λ-connectedness: method and application
Li Chen, Osei Adjei, Donald H. Cooley · 2002
Introduces a systematic approach, which we call a /spl lambda/-connectedness method, that can be applied to several real image processing problems, such as segmentation/classification, searching and data reconstruction. In this paper, we integrate previous research work into a network/graph-based system to build a unified framework for these processes. This technique is based on a graph G=(V,E) and an associated "potential function" /spl rho/ on the vertices of the graph. A measure C/sub /spl rho//(x,y) is defined for the /spl lambda/-connectedness on vertices x,y/spl isin/G with respect to /spl rho/. For a certain /spl lambda//spl isin/[0,1], x and y are said to be /spl lambda/-connected if C/sub /spl rho//(x,y)/spl ges//spl lambda/. If every pair of vertices are /spl lambda/-connected, then /spl rho/ is called /spl lambda/-connected on G. /spl lambda/-connectedness is a measure of continuity in discrete spaces or systems. In this paper, we introduce /spl lambda/-connected segmentation and /spl lambda/-connected fitting. The /spl lambda/-connected method can be viewed as a fuzzy system method, and it has a close relationship to rough sets. This method also has potential uses in network economics and resource management.