Communication method and platform for ultra-high-definition video E-band transmission

Abstract

The present invention discloses a communication method and platform for ultra-high-definition video E-band transmission, which belongs to the field of video transmission, wherein the method includes: monitoring environmental feature data, and obtaining transmission impact analysis results, when it is greater than a preset transmission impact threshold, obtaining an impact timeline, and obtaining multiple impact screen contents; according to multiple impact screen contents, deciding to obtain key content compensation ratios and key frame compensation ratios; when transmitting within the impact timeline, performing compensation generation, obtaining generated frames and generated content; deciding to obtain quality loss identification parameters, identifying generated frames and generated content, and adding transmission data for transmission and display. The method solves the technical problem in the prior art that when transmitting ultra-high-definition video in the E-band, the quality of video transmission is reduced and key frames and key content are lost due to the influence of the transmission environment, and achieves the technical effect of improving video transmission quality and reducing the loss of key frames and key content when transmitting ultra-high-definition video in the E-band.

Description

Technical Field

[0001] The present invention relates to the field of video transmission, and in particular to a communication method and platform for ultra-high-definition video E-band transmission. Background Art

[0002] At present, the transmission of ultra-high-definition video using the E-band is easily affected by the transmission environment, resulting in a decrease in video transmission quality and loss of key frames and key content. Specifically, E-band transmission is sensitive to environmental conditions. Environmental factors such as humidity, temperature, dust, and rainfall may interfere with signal transmission, thereby affecting the quality of video transmission. When the transmission quality drops to a certain level, the video content, especially the key frames and key content, will be lost, causing serious damage to the video quality at the receiving end and affecting the user's viewing experience. Summary of the invention

[0003] This application provides a communication method and platform for ultra-high-definition video E-band transmission, aiming to solve the technical problems in the prior art of ultra-high-definition video transmission in the E-band, such as the degradation of video transmission quality and the loss of key frames and key content due to the influence of the transmission environment.

[0004] In view of the above problems, the present application provides a communication method and platform for ultra-high-definition video E-band transmission.

[0005] The first aspect disclosed in the present application provides a communication method for E-band transmission of ultra-high-definition video, the method comprising: monitoring and collecting environmental feature data in a transmission environment for ultra-high-definition video transmission based on E-band transmission, and performing environmental feature data prediction and transmission impact analysis to obtain a transmission impact analysis result; when the transmission impact analysis result is greater than a preset transmission impact threshold, obtaining an impact time axis of the transmission impact analysis result, and indexing in the ultra-high-definition video being transmitted to obtain multiple impact picture contents in multiple impact frames; performing key frame analysis and key content analysis based on the multiple impact picture contents to obtain multiple key contents The ratio of key content and key frame ratio are calculated and the overall ratio of key content is obtained. According to the overall ratio of key content and key frame ratio, the key content compensation ratio and key frame compensation ratio are decided. When transmitting within the timeline, the transmission data is monitored to obtain lost frames and lost content. According to the key content compensation ratio and key frame compensation ratio, the lost frames and lost content are compensated and generated to obtain generated frames and generated content. According to the loss ratio of lost frames and lost content, combined with the overall ratio of key content and key frame ratio, the quality loss identification parameter is decided to identify the generated frames and generated content, and add them to the transmission data for transmission and display.

[0006] Another aspect disclosed in the present application provides a communication platform for E-band transmission of ultra-high-definition video, the platform comprising: a transmission impact analysis module, for monitoring and collecting environmental feature data in a transmission environment for ultra-high-definition video transmission based on E-band transmission, and performing environmental feature data prediction and transmission impact analysis to obtain a transmission impact analysis result; an impact identification module, for obtaining an impact timeline of the transmission impact analysis result when the transmission impact analysis result is greater than a preset transmission impact threshold, and for obtaining multiple impact picture contents in multiple impact frames affected by the transmission in the ultra-high-definition video being transmitted; a compensation ratio acquisition module, for performing key frame analysis and key content analysis based on multiple impact picture contents to obtain The proportions of multiple key contents and key frames are calculated, and the overall proportion of key contents is obtained. According to the overall proportion of key contents and the proportion of key frames, the key content compensation ratio and the key frame compensation ratio are decided. A compensation generation module is used to monitor the transmission data when transmitting within the impact timeline, obtain the lost frames and lost contents, and generate compensation for the lost frames and lost contents according to the key content compensation ratio and the key frame compensation ratio to obtain the generated frames and generated contents. A quality loss identification module is used to decide the quality loss identification parameters according to the loss ratio of the lost frames and lost contents, combined with the overall proportion of key contents and the proportion of key frames, identify the generated frames and generated contents, and add them to the transmission data for transmission and display.

[0007] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0008] By monitoring and collecting environmental characteristic data in the transmission environment, predicting the environmental characteristic data and analyzing the transmission impact, the transmission impact analysis result is obtained, and the degree of influence of the transmission environment on the video transmission quality is predicted, providing a basis for subsequent compensation processing; when the transmission impact analysis result is greater than the preset transmission impact threshold, it means that the transmission quality is affected. At this time, the impact timeline is obtained, and multiple affected frames and picture contents in the transmitted video within the impact timeline are indexed, and the video clips affected by the transmission environment are accurately located, providing a data basis for subsequent key content analysis; key frame analysis and key content analysis are performed on the indexed multiple affecting picture contents to obtain parameters such as the key content ratio and the key frame ratio, and the key content compensation ratio and the key frame compensation ratio are decided accordingly, the importance of the affected video clips is judged, and the key frames and contents that need to be protected and compensated are determined, so as to improve the subsequent compensation targeted; when transmitting within the affected timeline, real-time monitoring of transmission data, obtaining lost frames and content information, and compensating and generating lost key frames and content based on previously determined key content compensation ratio and key frame compensation ratio, to achieve dynamic repair of damaged video during transmission and ensure complete transmission of video content; based on the key content loss ratio, combined with the key content ratio and key frame ratio, generate quality loss identification parameters that characterize the degree of transmission quality impact, and then attach the parameters to the compensated video frames and content, output them together for transmission, and intuitively reflect the degree of transmission damage. This technical solution solves the technical problem in the prior art that when transmitting ultra-high-definition video in the E-band, the quality of video transmission is reduced and key frames and key content are lost due to the influence of the transmission environment, and achieves the technical effect of improving video transmission quality and reducing the loss of key frames and key content when transmitting ultra-high-definition video in the E-band.

[0009] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Figure 1 A flow chart of a communication method for ultra-high-definition video E-band transmission is provided for an embodiment of the present application;

[0011] Figure 2 A structural schematic diagram of a communication platform for ultra-high-definition video E-band transmission is provided for an embodiment of the present application.

[0012] Explanation of reference numerals: transmission impact analysis module 11 , impact identification module 12 , compensation ratio acquisition module 13 , compensation generation module 14 , quality loss identification module 15 . DETAILED DESCRIPTION

[0013] The overall idea of ​​the technical solution provided by this application is as follows:

[0014] The embodiment of the present application provides a communication method and platform for E-band transmission of ultra-high-definition video. First, by real-time monitoring and trend prediction of the transmission environment, the results of the analysis of the impact of environmental changes on the transmission quality are determined. When the impact exceeds the preset transmission impact threshold, a targeted quality compensation mechanism is initiated. During the compensation process, the impact timeline of the transmission is first located, the video frames within the impact timeline are extracted, and the video content that needs to be protected is determined through key frame analysis and key content analysis, so as to decide the proportion and strategy of subsequent compensation. Once the loss of key frames or key content is detected during the transmission process, the compensation process is triggered, and parameters such as the proportion of key content and the proportion of key frames are used to accurately and efficiently repair the damaged video. Afterwards, the video repair effect is quantified as a quality loss identification parameter and attached to the compensated video data to intuitively reflect the degree of transmission damage, so as to improve the video transmission quality and reduce the loss of key frames and key content when transmitting ultra-high-definition video in the E-band.

[0015] After introducing the basic principles of the present application, various non-limiting implementation methods of the present application will be specifically described below in conjunction with the drawings in the specification.

[0016] Embodiment 1, as Figure 1 As shown, an embodiment of the present application provides a communication method for ultra-high-definition video E-band transmission, the method comprising:

[0017] S1: Monitor and collect environmental characteristic data in the transmission environment for ultra-high-definition video transmission based on E-band transmission, and perform environmental characteristic data prediction and transmission impact analysis to obtain transmission impact analysis results.

[0018] Specifically, first, the transmission environment for ultra-high-definition video transmission based on E-band transmission is monitored and collected to obtain environmental characteristic data in the transmission environment. Among them, environmental characteristic data include but are not limited to temperature, humidity, dust concentration and other factor parameters that affect the E-band transmission quality. Then, the environmental characteristic data obtained are used to predict the environmental characteristic data and analyze the transmission impact. Specifically, based on the historically collected environmental characteristic data, an environmental characteristic data prediction model is established through machine learning and other technical means to predict the changing trend of environmental characteristic data in the future period of time, and judge the impact of environmental changes on transmission quality. At the same time, by analyzing the correlation between the current and predicted environmental characteristic data and the E-band transmission quality, the degree of influence of environmental characteristics on transmission quality is judged, and the transmission impact analysis results are obtained.

[0019] By collecting, predicting and analyzing environmental characteristic data and transmission impact, we can grasp the status and changing trends of the transmission environment in real time, predict the impact of environmental changes on the E-band transmission quality, and provide a basis for subsequent targeted transmission optimization and quality control, thereby improving the timeliness and pertinence of transmission processing, thereby minimizing the problem of transmission quality degradation caused by environmental changes when E-band transmission is susceptible to environmental influences.

[0020] S2: When the transmission impact analysis result is greater than a preset transmission impact threshold, the impact time axis of the transmission impact analysis result is obtained, and multiple impact picture contents in multiple affected frames are indexed in the ultra-high-definition video being transmitted.

[0021] Specifically, first, according to the obtained transmission impact analysis results, it is determined whether the impact of the current transmission environment on the transmission quality of the E-band ultra-high-definition video reaches the preset transmission impact threshold. The preset transmission impact threshold is set according to the actual application needs and transmission requirements, representing the critical point of the environmental impact degree. If it exceeds the threshold, it is considered that the transmission quality may be significantly affected and further optimization measures need to be taken. If the transmission impact analysis results show that the current environmental impact degree exceeds the preset transmission impact threshold, the transmission quality protection mechanism is triggered. First, the impact timeline in the transmission impact analysis results is obtained, which describes the start and end time range of the predicted environmental impact. Using the impact timeline, the affected video clips are accurately located in the ultra-high-definition video stream being transmitted. Next, the key frames and key contents of the video clips in the impact timeline are extracted. Through the frame index information of the video encoding format, multiple video frames in the affected clips are quickly located as the affected frames. In each affected frame, the picture content of the affected frame is further extracted, such as faces, text, target objects, etc., to obtain multiple affected picture contents.

[0022] By locating the video data affected by the transmission environment, the foundation is laid for subsequent targeted transmission quality optimization, providing support for minimizing the impact of environmental deterioration on the overall video transmission quality, so that optimization processing can intervene in advance to avoid a sharp deterioration in transmission quality.

[0023] S3: Perform key frame analysis and key content analysis based on the multiple influencing screen contents, obtain multiple key content proportions and key frame proportions, and calculate to obtain the overall key content proportion. According to the overall key content proportion and key frame proportion, decide to obtain the key content compensation ratio and key frame compensation ratio.

[0024] Specifically, for each influencing frame, key frame analysis and key content analysis are performed respectively. In key frame analysis, the importance of the current influencing frame relative to the entire video sequence is evaluated. Among them, key frames carry most of the video semantic information during video transmission and serve as a reference for subsequent inter-frame prediction and motion compensation. Once a key frame is lost or damaged, it will cause error propagation and affect the reconstruction quality of a series of subsequent frames. Therefore, the key frame ratio index is used to indicate the proportion of key frames in the influencing frame. The higher the ratio, the higher the criticality of the affected video segment. In key content analysis, the semantic importance of the key content area in the influencing frame is evaluated. Different key content areas, such as faces, texts, and target objects, have different degrees of influence on the expression and perception quality of video content. For example, the face area usually carries more video semantics, and its integrity has a greater impact on the perception of the character picture. Therefore, a fine-grained importance evaluation is performed on different types of key content areas. The key content ratio index is used to indicate the area ratio of each key content area in the influencing frame. The higher the ratio, the more important the key content of this type is.

[0025] After obtaining the key frame ratio and key content ratio of each affected frame, the criticality of the entire affected video clip is evaluated in a larger range. By calculating the weighted average of the key content ratio of each affected frame, the comprehensive overall key content ratio is obtained, which represents the degree of impact of the deterioration of the transmission environment on the semantic integrity of the entire video clip. The higher the ratio, the more key semantic information the affected video clip contains. Afterwards, the two indicators of key frame ratio and overall key content ratio are used to determine the strategy and parameters for subsequent transmission recovery processing. Specifically, through the set threshold interval and mapping function, the ratio index is mapped to the corresponding key frame compensation ratio and key content compensation ratio. The compensation ratio reflects the proportion of resources that need to be invested in the recovery process. The higher the ratio, the more redundant transmission and error correction coding measures are needed to focus on ensuring the recovery quality of key video data. The adaptive recovery strategy can save transmission and computing resources as much as possible while ensuring the recovery effect.

[0026] S4: When transmitting within the impact timeline, monitor the transmission data, obtain lost frames and lost content, and perform compensation generation of the lost frames and lost content according to the key content compensation ratio and the key frame compensation ratio to obtain generated frames and generated content.

[0027] Specifically, first, the transmission data of the E band is monitored in real time within the determined impact time axis. By parsing and verifying the transmission data, the data loss and bit error caused by environmental deterioration are found. Once data loss is detected, it is quickly determined which impact frame the lost data belongs to, and the specific content of the loss is further located, that is, part of the key content area in the key frame. Next, different strategies and methods are used to recover the lost data according to the determined key frame compensation ratio and key content compensation ratio. For the lost key frame, the video frame repair technology based on deep learning is adopted. By analyzing the spatiotemporal correlation before and after the key frame, the lost key frame is intelligently reconstructed using the successfully received frame as a reference. Specifically, a frame repair model based on a generative adversarial network is used to learn rich video content prior knowledge and inter-frame association rules by training massive video frame data; in the repair process, the frames before and after the lost frame are used as input, and the generator network synthesizes a repair result similar to the visual semantics of the original lost frame, and then the discriminator network is used for quality evaluation and optimization, so as to obtain a high-quality key frame repair result. Among them, according to the key frame compensation ratio, the complexity and generation scale of the frame repair model are dynamically adjusted. The higher the key frame compensation ratio, the deeper the network structure and the more reference frames the model will use to obtain better restoration effects, but it will also consume more computing resources.

[0028] For lost content, a semantic restoration technique based on an attention mechanism is used. Different from frame restoration, semantic restoration pays more attention to the semantic integrity of local areas, and performs targeted restoration optimization according to the content type of the lost content and the surrounding contextual information. For example, for the lost face area, the restoration focuses on maintaining the continuity and identity consistency of facial features; for the lost text area, the restoration focuses on the recognition and reconstruction of characters. To this end, a multi-task semantic restoration network is constructed, which dynamically focuses on different types of lost areas through an attention mechanism, adaptively extracts semantic features from adjacent successful reception areas, and then generates targeted restoration results through the corresponding decoder branches. Similarly, the parameter scale and computational complexity of the semantic restoration network will also be dynamically adjusted according to the key content compensation ratio to strike a balance between restoration quality and resource efficiency. The restoration results of the lost key frames and key content areas are obtained, which are compensations for the lost frames and lost content generation, thereby obtaining generated frames and generated content.

[0029] Through the adaptive, content-aware transmission recovery mechanism, the spatiotemporal correlation and semantic features of video data are fully utilized, and combined with deep learning technology, an intelligent trade-off is made between recovery quality and resource efficiency, improving the anti-interference ability and recovery performance of E-band transmission. Even in harsh transmission environments, through targeted compensation processing, the integrity of ultra-high-definition video is maximized to provide smooth and stable video.

[0030] S5: According to the loss ratio of the lost frames and lost content, combined with the overall key content ratio and key frame ratio, a decision is made to obtain quality loss identification parameters, the generated frames and generated content are identified, and added to the transmission data for transmission and display.

[0031] Specifically, first, according to the detected lost frames and lost contents, as well as the multiple affected picture contents in the obtained multiple affected frames, the loss ratio is calculated to reflect the overall loss degree of video data during transmission. The higher the loss ratio, the greater the impact on the transmission quality. Then, according to the calculated overall key content ratio and key frame ratio, the key ratio is obtained by weighted average to reflect the proportion of key content in the lost video data. The higher the key ratio, the greater the impact of the lost data on the video quality and semantic integrity. Then, the loss ratio and the key ratio are combined, and the quality loss identification parameter is obtained through a preset mapping function or lookup table, which comprehensively considers the total amount and key degree of lost data, and reflects the comprehensive impact of the deterioration of the transmission environment on the video quality. Among them, the quality loss identification parameter can be expressed as a percentage value or a level value as an intuitive indicator to measure the decline in transmission quality. Afterwards, the quality loss identification parameter is used to identify the quality of the generated frame and the generated content, clearly informing that the video clip has undergone transmission recovery processing and indicating the degree of quality loss before processing. The generated frames and generated content after quality identification will be merged with the originally transmitted video data to obtain ultra-high-definition video transmission output results, which improves the video transmission quality of ultra-high-definition video and reduces the loss of key frames and key content.

[0032] Furthermore, the embodiment of the present application also includes:

[0033] Monitor and collect environmental characteristic data within a transmission environment for ultra-high-definition video transmission based on E-band transmission; based on the historical environmental characteristic data monitoring records within the transmission environment, collect a sample environmental characteristic data set, and collect environmental characteristic data after a preset time length, and mark them as a sample predicted environmental characteristic data set; use the sample environmental characteristic data set and the sample predicted environmental characteristic data set to train an environmental characteristic data predictor, predict the environmental characteristic data, and obtain predicted environmental characteristic data; perform transmission impact analysis based on the predicted environmental characteristic data to obtain transmission impact analysis results.

[0034] In a feasible implementation, by learning and modeling historical environmental feature data, the future transmission environment can be predicted and the transmission quality can be estimated, providing a more accurate and timely decision-making basis for subsequent transmission optimization.

[0035] First, by deploying various sensors and detection instruments on the transmission equipment, the transmission environment of E-band ultra-high-definition video transmission is monitored in real time, and various environmental characteristic parameters are collected, including meteorological parameters such as temperature, humidity, air pressure, wind speed, rainfall, and transmission link related parameters such as geographical location, topography, and electromagnetic interference. At the same time, from the historical environmental characteristic data monitoring records in the transmission environment, a sample environmental characteristic data set is sampled in a targeted manner. This set contains environmental characteristic data of the transmission environment at different times and under different conditions, covering various change patterns and influencing factors of the transmission environment. At the same time, for each sample environmental characteristic data, its actual environmental characteristic data after a preset time length (such as the next 1 hour) is collected, and it is marked as the corresponding sample predicted environmental characteristic data to obtain a sample predicted environmental characteristic data set. Then, the collected sample environmental characteristic data set and the sample predicted environmental characteristic data set are used to construct and train an environmental characteristic data predictor. The environmental characteristic data predictor can adopt various time series prediction models, such as recurrent neural networks, long short-term memory networks, etc. By inputting sample environmental feature data into the predictor and using the sample predicted environmental feature data as a supervisory signal, the environmental feature data predictor learns the evolution law and change trend of the environmental feature data in the time dimension. The trained environmental feature data predictor predicts the newly collected environmental feature data to obtain the predicted environmental feature data for a period of time in the future. Afterwards, the predicted environmental feature data is used as input to perform transmission impact analysis and evaluate the potential impact of the future transmission environment on the transmission quality. For example, transmission impact analysis can be implemented through expert knowledge bases, influencing factor models, etc., such as pre-establishing a mapping relationship library between transmission environment characteristics and transmission quality indicators, and looking up the table to obtain the corresponding transmission bit error rate, signal-to-noise ratio and other estimated quality indicators based on the predicted environmental feature data as the transmission impact analysis result, indicating the degree of transmission quality degradation that may be caused by the future transmission environment.

[0036] Through the transmission environment prediction method based on machine learning, historical monitoring data is fully utilized to actively predict and analyze the future transmission environment, providing advance information for subsequent transmission optimization control, so that optimization measures can be timely intervened before the transmission quality deteriorates. Compared with traditional passive transmission control methods, it can perceive changes in the transmission environment more accurately and timely, reduce the risk of transmission interruption and quality degradation, and improve the stability and reliability of transmission.

[0037] Furthermore, the embodiment of the present application also includes:

[0038] According to the ultra-high-definition video transmission quality monitoring records in the transmission environment, a set of sample transmission impact analysis results is marked according to the degree of impact on the transmission quality under different sample environment characteristic data; the sample environment characteristic data set and the sample transmission impact analysis result set are used to train a transmission impact analyzer; the transmission impact analyzer is used to perform transmission impact analysis on the predicted environment characteristic data to obtain transmission impact analysis results.

[0039] In a preferred embodiment, a data-driven transmission impact analyzer is constructed by mining and learning historical transmission quality monitoring records, and the transmission impact analysis results are obtained by automatically evaluating and warning future transmission quality change trends based on predicted environmental characteristic data.

[0040] First, based on the transmission quality monitoring equipment and mechanisms deployed in the transmission environment, the ultra-high-definition video transmission quality monitoring records for a period of history are obtained, including various key quality indicators in the transmission process, such as bit error rate, packet loss rate, signal-to-noise ratio, delay jitter, etc. Then, these transmission quality monitoring records are associated and aligned with the environmental feature data collected in the same period, and the distribution law and degree of influence of transmission quality indicators under different environmental feature conditions are studied through statistical analysis and data mining technology. On this basis, an impact degree label is assigned to each historical environmental feature data sample to describe the impact of the environmental condition on the transmission quality, thereby obtaining a set of sample transmission impact analysis results. Then, using the sample environmental feature data set as input and the sample transmission impact analysis result set as output label, a transmission impact analyzer is constructed and trained. The transmission impact analyzer can use various supervised learning algorithms, such as support vector machine, random forest, neural network, etc. By inputting the sample environmental feature data into the transmission impact analyzer and using the sample transmission impact analysis results as supervision signals, the complex mapping relationship between environmental features and transmission quality impact is learned. Subsequently, the predicted environmental characteristic data is input into the trained transmission impact analyzer to automatically perform transmission impact analysis and estimation to obtain the transmission impact analysis results, which indicate the possible negative impact of transmission environment changes on transmission quality in the future, providing an important basis for transmission optimization.

[0041] Furthermore, the embodiment of the present application also includes:

[0042] According to the pre-marking of all video frames in the ultra-high-definition video, the number of key frames in the multiple influencing frames is obtained by screening, and the proportion of key contents in the multiple influencing frames is obtained to obtain the key frame proportion and multiple key content proportions; the average of the multiple key content proportions is calculated to obtain the overall key content proportion; a sample key frame proportion set and a sample overall key content proportion set are collected, and according to each sample key frame proportion, annotations are set to obtain a sample key frame compensation ratio set, and according to each sample overall key content proportion, annotations are set to obtain a sample key content compensation ratio set; the sample key frame proportion set and the sample key frame compensation ratio set are used to train a key frame compensation ratio classification branch, the sample overall key content proportion set and the sample key frame compensation ratio set are used to train a key content compensation ratio classification branch, and a compensation ratio classifier is obtained by combining the key frame compensation ratio classification branch; the compensation ratio classifier is used to classify the overall key content proportion and the key frame proportion to obtain the key content compensation ratio and the key frame compensation ratio.

[0043] In a feasible implementation, first, based on the pre-marking of all video frames in the ultra-high-definition video, the video frames that are key frames in the multiple influencing frames are quickly screened out. Pre-marking is the key frame indication metadata marked manually or by algorithm in the video production or encoding stage, which characterizes the importance of each video frame in the global semantics. On this basis, the ratio of the number of key frames in the influencing frames to the total number of influencing frames is counted as the key frame ratio. At the same time, for each picture in the influencing frame, the ratio of the number of pixels defined as the key content area according to the pre-marking to the total number of pixels is calculated as the key content ratio of the influencing frame, and multiple key content ratios are obtained. Then, the arithmetic average of the multiple key content ratios obtained from the multiple influencing frames is performed to obtain the overall key content ratio, which characterizes the overall value density of the content of the video clip affected by the transmission.

[0044] Then, a large amount of historical or simulated video transmission data is sampled to obtain representative key frame ratio and overall key content ratio combination samples, and a sample key frame ratio set and a sample overall key content ratio set are obtained. After that, according to expert experience and experimental statistics, the optimal key frame compensation ratio and key content compensation ratio are set and labeled for each sample ratio value to form a paired sample key frame compensation ratio set. Among them, the compensation ratio represents the relative proportion of key frames or key contents that need to be restored in the transmission recovery process. The higher the compensation ratio, the more data redundancy and computing resources are needed to ensure the recovery quality of this part of the semantics. After that, a compensation ratio classifier is constructed using the sample key frame ratio set, the sample overall key content ratio set and the sample key frame compensation ratio set. The compensation ratio classifier contains two parallel classification branches, namely the key frame compensation ratio classification branch and the key content compensation ratio classification branch, which are used to determine the key frame compensation ratio and the key content compensation ratio respectively. Each branch can use classification algorithms such as support vector machine, decision tree, neural network, etc. By inputting the sample ratio value into the classification branch and taking the sample compensation ratio as the training target, the compensation ratio classifier learns the corresponding relationship between the ratio value and the optimal compensation ratio. After the two branches are trained independently, they are combined and integrated to form a complete compensation ratio classifier. Subsequently, the key frame ratio and overall key content ratio that actually affect the video clip are input into the trained compensation ratio classifier to obtain the corresponding key frame compensation ratio and key content compensation ratio.

[0045] Furthermore, the embodiment of the present application also includes:

[0046] When the ultra-high-definition video is transmitted within the impact timeline, the transmission data is monitored, and packet loss detection is compared with the standard data to obtain lost frames and lost content; according to the transmission loss data of the ultra-high-definition video within the historical time, the frame compensator and the content compensator are trained, wherein the frame compensator and the content compensator respectively include M frame compensation branches and N content compensation branches, and M and N are integers greater than 1; according to the key content compensation ratio and the key frame compensation ratio, the number of frame compensation branches and the number of content compensation branches are obtained by rounding in combination with M and N; the frame data before and after the lost frame are input into the frame compensation branches of the number of frame compensation branches in the frame compensator to generate frame compensation, and multiple branch generated frames are obtained, and the generated frames are obtained by fusion processing; the successfully transmitted content in the frame where the lost content is located is input into the content compensation branches of the number of content compensation branches in the content compensator to generate content, and multiple branch generated content is obtained, and the generated content is obtained by fusion processing.

[0047] In a preferred embodiment, first, the transmission data of the ultra-high-definition video in the E band is monitored in real time within a determined impact timeline. By comparing the received transmission data with the standard data sent by the video encoding side, the error detection algorithm is used to locate the data loss caused by channel deterioration during the transmission process. Among them, data loss may occur at different granularity levels, which may be the loss of the entire video frame or the loss of content in a local area within the frame. By recording in detail the frame index, in-frame coordinate area and other information to which each lost position belongs during the detection process, structured lost frames and lost content are formed, providing an accurate positioning reference for subsequent targeted repairs.

[0048] Then, based on the historical video transmission data, a multi-branch, end-to-end frame compensator and content compensator are constructed. Among them, multiple parallel repair branches are set inside the frame compensator and the content compensator. The frame compensator contains M frame compensation branches, and the content compensator contains N content compensation branches, where M and N are integers greater than 1. Then, based on the obtained key content compensation ratio and key frame compensation ratio, the number of frame compensation branches and content compensation branches actually participating in the repair task is dynamically determined. Specifically, the compensation ratio is multiplied by the total number of branches M and N to obtain the number of enabled frame compensation branches and content compensation branches. The higher the compensation ratio, the greater the impact of the lost content on the video semantics, and more repair branches need to be called and more computing resources need to be invested to ensure the quality of repair and reconstruction.

[0049] After that, the data of several adjacent frames before and after the detected lost frame, as well as the context information such as the timestamp and encoding parameters of the lost frame, are input into a certain number of frame compensation branches, and the frame repair task is performed in parallel. Each branch independently estimates and generates a complete repair frame, which represents a possible repair result. Then, the weight coefficients of each branch, the quality score of the generated frame and other factors are comprehensively considered, and the final generated frame is obtained through fusion methods such as weighted average or voting, as a replacement and filler for the lost frame. At the same time, focusing on the incomplete frame where the lost content is located, the part of the content area that is successfully transmitted, together with the prior information such as the position, size, and surrounding correlation of the lost content, are input into a certain number of content compensation branches, and the content repair task is performed in parallel. Each branch uses the context clues provided by the residual area to propagate semantic information from different directions and scales, gradually fill in and refine the image content of the lost area, and obtain a natural and coherent repair result. Similarly, the content repair results generated by each branch are fused to obtain the final generated content, which is seamlessly embedded in the original incomplete frame to obtain a completely repaired video frame.

[0050] Through adaptive video transmission repair, the number and combination of repair branches are flexibly adjusted according to the semantic importance of the lost content, and targeted repair and reconstruction work is carried out in two dimensions: coding frames and content areas, to improve the estimation and recovery capabilities of lost data.

[0051] Furthermore, the embodiment of the present application also includes:

[0052] According to the transmission loss data of ultra-high-definition video in historical time, the sample before and after frame data set and the sample lost frame data set are collected; based on the generative adversarial network, the M frame compensation branches are constructed, each frame compensation branch includes a generative network and an adversarial network; randomly select M frames from the sample before and after frame data set and the sample lost frame data set to generate training data, and train the M frame compensation branches respectively until convergence, to obtain the trained M frame compensation branches, and obtain a frame compensator, wherein there is intersection data in the M frames generating training data; according to the transmission loss data of ultra-high-definition video in historical time, the sample successfully transmitted content set and the sample lost content set are collected; the sample successfully transmitted content set and the sample lost content set are used to construct and train the N content compensation branches to obtain a content compensator.

[0053] In a preferred embodiment, first, two types of training sample sets are collected based on the transmission loss data monitored in the historical ultra-high-definition video transmission records. One type is the sample front and back frame data set, which contains a large number of paired, consecutive frames before and after the lost frame, reflecting the temporal context information when the loss occurs. The other type is the sample lost frame data set, which contains a large number of lost frames themselves as a reference for repair and reconstruction. These two types of sample sets constitute the data basis for training frame compensators, including rich loss patterns and content features. Then, based on the generative adversarial network, a multi-branch frame compensator network structure is designed, which consists of M parallel frame compensation branches, each of which contains a generative network and an adversarial network. Among them, the generative network accepts the frames before and after the lost frame as conditional inputs, and learns to generate realistic and natural repair frames through a deep convolutional neural network structure to fill the position of the lost frame; the adversarial network accepts the output of the generative network and the original real lost frame, and learns to distinguish the difference between the generated frame and the real frame through adversarial training, guiding the generative network to continuously improve the repair quality, making the generated frame and the real frame visually difficult to distinguish. Multiple branches adopt different network structures and parameter configurations, representing complementary repair strategies and feature representation capabilities, mining and utilizing the contextual information of lost frames from different angles, and enhancing the diversity and robustness of the repair process. Subsequently, M independent, end-to-end frame compensation branches are constructed using the collected sample before and after frame data sets and sample lost frame data sets. Specifically, for the M frame compensation branches, M different training data subsets are randomly selected from the sample before and after frame data sets and the sample lost frame data sets, and each subset contains a certain number of before and after frame-lost frame paired samples. Among them, the M training data subsets are not completely independent, but there is a certain degree of intersection and overlap to promote knowledge sharing and collaborative learning between different branches. Then, based on each training data subset, the corresponding frame compensation branch is end-to-end generative adversarial training, that is, the before and after frames are input into the generative network, and the output of the generative network and the original lost frame are input into the adversarial network. By minimizing the adversarial loss between the generated frame and the original frame, the parameters of the generative network and the adversarial network are jointly optimized, so that the generative network can restore the lost frame as realistically as possible, and the adversarial network can accurately judge the authenticity of the generated result. The training processes of the M branches are independent of each other, achieving efficient parallelism, significantly improving training speed and resource utilization, and enhancing diversity and complementarity between branches by sharing samples. After sufficient training until convergence, M trained frame compensation branches are obtained, forming a complete and diverse frame compensator.

[0054] At the same time, for the content-level loss repair task, two types of key training sample sets are further collected based on historical transmission loss data. One type is the sample successfully transmitted content set, which contains a large number of successfully received, intact video frame content blocks, reflecting the local visual semantic information of successful transmission during the transmission process. The other type is the sample loss content set, which contains local video content blocks with data loss that are adjacent to the successfully transmitted content in the same frame. These two types of sample sets constitute the data basis for training the content compensator, including rich local semantic correlations and data missing patterns, providing necessary prior knowledge and reference information for regional-level content repair. Subsequently, the N parallel content compensation branches of the content compensator are constructed and trained using the sample successfully transmitted content set and the sample loss content set. Similar to the frame compensation branch, each content compensation branch also adopts an end-to-end structure based on a deep convolutional neural network, but focuses on learning the semantic representation and generation capabilities of local regional content. Specifically, the successfully transmitted content is taken as input, and high-level semantic features are extracted through the encoder network, and then the features are restored to complete content blocks through the decoder network. The restored content block is compared with the original lost content block, and the reconstruction loss is calculated, which is used as the optimization target for training the content compensation branch. At the same time, in order to further improve the realism and consistency of content compensation, attention mechanisms and adversarial training techniques can be introduced into the content compensation branch to strengthen the modeling of correlations between local regions and the generation of texture details. N content compensation branches are trained independently of each other, and the generalization and robustness of knowledge are enhanced by sharing sample data. Afterwards, N content compensation branches are trained to form a complete and diverse content compensator.

[0055] Furthermore, the embodiment of the present application also includes:

[0056] The loss ratio is calculated based on the lost frames and the lost content, as well as the multiple influencing picture contents in the multiple influencing frames; the key ratio is obtained by weighted calculation based on the overall key content ratio and the key frame ratio; the quality loss ratio is calculated based on the loss ratio and the key ratio, and is used as a quality loss identification parameter to identify the generated frames and generated content.

[0057] In a feasible implementation, first, according to the obtained statistical information of lost frames and lost content, and the determined overall scale of the affected video clips, an index reflecting the overall proportion of transmission loss is calculated. Specifically, the proportion of lost frames to the total number of affected frames and the proportion of lost content block area to the total area of ​​the affected video clips are used to calculate the loss ratio that represents the global loss level. For example, the proportion of lost frames and the proportion of lost content are multiplied to obtain a comprehensive loss ratio value. The larger the value, the more serious the data loss in this transmission, and the greater the overall impact on the video quality. Then, according to the obtained overall key content proportion and key frame proportion, the degree of influence of transmission loss on the semantic integrity of the video is evaluated. Since different video contents contribute to different degrees of perceived quality and semantic expression, the damage of lost key frames and key content to video quality is much greater than that of ordinary frames and content. Therefore, these two types of semantic importance indicators are weighted to highlight the content value of lost data. For example, the overall key content proportion and key frame proportion are assigned preset weight coefficients, and then weighted summation is performed to obtain a comprehensive key ratio. The larger the key ratio, the more semantic key information is contained in the data lost in this transmission, and the more serious the impact on the integrity and perception quality of the video content. After that, the obtained loss ratio and the obtained critical ratio are comprehensively considered to calculate the quality loss indicator parameter as a basis for intuitively quantifying the degree of transmission quality degradation. For example, the loss ratio and the critical ratio are weighted and summed, and the configuration of the weight coefficient reflects the relative importance of the global loss level and the local semantic importance on the video quality. The obtained quality loss ratio is expressed in the form of a percentage, which intuitively presents the relative magnitude of the transmission quality impairment.

[0058] By considering the total amount of lost data while focusing on the semantic importance of the lost content, we obtain the quality loss identification parameter, which intuitively quantifies the negative impact of transmission link deterioration on video quality, and provides a reference scale for judging transmission performance and fault-tolerant recovery effects. At the same time, by embedding the quality loss identification parameter into the metadata of the repaired video, it can be transmitted and stored together with the video, building a complete quality information identification and tracking mechanism, which greatly facilitates subsequent quality control and optimization processing.

[0059] In summary, the communication method for ultra-high-definition video E-band transmission provided by the embodiment of the present application has the following technical effects:

[0060] Monitor and collect environmental characteristic data in the transmission environment of ultra-high-definition video transmission based on E-band transmission, predict environmental characteristic data and analyze transmission impact, obtain transmission impact analysis results, discover environmental factors that cause transmission quality to decline, and provide decision-making basis for subsequent transmission compensation; when the transmission impact analysis result is greater than the preset transmission impact threshold, obtain the impact timeline of the transmission impact analysis result, and index the multiple impact screen contents in the multiple affected impact frames in the ultra-high-definition video being transmitted, index the affected timeline of transmission, extract the video frames in the timeline, narrow the scope of subsequent analysis and compensation, and improve processing efficiency; perform key frame analysis and key content analysis based on multiple impact screen contents, obtain multiple key content proportions and key frame proportions, and calculate the overall key content proportion. According to the overall key content proportion and key frame proportion, decide to obtain the key content compensation proportion and key frame compensation proportion, so as to determine the proportion of compensated key frames and key contents in a targeted manner and optimize the compensation strategy. When transmitting within the affected timeline, the transmission data is monitored to obtain lost frames and lost content. According to the key content compensation ratio and key frame compensation ratio, the lost frames and lost content are compensated and generated to obtain generated frames and generated content. The key video content damaged by transmission is dynamically repaired to minimize the impact of the transmission environment and ensure the quality of video transmission. According to the loss ratio of lost frames and lost content, combined with the overall key content ratio and key frame ratio, the quality loss identification parameters are determined, the generated frames and generated content are identified, and the transmission data is added for transmission display, which intuitively reflects the situation of transmission being affected, and provides a quantifiable reference for subsequent optimization of transmission plans and quality evaluation, thereby achieving the technical effect of improving video transmission quality and reducing the loss of key frames and key content when transmitting ultra-high-definition video in the E band.

[0061] Embodiment 2 is based on the same inventive concept as the communication method for ultra-high-definition video E-band transmission in the aforementioned embodiment. Figure 2 As shown, the embodiment of the present application provides a communication platform for ultra-high-definition video E-band transmission, the platform comprising:

[0062] The transmission impact analysis module 11 is used to monitor and collect environmental characteristic data in the transmission environment for ultra-high-definition video transmission based on E-band transmission, and to perform environmental characteristic data prediction and transmission impact analysis to obtain transmission impact analysis results;

[0063] The impact identification module 12 is used to obtain the impact time axis of the transmission impact analysis result when the transmission impact analysis result is greater than a preset transmission impact threshold, and to obtain multiple impact picture contents in the multiple impact frames affected by the index in the ultra-high-definition video being transmitted;

[0064] The compensation ratio acquisition module 13 is used to perform key frame analysis and key content analysis according to the multiple influencing screen contents, obtain multiple key content proportions and key frame proportions, and calculate the overall key content proportion, and decide to obtain the key content compensation ratio and the key frame compensation ratio according to the overall key content proportion and the key frame proportion;

[0065] The compensation generation module 14 is used to monitor the transmission data, obtain the lost frames and lost content, and perform compensation generation of the lost frames and lost content according to the key content compensation ratio and the key frame compensation ratio to obtain the generated frames and generated content when transmitting within the impact time axis;

[0066] The quality loss identification module 15 is used to determine the quality loss identification parameters according to the loss ratio of the lost frames and lost content, combined with the overall key content ratio and key frame ratio, identify the generated frames and generated content, and add them to the transmission data for transmission and display.

[0067] Furthermore, the transmission impact analysis module 11 includes the following execution steps:

[0068] Monitor and collect environmental characteristic data in the transmission environment for ultra-high-definition video transmission based on E-band transmission;

[0069] Based on the historical environmental characteristic data monitoring records in the transmission environment, a sample environmental characteristic data set is collected, and environmental characteristic data after a preset time length is collected and marked as a sample predicted environmental characteristic data set;

[0070] Using the sample environmental characteristic data set and the sample predicted environmental characteristic data set, training an environmental characteristic data predictor, predicting the environmental characteristic data, and obtaining predicted environmental characteristic data;

[0071] A transmission impact analysis is performed based on the predicted environmental characteristic data to obtain a transmission impact analysis result.

[0072] Furthermore, the transmission impact analysis module 11 further includes the following execution steps:

[0073] According to the ultra-high-definition video transmission quality monitoring record in the transmission environment, according to the degree of influence on the transmission quality under different sample environment characteristic data, a set of sample transmission impact analysis results are marked;

[0074] Using the sample environment feature data set and the sample transmission impact analysis result set to train a transmission impact analyzer;

[0075] The transmission impact analyzer is used to perform transmission impact analysis on the predicted environmental characteristic data to obtain a transmission impact analysis result.

[0076] Furthermore, the compensation ratio acquisition module 13 includes the following execution steps:

[0077] According to the pre-marking of all video frames in the ultra-high-definition video, the number of key frames in the multiple influencing frames is obtained by screening, and the proportion of key contents in the multiple influencing frames is obtained to obtain the key frame proportion and the proportion of multiple key contents;

[0078] Calculate the average of the multiple key content proportions to obtain the overall key content proportion;

[0079] Collect a set of sample key frame ratios and a set of sample overall key content ratios, and set annotations based on each sample key frame ratio to obtain a set of sample key frame compensation ratios, and set annotations based on each sample overall key content ratio to obtain a set of sample key content compensation ratios;

[0080] The sample key frame ratio set and the sample key frame compensation ratio set are used to train a key frame compensation ratio classification branch, the sample overall key content ratio set and the sample key frame compensation ratio set are used to train a key content compensation ratio classification branch, and the compensation ratio classifier is obtained by combining the key frame compensation ratio classification branch;

[0081] The compensation ratio classifier is used to classify the overall key content ratio and key frame ratio to obtain the key content compensation ratio and the key frame compensation ratio.

[0082] Furthermore, the compensation generation module 14 includes the following execution steps:

[0083] When the ultra-high-definition video is transmitted within the impact timeline, the transmission data is monitored, and packet loss detection and comparison is performed with standard data to obtain lost frames and lost content;

[0084] According to the transmission loss data of ultra-high-definition video in the historical time, the frame compensator and the content compensator are trained, wherein the frame compensator and the content compensator respectively include M frame compensation branches and N content compensation branches, where M and N are integers greater than 1;

[0085] According to the key content compensation ratio and the key frame compensation ratio, the number of frame compensation branches and the number of content compensation branches are obtained by combining M and N and rounding.

[0086] Inputting the frame data before and after the lost frame into the frame compensation branches of the frame compensation branch number in the frame compensator to generate frame compensation, obtaining multiple branch generated frames, and fusing the generated frames to obtain the generated frames;

[0087] The successfully transmitted content in the frame where the lost content is located is input into the content compensation branches of the content compensation branch number in the content compensator to generate content, obtain multiple branch generated contents, and perform fusion processing to obtain generated content.

[0088] Furthermore, the compensation generation module 14 further includes the following execution steps:

[0089] According to the transmission loss data of ultra-high-definition video in historical time, the frame data sets before and after the sample and the frame data sets with sample loss are collected;

[0090] Based on a generative adversarial network, construct the M frame compensation branches, each frame compensation branch including a generative network and an adversarial network;

[0091] Randomly select M frames from the sample before and after frame data set and the sample lost frame data set to generate training data, train the M frame compensation branches respectively until convergence, obtain the trained M frame compensation branches, and obtain a frame compensator, wherein there is intersection data in the M frames to generate training data;

[0092] According to the transmission loss data of ultra-high-definition video in historical time, a sample successful transmission content set and a sample loss content set are collected;

[0093] The sample successfully transmitted content set and the sample lost content set are used to construct and train the N content compensation branches to obtain a content compensator.

[0094] Furthermore, the quality loss identification module 15 includes the following execution steps:

[0095] Calculate and obtain a loss ratio according to the lost frame and the lost content, and multiple affected picture contents in the multiple affected frames;

[0096] According to the overall key content ratio and key frame ratio, a key ratio is obtained by weighted calculation;

[0097] According to the loss ratio and the key ratio, a quality loss ratio is calculated and obtained as a quality loss identification parameter to identify the generated frame and the generated content.

[0098] Any step of the method described above can be stored as a computer instruction or program in an unlimited computer memory, and can be called and recognized by an unlimited computer processor to implement any method in the embodiments of the present application, without any unnecessary restrictions.

[0099] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalents, the present application is intended to include these modifications and variations.

Claims

1. A communication method for ultra-high definition video E-band transmission, characterized in that: The method comprises: Monitor and collect environmental characteristic data in the transmission environment for ultra-high-definition video transmission based on E-band transmission, and perform environmental characteristic data prediction and transmission impact analysis to obtain transmission impact analysis results; When the transmission impact analysis result is greater than a preset transmission impact threshold, an impact time axis of the transmission impact analysis result is obtained, and multiple impact picture contents in multiple impact frames are obtained by indexing in the ultra-high-definition video being transmitted; According to the multiple influencing screen contents, key frame analysis and key content analysis are performed to obtain multiple key content proportions and key frame proportions, and the overall key content proportion is calculated, and according to the overall key content proportion and key frame proportion, a key content compensation ratio and a key frame compensation ratio are determined; When transmitting within the impact time axis, monitoring the transmission data, obtaining lost frames and lost content, and performing compensation generation of the lost frames and lost content according to the key content compensation ratio and the key frame compensation ratio to obtain generated frames and generated content; According to the loss ratio of the lost frames and lost content, combined with the overall key content ratio and key frame ratio, a decision is made to obtain quality loss identification parameters, the generated frames and generated content are identified, and added to the transmission data for transmission and display.

2. The communication method for ultra-high definition video E-band transmission according to claim 1, characterized in that: Monitor and collect environmental characteristic data in the transmission environment of ultra-high-definition video transmission based on E-band transmission, and perform environmental characteristic data prediction and transmission impact analysis to obtain transmission impact analysis results, including: Monitor and collect environmental characteristic data in the transmission environment for ultra-high-definition video transmission based on E-band transmission; Based on the historical environmental characteristic data monitoring records in the transmission environment, a sample environmental characteristic data set is collected, and environmental characteristic data after a preset time length is collected and marked as a sample predicted environmental characteristic data set; Using the sample environmental characteristic data set and the sample predicted environmental characteristic data set, training an environmental characteristic data predictor, predicting the environmental characteristic data, and obtaining predicted environmental characteristic data; A transmission impact analysis is performed based on the predicted environmental characteristic data to obtain a transmission impact analysis result.

3. The communication method for ultra-high definition video E-band transmission according to claim 2, characterized in that: According to the predicted environmental characteristic data, a transmission impact analysis is performed, including: According to the ultra-high-definition video transmission quality monitoring record in the transmission environment, according to the degree of influence on the transmission quality under different sample environment characteristic data, a set of sample transmission impact analysis results are marked; Using the sample environment feature data set and the sample transmission impact analysis result set to train a transmission impact analyzer; The transmission impact analyzer is used to perform transmission impact analysis on the predicted environmental characteristic data to obtain a transmission impact analysis result.

4. The communication method for ultra-high definition video E-band transmission according to claim 1, characterized in that: According to the multiple influencing screen contents, key frame analysis and key content analysis are performed to obtain multiple key content proportions and key frame proportions, and the overall key content proportion is calculated, and according to the overall key content proportion and key frame proportion, a key content compensation ratio and a key frame compensation ratio are determined, including: According to the pre-marking of all video frames in the ultra-high-definition video, the number of key frames in the multiple influencing frames is obtained by screening, and the proportion of key contents in the multiple influencing frames is obtained to obtain the key frame proportion and the proportion of multiple key contents; Calculate the average of the multiple key content proportions to obtain the overall key content proportion; Collect a set of sample key frame ratios and a set of sample overall key content ratios, and set annotations based on each sample key frame ratio to obtain a set of sample key frame compensation ratios, and set annotations based on each sample overall key content ratio to obtain a set of sample key content compensation ratios; The sample key frame ratio set and the sample key frame compensation ratio set are used to train a key frame compensation ratio classification branch, the sample overall key content ratio set and the sample key frame compensation ratio set are used to train a key content compensation ratio classification branch, and the compensation ratio classifier is obtained by combining the key frame compensation ratio classification branch; The compensation ratio classifier is used to classify the overall key content ratio and key frame ratio to obtain the key content compensation ratio and the key frame compensation ratio.

5. The communication method for ultra-high definition video E-band transmission according to claim 1, characterized in that: When transmitting within the impact time axis, monitoring the transmission data, acquiring the lost frames and lost content, and performing compensation generation of the lost frames and lost content according to the key content compensation ratio and the key frame compensation ratio, including: When the ultra-high-definition video is transmitted within the impact timeline, the transmission data is monitored, and packet loss detection is compared with standard data to obtain lost frames and lost content; According to the transmission loss data of ultra-high-definition video in the historical time, the frame compensator and the content compensator are trained, wherein the frame compensator and the content compensator respectively include M frame compensation branches and N content compensation branches, where M and N are integers greater than 1; According to the key content compensation ratio and the key frame compensation ratio, the number of frame compensation branches and the number of content compensation branches are obtained by combining M and N and rounding. Inputting the frame data before and after the lost frame into the frame compensation branches of the frame compensation branch number in the frame compensator to generate frame compensation, obtaining multiple branch generated frames, and fusing the generated frames to obtain the generated frames; The successfully transmitted content in the frame where the lost content is located is input into the content compensation branches of the content compensation branch number in the content compensator to generate content, obtain multiple branch generated contents, and perform fusion processing to obtain generated content.

6. The communication method for ultra-high definition video E-band transmission according to claim 5, characterized in that: Based on the transmission loss data of ultra-high-definition video in historical time, the frame compensator and content compensator are trained, including: According to the transmission loss data of ultra-high-definition video in historical time, the frame data sets before and after the sample and the frame data sets with sample loss are collected; Based on a generative adversarial network, construct the M frame compensation branches, each frame compensation branch including a generative network and an adversarial network; Randomly select M frames from the sample before and after frame data set and the sample lost frame data set to generate training data, train the M frame compensation branches respectively until convergence, obtain the trained M frame compensation branches, and obtain a frame compensator, wherein there is intersection data in the M frames to generate training data; According to the transmission loss data of ultra-high-definition video in historical time, a sample successful transmission content set and a sample loss content set are collected; The sample successfully transmitted content set and the sample lost content set are used to construct and train the N content compensation branches to obtain a content compensator.

7. The communication method for ultra-high definition video E-band transmission according to claim 1, characterized in that: According to the loss ratio of the lost frames and the lost content, combined with the overall key content ratio and the key frame ratio, a quality loss identification parameter is determined, including: Calculate and obtain a loss ratio according to the lost frame and the lost content, and multiple affected picture contents in the multiple affected frames; According to the overall key content ratio and key frame ratio, a key ratio is obtained by weighted calculation; According to the loss ratio and the key ratio, a quality loss ratio is calculated and obtained as a quality loss identification parameter to identify the generated frame and the generated content.

8. A communication platform for ultra-high definition video E-band transmission, characterized in that: For implementing the communication method for ultra-high-definition video E-band transmission according to any one of claims 1 to 7, the platform comprises: A transmission impact analysis module, which is used to monitor and collect environmental characteristic data in a transmission environment for ultra-high-definition video transmission based on E-band transmission, and to perform environmental characteristic data prediction and transmission impact analysis to obtain transmission impact analysis results; An impact identification module, wherein the impact identification module is used to obtain an impact time axis of the transmission impact analysis result when the transmission impact analysis result is greater than a preset transmission impact threshold, and to obtain multiple impact picture contents in multiple affected frames indexed in the ultra-high-definition video being transmitted; A compensation ratio acquisition module, the compensation ratio acquisition module is used to perform key frame analysis and key content analysis according to the multiple influencing screen contents, obtain multiple key content proportions and key frame proportions, and calculate the overall key content proportion, and decide to obtain the key content compensation proportion and the key frame compensation proportion according to the overall key content proportion and the key frame proportion; A compensation generation module, the compensation generation module is used to monitor the transmission data, obtain the lost frames and lost content, and perform compensation generation of the lost frames and lost content according to the key content compensation ratio and the key frame compensation ratio to obtain the generated frames and generated content when transmitting within the impact time axis; A quality loss identification module is used to determine the quality loss identification parameters according to the loss ratio of the lost frames and lost content, combined with the overall key content ratio and the key frame ratio, identify the generated frames and generated content, and add them to the transmission data for transmission and display.

Images