1. Region-Level Motion- Based Background Modeling and Subtraction Using MRFs Shih-Shinh Huang Li-Chen Fu Pei-Yung Hsiao 2007 IEEE

2. Abstract This paper presents a new approach to automatic segmentation of foreground objects from an image sequence by integrating techniques of background subtraction and motion-based foreground segmentation. This paper presents a new approach to automatic segmentation of foreground objects from an image sequence by integrating techniques of background subtraction and motion-based foreground segmentation.

3. Outline INTRODUCTION INTRODUCTION REGION-BASED MOTION SEGMENTATION REGION-BASED MOTION SEGMENTATION BACKGROUND MODELING BACKGROUND MODELING MRFS-BASED CLASSIFICATION MRFS-BASED CLASSIFICATION RESULTS RESULTS CONCLUSION CONCLUSION

4. INTRODUCTION In many applications, success of detecting foreground regions from a static background scene is an important step before high-level processing. In real-world situations, there exist several kinds of environment variations that will make the foreground detection more difficult.

5. Several kinds of environment variations Illumination Variation Gradual illumination variation Sudden illumination variation Shadow Motion Variation Global motion Local motion

6. System Overview

7. REGION-BASED MOTION SEGMENTATION motion vector

8. Region Projection Projecting regions in the previous frame to the current one, is to facilitate the segmentation.

9. Motion Marker Extraction The output of this step is a set of motion- coherent regions, all pixels within a region comply with a motion model.

11. Boundary Determination Merge uncertain pixels to one of the markers.

13. BACKGROUND MODELING A brief description of Stauffer and Grimson ’ s work is first given and then we introduce the Bhattacharyya distance as the difference measure between the region from the region-based motion segmentation and the one represented by the background model.

14. Adaptive Gaussian Mixture Models

15. Bhattacharyya Distance

16. Shadow effect However, the region similarity defined in this way will lead to misclassification of the background region where direct light is blocked by the foreground object.

17. An example of shadow effect

18. MRFS-BASED CLASSIFICATION Incorporate the background model to classify every region in the segmentation map SM into either a foreground object or a background one by MRFs.

19. MRFs Framework

20. Region Classification

22. RESULTS

26. CONCLUSION Experimental results demonstrate that our proposed method can successfully extract the foreground objects even under situations with illumination variation, shadow, and local motion. Our on-going research is to develop a tracking algorithm which can be used track the detected object.