Moving Object Tracking Method Under Traffic Scene

With the social and economic development, Intelligent Video Surveillance Systembecomes the effective means of traffic management. Tracking of moving object is the keytechnique of intelligent video surveillance system. It includes three parts: moving objectsdetection, feature extraction and moving target tracking.This paper detects moving objects from video sequence .And making use of thecharacteristics of wavelet transform which key information of video sequence are mainlydistributed in low frequency region, this paper proposes target tracking method based onwavelet transform and particle filter. The purpose is to eliminate noise , and reduce the featurematching region.We take some comparison and analysis among Hu moment, Zernike moment and Affinemoment invariants. The result is Affine moment invariants with the optimality .This paperproposes the target tracking method integrate particle filter and Affine moment invariants. Toreduce the computation, this paper gives a feature extraction method which based on edgeinformation.The results of tracking are related to characteristics of targets and tracking methods.This paper use multifeaturesand particle filter tracking method to achieve the purpose ofstable tracking in complex cases in the early study. This paper proposes the tracking methodwhich integrated color and Affine moment invariants. Experimental results show that thealgorithm is feasible and robust.