Region Based Moving Shadow Detection using Watershed Algorithm

Region-Based Moving Shadow Detection using Watershed Algorithm

Abstract:

Moving shadow detection is an important task in computer vision, with applications several fields, such as surveillance, video conference, visual tracking and object recognition. In this paper, we present a Spatio-Temporal based Moving Shadow Detection (STMSD) method, The main idea is: according to the gradient change of current image, we utilize the watershed algorithm to achieve adaptive segmentation foreground sub-regions, Then, the changes in texture of adjacent video frames and current video frame and background are extracted, and each sub-region is classified in terms of texture feature to obtain the final moving shadow and target area. Extensive experiments demonstrate that STMSD is superior to some well-known methods especially pixel-based methods. Especially, our method exhibits much better performance compared with fixed block method, which can maintain the texture consistency in one region adequately.

Comments are closed.