Video transcoding compression method and device, electronic equipment, storage medium and program

Abstract

Embodiments of the present application disclose a video transcoding compression method and device, electronic equipment, storage medium and program, the method comprising: performing segment feature extraction on a candidate video segment of a target processing video to obtain a segment feature vector; performing content understanding and scene recognition according to video frame association information and the segment feature vector of the candidate video segment to obtain a segment understanding and recognition result of the candidate video segment; performing video slicing processing on the target processing video according to video slicing reference information, and sending each video slice to each video compression processing node; calculating a slice preset volume of each video slice according to a target volume constraint; generating slice compression parameters of each video slice according to the segment understanding and recognition result, the slice preset volume and compression reference parameters; and sending the slice compression parameters of each video slice to the corresponding video compression processing node, so that each video compression processing node performs transcoding compression processing in parallel.

Description

Technical Field

[0001] The present invention relates to the field of image processing technology, and in particular to a video transcoding and compression method, apparatus, electronic device, storage medium and program. Background Technology

[0002] With the rapid development of services such as high-definition video, ultra-high-definition video, online education, government and enterprise meetings, security monitoring, industrial inspection, and media archiving, video files are characterized by longer durations, higher resolutions, and higher frame and bit rates. The demand for batch processing, efficient transcoding, and high-definition compression of ultra-large videos is becoming increasingly prominent.

[0003] Existing video processing systems generally employ single-machine sequential transcoding or fixed-parameter batch transcoding modes. These systems configure uniform encoding parameters for the entire video, with a single transcoding process handling the compression output. While this approach is logically simple, the use of globally fixed compression parameters leads to a rigid compression strategy that cannot adapt to the varying complexity of different video segments. Simple static scenes suffer from wasted bitrate resources, while complex dynamic scenes and scenes rich in detail exhibit over-compression and loss of detail. In large-scale video processing scenarios, there are also technical issues such as long processing cycles, uneven utilization of hardware resources, and significant fluctuations in compressed video quality. These systems cannot meet the demands for rapid processing of large batches and long-duration videos, and a single failed task requires complete reprocessing, failing to satisfy the platform's requirements for batch uploading, batch archiving, and rapid video publishing. Summary of the Invention

[0004] This invention provides a video transcoding and compression method, apparatus, electronic device, storage medium, and program that can improve the quality and efficiency of video transcoding and compression processing.

[0005] According to one aspect of the present invention, a video transcoding and compression method is provided, comprising: Extract segment features from candidate video segments of the target video to obtain segment feature vectors; Based on the video frame association information of the candidate video segments and the segment feature vector, content understanding and scene recognition are performed to obtain the segment understanding and recognition result of the candidate video segments; The target video is segmented according to the video segmentation reference information of the target video, and each video segment is sent to each video compression processing node. The preset volume of each video segment is calculated based on the target volume constraint of the video segment; Based on the segment understanding and recognition results, the segment preset volume, and the compression reference parameters, the segment compression parameters for each video segment are generated; The segment compression parameters of each video segment are sent to the corresponding video compression processing node, so that each video compression processing node performs transcoding and compression processing on the video segment in parallel according to the segment compression parameters.

[0006] According to another aspect of the present invention, a video transcoding and compression apparatus is provided, comprising: The segment feature vector processing module is used to extract segment features from candidate video segments of the target video and obtain segment feature vectors. The segment understanding and recognition result processing module is used to perform content understanding and scene recognition based on the video frame association information of the candidate video segment and the segment feature vector to obtain the segment understanding and recognition result of the candidate video segment; The video segmentation processing module is used to perform video segmentation processing on the target video according to the video segmentation reference information of the target video, and send each video segment to each video compression processing node; The segment preset volume calculation module is used to calculate the preset volume of each video segment according to the target volume constraint of the video segment; The segment compression parameter generation module is used to generate segment compression parameters for each video segment based on the segment understanding and recognition results, the preset segment volume, and the compression reference parameters. The segmented parallel transcoding and compression processing module is used to send the segmented compression parameters of each video segment to the corresponding video compression processing node, so that each video compression processing node performs parallel transcoding and compression processing on the video segment according to the segmented compression parameters.

[0007] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising: At least one processor; and a memory communicatively connected to said at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to perform the video transcoding and compression method according to any embodiment of the present invention.

[0008] According to another aspect of the present invention, a computer-readable storage medium is provided, the computer-readable storage medium storing computer instructions for causing a processor to execute and implement the video transcoding and compression method according to any embodiment of the present invention.

[0009] According to another aspect of the present invention, a computer program product is also provided, comprising a computer program that, when executed by a processor, implements the video transcoding and compression method described in any embodiment of the present invention.

[0010] This invention extracts segment features from candidate video segments of a target video to obtain segment feature vectors. Then, based on the video frame association information and segment feature vectors of the candidate video segments, content understanding and scene recognition are performed to obtain segment understanding and recognition results for the candidate video segments. Further, the target video is segmented according to video segmentation reference information, and each video segment is sent to a video compression processing node. Simultaneously, a preset volume for each video segment is calculated based on the target volume constraint. This allows for the generation of segment compression parameters for each video segment based on the segment understanding and recognition results, the preset volume, and compression reference parameters. After generation, the segment compression parameters are sent to the corresponding video compression processing nodes, enabling each node to perform parallel transcoding and compression processing on the video segments according to the segment compression parameters. This technical solution replaces traditional pixel-level fixed parameter compression by using content understanding and scene recognition results of candidate video segments to generate personalized compression parameters for video segments. This achieves differentiated compression of video segments with different scenes and content complexities, completely solving the problems of uneven image quality and loss of detail caused by global parameters. Meanwhile, the distributed, sharded, parallel processing architecture overcomes the limitations of single-machine computing power, adapts to large-scale video processing scenarios, exhibits strong task fault tolerance, and significantly improves the utilization rate of computing resources. Compared to single-machine transcoding, it greatly enhances parallel processing efficiency. In summary, the technical solution of this invention can improve the quality and efficiency of video transcoding and compression processing.

[0011] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of the present invention, nor is it intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description

[0012] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0013] Figure 1 This is a flowchart of a video transcoding and compression method provided in Embodiment 1 of the present invention; Figure 2 This is a flowchart of a video transcoding and compression method provided in Embodiment 2 of the present invention; Figure 3 A complete flowchart of video transcoding and compression processing is provided. Figure 4 A schematic diagram of a closed-loop process for video transcoding and compression is provided. Figure 5 This is a schematic diagram of a video transcoding and compression device provided in Embodiment 3 of the present invention; Figure 6 This is a schematic diagram of the structure of an electronic device provided in Embodiment 4 of the present invention. Detailed Implementation

[0014] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0015] It should be noted that the terms "comprising" and "having" and any variations thereof are intended to cover non-exclusive inclusion, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are explicitly listed, but may include other steps or units that are not explicitly listed or that are inherent to such process, method, product or device.

[0016] Example 1 Figure 1 This is a flowchart of a video transcoding and compression method provided in Embodiment 1 of the present invention. This embodiment is applicable to situations where personalized compression parameters for different video segments are generated based on video segment content understanding and scene recognition results, in order to perform transcoding and compression processing on the video segments. This method can be executed by a video transcoding and compression device, which can be implemented by software and / or hardware, and is generally integrated into an electronic device. This electronic device can be a terminal device or a server device, as long as it can execute the video transcoding and compression method. The embodiments of the present invention do not limit the specific device type of the electronic device. Accordingly, as... Figure 1 As shown, the method includes the following operations: S110. Extract segment features from candidate video segments of the target video to obtain segment feature vectors.

[0017] The target video for processing can be a video that requires transcoding and compression. Optionally, the target video can be a regular video, or it can be a large-scale or ultra-large video. Ultra-large videos refer to videos whose file size exceeds a set value (e.g., several GB to TB) or whose duration exceeds a set value (e.g., several hours). For example, the target video can be long-duration conference videos, online courses, or batch video archives. Candidate video segments can be video segments extracted from the target video for analysis and processing. The segment feature vector can be a vector composed of multi-dimensional features extracted from the candidate video segments, representing the feature composition of the candidate video segments.

[0018] In this embodiment of the invention, the video transcoding and compression method proposed in this embodiment can be implemented by a video transcoding and compression processing system. The video transcoding and compression processing system may include a video access module, a segment feature extraction module, a video content scene processing module, an intelligent segmentation planning module, transcoding middleware, a segment-level parameter generation module, and a video compression task distribution module.

[0019] The video access module can receive ultra-large video files or video object addresses to obtain the target video that needs transcoding and compression. The segment feature extraction module can obtain the target video from the video access module, further process it to generate multiple candidate video segments, extract corresponding segment features from each candidate video segment, and construct a segment feature vector for each candidate video segment based on the extracted segment features. The segment features are the features extracted from the candidate video segments.

[0020] S120. Based on the video frame association information of the candidate video segment and the segment feature vector, perform content understanding and scene recognition to obtain the segment understanding and recognition result of the candidate video segment.

[0021] The video frame association information can be attribute and content information of related video frames in the candidate video segment. The segment understanding and recognition result can be obtained after performing content understanding and scene recognition on the candidate video segment, and can reflect the differences, details, and scene types of the video segment content. This embodiment of the invention does not limit the specific content included in the video frame association information and the segment understanding and recognition result.

[0022] The core of narrowband high-definition video compression is to significantly reduce the output file size while ensuring the subjective clarity of the video, under constraints of limited bandwidth or target file size. This is achieved through intelligent content recognition, reasonable bitrate allocation, adaptive adjustment of encoding parameters, quality prediction, and feedback optimization, rather than simply reducing the video bitrate. Existing traditional transcoding systems rely heavily on human experience, completing compression with fixed resolution, fixed bitrate, and fixed quality levels. They cannot dynamically adapt compression strategies based on different video content, making it difficult to meet the diverse compression needs of different types of videos, such as course recordings, government and enterprise meetings, sports events, animations, surveillance videos, and movie footage.

[0023] To address the aforementioned technical challenges, after extracting the segment feature vectors of each candidate video segment, the video content scene processing module can acquire the video frame association information and segment feature vectors of the candidate video segments. Based on these information, the module performs content understanding and scene recognition on each candidate video segment, obtaining differentiated information on different candidate video segments in multiple dimensions such as content and scene. This results in segment understanding and recognition of each candidate video segment, serving as an important basis for generating subsequent adaptive compression strategies.

[0024] S130. Perform video segmentation processing on the target video according to the video segmentation reference information of the target video, and send each video segment to each video compression processing node.

[0025] The video segmentation reference information can be information used to segment the target video for processing. This information may include, but is not limited to, the total video duration, segment duration, key frame locations, scene change locations, and the computational resources of the video compression processing nodes. It may also include segment understanding and recognition results of candidate video segments. Any information that can be used as reference information for segmenting the target video for processing is acceptable. This embodiment of the invention does not limit the specific type and content of the video segmentation reference information. The video compression processing nodes can be computational nodes that perform transcoding and compression processing on the video segments. The number of video compression processing nodes can be multiple.

[0026] Current distributed video transcoding and compression systems improve transcoding throughput to some extent by segmenting video files into fixed-duration segments and distributing them across multiple nodes for parallel processing. However, this slicing method relies solely on time as the basis for slicing, neglecting the structural features of the video frame and ignoring key content characteristics such as critical frame boundaries, scene transitions, motion complexity, image noise intensity, subtitle text areas, main subjects, and quality-sensitive areas. This results in numerous image quality defects when different video segments are processed with the same compression parameters, including inconsistent image clarity, localized blurring, edge ringing, dark block artifacts, and abnormal seams between merged segments.

[0027] To address the aforementioned technical challenges, the video transcoding and compression system utilizes an intelligent segmentation planning module. This module combines diverse video segmentation reference information, such as the total video duration, segment understanding and recognition results, video segment duration, key frame locations, scene change locations, and the computational resources of the video compression processing nodes, to segment the target video and generate a segmentation plan, resulting in multiple video segments. These generated video segments take into account factors such as video content, structure, and keyframe boundaries, exhibiting characteristics such as stable content, inclusion of key frame content, or inclusion of content from the same scene. This reduces the risk of stuttering, duplicate frames, dropped frames, or audio-visual asynchrony during subsequent video segment synthesis.

[0028] Accordingly, each video segment can correspond to a video segment processing task, and these tasks can be distributed to the corresponding video compression processing nodes through a scheduling mechanism. Optionally, a video compression processing node can process one or more video segment processing tasks, performing transcoding and compression processing on the received video segments. Parallel transcoding and compression processing of video segments can significantly shorten the transcoding and compression cycle of ultra-large videos, improve the system's batch throughput, reduce task failure losses, and enhance system operational stability.

[0029] S140. Calculate the preset volume of each video segment according to the target volume constraint of the video segment.

[0030] The target volume constraint is the pre-set volume constraint condition for the video segment. The preset segment volume is the estimated volume for the video segment.

[0031] In narrowband high-definition scenarios, video transcoding and compression systems typically need to meet target size or target bandwidth constraints. Understandably, the size constraints required for video segments differ across different video compression scenarios. Therefore, to avoid the problem of insufficient image quality in complex segments and wasted bitrate in simple segments due to the average bitrate allocation across all video segments, the video transcoding and compression system can dynamically calculate the preset size of each video segment based on the target size constraints through transcoding middleware. This allows for prioritizing the allocation of limited bitrate to segments that are more visually sensitive or have more complex content, while keeping the overall video file size under control, thereby improving the overall subjective clarity of the video.

[0032] Since the size of a video segment is directly related to its duration and bitrate, calculating the preset size of each video segment is equivalent to budgeting the bitrate of the video segment.

[0033] S150. Based on the segment understanding and recognition results, the segment preset volume, and the compression reference parameters, generate segment compression parameters for each video segment.

[0034] The compression reference parameters can be any parameters that can be used for video transcoding and compression processing, other than the segment understanding and recognition results and the preset segment size. Examples include target resolution and service type. This embodiment of the invention does not limit the specific parameter type of the compression reference parameters. Segment compression parameters are the compression parameters for video segments.

[0035] Correspondingly, the video transcoding and compression system can use a segment-level parameter generation module to comprehensively analyze the segment understanding and recognition results, the preset segment size, and the compression reference parameters. It can then comprehensively reason about the differences in scenarios and content involved in different video segments, generating independent segment compression parameters for each segment. That is, the compression parameters for different video segments may also differ. Relying on the semantic-level content understanding capabilities of the video segments, a semantically driven segment-based differentiated compression strategy can be generated, ensuring accurate and controllable video compression quality.

[0036] For example, segmented compression parameters may include, but are not limited to, output resolution, quality control factor, noise reduction intensity, motion protection weight, text protection weight, subject area protection strategy, audio compression parameters, and encapsulation strategy. Output resolution is the resolution after video segmented compression; the quality control factor (CRF) is a core parameter controlling video compression clarity and balancing image quality and file size; noise reduction intensity refers to the strength of eliminating image noise such as graininess, snowflakes, and color distortion; motion protection weight is a temporal noise reduction parameter used to resolve conflicts between noise reduction and image motion; text protection weight and motion protection weight are a set of logic specifically designed to protect subtitles, text, signs, and road sign text in the image, preventing noise reduction from blurring, blurring, edge-shaping, or making the text illegible; the subject area protection strategy determines the uniform basic protection intensity for all subjects in the video image, uniformly controlling the basic noise reduction protection intensity for faces, objects, text, and moving subjects; audio compression parameters can be audio encoding, bitrate, sampling rate, and audio channels, etc.; the encapsulation strategy can be the method of packaging and encapsulating the image, sound, subtitles, chapters, and metadata in the video. The transcoding middleware performs video segmentation processing based on the aforementioned segmentation compression parameters. Because each video segment has an independent compression strategy, the video transcoding and compression system can achieve true segment-level adaptive narrowband high-definition compression, rather than using a single fixed strategy for the entire video.

[0037] S160. Send the segment compression parameters of each video segment to the corresponding video compression processing node, so that each video compression processing node performs transcoding and compression processing on the video segment in parallel according to the segment compression parameters.

[0038] After generating the segmented compression parameters, the video compression task distribution module sends the segmented compression parameters for each video segment to the corresponding video compression processing node. Upon receiving the video segment and its corresponding adapted compression parameters, each video compression processing node initiates a video transcoding and compression command through its transcoding middleware. Responding to the command, each node performs transcoding and compression operations according to the assigned parameters, including image scaling, quality control, noise reduction, encoding, audio compression, timestamp preservation, and intermediate encapsulation. After processing, the nodes can upload the transcoding and compression results and simultaneously return information such as processing logs, output size, output duration, average bitrate, initial quality assessment results, anomalies, and runtime. Each node can perform transcoding and compression operations in parallel.

[0039] This invention extracts segment features from candidate video segments of a target video to obtain segment feature vectors. Then, based on the video frame association information and segment feature vectors of the candidate video segments, content understanding and scene recognition are performed to obtain segment understanding and recognition results for the candidate video segments. Further, the target video is segmented according to video segmentation reference information, and each video segment is sent to a video compression processing node. Simultaneously, a preset volume for each video segment is calculated based on the target volume constraint. This allows for the generation of segment compression parameters for each video segment based on the segment understanding and recognition results, the preset volume, and compression reference parameters. After generation, the segment compression parameters are sent to the corresponding video compression processing nodes, enabling each node to perform parallel transcoding and compression processing on the video segments according to the segment compression parameters. This technical solution replaces traditional pixel-level fixed parameter compression by using content understanding and scene recognition results of candidate video segments to generate personalized compression parameters for video segments. This achieves differentiated compression of video segments with different scenes and content complexities, completely solving the problems of uneven image quality and loss of detail caused by global parameters. Meanwhile, the distributed, sharded, parallel processing architecture overcomes the limitations of single-machine computing power, adapts to large-scale video processing scenarios, exhibits strong task fault tolerance, and significantly improves the utilization rate of computing resources. Compared to single-machine transcoding, it greatly enhances parallel processing efficiency. In summary, the technical solution of this invention can improve the quality and efficiency of video transcoding and compression processing.

[0040] Example 2 Figure 2This is a flowchart of a video transcoding and compression method provided in Embodiment 2 of the present invention. This embodiment is based on the above embodiment and is further specified. In this embodiment, various specific and optional implementation methods are given for extracting segment features from candidate video segments of the target video, performing content understanding and scene recognition based on the video frame association information and segment feature vectors of the candidate video segments, performing video segmentation processing on the target video, and calculating the preset volume of each video segment. Implementation methods for controlling the video compression quality are also given. Correspondingly, as... Figure 2 As shown, the method in this embodiment may include: S210. Extract key sampled video frames from the target processed video based on the video basic information of the target processed video, and generate multiple candidate video segments based on the key sampled video frames.

[0041] Among them, the key sampled video frame can be the key video frame extracted from the target processing video.

[0042] Optionally, the video transcoding and compression processing system can also be configured with a media metadata parsing module. After receiving the target video through the video access module, the media metadata parsing module can read the basic video information of the target video. Optionally, the basic video information may include, but is not limited to, total video duration, resolution, frame rate, original bitrate, audio information, subtitle information, color information, image orientation information, and encapsulation information. Subsequently, the segment feature extraction module can extract several representative video frames from the video timeline of the target video according to a preset sampling interval based on the basic video information, and combine the extracted key sampled video frames according to the chronological order of the video frames to generate multiple candidate video segments. For example, every 10 key sampled video frames can be combined to generate one candidate video segment.

[0043] S220. Generate segment feature vectors for each candidate video segment based on the video frame association information of the key sampled video frames.

[0044] When the segment feature extraction module extracts key sampled video frames, it can simultaneously extract the audio transcription text, subtitle text, and basic image statistical features of the key sampled video frames as video frame association information. For each candidate video segment, the segment feature extraction module can extract features such as motion complexity, texture complexity, noise intensity, text density, subject sensitivity, and brightness anomaly of the entire candidate video segment to form a segment feature vector. Here, X_i represents the segment feature vector of the i-th candidate video segment, m_i represents motion complexity, u_i represents texture complexity, z_i represents noise intensity, l_i represents text density, s_i represents subject sensitivity, e_i represents brightness anomaly, and W, H, and F represent video width, height, and frame rate, respectively. Motion complexity, texture complexity, noise intensity, text density, subject sensitivity, and brightness anomaly can all be calculated using relevant methods. In this way, the original video can be transformed from a single media file into a computable, analyzable, and schedulable set of segment-level features, providing input for subsequent semantic judgment of video segments, intelligent segmentation planning, and narrowband high-definition compression parameter generation.

[0045] S230. Input the candidate video segment, the video frame association information of the candidate video segment, and the segment feature vector into the target large model, so as to output the scene type to which the candidate video segment belongs through the target large model.

[0046] The target large model can be any available large language model (LLM) used for content understanding and scene recognition of candidate video clips. Large models, also known as large language models or large-scale language models, refer to deep learning models trained with a large amount of relevant data (such as text data, speech data, or combined text-image data) capable of processing a sequence (such as text, speech, image, audio-visual, or combined text-image sequences). They can generate natural language text or understand the meaning of language text. These models typically have billions of parameters. Large models can handle various natural language tasks, such as text classification, question answering, and dialogue, and have wide applications. The input to a large model is data, such as text data, speech data, or combined text-image data. The large model encodes the input data to obtain corresponding word vector representations, and then decodes the encoded word vectors to automatically complete the input data processing and obtain the corresponding output data. For example, text can be input into a large model, which processes and predicts the input text, outputting the corresponding response text.

[0047] Multimodal large-scale models possess excellent capabilities in video semantic understanding, scene classification, content recognition, and strategy generation. They can comprehensively analyze multi-dimensional information such as video frame extraction, audio transcription, subtitle text, screen motion states, and image quality metrics. Therefore, in this embodiment of the invention, the video content scene processing module can be a large-scale model content understanding module, which can call the target large-scale model to perform content understanding and scene recognition, and output the segment understanding and recognition results of candidate video segments. Specifically, the video frame association information (such as subtitle text and audio transcription text) and segment feature vectors of subsequent video segments and candidate video segments can be input into the target large-scale model to perform semantic analysis on the candidate video segments, identify, and output the scene type to which the candidate video segments belong. For example, scene types may include, but are not limited to, meeting portraits, course screen recordings, screen sharing, sports, animation, surveillance footage, natural scenery, and dark scenes in movies.

[0048] S240. Input the video image information of the candidate video segment into the target large model, so as to output the segment complexity, image quality sensitivity, compression tolerance, subject protection suggestions, text protection suggestions and motion preservation suggestions of the candidate video segment through the target large model.

[0049] Simultaneously, video image information from candidate video clips can be input into the target large model to identify visually sensitive areas in the video image, including but not limited to human faces, text areas, product details, screen-shared areas, and moving subject areas. Ultimately, the target large model can output the segment complexity, image quality sensitivity, compression tolerance, subject protection suggestions, text protection suggestions, and motion preservation suggestions for the candidate video clips based on the identification results of visually sensitive areas. Segment complexity refers to the complexity of the video content of the candidate video clip, which can include spatial and temporal complexity. Spatial complexity reflects the complexity of the texture of a single frame of video image, while temporal complexity reflects the degree of change between frames. Image quality sensitivity can be an adjustment parameter simulating human visual judgment, used to simulate the strength of human perception of image distortion, thereby determining the differentiated bitrate allocation. Compression tolerance is the maximum allowable degree of compression defects (color blocks, blur, noise, and smearing, etc.) in the image. Compression tolerance controls the overall compression limit; a higher value means more aggressive compression and smaller file size, but also a higher chance of blurry color blocks in the image. A lower value means more conservative compression, resulting in a cleaner and clearer image, but a larger file size. Subject protection suggestions can provide noise reduction and compression protection levels tailored to different scene scenarios, targeting core subjects such as people, animals, products, and objects. Text protection suggestions can provide protection recommendations for various types of text generated in the video, specifically for video subtitles, PPT slides, and logo text, adapting to different materials to avoid blurry text. Motion retention suggestions can provide references for preventing motion blur and ghosting caused by noise reduction in moving subjects and camera movements.

[0050] The above technical solution combines the intelligent content understanding capabilities of large models with video transcoding technology, which can overcome the limitations of traditional transcoding technology and build a new ultra-large video transcoding and compression solution with intelligent content recognition, fine scheduling, multi-node parallel processing and adaptive narrowband high-definition compression capabilities.

[0051] S250. Generate video segmentation points for the target processed video based on the video segmentation reference information of the target processed video.

[0052] The video segmentation reference information may include the total video duration, the segment understanding and recognition results, the video segment duration, key frame locations, scene change locations, and the computational resources of the video compression processing nodes. The video segmentation point can be a time reference point for segmenting the target video.

[0053] S260. Perform video segmentation processing on the target video according to the video segmentation points, and send each video segment to each video compression processing node.

[0054] The video segmentation point can be a stable time point in the target video, a time point near a key scene, or a scene transition boundary time point. A stable time point is the point in time where stable video content can be divided into a video segment. A time point near a key scene can be a time point near a key scene. A scene transition boundary time point can be a time point near a scene transition boundary in the video frame.

[0055] The intelligent segmentation planning module generates video segmentation points for the target video based on the total video duration, segment understanding and recognition results, video segment duration, key frame positions, scene change positions, and the computing resources of the video compression processing nodes. It then performs video segmentation processing on the target video according to these determined segmentation points and sends each generated video segment to its corresponding video compression processing node. For example, the intelligent segmentation planning module can divide consecutive video frames with high image quality sensitivity into a single video segment, such as consecutive video frames containing portraits or PPT (PowerPoint) content; it can also divide consecutive video frames with low compression tolerance into a single video segment, such as consecutive video frames with dynamic backgrounds, high motion, high texture, or minimal repetition.

[0056] Currently, existing video transcoding and compression methods typically use fixed-duration video slicing. This method of video slicing does not take into account the video frame structure and keyframe boundaries, which can easily lead to problems such as discontinuous timestamps at video slice splicing points, audio-visual synchronization offset, and sudden changes in image quality.

[0057] Unlike traditional fixed-duration video slicing, this invention prioritizes video segmentation points based on stable content, key frames, or scene transition boundaries. This reduces the risk of stuttering, duplicate frames, dropped frames, or audio-visual desynchronization during subsequent video compositing. The intelligent segmentation planning module records the start and end times, original timestamp, segment features, scene type, compression strategy, target volume budget, target quality threshold, task priority, and retry strategy for each video segment during segment generation. Each video segment has independent processing capabilities and an independent processing state. It can be distributed to different video compression processing nodes for execution and can also be retried individually on a segment-by-segment basis after transcoding and compression failures, without needing to reprocess the entire video, thus significantly improving overall efficiency and system stability.

[0058] In an optional embodiment of the present invention, before sending each video segment to each of the video compression processing nodes, the method may further include: obtaining node association information of the video compression processing nodes; wherein the node association information includes the load and computing power of the video compression processing nodes; and determining the video compression processing node corresponding to each video segment based on the node association information and the expected processing cost of the video segment.

[0059] Among them, node association information refers to the status and performance information of the video compression processing node.

[0060] Optionally, the video transcoding and compression processing system can also be configured with a distributed task scheduling module, used to achieve distributed task scheduling and node load balancing based on the node association information of each video compression processing node. Specifically, the distributed task scheduling module can obtain node association information such as the current load and computing power of each video compression processing node, and determine the estimated processing cost of video segments by combining the queue length and historical processing speed of each video compression processing node. Thus, based on the load and computing power of the video compression processing nodes and the estimated processing cost of the video segments, the corresponding video compression processing nodes are determined.

[0061] For example, the distributed task scheduling module can determine the video compression processing node assigned to each video segment based on the following formula: Here, node(i) represents the video compression processing node to which the i-th video slice is assigned, argmin() represents the independent variable (input value) or its index position when the objective function reaches its minimum value, Load_j represents the current load of video compression processing node j, cost_i represents the expected processing cost of video slice i, and cap_j represents the computing power of video compression processing node j. Through this scheduling method, the video transcoding and compression processing system can avoid the trailing problem caused by complex video slices being concentrated on the same video transcoding and compression processing node, and can also avoid the situation where high-performance video transcoding and compression processing nodes are idle while low-performance video transcoding and compression processing nodes are congested. If a video transcoding and compression processing node fails, times out, or returns an abnormal result, the distributed task scheduling module can resubmit the corresponding video slice to other video transcoding and compression processing nodes and reuse completed qualified video slices, thereby reducing the failure cost in the process of processing ultra-large videos.

[0062] S270. Calculate the preset volume of each video segment based on the target volume constraint of the video segment.

[0063] In an optional embodiment of the present invention, calculating the preset volume of each video segment based on the target volume constraint of the video segment may include: determining the segment duration of each video segment and calculating the overall complexity of each video segment; wherein the overall complexity includes motion complexity, texture complexity, noise intensity, text density, subject sensitivity, and image quality sensitivity; and calculating the preset volume of each video segment based on the target volume budget, segment duration, and overall complexity of the video segment.

[0064] For example, the preset size of each video segment can be calculated based on the following formula: Where B_i represents the preset size of the i-th video segment, i.e., the target size budget; B_total represents the target output size of the entire target video segment; T_i represents the duration of the i-th video segment; and C_i represents the overall complexity of the i-th video segment. T_j represents the duration of the j-th video segment, and C_j represents the overall complexity of the j-th video segment. This represents the overall complexity of the target video. Overall complexity is determined by motion complexity, texture complexity, noise intensity, text density, subject sensitivity, and image quality sensitivity. For example, the values ​​of motion complexity, texture complexity, noise intensity, text density, subject sensitivity, and image quality sensitivity can be calculated using relevant formulas, and then weighted and fused to obtain the overall complexity value. Video segments with higher complexity, denser text, more important subjects, or more intense motion will receive a higher volume budget; static images, repetitive images, and low-texture video segments will be allocated a lower volume budget. The preset volume of a segment is related to the duration and bitrate of the video segment.

[0065] S280. Based on the segment understanding and recognition results, the segment preset volume, and the compression reference parameters, generate segment compression parameters for each video segment.

[0066] For example, compression reference parameters may specifically include parameters such as target clarity and business type.

[0067] For example, for video segments with high text density or obvious main characters, the video transcoding and compression system will increase the priority of image quality protection, thereby increasing the sensitivity of image quality; for video segments with static backgrounds, low motion, low texture, or many repeated images, the video transcoding and compression system will increase the compression tolerance of the video segments.

[0068] Based on the segment understanding and recognition results, the preset segment size, and compression reference parameters, the system generates differentiated segment compression parameters for each video segment. For videos such as meetings, courses, and screen sharing, the video transcoding and compression system prioritizes ensuring the clarity of faces, subtitles, PPT text, and screen content. For high-motion videos such as sports, performances, and games, the system improves motion retention to avoid ghosting and blocky distortion in fast-moving areas. For surveillance, fixed-camera, and low-motion videos, the system appropriately increases compression intensity to achieve a smaller file size. For noisy videos, the system performs moderate noise reduction before compression to reduce noise's impact on bitrate.

[0069] The personalized processing of the above-mentioned segmented compression parameters can significantly improve the subjective clarity of the video under the same compressed volume, eliminate the problem of wasted bitrate in simple scenes and blurred image quality in complex scenes, and significantly improve the uniformity of image quality.

[0070] S290. Send the segment compression parameters of each video segment to the corresponding video compression processing node, so that each video compression processing node performs transcoding and compression processing on the video segment in parallel according to the segment compression parameters.

[0071] In an optional embodiment of the present invention, the above method may further include: obtaining the segmented compression processing association results of each of the video compression processing nodes; calculating the objective image quality index, sharpness preservation index, compression quality judgment result, bitrate saving benefit, and deviation penalty item of the video segmented compression result based on the segmented compression processing association results; comprehensively calculating the comprehensive compression quality score of each of the video segments based on the objective image quality index, sharpness preservation index, compression quality judgment result, bitrate saving benefit, and deviation penalty item; and performing parameter rollback processing on the segmented compression parameters of the video segment corresponding to the target comprehensive compression quality score when it is determined that the target comprehensive compression quality score is lower than a set threshold.

[0072] The target overall compression quality score can be an overall compression quality score that is lower than a set threshold.

[0073] For example, the associated results of segmented compression processing may include, but are not limited to, processing logs, output volume, output duration, average bitrate, initial quality assessment results, anomaly information, and running time generated synchronously when video segments are transcoded and compressed.

[0074] Currently, existing video transcoding and compression methods lack segment-level quality prediction, anomaly retry, and parameter self-learning optimization mechanisms. Video quality issues can only be detected after the entire transcoding process is completed, resulting in extremely high costs for troubleshooting and rework.

[0075] To address the aforementioned technical challenges, the video transcoding and compression system can write the processing logs, output size, output duration, average bitrate, initial quality assessment results, anomaly information, and runtime—all synchronously generated during the transcoding and compression of video segments—into a task status table. This data is used for subsequent quality assessment, anomaly tracking, and scheduling optimization. Correspondingly, the system can perform quality assessments on each video segment based on the associated compression results. Assessment metrics may include, but are not limited to, the degree of sharpness preservation, subject area quality, text readability, motion area stability, bitrate savings, output duration deviation, timestamp continuity, and compression quality judgment results.

[0076] Specifically, the video transcoding and compression system can calculate objective image quality indicators, sharpness preservation indicators, compression quality judgment results, bitrate savings, and deviation penalties for the video segment compression results based on the correlation results of segment compression processing. Then, based on these objective image quality indicators, sharpness preservation indicators, compression quality judgment results, bitrate savings, and deviation penalties, it comprehensively calculates the overall compression quality score for each video segment. For example, the formula for calculating the overall compression quality score can be: Wherein, S_i represents the overall compression quality score of the i-th video segment; V_i represents the objective image quality index, which is a standardized value calculated by an algorithm without relying on human subjective perception to quantify the clarity and distortion of the video image; M_i represents the sharpness preservation index, which is a quantitative reference parameter used to measure the degree of preservation of image details and edge sharpness during transcoding and noise reduction; L_i represents the compression quality judgment result, which can be the subjective quality judgment result of a large model; R_i represents the bitrate saving benefit, which can be the percentage of data reduction after optimizing compression parameters compared to the original bitrate under the same visual quality, used to measure the effect of compression saving traffic and storage space; D_i represents the time axis or quality deviation penalty term, where α, β, γ, δ and ξ are weight parameters, and the sum of the weight parameters is 1.

[0077] When the overall compression quality score of a video segment falls below a set threshold, the video transcoding and compression system will not reprocess the entire target video. Instead, it will only perform parameter rollback operations on the abnormal video segment. Rollback methods may include, but are not limited to, reducing compression intensity, increasing target bitrate, reducing noise reduction intensity, enhancing text protection, enhancing main area protection, adjusting segment boundaries, or reselecting the output resolution. Through this quality closed-loop mechanism, low-quality video segments can be identified and locally repaired before the final synthesis of each video segment, avoiding large-scale rework caused by discovering quality problems only after the entire video has been transcoded.

[0078] After all video segments pass quality assessment, the video transcoding and compression system sorts the video segments according to the original timeline order and verifies the continuity of video frames, audio sampling points, and timestamps between adjacent video segments. The system focuses on checking for issues such as duplicate frames, dropped frames, audio holes, audio-visual misalignment, abnormal segment durations, and inconsistent encapsulation. If a slight time deviation exists at a seam, the system can align it using a timeline correction mechanism; if the deviation exceeds a threshold, the corresponding video segment is reprocessed. For the final output video, the system verifies that the total output duration matches the original video duration and generates a final quality report. These operations ensure that video segments processed in parallel across multiple nodes can be stably synthesized into a continuous video, avoiding playback seam anomalies caused by distributed processing, completely resolving video segment splicing defects, and resulting in smooth and stable output video playback without stuttering, dropped frames, audio-visual misalignment, or seam flickering.

[0079] After completing video synthesis, the video transcoding and compression system can output a single video file or segmented on-demand video products according to business needs. Simultaneously, it generates a quality report, which may include, but is not limited to, original video information, output video information, comparison of file sizes before and after compression, overall compression rate, average bitrate, quality scores for each segment, abnormal retry records, processing time, node usage, and final verification results. For business platforms, this quality report can be used for transcoding result acceptance, problem tracking, quality auditing, and subsequent strategy optimization. For large-scale video processing scenarios, this report can also serve as a data source for model training and scheduling optimization, enabling the video transcoding and compression system to continuously accumulate optimal compression strategies for different content types during continuous operation.

[0080] To more clearly illustrate the technical solutions provided in the embodiments of the present invention, Figure 3 This document provides a complete flowchart illustrating the video transcoding and compression process. In a specific example, such as... Figure 3As shown, the video transcoding and compression system first receives a large video and parses its basic media information. Then, it performs multimodal sampling to extract segment features such as sampled frames, subtitles, audio transcription, motion, texture, and noise. Next, a large model identifies video scenes and quality-sensitive areas, generating content understanding results. Processing the generated content understanding results first avoids using a fixed compression strategy for the entire video. Based on the recognition results, the system intelligently plans segmentation, forming independently processable segment tasks, which are then distributed to different video compression processing nodes (referred to as nodes) via a distributed scheduling module. Each node calls the transcoding middleware to complete segment processing in parallel, while simultaneously generating segment-level narrowband high-definition parameters. Video segmentation tasks are independent, retryable, and transferable to reduce trailing time. After processing, the system evaluates the segment quality, performing feedback rollback and partial retransmission on low-quality video segments, while qualified video segments can be directly reused. Finally, the system completes timeline alignment, continuous compositing, and outputs the finished video and a quality report.

[0081] Figure 4 A schematic diagram of a closed-loop process for video transcoding and compression is provided. In a specific example, such as... Figure 4 As shown, the video transcoding and compression system first inputs features such as motion, texture, noise, text density, and subject sensitivity for each video segment, and then combines these features with a large model to predict semantic sensitivity and model segment complexity. Subsequently, under a target volume constraint, the system allocates bitrate budgets to different segments and generates corresponding parameters such as resolution, quality factor, noise reduction intensity, motion protection, and text protection. The transcoding middleware performs compression processing based on the segment parameters. After completion, the video transcoding and compression system performs a comprehensive quality score on the output segments and determines whether they meet the quality threshold. If the segment meets the threshold, it is stored in the database for compositing; if it does not, a parameter rollback process is initiated, adjusting the compression parameters and reprocessing the segment. For example, the compression intensity can be reduced, the bitrate increased, or noise reduction and text protection parameters adjusted, thereby ensuring overall image quality while controlling file size.

[0082] The aforementioned technical solution achieves semantic-driven segmented differential compression strategy generation by evaluating video segment complexity based on large-model content understanding and integrating multi-dimensional information such as image motion, texture, noise, text density, subject sensitivity, subtitles, and audio semantics. It allocates video segment volume budgets under the dual constraints of target volume and target clarity, and dynamically allocates bitrate resources based on segment complexity and duration, achieving refined and rational bitrate utilization. Through an intelligent video segmentation planning method, within the standard segment duration, it comprehensively selects the optimal segmentation boundary by considering key image boundaries, scene stability, and cluster resource load, avoiding splicing anomalies from the source. Combined with dynamic optimization through a quality closed-loop system, it maximizes the preservation of video image details and achieves high-definition compression effects in narrowband scenarios with limited bandwidth and file size. Through a segment-level adaptive parameter generation and execution mechanism, it adaptively generates exclusive encoding and compression parameters such as resolution, quality factor, noise reduction intensity, and text and motion protection weights based on image quality sensitivity, compression tolerance, and video scene type. Simultaneously, by combining segment-level quality assessment and feedback rollback mechanisms, the system automatically identifies low-quality segments and completes local retransmission and repair through parameter iterative optimization, avoiding complete rework and eliminating the need for overall re-transcoding and compression. This significantly reduces the cost and time loss in repairing video quality issues. Employing a multi-node parallel transcoding and continuous timeline synthesis mechanism ensures seamless splicing of segment data processed independently by multiple nodes, strictly controlling seam and duration deviations to improve the stability of the final video. Through end-to-end segment boundary optimization, timeline alignment verification, and splicing deviation control, common issues in multi-node parallel transcoding and compression, such as seam anomalies, audio-visual misalignment, and frame drops, are completely resolved, resulting in stable video output quality. The aforementioned video transcoding and compression processing system adopts a modular architecture design, with each functional module being independent, adaptable, replaceable, and iterative. This facilitates customized deployment and functional expansion for different business scenarios, making it suitable for applications in various industries such as government, enterprise, education, security, and media.

[0083] It should be noted that the relevant information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for display, data used for analysis, etc.) involved in this disclosure are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data comply with the relevant laws, regulations and standards of the relevant regions, and do not violate public order and good morals.

[0084] It should be noted that any arrangement or combination of the technical features in the above embodiments also falls within the protection scope of this invention.

[0085] Example 3 Figure 5 This is a schematic diagram of a video transcoding and compression device provided in Embodiment 3 of the present invention, as shown below. Figure 5As shown, the device includes: a segment feature vector processing module 510, a segment understanding and recognition result processing module 520, a video segmentation processing module 530, a segment preset volume calculation module 540, a segment compression parameter generation module 550, and a segment parallel transcoding and compression processing module 560, wherein: The segment feature vector processing module 510 is used to extract segment features from candidate video segments of the target video to obtain segment feature vectors; The segment understanding and recognition result processing module 520 is used to perform content understanding and scene recognition based on the video frame association information of the candidate video segment and the segment feature vector to obtain the segment understanding and recognition result of the candidate video segment; The video segmentation processing module 530 is used to perform video segmentation processing on the target video according to the video segmentation reference information of the target video, and send each video segment to each video compression processing node; The segment preset volume calculation module 540 is used to calculate the segment preset volume of each video segment according to the target volume constraint of the video segment; The segment compression parameter generation module 550 is used to generate segment compression parameters for each video segment based on the segment understanding and recognition results, the segment preset volume, and the compression reference parameters. The segmented parallel transcoding and compression processing module 560 is used to send the segmented compression parameters of each video segment to the corresponding video compression processing node, so that each video compression processing node performs parallel transcoding and compression processing on the video segment according to the segmented compression parameters.

[0086] This invention extracts segment features from candidate video segments of a target video to obtain segment feature vectors. Then, based on the video frame association information and segment feature vectors of the candidate video segments, content understanding and scene recognition are performed to obtain segment understanding and recognition results for the candidate video segments. Further, the target video is segmented according to video segmentation reference information, and each video segment is sent to a video compression processing node. Simultaneously, a preset volume for each video segment is calculated based on the target volume constraint. This allows for the generation of segment compression parameters for each video segment based on the segment understanding and recognition results, the preset volume, and compression reference parameters. After generation, the segment compression parameters are sent to the corresponding video compression processing nodes, enabling each node to perform parallel transcoding and compression processing on the video segments according to the segment compression parameters. This technical solution replaces traditional pixel-level fixed parameter compression by using content understanding and scene recognition results of candidate video segments to generate personalized compression parameters for video segments. This achieves differentiated compression of video segments with different scenes and content complexities, completely solving the problems of uneven image quality and loss of detail caused by global parameters. Meanwhile, the distributed, sharded, parallel processing architecture overcomes the limitations of single-machine computing power, adapts to large-scale video processing scenarios, exhibits strong task fault tolerance, and significantly improves the utilization rate of computing resources. Compared to single-machine transcoding, it greatly enhances parallel processing efficiency. In summary, the technical solution of this invention can improve the quality and efficiency of video transcoding and compression processing.

[0087] Optionally, the segment feature vector processing module 510 is further configured to: extract key sampled video frames from the target processed video based on the video basic information of the target processed video, and generate multiple candidate video segments based on the key sampled video frames; generate segment feature vectors for each candidate video segment based on the video frame association information of the key sampled video frames; wherein the segment feature vectors include motion complexity, texture complexity, noise intensity, text density, subject sensitivity, and brightness anomaly.

[0088] Optionally, the segment understanding and recognition result processing module 520 is further configured to: input the candidate video segment, the video frame association information of the candidate video segment, and the segment feature vector into the target large model, so as to output the scene type to which the candidate video segment belongs through the target large model; input the video screen information of the candidate video segment into the target large model, so as to output the segment complexity, image quality sensitivity, compression tolerance, subject protection suggestions, text protection suggestions, and motion preservation suggestions of the candidate video segment through the target large model.

[0089] Optionally, the video segmentation processing module 530 is further configured to: generate video segmentation points for the target processed video based on the video segmentation reference information of the target processed video; wherein, the video segmentation reference information includes the total video duration, the segment understanding and recognition result, the video segment duration, key frame positions, scene change positions, and the computing resources of the video compression processing nodes; and perform video segmentation processing on the target processed video based on the video segmentation points; wherein, the video segmentation points are the stable time points of the screen content of the target processed video, the time points near the key frames, or the time points at the scene switching boundaries.

[0090] Optionally, the above apparatus further includes a video compression processing node allocation module, configured to: obtain node association information of video compression processing nodes; wherein the node association information includes the load and computing power of the video compression processing node; and determine the video compression processing node allocated to the video segment based on the node association information and the expected processing cost of the video segment.

[0091] Optionally, the segment preset volume calculation module 540 is further configured to: determine the segment duration of each video segment and calculate the comprehensive complexity of each video segment; wherein the comprehensive complexity includes motion complexity, texture complexity, noise intensity, text density, subject sensitivity and image quality sensitivity; and calculate the segment preset volume of each video segment based on the target volume budget, segment duration and comprehensive complexity of the video segment.

[0092] Optionally, the above apparatus further includes a compression quality evaluation module, configured to: obtain the segmented compression processing association results of each of the video compression processing nodes; calculate the objective image quality index, sharpness preservation index, compression quality judgment result, bitrate saving benefit, and deviation penalty item of the video segmented compression result based on the segmented compression processing association results; comprehensively calculate the comprehensive compression quality score of each of the video segments based on the objective image quality index, sharpness preservation index, compression quality judgment result, bitrate saving benefit, and deviation penalty item; and perform parameter rollback processing on the segmented compression parameters of the video segment corresponding to the target comprehensive compression quality score when the target comprehensive compression quality score is determined to be lower than a set threshold.

[0093] The video transcoding and compression apparatus described above can execute the video transcoding and compression method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects for executing the method. Technical details not described in detail in this embodiment can be found in the video transcoding and compression method provided in any embodiment of the present invention.

[0094] Since the video transcoding and compression apparatus described above is capable of executing the video transcoding and compression method in the embodiments of the present invention, those skilled in the art can understand the specific implementation methods and various variations of the video transcoding and compression apparatus in this embodiment based on the video transcoding and compression method described in the embodiments of the present invention. Therefore, how the video transcoding and compression apparatus implements the video transcoding and compression method in the embodiments of the present invention will not be described in detail here. Any apparatus used by those skilled in the art to implement the video transcoding and compression method in the embodiments of the present invention falls within the scope of protection of this application.

[0095] Example 4 Figure 6 A schematic diagram of an electronic device 10, which can be used to implement embodiments of the present invention, is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices (e.g., helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0096] like Figure 6 As shown, the electronic device 10 includes at least one processor 11 and a memory, such as a read-only memory (ROM) 12 or a random access memory (RAM) 13, communicatively connected to the at least one processor 11. The memory stores computer programs executable by the at least one processor. The processor 11 can perform various appropriate actions and processes based on the computer program stored in the ROM 12 or loaded into the RAM 13 from storage unit 18. The RAM 13 can also store various programs and data required for the operation of the electronic device 10. The processor 11, ROM 12, and RAM 13 are interconnected via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0097] Multiple components in electronic device 10 are connected to I / O interface 15, including: input unit 16, such as keyboard, mouse, etc.; output unit 17, such as various types of displays, speakers, etc.; storage unit 18, such as disk, optical disk, etc.; and communication unit 19, such as network card, modem, wireless transceiver, etc. Communication unit 19 allows electronic device 10 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0098] Processor 11 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. Processor 11 performs the various methods and processes described above, such as video transcoding and compression methods.

[0099] Optionally, the video transcoding and compression method may include: extracting segment features from candidate video segments of the target video to obtain segment feature vectors; performing content understanding and scene recognition based on the video frame association information of the candidate video segments and the segment feature vectors to obtain segment understanding and recognition results for the candidate video segments; performing video segmentation processing on the target video based on video segmentation reference information of the target video, and sending each video segment to each of the video compression processing nodes; calculating the preset volume of each video segment based on the target volume constraint of the video segments; generating segmentation compression parameters for each video segment based on the segment understanding and recognition results, the preset volume of the segments, and compression reference parameters; and sending the segmentation compression parameters of each video segment to the corresponding video compression processing node, so that each of the video compression processing nodes performs transcoding and compression processing on the video segments in parallel according to the segmentation compression parameters.

[0100] In some embodiments, the video transcoding and compression method may be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or mounted on electronic device 10 via ROM 12 and / or communication unit 19. When the computer program is loaded into RAM 13 and executed by processor 11, one or more steps of the video transcoding and compression method described above may be performed. Alternatively, in other embodiments, processor 11 may be configured to perform the video transcoding and compression method by any other suitable means (e.g., by means of firmware).

[0101] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0102] Computer programs used to implement the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the processor, the computer programs cause the functions / operations specified in the flowcharts and / or block diagrams to be performed. The computer programs may be executed entirely on a machine, partially on a machine, or as a standalone software package, partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0103] In the context of this invention, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination thereof. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0104] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device for displaying information to the user (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor); and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0105] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or middleware components (e.g., application servers), or frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.

[0106] A computing system can include clients and servers. Clients and servers are generally located far apart and typically interact through communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a hosting product within the cloud computing service system to address the shortcomings of traditional physical hosts and VPS services, such as high management difficulty and weak business scalability.

[0107] This invention also discloses a computer program product, which includes a computer program that, when executed by a processor, implements the video transcoding and compression method provided in any embodiment of this invention. This program product shares the same inventive concept as the video transcoding and compression methods disclosed in the embodiments of this invention, and therefore will not be described in detail here.

[0108] It should be understood that the various forms of processes shown above can be used to reorder, add, or delete steps. For example, the steps described in the embodiments of the present invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution disclosed in the embodiments of the present invention can be achieved, and this is not limited herein.

[0109] The specific embodiments described above do not constitute a limitation on the scope of protection of this disclosure. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this disclosure should be included within the scope of protection of this disclosure.

Claims

1. A video transcoding and compression method, characterized in that, include: Extract segment features from candidate video segments of the target video to obtain segment feature vectors; Based on the video frame association information of the candidate video segments and the segment feature vector, content understanding and scene recognition are performed to obtain the segment understanding and recognition result of the candidate video segments; The target video is segmented according to the video segmentation reference information of the target video, and each video segment is sent to each video compression processing node. The preset volume of each video segment is calculated based on the target volume constraint of the video segment; Based on the segment understanding and recognition results, the segment preset volume, and the compression reference parameters, the segment compression parameters for each video segment are generated. The segment compression parameters of each video segment are sent to the corresponding video compression processing node, so that each video compression processing node performs transcoding and compression processing on the video segment in parallel according to the segment compression parameters.

2. The method according to claim 1, characterized in that, The step of extracting segment features from candidate video segments of the target video to obtain segment feature vectors includes: Based on the video basic information of the target video, key sampled video frames are extracted from the target video, and multiple candidate video segments are generated based on the key sampled video frames. Based on the video frame association information of the key sampled video frames, generate segment feature vectors for each of the candidate video segments; The fragment feature vector includes motion complexity, texture complexity, noise intensity, text density, subject sensitivity, and brightness anomaly.

3. The method according to claim 1, characterized in that, The step of performing content understanding and scene recognition based on the video frame association information and the segment feature vector of the candidate video segment to obtain the segment understanding and recognition result of the candidate video segment includes: The candidate video segment, the video frame association information of the candidate video segment, and the segment feature vector are input into the target large model, so that the target large model can output the scene type to which the candidate video segment belongs; The video image information of the candidate video segments is input into the target large model, so that the target large model can output the segment complexity, image quality sensitivity, compression tolerance, subject protection suggestions, text protection suggestions and motion preservation suggestions of the candidate video segments.

4. The method according to claim 1, characterized in that, The step of performing video segmentation processing on the target video based on the video segmentation reference information of the target video includes: The video segmentation points of the target processed video are generated based on the video segmentation reference information of the target processed video; wherein, the video segmentation reference information includes the total video duration, the segment understanding and recognition result, the video segment duration, the key frame position, the scene change position, and the computing resources of the video compression processing node; The target video is segmented based on the video segmentation points; wherein, the video segmentation points are the stable time points of the video content, the time points near the key scenes, or the time points at the scene switching boundaries of the target video.

5. The method according to claim 1, characterized in that, Before sending each video segment to each of the video compression processing nodes, the method further includes: Obtain the node association information of the video compression processing node; wherein, the node association information includes the load and computing power of the video compression processing node; The video compression processing node corresponding to the video segment is determined based on the node association information and the estimated processing cost of the video segment.

6. The method according to claim 4, characterized in that, The step of calculating the preset volume of each video segment based on the target volume constraint of the video segment includes: The duration of each video segment is determined and the overall complexity of each video segment is calculated; wherein, the overall complexity includes motion complexity, texture complexity, noise intensity, text density, subject sensitivity, and image quality sensitivity; The preset volume of each video segment is calculated based on the target volume budget, segment duration, and overall complexity of the video segment.

7. The method according to claim 1, characterized in that, The method further includes: Obtain the fragment compression processing association results of each of the aforementioned video compression processing nodes; Based on the correlation results of the segmented compression processing, calculate the objective image quality index, sharpness preservation index, compression quality judgment result, bitrate saving benefit and deviation penalty of the video segmented compression result; Based on the objective image quality indicators, clarity preservation indicators, compression quality judgment results, bitrate saving benefits, and deviation penalties, the comprehensive compression quality score of each video segment is calculated. If the target overall compression quality score is determined to be lower than a set threshold, the compression parameters of the video segments corresponding to the target overall compression quality score are rolled back.

8. A video transcoding and compression device, characterized in that, include: The segment feature vector processing module is used to extract segment features from candidate video segments of the target video and obtain segment feature vectors. The segment understanding and recognition result processing module is used to perform content understanding and scene recognition based on the video frame association information of the candidate video segment and the segment feature vector to obtain the segment understanding and recognition result of the candidate video segment; The video segmentation processing module is used to perform video segmentation processing on the target video according to the video segmentation reference information of the target video, and send each video segment to each video compression processing node; The segment preset volume calculation module is used to calculate the preset volume of each video segment according to the target volume constraint of the video segment; The segment compression parameter generation module is used to generate segment compression parameters for each video segment based on the segment understanding and recognition results, the preset segment volume, and the compression reference parameters. The segmented parallel transcoding and compression processing module is used to send the segmented compression parameters of each video segment to the corresponding video compression processing node, so that each video compression processing node performs parallel transcoding and compression processing on the video segment according to the segmented compression parameters.

9. An electronic device, characterized in that, The electronic device includes: At least one processor; and a memory communicatively connected to said at least one processor; wherein, The memory stores a computer program that is executed by the at least one processor to enable the at least one processor to perform the video transcoding and compression method according to any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that cause a processor to execute the video transcoding and compression method according to any one of claims 1-7.

11. A computer program product, characterized in that, Includes a computer program / instruction, wherein the computer program / instruction, when executed by a processor, implements the video transcoding and compression method according to any one of claims 1-7.

Images