Stop Rule for Image Hierarchical Segmentation Algorithm
David G. Asatryan · Mathematical Problems of Computer Science · 2008
In this paper we consider an important problem of stopping the hierarchical segmentation procedure when the appropriate segmentation is achieved. This problem arises at every segmentation procedure, which uses a searching algorithmfor selection of acceptable decision. We propose an algorithm for stopping the hierarchical segmentation procedure. Stop-rule is based on the segmentation homogeneity measure, and uses a ratio of special sum of squares. The first sumequals to summarized variance of pixel intensity relative to the centers of intervals, being determined by thresholds, the second one expresses variance of mean values of the segments relative to the same centers. Examples of segmentation results to demonstrate the features and properties of proposed technique are considered.