Mobile Robot Vision for Object Tracking using Modified SIFT

Abstract

For a Mobile robot vision object tracking is the important application task. The main problem which has been faced in Image Based Visual Servo (IBVS) is that to detect and track objects in a series of images. In this present paper, we are going to implement a Modified SIFT algorithm to detect & track the object in various conditions such as changes in intensity, scaling factor and circular rotation. This algorithm first of all find all local feature keypoints from the image taken by the robot while is moving. In the next step, it compares these local features with the reference template. At the end based upon matched local keypoints, it approximates the position of the object, so that it can easily track the object.