Moving shadow detection based on normalized eigenvalue of Wishart matrix

Moving objects detection is one basic problem in computer vision and various methods have been proposed. However, the performance of these methods may be deteriorated by the moving shadows. In this paper we present a novel method for the moving shadow detection. Based on the analysis of the illumination model, we prove that the normalized eigenvalue of image block is illumination invariant. The distribution of the normalized eigenvalue is discussed and a significance test is performed to classify each image pixel into foreground object or moving shadow. Experimental results on typical scenes show that the proposed method can detect moving shadows correctly. Quantitative comparison with the state-of-the-art demonstrates that the proposed method provides much improved results.