Shadow detection based on texture features in gray sequence images

Moving object detection is usually effected by moving shadows from the object and it is easy to produce serious problems because the object and the shadow have the same moving characteristic.This paper proposes a method for moving shadow detection based on texture feature under the assumption that shadow regions in the current image have the same texture feature with a corresponding background image.For the discrimination between shadows and foregrounds,several texture similarity measurements are used and the shadows and foregrounds in current image are distinguished by computing the similarity of current image and background image.Furthermore,an adaptive method for segmentation of shadows and foregrounds is presented based on Ostu algorithm.In the end,experiments are carried on the video extracted from indoor or outdoor environments.Analysis indicates that the method has a excellent detection result in a video sequence with outstanding texture features.However,detection accuracy will be reduced when the image suffers blurring or noising.The obtained results verify the efficiency of the proposed method by using both of subjective evaluation and objective evaluation.