Comparative assessment of the Image Divide and Link Algorithm in different color spaces

Abstract

In this paper, a comparative assessment of the Image Divide and
Link Algorithm (ID&L) in different color spaces is presented. This, in
order to show the signicance of choosing a specic color space when
the algorithm computes the dissimilarity measure between adjacent
pixels. Specically, the algorithm procedure is based on treating a dig-
ital image as a graph, assigning a weight to each edge based on the
dissimilarity measure between adjacent pixels. Then, the algorithm
constructs a spanning forest through a Kruskal scheme to order the
edges successively while partitions are obtained. This process is driven
until all the pixels of the image are segmented, that is, there are as
many regions as pixels. The results of the algorithm have been com-
pared with those generated using dierent color spaces are shown.