A Methodology for Evaluating Range Image Segmentation Techniques

. This paper describes a definition of the range image segmentation (of polyhedral scenes) problem, a data set to use in evaluation, a method for specifying ground truth, and a set of metrics to classify segmentation results against ground truths. 1 Introduction "In general, standardized segmentation error metrics are needed to help advance the state-of-the-art. No quantitative metrics are measured on standard test images in most of today's research environments." (page 11, 1988 NSF Range Image Understanding Workshop proceedings [9]) "Comparison of segmentation results is difficult. This is because of the difficulty in implementing other people's algorithms due to lack of necessary details. In many cases, we have not been able to reproduce the published results by using the authors algorithm. This is further complicated by the fact that there is no standard evaluation criterion." (PAMI, May 1994 [12]) Interesting comparative studies have recently been performed for stereo analysis [4]..