Simulation of Object Detection Algorithms for Video Survillance Applications

In recent days, video observation frameworks have turned into a greatly dynamic research region because of a sharp expansion in the levels of security and surveillance. Over the last couple of years, object detection researchers have been developing many new techniques for this purpose. In this paper, objects are detected from video sequence by making use of object detection algorithms like Gaussian Mixture Model, Haar algorithm, Histogram of oriented gradients and Local binary patterns. As the field of image processing is very application specific, different algorithms work well for different circumstances of video sequencing. The results obtained from simulations reflect that the Gaussian mixture model provides a very good detection rates for fixed type. But the idea of training a cascade detector provides more generic type of detection. For this type of detection, exclusive databases for those objects which are required to be detected are created. Local binary pattern algorithm for object detection required more number of training stages than Haar and HOG but provides accurate detection of specified object.