Target tracking method based on rotor UAV platform

Abstract

The invention discloses a target tracking method based on a rotor UAV platform. Through multi-scale sample collection and integration of a classifier updated in real time, the method realizes quick and accurate online real-time tracking of a moving target by a rotor UAV platform. During the tracking process, according to the maximum classifier response value of a current frame and the change of the maximum classifier response value of the current frame relative to that of a previous frame, the method integrates a tracking performance judging result of the previous frame to judge whether the tracking performance of the current frame is stable. When the tracking performance is not stable, the classification result of the classifier is corrected in time, so that the method can effectively prevent a tracked target from being lost during target tracking process due to shielding. Compared with the prior art, the method achieves quick and accurate online real-time tracking of a moving target by a rotor UAV platform. Besides, the method has multi-scale tracking characteristics, good tracking performance, lower algorithm complexity, better real-time performance, and lower requirement on hardware resources.

Description

Technical field [0001] The invention relates to the technical field of image processing and pattern recognition, in particular to a target tracking method based on a rotary wing drone platform. Background technique [0002] In recent years, with the increasing level of science and technology, the development of the UAV business has been flourishing. Due to its simple structure, low cost, high flexibility, strong environmental adaptability, and easy loading of various sensors and equipment, coupled with the rapid advancement of digital communications, sensors, microelectronics, virtual reality and other technologies, UAVs It has broad application prospects and development space in military and civilian fields such as air combat, urban armed counter-terrorism, maritime search and rescue, and the use of video captured by drones to track targets is one of the important applications. [0003] According to different backgrounds, the recognition and tracking of moving targets can be divi...

AI Technical Summary

Problems solved by technology

However, it is difficult to obtain an accurate geometric model of a moving target in practical applications, and the amount of calculation is large, and the algorithm has poor real-time performance.

[0005] After the UAV is equipped with a camera, the moving target detection is based on the dynamic background. Although the field of view of the camera is expanded, there is a complex relative motion between the target and the camera, which puts forward higher requirements for the target tracking algorithm.

Images