Background subtraction method using codebook-GMM model

Abstract

In this paper, we present a new practical background subtraction method taking advantages of the conventional codebook and GMM-based approaches. The fundamental idea is approximating GMM parameters based on color statistics of background pixels which are clustered by the computationally efficient codebook scheme. The experiments on real visual surveillance dataset demonstrate that the performance of the proposed method is excellent in the aspects of subtraction accuracy and processing time.