Efficient edge extraction of images by directional tracing
S. S. Iyengar, Yuyan Wu, Hla Min · 2002
The ability to recognize edges of an object fast and accurately is a fundamental goal in computer vision and image processing. The facility is important in automatic manufacturing environment in industry. This paper defines and investigates an efficient method of edge extraction of image using directional tracing algorithm. Based on the approach of linear feature extraction presented by R. Nevatia and K.R. Babu (1980), our algorithm is developed for the general situation of edge extraction. The technique employs the basic and intuitive principle that an edge pixel should possess local maximal gradient but no more domain information or knowledge is required. An effective and robust tracing strategy is proposed. Our algorithm has been tested on different kinds of images, and the results are satisfactory.