21. Machine Vision

The goal of computer vision is to extract information from images—a method that can produce a structure from motion, can recover a three-dimensional model of an object from a sequence of views, e.g., grasping by robot, medical imaging, and graphical modeling. This chapter presents the machine vision applications, the basic principle of vision, cognition and classification, and cognitive architecture. The cognition is—going from image-to-scene, which can be achieved through inversion by fixing scene parameters, inversion by restricting the problem domain, and inversion by acquiring additional images. The machine vision techniques are presented here for low-level, middle-level, and high-level vision. The last one requires indexing of images which can be searched through geometric hashing. One of advanced areas of vision—the object tracking, is presented in-depth, which requires the sequences—subtraction of image from background, segmentation of the image, and learning, followed with tracking. Finally, the axioms of vision, tools for computer vision, chapter summary, and practice exercises are presented.