Research And Implement On Pedestrian Counting In Intelligent Video Surveillance

With the rapid development of economy, large-scale events and activities held,pedestrian counting in intelligent video surveillance is widely used in security. In thispaper, pedestrian counting based on moving targets detection, classification and trackingis implemented.On the research of moving targets detection, a method based on adaptivebackground model is proposed in this paper. This method can detect and updatechanging regions in background. According to the characteristics of shadow in the grayimage, we detect and suppress the shadow in foreground to improve the accuracy ofmoving targets detection.In the classification of the moving targets, we use aspect ratios and dispersions ofmoving targets to classify moving targets effectively. For the group target classified, thecurve fit method based on the linear relationship between Canny edge features and thenumber of people is adopted to count the number of people in it, which will be helpfulfor the following tracking.On the aspect of targets tracking, the individual target is difficult to track in thecrowd. Therefore, individual targets and group targets are extracted features andmodeled similarly in this paper. The tracking of moving targets is accomplished reliablythrough the Kalman filter.Experiment results show that the proposed algorithms meet the needs of real-timeand have high accuracy in pedestrian counting.