Image clustering by incorporating adaptive spatial connectivity

In this paper, we present a novel image clustering algorithm that has a new dissimilarity measure which incorporates the adaptive spatial information. The spatial connectivity of an image is controlled by a weighting factor so that it enhances the smoothness towards piecewise-homogeneous region and reduces the edge-blurring effect. Our method also utilizes the capacity maximization to evaluate the quality of the clustering result via mutual information maximization. The unreliable data points will be further processed to improve the clustering results. Experimental results with synthetic and real images demonstrate the effectiveness of our algorithm.