The application of cosine transform and principal components for foreground detection in video

Detection and tracking of moving objects in video sequences are essential for many computer vision applications & it is considered as a challenging research issue due to dynamic changes in objects or background appearance, illumination, shape and occlusions. In this article, we proposed a robust algorithm to detect moving objects in a video, based on the combination of discrete cosine transform (DCT) and principal component analysis (PCA). The elementary frequency components are obtained from separable property of DCT and the dimensionality of these components is usually high. In order to reduce dimensionality and to extract effective features of the elementary frequencies, PCA approach is used. The proposed method is tested on standard PETS dataset and other real time video sequences collected from various sources. Experimental results obtained for the proposed method are encouraging.