Estimating the density of the people and counting the number of people in a crowd environment for human safety

The video-surveillance systems are popularly used in crowd monitoring and people detection in that crowd is estimated for security of humans in public places and also managing the resources. In India number of accidents have occurred due to crowd environment in public areas like Shopping malls, Airports, pilgrimage, Temples, Political Meetings etc.,. In order to avoid such accidents, we need to estimate the density of people in crowd environment. In this paper, our proposed method is divided into two fold. In first, we propose density estimation of the crowd size. Secondly, count the number of people in the crowd. As crowd density increases, the occlusion between the people also increases. In order to avoid such problem in crowd we can use Improved Adaptive K-GMM Background subtraction method to extract the exact foreground in real time applications to avoid the estimation problem. By applying boundary detection algorithm, we can estimate the size of the crowd. The number of people in a crowd is counted by using algorithm “canny edge detector”, “connected component labeling” method and “bounding box with centroid” method. This paper proposes a real time video surveillance system. Our proposed method is compared with existing method. The above proposed works are compared with different datasets like IBM, KTH, CAVIAR, PETS2009 and CROWD etc. It can be used for both testing and training phases. The aim of this work is to analyze performance of estimation and counting peoples with different datasets.