologies are as follows: by using artificial intelligence to intelligently analyze video content, and through intelligent compression strategies and optimized compression algorithms, compression can be performed while maintaining video quality, avoiding image quality loss. At the same time, the compression loading speed can be adjusted according to user needs, thus making the compression of video data in the industrial field more efficient and accurate.

[0251] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for efficient compression of video data based on artificial intelligence, characterized in that, include: S1: Acquire video data of the target area based on a preset sensor, process the data to obtain the first processed video frame, and then perform frame-by-frame analysis on the first processed video frame based on a preset deep learning model to determine the video analysis result; S2: Determine the intelligent compression strategy based on the video analysis results, and dynamically adjust it in combination with real-time device characteristics and real-time user needs to obtain the optimized compression strategy; S3: Select a matching compression algorithm based on the optimized compression strategy, and optimize the compression parameters based on the preset machine learning algorithm, thereby compressing the first video frame. S4: Evaluate the quality of the compressed video stream after compression, and optimize the compression strategy based on the evaluation quality and user feedback to achieve efficient compression; S2: Based on video analysis results, an intelligent compression strategy is determined and dynamically adjusted according to real-time device characteristics and real-time user needs to obtain an optimized compression strategy, including: Based on the video analysis results, a comprehensive compression target is determined, and in combination with real-time user needs, it is judged whether the comprehensive compression target can meet the video's watchability. If the video compression of the first processed video frame achieves the comprehensive compression target and satisfies the video's watchability, then based on the comprehensive compression target and combined with keyframe data, detail area data, and redundant information, an intelligent compression strategy for the first processed video frame is obtained. Based on the content type of the first processed video frame and the real-time device characteristics of the playback device, the intelligent compression strategy is dynamically adjusted to obtain an optimized compression strategy. The intelligent compression strategy is dynamically adjusted based on the content type of the first processed video frame and the real-time device characteristics of the playback device to obtain an optimized compression strategy, including: The application scenario and real-time performance of the first processed video frame are determined based on the user's historical requirements. Determine whether the intelligent compression strategy can meet the application scenario and real-time performance requirements of the first processed video frame. If the intelligent compression strategy can meet the application scenario and real-time performance requirements of the first processed video frame, then the intelligent compression strategy will be used as the first compression strategy. If the intelligent compression strategy cannot meet the application scenario or real-time performance requirements of the first processed video frame, then extract the first compression scheme that can meet the application scenario or real-time performance requirements. Compare the corresponding compression parameters in the intelligent compression strategy with those in the first compression scheme, and replace the compression parameters in the intelligent compression strategy that are not greater than the corresponding compression parameters in the first compression scheme with the compression parameters of the first compression scheme to obtain the initial compression strategy; Determine the feasibility of the initial compression strategy, and optimize the compression parameters that conflict with the strategy to obtain the first compression strategy; Obtain the content type of the first processed video frame, obtain the historical compression degree in the effective historical compression process corresponding to the content type, and calculate the average value of the historical compression degree to obtain the average compression degree range corresponding to the current content type. The first compression strategy and the first processed video frame are input into the virtual machine, and compression simulation is performed in combination with the real-time device characteristics of the playback device to obtain the first compression degree of the first compression strategy. Compare and determine whether the first degree of compression is within the range of average compression. If the first compression level is within the range of the average compression level, then the first compression strategy will be used as the optimized compression strategy. If the first compression degree is less than the minimum compression threshold of the average compression degree range, it is determined that the first compression strategy is under-compressed. The compression parameter adjustment strategy that increases the compression parameter is obtained from the compression strategy database, the amount of compression parameter adjustment is determined, and the first compression strategy is optimized to obtain the optimized compression strategy. If the first compression degree is greater than the maximum compression threshold of the average compression degree range, it is determined that the first compression strategy is in an over-compression state. The compression parameter adjustment strategy that reduces the compression parameter is obtained from the compression strategy database, the compression parameter adjustment amount is determined, and the first compression strategy is optimized to obtain an optimized compression strategy.

2. The method for efficient compression of video data based on artificial intelligence according to claim 1, characterized in that, S1: Acquire video data of the target area based on a preset sensor, process the data to obtain a first processed video frame, and then perform frame-by-frame analysis on the first processed video frame based on a preset deep learning model to determine the video analysis results, including: The system acquires real-time video of the device's operating status in the target area based on preset sensors, and then decodes and generates continuous initial video frames. The initial video frame is denoised and contrast-enhanced to obtain the first processed video frame. Based on the video features extracted from the first processed video frame, a deep learning model is pre-built and trained in the video model database. The first processed video frame is analyzed frame by frame based on a deep learning model to determine the video analysis results in the first processed video frame. The video analysis results include the analysis results of keyframe data, detailed region data, and redundant information.

3. The method for efficient compression of video data based on artificial intelligence according to claim 2, characterized in that, The first processed video frame is analyzed frame by frame based on a deep learning model to determine the video analysis results in the first processed video frame, including: Based on a deep learning model, video analysis is performed on the inter-frame differences between each adjacent video frame of the first processed video frame to determine key frame data. The video compression accuracy is determined based on the real-time user demand in the target area, thereby obtaining the standard contrast of the first processed video frame. Extract the region with a contrast higher than the standard contrast from each video frame of the first processed video frame as the detail region data of the current video frame; Compare the inter-frame similarity between each adjacent video frame of the first processed video frame to identify and determine redundant information in the first processed video frame; The video analysis results in the first processed video frame are obtained by combining the keyframe data, detail region data, and redundancy information of the first processed video frame.

4. The method for efficient compression of video data based on artificial intelligence according to claim 1, characterized in that, S3: Based on the optimized compression strategy, a matching compression algorithm is selected, and compression parameters are optimized based on a preset machine learning algorithm to compress the first video frame, including: A comprehensive compression algorithm based on optimized compression strategy selection and matching of the first processed video frame; Based on the types of historical compression parameters that affect the video compression effect during the historical video compression process, the key parameter types of the first processed video frame are determined. The initial parameter values ​​for key parameter types are determined based on the video characteristics of the first processed video frame and the overall compression target. Input the historical compressed video of the target area and the corresponding historical compression parameters into the machine learning algorithm for training, and determine the initial set of compression parameters; Based on the real-time characteristics of the device, the compression parameters of the initial compression parameter set are optimized in real time to obtain the optimal compression parameter set. The comprehensive compression algorithm is configured based on the optimal compression parameter set and deployed to the preset compression tool to compress the first video frame.

5. The method for efficient compression of video data based on artificial intelligence according to claim 4, characterized in that, S4: Evaluate the quality of the compressed video stream after compression, and optimize the compression strategy based on the evaluation quality and user feedback to achieve efficient compression, including: The quality of the compressed video stream after compression is evaluated based on a preset subjective and objective quality evaluation scheme. Real-time feedback ratings from target users on compressed video streams, and overall user experience based on these ratings; Based on the quality assessment results and the overall user experience, the real-time compression defects of the optimized compression strategy are analyzed and optimized to obtain the second optimized compression strategy. The second optimized compression strategy is applied to the first processed video frame to obtain the second compression result; A second quality assessment is performed based on the second compression result, and the second quality assessment result is compared with the quality assessment result to obtain the optimal compression strategy for the first processed video frame, thereby achieving efficient compression of the first processed video frame.

6. The method for efficient compression of video data based on artificial intelligence according to claim 5, characterized in that, After achieving efficient compression, the process also includes: verifying the compression effect, specifically including: Obtain the compression time for video compression based on the optimal compression strategy, and compare the compression time with the historical compression time of the video. If the compression time is less than the historical compression time, the optimal compression scheme is preliminarily judged to be qualified. Randomly extract any time segment of the compressed video from the compressed video stream that has been preliminarily judged to be qualified by the optimal compression scheme, and obtain the first compressed video; Determine whether there is missing key video information in the first compressed video; If the first compressed video does not contain missing key video information, then the optimal compression strategy is deemed acceptable. Conversely, if the optimal compression strategy is deemed unqualified, the first video frame needs to be reprocessed and compressed.

7. A high-efficiency video data compression system based on artificial intelligence, characterized in that, The method for performing efficient video data compression based on artificial intelligence according to any one of claims 1 to 6 includes: The content analysis module is used to acquire video data of the target area based on a preset sensor, and to process the data to obtain the first processed video frame. Then, based on a preset deep learning model, the first processed video frame is analyzed frame by frame to determine the video analysis results. Strategy formulation module: used to determine intelligent compression strategies based on video analysis results, and dynamically adjust them in combination with real-time device characteristics and real-time user needs to obtain optimized compression strategies; Compression Implementation Module: Used to select a matching compression algorithm based on an optimized compression strategy and optimize compression parameters based on a preset machine learning algorithm, thereby compressing the first video frame. Evaluation and optimization module: Used to evaluate the quality of the compressed video stream after compression, and to optimize and adjust the compression strategy based on the evaluation quality and user feedback, so as to achieve efficient compression; The strategy formulation module is used for: Based on the video analysis results, a comprehensive compression target is determined, and in combination with real-time user needs, it is judged whether the comprehensive compression target can meet the video's watchability. If the video compression of the first processed video frame achieves the comprehensive compression target and satisfies the video's watchability, then based on the comprehensive compression target and combined with keyframe data, detail area data, and redundant information, an intelligent compression strategy for the first processed video frame is obtained. Based on the content type of the first processed video frame and the real-time device characteristics of the playback device, the intelligent compression strategy is dynamically adjusted to obtain an optimized compression strategy. The intelligent compression strategy is dynamically adjusted based on the content type of the first processed video frame and the real-time device characteristics of the playback device to obtain an optimized compression strategy, including: The application scenario and real-time performance of the first processed video frame are determined based on the user's historical requirements. Determine whether the intelligent compression strategy can meet the application scenario and real-time performance requirements of the first processed video frame. If the intelligent compression strategy can meet the application scenario and real-time performance requirements of the first processed video frame, then the intelligent compression strategy will be used as the first compression strategy. If the intelligent compression strategy cannot meet the application scenario or real-time performance requirements of the first processed video frame, then extract the first compression scheme that can meet the application scenario or real-time performance requirements. Compare the corresponding compression parameters in the intelligent compression strategy with those in the first compression scheme, and replace the compression parameters in the intelligent compression strategy that are not greater than the corresponding compression parameters in the first compression scheme with the compression parameters of the first compression scheme to obtain the initial compression strategy; Determine the feasibility of the initial compression strategy, and optimize the compression parameters that conflict with the strategy to obtain the first compression strategy; Obtain the content type of the first processed video frame, obtain the historical compression degree in the effective historical compression process corresponding to the content type, and calculate the average value of the historical compression degree to obtain the average compression degree range corresponding to the current content type. The first compression strategy and the first processed video frame are input into the virtual machine, and compression simulation is performed in combination with the real-time device characteristics of the playback device to obtain the first compression degree of the first compression strategy. Compare and determine whether the first degree of compression is within the range of average compression. If the first compression level is within the range of the average compression level, then the first compression strategy will be used as the optimized compression strategy. If the first compression degree is less than the minimum compression threshold of the average compression degree range, it is determined that the first compression strategy is under-compressed. The compression parameter adjustment strategy that increases the compression parameter is obtained from the compression strategy database, the amount of compression parameter adjustment is determined, and the first compression strategy is optimized to obtain the optimized compression strategy. If the first compression degree is greater than the maximum compression threshold of the average compression degree range, it is determined that the first compression strategy is in an over-compression state. The compression parameter adjustment strategy that reduces the compression parameter is obtained from the compression strategy database, the compression parameter adjustment amount is determined, and the first compression strategy is optimized to obtain an optimized compression strategy.

Images