Background Modeling and Foreground Object Detection for Indoor Video Sequence

This paper proposed efficient and reliable algorithm for background subtraction for indoor video sequences. The paper mainly focused on compensating the illumination variation in the frame and then applying improved Gaussian mixture model to build the background model and then detect the moving foreign objects in indoor video sequences by using the Euclidian distance as metric. To compensate the illumination variation, homomorphic filtering algorithm for color image in HSV color space was proposed in the paper. The paper also reported the performance evaluation of the proposed method and existing and found that the proposed algorithm achieved 80 % of improvement in detecting actual foreground objects in indoor video sequences having illumination variations.