Contextual clustering for satellite image segmentation

Several interesting strategies are adopted by the well-known Pappas clustering algorithm to segment smooth images. These include exploitation of contextual information to model both class conditional densities and a priori knowledge in a Bayesian framework. Deficiencies of this algorithm are that: i) it removes from the scene any genuine but small region; and ii) its feature-preserving capability largely depends on a user-defined smoothing (regularization) parameter. For these reasons Pappas´ algorithm is employed to provide sketches or caricatures of the original image. A modified version of the Pappas algorithm is proposed to segment smooth and noiseless images when enhanced pattern-preserving capability is required. Results show that the authors´ contextual algorithm can be employed: iii) in cascade to any noncontextual (pixel-wise) crisp c-means clustering algorithm, to enhance detection of small image features; and iv) as the initialization stage of any crisp and iterative segmentation algorithm requiring priors to be neglected on earlier iterations (such as the Iterative Conditional Modes algorithm)