Physics-Motivated Algorithms for Image Clustering

Image clustering, defined as the task of finding natural grouping of similar items, is one of the key tasks in computer vision and pattern recognition. This problem has many incarnations in mathematics and applied sciences. Recently many papers have been proposed to tackle clustering solutions using methods in physics, in particular statistical thermodynamics. Some of such method like Deternministic and Simulated Annealing, Weak Membrane, and Watershed Transform, aquired status of classical in the realm of Computer Vision. The goal of this research paper is to analyze two of the lesser known methods in great details after a short survey on clustering techniques is presented.