Critical Performance Analysis of Object Tracking Algorithm for Indoor Surveillance using modified GMM and Kalman Filtering

Abstract

Our objective is to ensure high level of security in public places using static PTZ camera and robust detection and tracking algorithm for video sequences and also to generate the multi model background subtraction approach that can handle dynamic scenes. In this paper we are focussing to design the robust foreground and background detection technique using the statistical approach and implement it on the different indoor environments. Among all familiar background subtraction approach our proposed method used Gaussian mixture model and predictive filters to detect and track objection successive frames. To test the performance of our algorithm we will take the standard test sequences and own datasets. We will analyze our algorithm with the qualitative and quantitative approaches. so, our aim is to develop such an smart surveillance system that can not only analyze but also to interpret and act with reference to the object beheviour against illumination changes, clutter background, moving background, occlusions and complex silhouette.