REASONING ABOUT EDGES IN SCALE SPACE

The importance of applying different knowledge has been recognized since the early days of computer vision. A common belief in the field is that the low-level processes are dominated by data-driven operations such as edge detection, and the high-level processes use explicit knowledge. This belief has resulted in emphasis on filtering operations in the low-level processes and on reasoning approaches in the high-level processes. Many techniques have been developed for low-level vision processing, but their performance on real images is far from satisfactory. This paper explores the role of reasoning in early vision processing. In particular, we address the problem of detecting edges. We do not try to develop one more edge detector, but rather, we study an edge detector rigorously to understand its behavior well enough to formulate a reasoning process that will allow appliance of the detector judiciously to recover useful information. We present a multiscale reasoning algorithm for edge recovery: reasoning about edges in scale space (RESS). The knowledge in RESS is acquired from the theory of edge behavior in scale space and represented by a number of procedures. RESS recovers desired edge curves through a number of reasoning processes on zero crossing images at various scales. The knowledge of edge behavior in scale space enables RESS to select proper scale parameters, recover missing edges, eliminate noise or false edges, and correct the locations of edges. A brief evaluation of RESS is performed by comparing it with two well-known multistage edge detection algorithms.