Method for rapidly retrieving motional trajectory of interested video target

Abstract

The invention relates to a method for rapidly retrieving a motional trajectory of an interested video target. The method comprises the following steps: (1) integrating massive videos, and correlating geospatial information; (2) converting heterogeneous data to generate a video stream with a unified format for video viewing; (3) presetting a topological camera relationship table in a database, traversing all cameras shooting the target; (4) editing and recording elements of video pictures of all the cameras shooting the target, and storing the elements in the database; (5) according to the recorded elements, splicing the videos of the whole motional trajectory of the interested target according to the time sequence; forming a motional trajectory route of the target on a GIS map according to the time sequence; (6) outputting a video route graph of the motional trajectory of the target with the same format so as to facilitate archiving and later viewing. By the method, the target can be found rapidly and the spatial motional trajectory of the target can be intuitively acquired, so that the retrieving efficiency is improved, and the viewing time is saved.

Description

technical field [0001] The invention relates to a method for quickly retrieving the movement track of a video attention target. When the target moves continuously in a certain monitoring area, the movement track of the focus target can be directly grasped by viewing several camera videos at the same time, which belongs to the field of video monitoring. Background technique [0002] With the continuous development of systems such as "Safe City" and "Smart City", video surveillance has been applied on a large scale, accumulating massive video data resources, providing effective resources for relevant departments to find targets of concern, and playing an important role in maintaining public safety. To a certain effect. However, when it is necessary to grasp the trajectory of a target of interest, how to quickly find the actual spatial trajectory of the target of interest in massive video data resources has become a huge challenge for the video surveillance industry. Due to th...

AI Technical Summary

Problems solved by technology

[0003] (1) Surveillance videos are scattered and have different formats, and it is difficult to import and export videos uniformly: the surveillance videos used in the society come from different manufacturers, and the data encoding and packaging formats they adopt are not the same, and sometimes different players need to be used To view the video, and the distribution of the monitoring network is relatively scattered, it is necessary to export the video one by one on site

[0004] (2) Manual viewing of videos is heavy work and inefficient: a large number of video resources need to adopt "human sea tactics". Manually focusing on the same video resource for a long time, it is easy to get tired and unable to concentrate, and it is possible to ignore some important points in the video. Target

[0005] (3) It is difficult to archive video resource data, and the correspondence between clues and videos is confusing: the current file directory management method is mainly used for case filing, which requires other methods to record important content of videos, resulting in the inability to correspond between clues and videos

[0006] (4) Insufficient spatio-temporal information of video clues makes it impossible to intuitively obtain the target's trajectory: Usually, the behavior process of focusing on the target involves multiple cameras, and manual participation is required when analyzing the target's trajectory. There is a lack of effective geographical analysis tools and a schematic diagram of monitoring distribution points

Images