D.A. Lyon (USA)
Edge detection, heuristic search, A*.
We describe an interactive edge detection algorithm that heuristically searches a pixel space. The interactive algorithm uses operator guidance to perform computer assisted edge detection. We show such search techniques have path-planning applications, using a morphological skeleton as a heursitic cost function in the formulation of an evaluation function. The algorithm is suitable for applications where the quality of the edge is as important as the speed. The interactive nature of the algorithm makes it well suited to workstation environments. The algorithm uses human operators to speed the search by making it better informed. We find that humans are better at identifying important edges than unattended algorithms. The algorithm has a non-deterministic execution time, and can take more time on difficult images than on easy ones. Further, more human interaction can speed the search and improve the edge detection. Our algorithm is much faster than the dynamic programming approach and the edges appear generally as good.
Important Links:
Go Back
