On the Stability of the K-Max Distance to the Position of Seeds
Michael Holuša, Eduard Sojka · 2018
The position of seeds plays a key role in the quality of image segmentation, especially if the segmentation is based on computing the distances from the seeds. If the seeds placed in different positions lead to the stable segmentation result, it simplifies the user interaction. In this paper, we investigate how the k-max distance, which we introduced in [1], is stable, and how this stability influences the results of a distance-based image segmentation. We compare the results of the k-max distance with the classic geodesic distance, and with the minimum barrier distance. We note that the k-max distance seems to be a logical missing link between the max-arc distance (k=1) and the geodesic distance (k→∞).