Research on CAMshift Algorithm Based on Feature Matching and Prediction Mechanism

摘要

Aiming at the problem of target loss caused by background interference and occlusion in the traditional CAMshift algorithm for target tracking, a CAMshift tracking algorithm based on feature matching and prediction mechanism is designed. The algorithm re-locates the size and position of the target by ORB feature matching between the template target and the frame to be tracked, and realizes the accurate tracking of the target under the interference of similar background and complex background. In order to solve the problem of inaccurate tracking and feature matching under occlusion, the Kalman filter is used to predict the position of the occluded target. The experimental results show that the algorithm can accurately track the target under complex background, similar background and occlusion.