Paired Region Approach based Shadow Detection and Removal

Shadow detection and removal in various real life scenarios including surveillance system, indoor outdoor scenes, and computer vision system remained a challenging task. Shadow in traffic surveillance system may misclassify the actual object, reducing the system performance. However, shadow causes problems in computer vision applications such as segmentation, object detection and object counting. That’s why shadow detection and removal is a pre-processing task in many computer vision applications. So we propose a simple method to detect and remove shadows from a single natural scene image. The proposed method begins by selecting natural scene image and by segmenting method we focus only on shadow part. In image classification, we distinguish between the shadow and non shadow pixels. So that we able to label shadow and non shadow regions of the image. Once shadow is detected that detection results are later refined by image matting and the shadow-free image is recovered by removing shadow region. Examination of number of examples indicates that this method yields a significant improvement over the previous methods. IndexTerms—Shadow detection, segmentation, region classification, shadow removal, matting. ________________________________________________________________________________________________________