Supercompression storage method based on content video and system applying same

Abstract

The invention relates to a content video-based hypercompression storage method. The method comprises the following steps of S1, acquiring video stream data of a camera through a network; s2, for the video stream data, performing corresponding decoding according to the code stream type to generate YUV data; s3, analyzing the YUV data, determining an ROI (Region of Interest), and extracting content feature values of a plurality of moving targets and position data of images where the moving targets are located as element data of the video; the ROI is logically separated from the non-ROI, secondary high-fidelity compression is carried out on the ROI, high compression is carried out on the non-ROI, and the high compression rate of the ROI under the high-fidelity condition is achieved; s4, combining the code streams of the ROI and the non-ROI into the same frame of code stream; s5, taking the ROI metadata of each region of interest as structured description data of the frame data; and S6, sending the coded code stream and the ROI metadata to a video storage engine for intelligent video storage.

Description

Technical Field

[0001] The present disclosure relates to the field of ultra-compressed video storage in the security field, and in particular to an ultra-compressed storage method based on content video and a system using the same. Background Art

[0002] In current video surveillance systems, as resolutions become higher and higher, bit rates also become higher and higher, posing huge challenges to network transmission and video storage. Therefore, how to improve video compression rates while ensuring unchanged image quality has become a rigid demand in the current market.

[0003] The current mainstream video compression technology is H.265, which is a new video coding standard developed by ITU-T VCEG after H.264. The H.265 standard revolves around the existing video coding standard H.264, retains some of the original technologies, and improves some related technologies. The new technology uses advanced technology to improve the relationship between bit stream, encoding quality, delay and algorithm complexity to achieve the optimal setting. The specific contents include: improving compression efficiency, reducing real-time delay, reducing channel acquisition time and random access delay, reducing complexity, etc.

[0004] The main idea of ​​the current mainstream video compression technology is to remove redundant information from several main angles such as space, time, coding, and vision. Due to its excellent data compression ratio and video quality, H.264 has become the most popular codec standard in the current market. H.265 is based on H.264, while ensuring the same video quality, the bit rate of the video stream can be reduced by 50%. As the H.265 encoding format becomes more and more popular, it has become the mainstream video compression standard in the current market.

[0005] However, the H.265 encoding format still contains a lot of redundant information, which results in a large amount of video data and high requirements for storage space and transmission bandwidth, affecting network transmission and video storage efficiency.

[0006] A Chinese patent document with publication number CN105786975A discloses a method for storing massive videos for rapid retrieval, but its video content lacks metadata storage and retrieval functions, and cannot be quickly retrieved and located based on the video content, so it still needs to be further improved. Summary of the invention

[0007] In view of the above technical problems in the related art, the present disclosure proposes an ultra-compression storage method based on content video, which can overcome the above shortcomings of the prior art.

[0008] To achieve the above technical objectives, the technical solution of the present disclosure is implemented as follows: The first object of the present disclosure is to provide a method for ultra-compressed storage based on content video, comprising the following steps: S1: Obtain the video stream data of the camera through the network; S2: For the video stream data, decode it according to the bitstream type and generate YUV data; S3: Analyze the YUV data, determine the region of interest ROI, and extract the content feature values ​​of several moving targets and the position data of the image where they are located as the element data of the video; logically separate the region of interest ROI from the region of non-interest, perform secondary high-fidelity compression on the region of interest ROI, and perform high compression on the region of non-interest, so as to achieve a high compression rate with high fidelity for the region of interest ROI; S4: The bitstreams of the region of interest (ROI) and the non-region of interest (ROI) are merged into the same frame bitstream; S5: metadata of each region of interest ROI is used as structured description data of the frame data; S6: Send the encoded bitstream and the region of interest ROI metadata to the video storage engine for intelligent video storage.

[0009] Preferably, in S1, the camera is preferably a front-end high-definition camera.

[0010] Preferably, in S1 and S2, the code stream type of the video stream data is H.264 / H.265.

[0011] Preferably, in S2, the YUV data adopts YUV (4:2:0) data.

[0012] Preferably, the specific steps of S3 are: S3.1: Analyze the YUV data and determine the region of interest (ROI); S3.2: extracting content feature values ​​of several moving objects and location data of the images in which they are located as element data of the video; S3.3: Logically separate the region of interest (ROI) from the region of no interest; S3.4: Perform secondary high-fidelity compression on the region of interest ROI and perform high compression on the non-region of interest to achieve a high compression rate with high fidelity for the region of interest ROI.

[0013] Preferably, in S3.1, deep learning technology is used to analyze the YUV data.

[0014] Preferably, in S3.2, the moving targets include motor vehicles, non-motor vehicles, people and other types of moving targets.

[0015] Preferably, in S6, the video storage engine adopts an SVDSE video storage engine. The prior art of the SVDSE video storage engine is a patent document (publication number: CN105786975A).

[0016] The second object of the present disclosure is to apply the system of the ultra-compression storage method, including a camera and an end-side device connected to each other, the end-side device including an intelligent analysis module, a storage control module and a plurality of decoding units; after the end-side device obtains the video stream data from the camera, it selects the corresponding decoding unit to parse and decode the video stream data according to the bitstream type of the video stream data to generate the YUV data; then, the intelligent analysis module receives and analyzes the YUV data, separates the region of interest ROI and parses the corresponding structured metadata information; then, the intelligent analysis module encodes the parsed and separated data into a bitstream according to the bitstream type of the video stream data; finally, the structured metadata information and the encoded bitstream are output to the storage control module, thereby completing the disk storage of the video data and metadata.

[0017] During implementation, the access part of the camera completes access initialization to ensure connection with the end-side device; the encoding and decoding part of the end-side device completes initialization to ensure that sufficient memory space is opened up. If hard encoding and decoding is used, ensure that related hardware is started correctly; the intelligent analysis part of the end-side device completes initialization to ensure that the model data is loaded correctly; the storage control module completes initialization to ensure that the storage space is correctly connected.

[0018] Beneficial effects of the present disclosure: The present disclosure is a content-aware compression method based on deep learning on the basis of existing video compression technology, which uses deep learning technology to extract video content features and then compresses according to these features. The present disclosure effectively reduces the amount of video data by identifying and removing redundant information in the video, thereby reducing the demand for storage space and transmission bandwidth. The present disclosure is based on video storage services, and by adding element data storage and retrieval of video content on the basis of the video storage method disclosed in CN105786975A, it can achieve rapid retrieval and positioning based on video content, and realize an efficient intelligent video storage and retrieval system.

[0019] Based on the existing popular video compression algorithm (H.264 / H.265), the present invention recognizes and removes redundant information in the video, adds video compression based on ROI perception, realizes efficient video compression, effectively reduces the amount of video data, and thus reduces the demand for storage space and transmission bandwidth. Based on the video storage service, the video storage method published in CN105786975A is added with the element data storage and retrieval of the video content, which can realize fast retrieval and positioning based on the video content, and realize an efficient intelligent video storage and retrieval system.

[0020] The present invention uses video coding and decoding technology to decode the compressed standard video stream (H264 / H265), uses intelligent analysis technology to intelligently analyze the decoded video content, extracts the focus area (ROI) in the video as the foreground, and generates metadata for retrieval. At the same time, the video encoding technology is used to losslessly restore the focus area (ROI) and deeply compress the background image. While ensuring high fidelity, ultra-high video compression effect is achieved, and the compressed video stream is stored in a serialized manner.

[0021] The present disclosure adopts deep learning technology to intelligently extract the region of interest (ROI) of the video, and through different compression strategies, realizes high-rate video compression while ensuring high fidelity of the region of interest (ROI). Realizing high-rate video compression can effectively reduce the amount of video data and reduce the demand for storage space and transmission bandwidth. It is a rigid demand in current video surveillance and other related industries, and has great market expansion space. The present disclosure can also adopt but is not limited to video compression based on region of interest (ROI) perception, which can increase the compression rate by two times when it is basically consistent with the subjective evaluation of the original video stream. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the embodiments of the present disclosure or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present disclosure. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0023] Figure 1 The present invention is a flowchart of the method for ultra-compressed storage of content-based video.

[0024] Figure 2 This is a module logic diagram of a system that applies the ultra-compression storage method described in the present disclosure. DETAILED DESCRIPTION

[0025] The technical solutions in the embodiments of the present disclosure will be clearly and completely described below in conjunction with the drawings in the embodiments of the present disclosure.

[0026] like Figure 1-2 As shown, in order to facilitate the understanding of the above technical solutions of the present disclosure, the above technical solutions of the present disclosure are described in detail below through specific usage methods.

[0027] The first object of the present disclosure is to provide a method for ultra-compressed storage based on content video, comprising the following steps: S1: Obtain the video stream data of the camera through the network; S2: For the video stream data, decode it according to the bitstream type and generate YUV data; S3: Analyze the YUV data, determine the region of interest ROI, and extract the content feature values ​​of several moving targets and the position data of the image where they are located as the element data of the video; logically separate the region of interest ROI from the region of non-interest, perform secondary high-fidelity compression on the region of interest ROI, and perform high compression on the region of non-interest, so as to achieve a high compression rate with high fidelity for the region of interest ROI; S4: The bitstreams of the region of interest (ROI) and the non-region of interest (ROI) are merged into the same frame bitstream; S5: metadata of each region of interest ROI is used as structured description data of the frame data; S6: Send the encoded bitstream and the region of interest ROI metadata to the video storage engine for intelligent video storage.

[0028] In one embodiment, in S1, the camera is preferably a front-end high-definition camera.

[0029] In one embodiment, in S1 and S2, the bitstream type of the video stream data is H.264 / H.265.

[0030] In one embodiment, in S2, the YUV data adopts YUV (4:2:0) data.

[0031] In one embodiment, the specific steps of S3 are: S3.1: Analyze the YUV data and determine the region of interest (ROI); S3.2: extracting content feature values ​​of several moving objects and location data of the images in which they are located as element data of the video; S3.3: Logically separate the region of interest (ROI) from the region of no interest; S3.4: Perform secondary high-fidelity compression on the region of interest ROI and perform high compression on the non-region of interest to achieve a high compression rate with high fidelity for the region of interest ROI.

[0032] In one embodiment, in S3.1, deep learning technology is used to analyze the YUV data.

[0033] In one embodiment, in S3.2, the moving targets include motor vehicles, non-motor vehicles, people, and other types of moving targets.

[0034] In one embodiment, in S6, the video storage engine uses an SVDSE video storage engine. The prior art of the SVDSE video storage engine is a patent document (publication number: CN105786975A).

[0035] The second object of the present disclosure is to apply the system of the ultra-compression storage method, including a camera and an end-side device connected to each other, the end-side device including an intelligent analysis module, a storage control module and a plurality of decoding units; after the end-side device obtains the video stream data from the camera, it selects the corresponding decoding unit to parse and decode the video stream data according to the bitstream type of the video stream data to generate the YUV data; then, the intelligent analysis module receives and analyzes the YUV data, separates the region of interest ROI and parses the corresponding structured metadata information; then, the intelligent analysis module encodes the parsed and separated data into a bitstream according to the bitstream type of the video stream data; finally, the structured metadata information and the encoded bitstream are output to the storage control module, thereby completing the disk storage of the video data and metadata.

[0036] During implementation, the access part of the camera completes access initialization to ensure connection with the end-side device; the encoding and decoding part of the end-side device completes initialization to ensure that sufficient memory space is opened up. If hard encoding and decoding is used, ensure that related hardware is started correctly; the intelligent analysis part of the end-side device completes initialization to ensure that the model data is loaded correctly; the storage control module completes initialization to ensure that the storage space is correctly connected.

[0037] In summary, through the unique design of the present disclosure, the present disclosure adopts deep learning technology to intelligently extract the region of interest ROI of the video, and through different compression strategies, realizes high-rate video compression while ensuring high fidelity of the region of interest ROI. Realizing high-rate video compression can effectively reduce the amount of video data and reduce the demand for storage space and transmission bandwidth. It is a rigid demand in current video surveillance and other related industries, and has great market expansion space. The present disclosure can also adopt and is not limited to video compression based on region of interest (ROI) perception, and can increase the compression rate by two times when it is basically consistent with the subjective evaluation of the original video stream.

[0038] The above description is only a preferred embodiment of the present disclosure and is not intended to limit the present disclosure. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present disclosure should be included in the protection scope of the present disclosure.

Claims

1. A method for ultra-compressed storage based on content video, characterized in that: The following steps are involved: S1: Obtain the video stream data of the camera through the network; S2: For the video stream data, decode it according to the bitstream type and generate YUV data; S3: Analyze the YUV data, determine the region of interest (ROI), and extract the content feature values ​​of several moving targets and the position data of the image where they are located as the element data of the video; The region of interest (ROI) is logically separated from the region of non-interest, and the region of interest (ROI) is compressed twice with high fidelity, and the region of non-interest is highly compressed, so as to achieve a high compression rate with high fidelity for the region of interest (ROI); S4: The bitstreams of the region of interest (ROI) and the non-region of interest (ROI) are merged into the same frame bitstream; S5: metadata of each region of interest ROI is used as structured description data of the frame data; S6: Send the encoded bitstream and the region of interest ROI metadata to the video storage engine for intelligent video storage.

2. The ultra-compression storage method according to claim 1, characterized in that: In S1, the camera is a front-end high-definition camera.

3. The ultra-compression storage method according to claim 1, characterized in that: In S1 and S2, the code stream type of the video stream data is H.264 / H.

265.

4. The ultra-compression storage method according to claim 1, characterized in that: In the S2, the YUV data adopts YUV (4:2:0) data.

5. The ultra-compression storage method according to claim 1, characterized in that: The specific steps of S3 are: S3.1: Analyze the YUV data and determine the region of interest (ROI); S3.2: extracting content feature values ​​of several moving objects and location data of the images in which they are located as element data of the video; S3.3: Logically separate the region of interest (ROI) from the region of no interest; S3.4: Perform secondary high-fidelity compression on the region of interest ROI and perform high compression on the non-region of interest to achieve a high compression rate with high fidelity for the region of interest ROI.

6. The ultra-compression storage method according to claim 5, characterized in that: In S3.1, deep learning technology is used to analyze the YUV data.

7. The ultra-compression storage method according to claim 5, characterized in that: In S3.2, the moving targets include motor vehicles, non-motor vehicles, people and other types of moving targets.

8. The ultra-compression storage method according to claim 1, characterized in that: In S6, the video storage engine adopts the SVDSE video storage engine.

9. A system using the ultra-compression storage method according to claims 1 to 8, characterized in that: It includes interconnected cameras and end-side devices, and the end-side devices include an intelligent analysis module, a storage control module and several decoding units; after the end-side device obtains the video stream data from the camera, it selects the corresponding decoding unit to parse and decode the video stream data according to the bitstream type of the video stream data to generate the YUV data; then, the intelligent analysis module receives and analyzes the YUV data, separates the region of interest ROI and parses the corresponding structured metadata information; then, the intelligent analysis module encodes the parsed and separated data into a bitstream according to the bitstream type of the video stream data; finally, the structured metadata information and the encoded bitstream are output to the storage control module, thereby completing the disk storage of the video data and metadata.

Images