Usability of Cluster Based Co-Saliency in Video Foreground Detection

Merin Joseph, Anil A. R.

Abstract: Ability of human visual system to detect prominent regions in an image is fast, reliable and efficient. Computational modeling of this extraordinary behavior is termed as saliency detection. Saliency detection identifies salient regions in an image and is relevant in many computer vision applications such as object recognition, image segmentation, image retrieval etc. The term co-saliency can be explained as identifying common prominent regions in multiple images. In this paper we analysis various saliency and co-saliency detection mechanisms and studies the usability of cluster based co-saliency detection method in video foreground detection. The paper also explains design and implementation of cluster based co-saliency in video foreground detection.

Keywords: Saliency, Co-saliency, Clustering, K-means, Video Foreground Detection