Detection and Tracking of People's group

This paper presents a method for monitoring human activities in real time on video sequences. The monitoring process consists in detecting people in video sequences by motion detection using a background model, where each pixel is modeled as a mixture of Gaussian. The detection of people is performed by skeletonization of the movement objects. The segmentation of groups of people is obtained by analyzing the minimum vertical histogram of the moving objects. The results show that the scheme is suitable to detect activity in specific scenarios, even when lighting changes have on the scene or the video quality is low it acquired. Index Terms—Detection, segmentation, background, tracking.