An improved moving object tracking algorithm based on CAMSHIFT
Xiaoyan Yang, Jinyu Yang · Journal of Physics Conference Series · 2021
Abstract There is a problem that Camshift algorithm relies too much on color as a criterion, in order to solve it, this paper presented an improved moving object tracking algorithm which combined Kalman filter predictor with Camshift algorithm. Firstly, the object model was built by Camshift Algorithm, and the centroid coordinates of the moving objectwere extracted, then the position of the object in the next frame was predicted by using Kalman filter, and finally, the value was used as the centroid coordinates of the object searched by Camshift algorithm in the next frame, proceeding with target match. The simulation results showed that this method can effectively solve the problem of target loss caused by the similar color between the target and the back. The new algorithm can effectively reduce the tracking error accumulation in the tracking process, greatly reduce the template drift, shorten the tracking time, and can have a better practicability and real-time performance.