Removal of object's shadow

Abstract

During bright day light or under strong lighting condition, shadows will appear and be part of an object in image. Object’ shadow in images may cause problem to several important algorithm in the fields of image processing such as object recognition, segmentation and object tracking. Removing the shadows can greatly improve the result of the image processing. To tackle this problem then, this project is divided into three parts; eliminating the surrounding objects of object interest, detecting shadows region and removing the object’s shadow. This project used moving object images to distinguish between object focusing and objects will be eliminated in image. Support Vector Machine (SVM) was used to classify the image into shadow and non-shadow region based on texture features. Then, the image of detected shadow region was differentiated with original image to remove the shadow. In this project, Matlab and Image Processing Toolbox were utilized and OSU-SVM was used as interface SVM to the Matlab environment