Video processing method and system

Abstract

The invention provides a video processing method and system. The method comprises the steps of: extracting each motion target and motion track thereof from a video stream; for each motion target, determining video frames of each motion target during emergence and disappearance according to the motion track respectively; and searching the video frames between the video frame on motion target emergence and the video frame on motion target disappearance in the video stream, labeling the searched video frames and the video fames of the motion target during emergence and disappearance together to be a video sequence corresponding to the motion target. According to the invention, video data can be located more rapidly and accurately and are easier to look up and discriminate, and data search management is more convenient.

Description

technical field [0001] The invention relates to video analysis and processing technology, in particular to a video processing method and system. Background technique [0002] With the gradual popularization of public security and social video surveillance systems, a large amount of surveillance video data has accumulated. These video data can play an important role in applications, such as providing strong support for the detection of a large number of cases, and providing rich resources for criminal investigation applications. Clues and evidence materials; while monitoring video data brings great help to the application, it also presents obstacles and technical bottlenecks in the application: [0003] 1. The video content information is huge, and it is inconvenient to review. For example, usually a simple case needs to review the video data which usually takes several to ten hours in total, while some important cases may be related to the video data collected by more than ...

AI Technical Summary

Problems solved by technology

However, in the investigation of criminal investigations, it often takes more than four hours for one person to review a one-hour video. Therefore, such a case often requires more than a dozen people to work overtime for more than a week to review.

At the same time, because surveillance video content is often not continuous and logical, the viewing process can easily cause fatigue, causing some important details to be missed during a single viewing process, and it takes more time to review repeatedly, which greatly increases workload

[0004] 2. The sources of data are complex and inconvenient to play

However, due to the different DVR manufacturers and models, the collected video formats are different. There is no unified player for various video materials. It needs to be played by the DVR manufacturer's own player, and there is no unified decoding player. most prominent issue

Moreover, the DVR file player has weak functions, only has the function of single-channel video playback and related basic controls, and cannot achieve multi-channel video synchronous playback, making it difficult to compare and review, and even the vast majority of DVR file players cannot accurately position to frame

[0005] 3. The number of video files is huge and inconvenient to manage

As the number of case videos gradually increases, it is inconvenient to retrieve single or multiple videos related to the case, and it is more difficult to retrieve important clips and description information of important clips in case-related videos

[0006] 4. There are many related documents, and it is impossible to establish an accurate correspondence with the video content

[0007] 5. The post-processing is complicated and the data is inconvenient to use

After the investigator looks at the video to find the suspect, it is usually necessary to distribute the relevant video or video clips and related information description files to various places for identification. After the identification is confirmed, it needs to be exported as an evidence file. The management method requires a series of processes such as interception, code conversion, and non-linear editing software splicing from large segments of video through the original DVR software, which is very time-consuming; while the management of video-related information description files is through copying Simple method, it is easy to misuse and cause inconsistent or missing video related information description files

This requires a high level of computer software use by the operators, and there is no countermeasure for the human errors of the operators, which also affects the efficiency of case handling to a certain extent.

[0008] The above problems have seriously affected the efficiency of the application of monitoring data in the fields of public security criminal investigation and other fields, and at the same time caused a lot of manpower consumption

Images