Texture based classification and segmentation using the Laguerre-Gauss transform

In this paper, we introduce a new tool for the image segmentation problem. For this purpose, we have developed a texture based segmentation and classification strategy based on a representation of the image components by a family of circular harmonic functions (CHF) defining the Laguerre-Gauss transform (LGT). LGT is the polar counterpart of the Hermite transform. By utilizing the outputs of CHF operators at different scales, we use a multiscale segmentation and investigate the performance of CHF texture features.