Research And Realization Of Moving Objects Tracking And Classification Algorithm Based On Video

Intelligent video analysis technology is mainly used for processing a sequence of videoframes which contain all kinds of moving objects, and detecting, tracking and classifing targetsfrom complex scenes, and then analysising and understanding their behavior. Classification is asignificant part of intelligent video surveillance, and it is the basis of behavior analysis andunderstanding of object. The content of the research is to select discriminative feature to classify theextracted moving objects based on the detection and tracking of moving targets.This study is based on classification technology of moving targets on the roads with static andsingle camera. On the basis of summing up existing algorithms, the study puts forward an improvedalgorithm of target classification based on shape features and support vector machines. It classes theobjects detected in video into two categories: vehicle and person. In the detection of the movingobjects, this study makes use of moving object detection algorithm based on background subtractionalgorithm, and deal with the detected moving targets. These processing methods include shadowdetection and subtraction, morphological filtering, splitting and marking of connecting area, andrationality judgment, thus improving target detection results effectively. At the same time, in viewof the situation that there are static targets in Original video frames, the study optimizes traditionalbackground updating algorithm.The study checks and verifies programs in the condition of live video of roads. Experimentsproved that the method used in this study could achieve accurate classification of moving objectsand obtain good classification accuracy. Thus the study has huge practical significance.