Multi-targets Detection and Tracking Algorithm under Complicated Movements

Zhihong Xi,Daiyu Xing,Xichao Xu

DOI: https://doi.org/10.3969/j.issn.1671-4598.2014.01.017

2014-01-01

Abstract:Owing to the problem of too slow movements and inter-block may come up in video sequences,a new algorithm is presented based on improved Gaussian mixture model and Kalman filter.In the background motion detection,a fusion of three frame difference method and Gaussian mixture model is proposed to solve the problem that slowly moving target detection area is incomplete.In the trace,Kalman fil-ter use the target detection results to complete forecast tracking and the adaptively observation noise matrix to reinforce the stability of the tracking.Experimental results show that the algorithm can well ensure the detection and tracking of the slowly,blocky and irregular Multi-targets by comparing the previous algorithms.