Introduction

Detection and tracking tasks can be formulated in terms of comparing areas in images. If we're able to find special points in the images and build up descriptors for these points, we can just compare the descriptors and arrive at a conclusion about the similarity of the objects in the images. In Computer Vision, these special points are called keypoints, but several questions arise around this concept: how do you find truly special locations in the images? How do you compute the robust and unique descriptors? And how do you compare these descriptors rapidly and accurately? This chapter addresses all these queries and leads you through all the steps from finding the keypoints to comparing them using OpenCV.