Major contribution of this article is to devise an efficient moving cast shadow segmentation technique that separates out the moving objects from their shadows casted on the background. It follows two major steps: background separation and shadow detection. For background separation, initially a background model is built. For a particular pixel location we construct the background model by taking the median of pixel values at the corresponding pixel locations in the temporal direction. To suppress the effects of quick change in illumination, and color frequency variation of texture background, in the proposed scheme we have extracted the RGB color features and ten local features at each pixel location in the considered target image frame and the constructed reference image frame. For background separation, a difference image is generated by considering pixel by pixel absolute difference of the thirteen dimensional target image frame and the constructed background model. This is followed by a spatial MRF constrained fuzzy clustering to find the moving regions in the considered scene. The maximum a'posteriori probability (MAP) estimate of the fuzzy statistic based MRF are obtained by fuzzy clustering. The MAP of the MRF constrained fuzzy clustering provides a binary image, where the moving objects with the moving cast shadow are identified as one group and the background is obtained as another group. To segment the moving object from its shadow we explore a three stage shadow analysis technique. It uses analysis of rg color chrominance property of shadow, local gray level co-occurrence based shadow processing followed by boundary refinement to separate out the regions corresponding to the moving cast shadow and moving objects. Performance of the proposed scheme is tested on several test video sequences. Effectiveness of the proposed scheme is verified by comparing the results obtained with those of some of the state-of-the-art techniques.
Bibliographical notePublisher Copyright:
© 2015 Elsevier Inc. All rights reserved.
All Science Journal Classification (ASJC) codes
- Control and Systems Engineering
- Theoretical Computer Science
- Computer Science Applications
- Information Systems and Management
- Artificial Intelligence