Texture boundary detection using 2-D Gabor elementary functions

A framework is proposed for segmenting image textures by using Gabor filter s to detect boundaries between adjacent textured regions. By performing a m ulti-channel filtering of the input image with a small set of adaptively se lected Gabor filters, tuned to underlying textures, feature images are obta ined. To reduce the variance of the filter output for better texture bounda ry detection, a Gaussian post-filter is applied to the Gabor filter respons e over each channel. Significant local variations in each channel response are detected using a gradient operator, and combined through channel groupi ng to produce the texture gradient. A subsequent post-processing produces e xpected texture boundaries. The effectiveness of the proposed technique is demonstrated through experiments, on synthetic and natural textures.