Skip to Main Content
Image segmentation is one of the most fundamental steps of image analysis. Almost all vision based systems need segmented images using a good segmentation algorithm. The problem of automatically finding the best algorithm, from a pool of algorithms, on a per image basis has been largely ignored in the vision community. In this paper we present a novel solution to this problem based on classification complexity and image edge analysis.
Date of Conference: 25-27 Aug. 2011