Distance measures for image segmentation evaluation

Abstract

The task considered in this paper is performance evaluation ofregion segmentation algorithms in the ground-truth-based paradigm.Given a machine segmentation and a ground-truth segmentation,performance measures are needed. We propose to consider the imagesegmentation problem as one of data clustering and, as aconsequence, to use measures for comparing clusterings developedin statistics and machine learning. By doing so, we obtain avariety of performance measures which have not been used before inimage processing. In particular, some of these measures have thehighly desired property of being a metric. Experimental resultsare reported on both synthetic and real data to validate themeasures and compare them with others. Copyright © 2006 Hindawi Publishing Corporation. All rights reserved.