Behavior of the CIE L*a*b* Color Space in the Detection of Saturation Variations During Color Image Segmentation

In this paper, a study of the behavior of the CIE L*a*b* color space to detect subtle changes of saturation during image segmentation is presented. It was performed a comparative study of some basic segmentation techniques implemented in the L*a*b*, RGB color space and in a modified HSI color space using a recently published adaptive color similarity function. In the CIE L*a*b* color space we have studied the behavior of: (1) the Euclidean metric of a* and b* color components rejecting L* and (2) a probabilistic approach on a* and b*. From the results it was obtained that the CIE L*a*b* color space is not adequate to distinguish subtle changes of color saturation under illumination variations. In some high saturated color regions the CIE L*a*b* is not useful to distinguish saturation variations at all. It can be observed that the CIE L*a*b* has better performance than the RGB color space in low saturated regions but it has worse performance in most high saturated color regions; all high saturation regions are very sensitive to changes in illumination and a minimum change causes failures during segmentation. The improvement in quality of the recently published color segmentation technique to distinguish subtle saturation variations is substantially significant.