Target tracking optimization method under high-speed dome camera

Abstract

The invention relates to the field of target tracking, and discloses a target tracking optimization method under a high-speed dome camera, which comprises the following steps of: triggering selectionof a tracking target; initializing the target by using a kcf operator; a kcf operator is used for selecting a target, and the selected target has similar characteristics with the target during initialization and is subjected to nearest neighbor matching with the target detected by a target detection algorithm; carrying out IOU matching on the selected target and the tracking target of the previousframe; if matching fails, if so, taking the target center selected in the step S3 as a circle center; searching targets within radius gamma, otherwise, taking the target detected by the target detection algorithm as a final tracking target; preserving relevant location information, according to the method, the advantages of the kcf operator and the advantages of the deep learning target detectionalgorithm are integrated, the tracking algorithm is optimized, the target can be accurately tracked in the actual complex and variable environment, and therefore the problem that a high-speed spherical tracking camera is difficult to track and apply in the complex environment is solved.

Description

technical field [0001] The invention relates to the technical field of target tracking, in particular to a target tracking optimization method under a high-speed dome camera. Background technique [0002] In the target tracking implementation of the high-speed dome tracking camera, since the surveillance camera will rotate with the tracking target, the position information of the tracking target will change continuously in the image, and in the case of multiple targets appearing at the same time, it is difficult to lock the initial position. Track the target, and thus lose the track target. In the traditional tracking methods, a single algorithm structure is mostly used, such as the kcf operator to track the target, or the two-frame image difference method is used to obtain the foreground target information. However, this method is difficult to continuously track the target when the environment is complex, there are many targets and not a single case, and it is easily affec...

AI Technical Summary

Problems solved by technology

[0003] The purpose of the present invention is to solve the disadvantages of the traditional tracking method in the prior art that it is difficult to continuously track the target when the environment is complex, there are many targets and not a single one, and it is easily affected by ambient light, and proposes a high-speed dome Object Tracking Optimization Method Based on Large Camera

Images