Cross-camera trajectory tracking method and device based on cooperation of multiple identification technologies, and storage medium

Abstract

The invention provides a cross-camera trajectory tracking method and device based on cooperation of multiple identification technologies, and a storage medium. The method comprises the steps of obtaining information of a to-be-tracked target obj and a spatial position Pos0; setting a monitoring video range region0 to be analyzed, wherein the region0 is a range taking the spatial position Pos0 as a circle center and the radius of the range is R; performing cooperative positioning on the to-be-tracked target based on the information of the to-be-tracked target obj and the obtained video data in the monitoring video range regon0 by using multiple identification technologies to obtain a positioning result Pos1; and a plurality of positioning results Posn-1... Posn are obtained by repeating the setting step and the multi-identification technology cooperative positioning step, and Pos0, Pos1... Posn are used as the real-time trajectory of the to-be-tracked target obj. According to the invention, through cooperative work of a plurality of identification technologies, positioning is accurate, tracking trajectory is complete, hardware cost is reduced by using a small amount of hardware equipment resources in a video relay mode, and a specific conflict processing strategy is provided.

Description

technical field [0001] The present invention relates to the technical field of image recognition, in particular to a cross-camera trajectory tracking method, device and storage medium based on the collaboration of multiple recognition technologies. Background technique [0002] In the field of computer vision, the problem of video-based object tracking has always been an important topic and research hotspot. With the increase in the number and coverage of urban public security video surveillance cameras, the target trajectory tracking technology based on video images has received more attention and application in social governance and maintenance of public security. [0003] At present, for the tracking and control of key objects, the methods of video image target (pedestrian) trajectory tracking technology are as follows: 1. Through people staring at the video surveillance screen, the target object is positioned to form trajectory tracking; 2. Based on human Face capture c...

AI Technical Summary

Problems solved by technology

[0004] There are many deficiencies and bottlenecks in the three methods of the above-mentioned video image target (pedestrian) trajectory tracking technology, mainly: Method 1. People staring at the monitoring consumes a lot of manpower and material resources, people's energy is limited, tired, and it is easy to cause the target to lose track. It is difficult to popularize and use; method 2 is currently a very mature face recognition technology, which can accurately locate the target object, but this method relies on face capture cameras. Loopholes and holes in the spatial position are easy to cause the target object to be lost, and eventually the object trajectory is incomplete; method 3 uses the video structure technology to extract the pedestrian (human body) object and its attribute characteristics in the video image, and then uses the pedestrian re-identification technology and Pedestrian attribute feature comparison technology, locating whether key objects appear at the positions of each surveillance camera, this method requires structural analysis of all surveillance videos in an area, and in order to achieve real-time tracking targets, this requires a large number of hardware server equipment, The use cost of this method is greatly increased, and in terms of the maturity of the current technology development and the actual combat effect, the accuracy of pedestrian recognition by this method will be slightly lower. These two reasons lead to the lack of widespread application of this technology at present.

Images