Time-variable-parametric relaxation labeling and its application in texture segmentation

Relaxation labeling (RL) is a very important tool applied for exploitation of contextual information to reduce local ambiguity and achieve global consistency. In image segmentation, RL is used for post-processing the initial segmentation results, based on region or boundary information. Time-variable-parametric RL (TVP-RL) is proposed in this paper, aiming at the incorporation of a boundary strength map into the region based relaxation process by a time-variable compatibility coefficient and neighboring size. This method can remove scattered false segmented regions and maintain real boundaries at the same time. Compared with other region and boundary information hybrid methods, TVP-RL is very simple and easy to realize on a computer. Its application to texture image segmentation is also discussed.