A kind of target tracking method, system and pan-tilt camera used therefor

Abstract

The invention discloses a target tracking method comprising: acquiring a preliminary tracking window in a current frame image, extracting FAST feature points of a tracking target in the preliminary tracking window; calculating all the FAST feature points when there is a previous frame image The Gaussian weighted sum of the optical flow information; Utilize the preliminary tracking window and the Gaussian weighted sum of the optical flow information to perform Kalman filter prediction to obtain a predicted tracking window; filter the preliminary tracking window and the predicted tracking window to obtain a final tracking window; Sending a control command to the steering gear, so that the steering gear adjusts the angle of the pan / tilt camera according to the final tracking window until the tracking target is at the center of the final tracking window. It can be seen that the object tracking method disclosed in the present invention realizes object tracking on the mobile platform. The invention also discloses a target tracking system, a pan-tilt camera and a computer-readable storage medium, which can also achieve the above-mentioned technical effects.

Description

technical field [0001] The present invention relates to the technical field of image processing, and more specifically, relates to a target tracking method and system, a pan / tilt camera device capable of computing, data sending, and program storage, and a computer-readable storage medium. Background technique [0002] Currently, moving object tracking, as a branch of video data analysis technology, provides a basis for high-level video understanding and scene interpretation, and has broad application prospects in many fields such as traffic control, visual navigation, and unmanned equipment detection and monitoring. [0003] In the prior art, a target tracking method based on FAST corners and pyramid KLT, the camera equipment is fixed, and for fast moving targets, real-time tracking cannot be achieved, and the target is prone to be lost, or the camera cannot be kept facing the target. state. At the same time, extraction algorithms based on high-dimensional features (such as...

AI Technical Summary

Problems solved by technology

[0003] In the prior art, a target tracking method based on FAST corners and pyramid KLT, the camera equipment is fixed, and for fast moving targets, real-time tracking cannot be achieved, and the target is prone to be lost, or the camera cannot be kept facing the target. state

At the same time, extraction algorithms based on high-dimensional features (such as scale-invariant feature SIFT) have low time efficiency and high hardware requirements

[0004] In addition, although most of the background effects can be eliminated by using the background compensation method, it is only suitable for scenes with a relatively simple background. Once the background is complex, a large amount of noise information will be generated, the target cannot be determined, and the real-time performance is not high.

Images