IMAGE BLOCK CLASSIFICATION USING STOCHASTIC IMAGE SEGMENTATION

The authors propose a new image block classification method. The propo sed algorithm incorporates image context into the classification via p ixel-based segmentation. To obtain a segmented image they adopt the st ochastic model-based unsupervised image segmentation algorithm. Since the block classifier considers the grey level distribution in the bloc k, it can differentiate edges from textures. Also, since the segmentat ion is executed independently at each small block, a parallel processo r can be applied to obtain a real-time block classification.