This chapter describes the need for image features, categorizes them in several ways, presents the constraints that may determine which are used in a given application, defines some of them mathematically, and gives examples of their use in research and in clinical settings. Features can be based on individual pixels (e.g., the number having an intensity greater than x; the distance between two points), on areas (the detection of regions having specific shapes), on time (the flow in a vessel, the change in an image since the last examination), and on transformations (wavelet, Fourier, and many others) of the original data.
Excluded from this chapter are discussions of the many methods of image enhancement and preprocessing (for example, noise removal, contrast improvement, edge detection) used in improving human visual understanding. A large literature exists, and reference to it for those methods will be made as needed.
Online access to SPIE eBooks is limited to subscribing institutions.