Machine Vision, 3rd Edition

CHAPTER 15 Abstract Pattern Matching Techniques

15.1 Introduction

We have seen how objects having quite simple shapes may be located in digital images via the Hough transform. For more complex shapes this approach tends to require excessive computation. In general, this problem may be overcome by locating objects from their features. Suitable salient features include small holes, corners, straight, circular, or elliptical segments, and any readily localizable subpatterns. Earlier chapters have shown how such features may be located. However, at some stage it becomes necessary to find methods for collating the information from the various features in order to recognize and locate the objects containing them. This task is studied in the present ...