Media management system based on AIGC
By introducing technologies such as upper confidence limit algorithm, knowledge graph, AI model and blockchain in the media management system, the problem of low file type identification and storage utilization in the existing technology is solved, and high-quality file conversion and compression are achieved , reduces dependence on manual operations, improves user experience and storage efficiency.
Patent Information
- Application Number
- CN202411214837.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-02
- Publication Date
- 2024-11-26
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
The existing AIGC-based media management system cannot accurately identify file types, resulting in poor quality and compatibility of converted files, and low overall storage utilization, so it cannot perform maximum compression while ensuring file quality.
A media management system based on AIGC is designed, including authentication access module, file processing module, search query module, content enhancement module, compression optimization module, record audit module, content distribution module, statistical analysis module and backup and recovery module. The file processing module uses the upper confidence limit algorithm to identify file types and convert formats. It searches and query modules to build a knowledge graph. The content enhancement module optimizes file quality through AI models. The compression optimization module uses historical data to select the optimal compression algorithm, and records the audit module’s pass area. Blockchain record and audit operations.
It realizes more accurate file type recognition, improves the quality and compatibility after file conversion, reduces manual operation dependence, improves operation convenience, and improves storage utilization while ensuring file quality, ensuring user acquisition Good compressed file experience.
Smart Images

Figure FT_1
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of file management, and in particular to a media management system based on AIGC. Background Art
[0002] With the rapid development of digital media technology, the generation and use of media files have increased dramatically. How to efficiently manage these media files has become an urgent problem to be solved. Traditional media management systems mainly rely on simple folder structures and basic keyword searches, which are difficult to cope with the management needs of massive and diverse media files. In addition, the operation records and security management of media files also face challenges. How to ensure the transparency and immutability of file operations, and how to efficiently compress and optimize while ensuring file quality are important issues facing the current system.
[0003] The existing AIGC-based media management system cannot accurately identify file types, resulting in poor quality and compatibility of converted files and increased reliance on manual operations. In addition, the overall storage utilization of the existing AIGC-based media management system is low, and it is impossible to perform maximum compression while ensuring file quality. Therefore, we propose a media management system based on AIGC. Summary of the invention
[0004] The purpose of the present invention is to solve the defects in the prior art and to propose a media management system based on AIGC.
[0005] In order to achieve the above object, the present invention adopts the following technical solutions: An AIGC-based media management system includes an authentication access module, a file processing module, a retrieval query module, a content enhancement module, a compression optimization module, a personalized recommendation module, a record review module, a content distribution module, a statistical analysis module, and a backup and recovery module; The authentication access module is used for user identity authentication and authority management, and for accessing each group of media files; The file processing module is used to convert the format of the media file type uploaded by the user; The search query module is used to automatically extract metadata from media files and build a knowledge graph in combination with external data sources to provide multi-dimensional search; The content enhancement module is used to optimize and enhance each group of stored media files; The compression optimization module is used to compress and optimize the media files; The personalized recommendation module is used to provide personalized recommendation services based on user information; The recording and auditing module is used to record and audit various operations on all media files; The content distribution module is used to distribute media content to different platforms and devices and provide multiple sharing methods; The statistical analysis module is used for statistical analysis of various operations and data; The backup and restore module is used to provide safe backup and restore services for media files.
[0006] As a further solution of the present invention, the specific steps of the file processing module format conversion are as follows: P1.1: Receive the media files uploaded by the user, collect the basic information of each group of files, the current file types and format conversion paths, and use the collected data as the initial state of each file, each group of media files as the initial node, and each convertible file type as the child node of the corresponding initial node; P1.2: Starting from the initial node, calculate the UCB value of each child node of the initial node, select the child node with the highest UCB value as the next choice through the upper confidence limit algorithm, and repeat the selection process until reaching the leaf node or the preset maximum search depth; P1.3: If the file type is not determined at the finally selected node or the format conversion path is not completed, the node is expanded to generate a new child node to represent the new file type or conversion step. A random simulation is performed starting from the expanded new node until the file type is successfully identified and the format conversion is completed, or it is determined that the path is not suitable for the current file, and the simulation is terminated; P1.4: Based on the simulation results, update the information of all nodes in the path, and then repeat the selection, expansion, simulation and backtracking multiple times until the preset maximum number of iterations is reached. The total winning rate and total number of visits of each complete node path are calculated. Two sets of tables sorted from high to low are generated according to the total winning rate and total number of visits of each path; P1.5: Select the path with the highest winning rate and the most visits from the sorted path list as the optimal path, find the end node of the optimal path, and read the file type represented by the end node. Then, according to the selected optimal path, perform the corresponding format conversion operation, save the converted file and return it to the user.
[0007] As a further solution of the present invention, the specific steps of constructing the knowledge graph by the retrieval query module are as follows: P2.1: Receive and analyze the converted media file header information, extract the basic information of the file, and classify each media file into audio files, video files, and image files according to the file type. Then, decode each type of file through the audio decoder, video decoder, and image decoder to obtain the original audio signal, original video frame sequence, and original pixel data. P2.2: Extract audio features from raw audio signals by using spectrum analysis or time domain analysis methods, classify audio features, identify audio content types, extract image features from each raw video frame sequence, detect scene switching points in the video, detect different scene segments, identify objects in the frame, track the movement of each object in the video, collect actions and behaviors in the video, extract image features from raw pixel data through image processing algorithms, and identify objects and scene types in the image; P2.3: Use NLP technology to extract text contained in various types of media files. According to the extracted metadata of various types of media files, search for corresponding external data sources, integrate the external data with the initially extracted metadata, remove duplicates, clean and normalize them, and define the relationship between entities based on the metadata. P2.4: Use the Neo4j graph database to construct a knowledge graph using entities as nodes and relationships between entities as edges. Store the entities and relationships in the graph database and create a keyword index based on all extracted metadata keywords and their corresponding media files.
[0008] As a further solution of the present invention, the specific steps of optimizing and enhancing the content enhancement module are as follows: P3.1: Receive and record the converted media files, check the basic integrity of the file format and content, read the file metadata, determine the file type, and adjust the resolution, frame rate, and audio sampling rate of each group of media files to be processed; P3.2: Segment the file into multiple groups of small segments, and input each group of file segments into the pre-trained AI model. The AI model extracts important features from the file segments, and the model performs enhancement processing based on the feature data to improve the quality of the file segments. The processed file segments are then restored from the internal representation of the AI model to the standard format. P3.3: Merge the clips into a complete media file, post-process the enhanced media files, and verify the quality of each group of post-processed media files. At the same time, feed the media files that do not meet the verification standards back to the AI model for reprocessing and verification, and save the enhanced high-quality files.
[0009] As a further solution of the present invention, the compression optimization module compresses and optimizes the media file in the following specific steps: P4.1: The compression optimization module extracts key features from media files, normalizes and standardizes the extracted feature data, collects historical compression data, and calculates the prior probability of each compression algorithm from the historical data to obtain the overall performance of each algorithm under different feature data; P4.2: Based on the processed feature data and historical compression data, calculate the likelihood function of the feature data under each compression algorithm, evaluate the matching degree between the feature data and each algorithm, calculate the posterior probability of each compression algorithm based on the prior probability and likelihood function, then compare the posterior probabilities of all compression algorithms and select the algorithm with the highest posterior probability as the optimal algorithm; P4.3: Apply the selected optimal compression algorithm to the media file for compression processing, then perform image denoising, video stabilization and audio noise reduction optimization processing on the compressed file, record the optimized media file, convert the optimized media file back to the original format, and save the optimized media file or the optimized media file in the original format according to user needs.
[0010] As a further solution of the present invention, the specific steps of recording and reviewing various operations of all media files by the recording and reviewing module are as follows: P5.1: Collect user operation request information, and select the features of operation frequency, file size, and modification time for anomaly detection from the collected data to build an antibody library, where each antibody represents a normal operation mode and uses known abnormal operations or potential attack modes as antigens; P5.2: Match the feature data of the new operation with the normal pattern in the antibody library. If the feature data is similar to the normal pattern in the antibody library, the operation is considered normal; if it is similar to the antigen pattern, the operation is considered abnormal; P5.3: Score detected abnormal operations based on the severity of the abnormality and the similarity with known attack patterns, and generate abnormal operation reports that describe the detected abnormalities, scores, and risks in detail. At the same time, process the abnormal operation data and update the antibody library and antigen pattern based on the processing results; P5.4: Package the information of the filtered operation request into blocks, and add the created blocks to the blockchain through the consensus mechanism. During the audit process, retrieve and verify the operation history of all media files, feedback the operation results to the user, and provide detailed records or audit results of the operation.
[0011] Compared with the prior art, the present invention has the following beneficial effects: 1. The present invention receives each media file uploaded by the user and uses it as the initial node. At the same time, each convertible file type is used as a child node of the corresponding initial node. Starting from the initial node, the UCB value of each child node of the initial node is calculated. The child node with the highest UCB value is selected as the next selection through the upper confidence limit algorithm. The selection process is repeated until a leaf node or a preset maximum search depth is reached. If the file type of the finally selected node is not determined or the format conversion path is not completed, the node is expanded to generate a new child node to represent a new file type or conversion step. Starting from the expanded new node, a random simulation is performed until the file type is successfully identified and the format conversion is completed, or it is determined that the path is not suitable for the current file, the simulation is terminated, and the simulation result is determined according to the simulation result. , update the information of all nodes in the path, and then repeat the selection, expansion, simulation and backtracking many times until the preset maximum number of iterations is reached. The total winning rate and total access times of each complete node path are calculated. According to the total winning rate and total access times of each path, two sets of tables sorted from high to low are generated. The path with the highest winning rate and the most access times is selected from the sorted path list as the optimal path, and the end node of the optimal path is read, and the file type represented by the end node is read. Then, according to the selected optimal path, the corresponding format conversion operation is performed, and the converted file is saved and returned to the user. It can identify the file type more accurately, ensure the quality and compatibility of the converted file, reduce the dependence on manual operation, and improve the convenience of operation and user experience.
[0012] 2. The present invention extracts key features in a media file, normalizes and standardizes the extracted feature data, collects historical compression data, and calculates the prior probability of each compression algorithm from the historical data, obtains the overall performance of each algorithm under different feature data, calculates the likelihood function of the feature data under each compression algorithm based on the processed feature data and the historical compression data, evaluates the degree of matching between the feature data and each algorithm, calculates the posterior probability of each compression algorithm based on the prior probability and the likelihood function, then compares the posterior probabilities of all compression algorithms, selects the algorithm with the highest posterior probability as the optimal algorithm, applies the selected optimal compression algorithm to the media file, performs compression processing, and then performs image denoising, video stabilization, and audio noise reduction optimization processing on the compressed file, records the optimized media file, converts the optimized media file back to the original format, and saves the optimized media file or the optimized media file in the original format according to user needs, thereby improving the overall storage utilization of the system, being able to perform maximum compression while ensuring file quality, and ensuring that users can still get a good experience when using compressed files. BRIEF DESCRIPTION OF THE DRAWINGS
[0013] The accompanying drawings are used to provide further understanding of the present invention and constitute a part of the specification. They are used to explain the present invention together with the embodiments of the present invention and do not constitute a limitation of the present invention.
[0014] Figure 1 This is a system block diagram of an AIGC-based media management system proposed by the present invention. DETAILED DESCRIPTION
[0015] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments. Example
[0016] Reference Figure 1 , a media management system based on AIGC, includes an authentication access module, a file processing module, a retrieval query module, a content enhancement module, a compression optimization module, a personalized recommendation module, a record review module, a content distribution module, a statistical analysis module and a backup and recovery module.
[0017] The authentication access module is used for user identity authentication and permission management, and access to each group of media files; the file processing module is used to convert the format of the media file type uploaded by the user.
[0018] Specifically, the media files uploaded by the user are received, the basic information of each group of files and the current file types and format conversion paths are collected, and the collected data is used as the initial state of each file, each group of media files is used as the initial node, and each convertible file type is used as the child node of the corresponding initial node. Starting from the initial node, the UCB value of each child node of the initial node is calculated, and the child node with the highest UCB value is selected as the next selection through the upper confidence limit algorithm. The selection process is repeated until a leaf node or a preset maximum search depth is reached. If the file type of the node finally selected is not determined or the format conversion path is not completed, the node is expanded to generate a new child node to represent the new file type or conversion step, and a search is performed starting from the expanded new node. Random simulation is performed until the file type is successfully identified and the format conversion is completed, or if it is determined that the path is not suitable for the current file, the simulation is terminated. According to the simulation results, the information of all nodes in the path is updated, and then selection, expansion, simulation and backtracking are repeated many times until the preset maximum number of iterations is reached. The total winning rate and total number of visits of each complete node path are calculated. Two sets of tables sorted from high to low are generated according to the total winning rate and total number of visits of each path. The path with the highest winning rate and the most visits is selected from the sorted path list as the optimal path. The end node of the optimal path is selected and the file type represented by the end node is read. Then, according to the selected optimal path, the corresponding format conversion operation is performed, and the converted file is saved and returned to the user.
[0019] The retrieval query module is used to automatically extract metadata from media files and build a knowledge graph in combination with external data sources to provide multi-dimensional retrieval.
[0020] Specifically, the converted media file header information is received and analyzed, the basic file information is extracted, and each media file is divided into audio files, video files and image files according to the file type. Each type of file is decoded by an audio decoder, a video decoder and an image decoder respectively to obtain the original audio signal, the original video frame sequence and the original pixel data. By using spectrum analysis or time domain analysis methods, audio features are extracted from the original audio signal, and then the audio features are classified to identify the audio content type. Image features are extracted for each frame of the original video frame sequence, and scene switching points in the video are detected, different scene segments are detected, objects in the frame are identified, and each object in the video is tracked. Move, collect the actions and behaviors that occur in the video, extract the image features of the original pixel data through image processing algorithms, and identify the objects and scene types in the image, use NLP technology to extract the text contained in various types of media files, retrieve the corresponding external data sources based on the extracted metadata of various media files, and then integrate the external data with the preliminary extracted metadata, remove duplication, clean and normalize them, define the relationship between entities based on metadata, use the Neo4j graph database to use entities as nodes and the relationship between entities as edges, build a knowledge graph, and store the entities and relationships in the graph database, and create a keyword index based on all extracted metadata keywords and their corresponding media files.
[0021] The content enhancement module is used to optimize and enhance each group of stored media files.
[0022] Specifically, the converted media files are received and recorded, and the basic integrity of the file format and content is checked. The file metadata is then read to determine the file type, and the resolution, frame rate, and audio sampling rate of each group of media files to be processed are adjusted. The files are divided into multiple groups of small segments, and each group of file segments is input into a pre-trained AI model. The important features in the file segments are extracted through the AI model, and the model performs enhancement processing based on the feature data to improve the quality of the file segments. The processed file segments are then restored from the internal representation of the AI model to a standard format, and the segments are merged into complete media files. The enhanced media files are then post-processed, and the quality of each group of post-processed media files is verified. At the same time, the media files that do not meet the verification standards are fed back to the AI model for reprocessing and verification, and the enhanced high-quality files are saved.
[0023] The compression optimization module is used to compress and optimize media files.
[0024] Specifically, the compression optimization module extracts key features from the media files, normalizes and standardizes the extracted feature data, collects historical compression data, and calculates the prior probability of each compression algorithm from the historical data, obtains the overall performance of each algorithm under different feature data, calculates the likelihood function of the feature data under each compression algorithm based on the processed feature data and the historical compression data, evaluates the degree of matching between the feature data and each algorithm, calculates the posterior probability of each compression algorithm based on the prior probability and the likelihood function, then compares the posterior probabilities of all compression algorithms, selects the algorithm with the highest posterior probability as the optimal algorithm, applies the selected optimal compression algorithm to the media files for compression processing, performs image denoising, video stabilization, and audio noise reduction optimization processing on the compressed files, records the optimized media files, converts the optimized media files back to the original format, and saves the optimized media files or the optimized media files in the original format according to user needs. Example
[0025] Reference Figure 1 , a media management system based on AIGC, includes an authentication access module, a file processing module, a retrieval query module, a content enhancement module, a compression optimization module, a personalized recommendation module, a record review module, a content distribution module, a statistical analysis module and a backup and recovery module.
[0026] The personalized recommendation module is used to provide personalized recommendation services based on user information; the record and review module is used to record and review various operations on all media files.
[0027] Specifically, the user's operation request information is collected, and the features of each abnormal detection such as operation frequency, file size and modification time are selected from the collected data to construct an antibody library, where each antibody represents a normal operation mode, and the known abnormal operation or potential attack mode is used as an antigen, and the feature data of the new operation is matched with the normal mode in the antibody library. If the feature data is similar to the normal mode in the antibody library, the operation is considered normal; if it is similar to the antigen mode, the operation is considered abnormal, and the detected abnormal operation is scored according to the severity of the abnormality and the similarity with the known attack mode, and an abnormal operation report is generated, which describes the detected anomaly, score and risk in detail. At the same time, the abnormal operation data is processed, and the antibody library and antigen mode are updated according to the processing results. The information of the operation request after screening is packaged into blocks, and the created blocks are added to the blockchain through a consensus mechanism. In the audit process, the operation history of all media files is retrieved and verified, the operation results are fed back to the user, and a detailed record or audit result of the operation is provided.
[0028] The content distribution module is used to distribute media content to different platforms and devices, and provides multiple sharing methods; the statistical analysis module is used to perform statistical analysis on various operations and data; the backup and recovery module is used to provide secure backup and recovery services for media files.
Claims
1. A media management system based on AIGC, characterized in that: It includes authentication access module, file processing module, search and query module, content enhancement module, compression optimization module, personalized recommendation module, record review module, content distribution module, statistical analysis module and backup and recovery module; The authentication access module is used for user identity authentication and authority management, and for accessing each group of media files; The file processing module is used to convert the format of the media file type uploaded by the user; The search query module is used to automatically extract metadata from media files and build a knowledge graph in combination with external data sources to provide multi-dimensional search; The content enhancement module is used to optimize and enhance each group of stored media files; The compression optimization module is used to compress and optimize the media files; The personalized recommendation module is used to provide personalized recommendation services based on user information; The recording and auditing module is used to record and audit various operations on all media files; The content distribution module is used to distribute media content to different platforms and devices and provide multiple sharing methods; The statistical analysis module is used for statistical analysis of various operations and data; The backup and restore module is used to provide safe backup and restore services for media files.
2. The AIGC-based media management system according to claim 1, characterized in that: The specific steps of the file processing module format conversion are as follows: P1.1: Receive the media files uploaded by the user, collect the basic information of each group of files, the current file types and format conversion paths, and use the collected data as the initial state of each file, each group of media files as the initial node, and each convertible file type as the child node of the corresponding initial node; P1.2: Starting from the initial node, calculate the UCB value of each child node of the initial node, select the child node with the highest UCB value as the next choice through the upper confidence limit algorithm, and repeat the selection process until reaching the leaf node or the preset maximum search depth; P1.3: If the file type is not determined at the finally selected node or the format conversion path is not completed, the node is expanded to generate a new child node to represent the new file type or conversion step. A random simulation is performed starting from the expanded new node until the file type is successfully identified and the format conversion is completed, or it is determined that the path is not suitable for the current file, and the simulation is terminated; P1.4: Based on the simulation results, update the information of all nodes in the path, and then repeat the selection, expansion, simulation and backtracking multiple times until the preset maximum number of iterations is reached. The total winning rate and total number of visits of each complete node path are calculated. Two sets of tables sorted from high to low are generated according to the total winning rate and total number of visits of each path; P1.5: Select the path with the highest winning rate and the most visits from the sorted path list as the optimal path, find the end node of the optimal path, and read the file type represented by the end node. Then, according to the selected optimal path, perform the corresponding format conversion operation, save the converted file and return it to the user.
3. The AIGC-based media management system according to claim 2, characterized in that: The specific steps of constructing the knowledge graph by the retrieval query module are as follows: P2.1: Receive and analyze the converted media file header information, extract the basic information of the file, and classify each media file into audio files, video files, and image files according to the file type. Then, decode each type of file through the audio decoder, video decoder, and image decoder to obtain the original audio signal, original video frame sequence, and original pixel data. P2.2: Extract audio features from raw audio signals by using spectrum analysis or time domain analysis methods, classify audio features, identify audio content types, extract image features from each raw video frame sequence, detect scene switching points in the video, detect different scene segments, identify objects in the frame, track the movement of each object in the video, collect actions and behaviors in the video, extract image features from raw pixel data through image processing algorithms, and identify objects and scene types in the image; P2.3: Use NLP technology to extract text contained in various types of media files. According to the extracted metadata of various types of media files, search for corresponding external data sources, integrate the external data with the initially extracted metadata, remove duplicates, clean and normalize them, and define the relationship between entities based on the metadata. P2.4: Use the Neo4j graph database to construct a knowledge graph using entities as nodes and relationships between entities as edges. Store the entities and relationships in the graph database and create a keyword index based on all extracted metadata keywords and their corresponding media files.
4. The AIGC-based media management system according to claim 2, characterized in that: The specific steps of optimizing and enhancing the content enhancement module are as follows: P3.1: Receive and record the converted media files, check the basic integrity of the file format and content, read the file metadata, determine the file type, and adjust the resolution, frame rate, and audio sampling rate of each group of media files to be processed; P3.2: Segment the file into multiple groups of small segments, and input each group of file segments into the pre-trained AI model. The AI model extracts important features from the file segments, and the model performs enhancement processing based on the feature data to improve the quality of the file segments. The processed file segments are then restored from the internal representation of the AI model to the standard format. P3.3: Merge the clips into a complete media file, post-process the enhanced media files, and verify the quality of each group of post-processed media files. At the same time, feed the media files that do not meet the verification standards back to the AI model for reprocessing and verification, and save the enhanced high-quality files.
5. The AIGC-based media management system according to claim 1, characterized in that: The specific steps of the compression optimization module compressing and optimizing the media files are as follows: P4.1: The compression optimization module extracts key features from media files, normalizes and standardizes the extracted feature data, collects historical compression data, and calculates the prior probability of each compression algorithm from the historical data to obtain the overall performance of each algorithm under different feature data; P4.2: Based on the processed feature data and historical compression data, calculate the likelihood function of the feature data under each compression algorithm, evaluate the matching degree between the feature data and each algorithm, calculate the posterior probability of each compression algorithm based on the prior probability and likelihood function, then compare the posterior probabilities of all compression algorithms and select the algorithm with the highest posterior probability as the optimal algorithm; P4.3: Apply the selected optimal compression algorithm to the media file for compression processing, then perform image denoising, video stabilization and audio noise reduction optimization processing on the compressed file, record the optimized media file, convert the optimized media file back to the original format, and save the optimized media file or the optimized media file in the original format according to user needs.
6. The AIGC-based media management system according to claim 1, characterized in that: The specific steps of recording and reviewing various operations of all media files by the recording and reviewing module are as follows: P5.1: Collect user operation request information, and select the features of operation frequency, file size, and modification time for anomaly detection from the collected data to build an antibody library, where each antibody represents a normal operation mode and uses known abnormal operations or potential attack modes as antigens; P5.2: Match the feature data of the new operation with the normal pattern in the antibody library. If the feature data is similar to the normal pattern in the antibody library, the operation is considered normal; if it is similar to the antigen pattern, the operation is considered abnormal; P5.3: Score detected abnormal operations based on the severity of the abnormality and the similarity with known attack patterns, and generate abnormal operation reports that describe the detected abnormalities, scores, and risks in detail. At the same time, process the abnormal operation data and update the antibody library and antigen pattern based on the processing results; P5.4: Package the information of the filtered operation request into blocks, and add the created blocks to the blockchain through the consensus mechanism. During the audit process, retrieve and verify the operation history of all media files, feedback the operation results to the user, and provide detailed records or audit results of the operation.
Citation Information
Cited By
Online visual demonstration document editing method
CN119378504A
