Browsed-Back Post Event Analyzer

Abstract

Face recognition is considered as the integral part of the video surveillance or video security system. Most of the video surveillance systems recognize human on the basis of face, but the careful study shows that faces become vague from cameras either due to mask worn by terrorists or thieves, or due to distance from camera. This constraint makes thorny the investigation of an event after it has occurred. The objective of this paper is to propose a system that can recognize humans on the basis of their cloth color irrespective of their face ambiguity so that wearing of mask and distance of people from camera does not make any difference. The system is composed of two parts. One part detect the humans present in the video, segment the detected humans according to their trouser and shirt color and store the parameters of these clusters found with color based K-mean clustering. The second part is the query system. It provides an interface to the user to enter the query to get the required information from the video. By using this system we can convert a lot of available data (in the form of video) in to fruitful information. The system can also be used as image indexing with some modifications.