Learning a color distance metric for region-based image segmentation

In this paper we describe an experiment where we studied empirically and in depth the application of a global adaptive color space to be used in the similarity function of an established region-growing color image segmentation algorithm. To perform this experiment we chose the Mumford-Shah energy functional and the Mahalanobis distance metric. The objective was to test the approach empirically and in an objective and quantifiable way on this specific algorithm when using this particular distance model, without making any generalization claims. The empirical validation of the results was performed applying the resulting segmentation method on a subset of the Berkeley Image Database, an exemplar image set possessing groundtruths and validating the results against the ground-truths using two well-known segmentation validation methods, the Rand and BGM indexes. The obtained results suggest that to employ this adaptive color distance approach provides better and more robust segmentations, even if no other modification of the segmentation algorithm is performed.