SEGMENTATION OF TEXTURED IMAGES AND GESTALT ORGANIZATION USING SPATIAL SPATIAL-FREQUENCY REPRESENTATIONS

We address in this paper the generic issue of clustering/ grouping. Recent research, both in computer and human vision, suggests the use of joint spatial/spatial-frequency (s/sf) representations. Discussion on the spectrogram, the difference of Gaussians (DOG) representation, the Gabor representation, and the Wigner distribution is presented and a comparison is made. Experimental results in the area of texture segmentation and Gestalt grouping using the Wigner distribution are presented, proving the feasibility of using s/sf representations for low-level (early, preattentive) vision. © 1990 IEEE.