A comparison study of image segmentation by clustering techniques

Many image segmentation methods based on clustering are available in the literature. Some of these techniques use classical clustering, some use fuzzy sets. Most of these techniques are not suitable for noisy environments. Some work has been done using the possibilist clustering approach which is robust to noise, but requires knowledge of the number of clusters. The probabilistic approach can be used to assess the number of components. This paper reviews and summarizes some of these techniques. Attempts have been made to cover hard, fuzzy and possibilist approaches as well as mixture model clustering. Adequate attention is paid to possibilist clustering in discarding noisy pixels and to entropy criteria in assessing the number of clusters. We also propose a quantitative evaluation of segmentation results.