This research reports on a study of two methods of recognizing objects in images: the generalized Hough transform and the Procrustean transform. The Hough transform is an established technique which requires much computer time and memory for its implementation. Improvements have been made to the generalized Hough transform which increases speed and decreases the required memory. The Procrustean transform is new and it belongs to the family of least squared error techniques. We report on an experimental investigation to validate the technique, and to extend it to recognize objects of various sizes and orientations.
展开▼
