A sparse error compensation based incremental principal component analysis method for foreground detection

Abstract

Foreground detection is a fundamental task in video processing. Recently, many background subspace estimation based foreground detection methods have been proposed. In this paper, a sparse error compensation based incremental principal component analysis method, which robustly updates background subspace and estimates foreground, is proposed for foreground detection. There are mainly two notable features in our method. First, a sparse error compensation process via a probability sampling procedure is designed for subspace updating, which reduces the interference of undesirable foreground signal. Second, the proposed foreground detection method could operate without an initial background subspace estimation, which enlarges the application scope of our method. Extensive experiments on multiple real video sequences show the superiority of our method.