Moving Cast Shadow Detection by Energy Minimization
Yufei Zha, Yuan Yang, Ming Zhang, BI Du-yan · 2007
Moving cast shadow causes serious problem while segmenting and extracting true foreground from image sequences, due to the misclassification of moving shadow as foreground. This paper proposes a novel approach for moving cast shadow detection. Firstly, the color and texture information of moving cast shadow is employed with the temporal-spatial coherence of shadow and foreground in a probability framework; afier that, the maximization of posterior probability can be converted to energy minimization problem through Gibbs energy; finally, moving cast shadow and foreground can be segmented accurately through graph cut. Results show that the proposed method excels classical method both in indoor and outdoor scene.