Interactive approach to multiobjective genetic fuzzy clustering for satellite image segmentation

The problem of segmenting a satellite image can be posed as the task of clustering the pixels in the intensity space. Some recent studies have posed the problem of data clustering as a multiobjective optimization problem, where several cluster validity indices are simultaneously optimized to obtain robust clustering solutions. Since no validity index performs equally well in all kinds of images, identifying the best set of validity indices to be optimized simultaneously is therefore an important problem. In this article, we study an interactive genetic algorithm based multiobjective fuzzy clustering technique for satellite im-age clustering problem. The algorithm simultaneously finds the clustering solution as well as evolves the set of validity measures that are to be optimized simultaneously. The method periodically interacts with a human decision maker (DM) and adaptively learns to obtain the optimum set of validity measures along with the final clustering result. The performance of the technique has been demonstrated on an Indian city, Kolkata and compared with that of some other existing clustering techniques.