Abnormal moving target tracking method, system and device

Abstract

The invention relates to an abnormal target tracking method, system and device. Two cameras are used, large-range and wide-viewing-angle scanning type monitoring is completed by one of the two cameras,; when an abnormal moving target is found, the second camera is rapidly used for completing close-up of the abnormal moving target; therefore, an ultra-high-definition abnormal target image is obtained, thereby increasing identification accuracy. A first image acquisition module and a second image acquisition module rotate coaxially so as to be substantially consistent in the horizontal view angle, the second image acquisition module can be aligned with an abnormal target to perform image acquisition only by performing corresponding pitching view angle adjustment after the abnormal target isfound, so that the response time of information return and horizontal view angle adjustment is saved, and the response sensitivity is greatly improved.

Description

technical field [0001] The present invention relates to the field of video acquisition, in particular to a method, system and device for tracking abnormal objects. Background technique [0002] With the advancement of science and technology, in various fields such as security monitoring, border inspection, traffic maintenance, crowd monitoring, and explosion-proof emergency management, the intelligent requirements for video surveillance are gradually increasing, and the existing intelligent tracking pan / tilt is mainly based on the radar monitoring function. The intelligent tracking pan-tilt and AI-based enhanced mobile detection pan-tilt, but due to the limitations of related equipment and technology, there are mainly the following problems: [0003] 1) The intelligent tracking gimbal based on the radar monitoring function needs a time interval for the radar signal to be sent and returned. If the subject that needs to be tracked moves too fast in a unit of time, the received...

AI Technical Summary

Problems solved by technology

[0003] 1) The intelligent tracking gimbal based on the radar monitoring function needs a time interval for the radar signal to be sent and returned. If the subject that needs to be tracked moves too fast in a unit of time, the received radar signal may deviate greatly from the actual position of the object; and because Due to the particularity of radar technology, the degree of discrimination of moving objects is low, and video AI is usually used for processing and analysis, and there is a time delay between the video link and the radar link, which will affect the real-time performance and accuracy of the tracking effect

[0004] 2) The pan / tilt based on AI-enhanced mobile monitoring usually has an artificial intelligence tracking algorithm directly built into the pan / tilt. Due to the shooting resolution and hardware configuration for the algorithm, this method can only be applied to simple monitoring scenarios and cannot adapt to A more complex and sophisticated monitoring environment

To meet the needs of a variety of monitoring environments, it is necessary to make corresponding customizations for this environment, which is costly and difficult to maintain, which is not conducive to large-scale implementation

[0005] 3) Monitoring with a single use of the gimbal may cause incomplete tracking and detection when multiple changes occur at the same time because the gimbal can only detect and track one change; if multiple gimbals are operated at the same time, it may cause different gimbals Track the same subject, reducing the efficiency of monitoring and tracking

[0007] In the above-mentioned patents, in order to identify abnormal targets, two sets of monitoring are used, one of which finds abnormal targets, and the other zooms in on abnormal targets. The method of zooming in is to adjust the focal length for image acquisition. When the camera is at the focal length, especially for long-distance targets, the image obtained after adjusting the focal length is enlarged, but its clarity is significantly reduced, so it is impossible to effectively acquire ultra-high-definition images of abnormal targets

Images