(A) new edge gap filling algorithm based on local operations

Abstract

One of the key problems of edge-based digital picture processing is the filling of small gaps between edge segments after edge extraction. We describe a new solution to this problem which draws its power from the locality of gap filling procedure. This efficient algorithm finds by raster scanning the edge end-points from which filling operations are started and propagated only to the neighborhoods. Specifically, it restricts the search area for an edge extension to be filled to the direction in which each edge would most likely extend. We tested three algorithms including ours with a number of sample images to compare the performance. The experimental results confirm the superiority of our algorithm to others in terms of the time and storage complexity.