A constrained clustering algorithm for shape analysis with multiple features
Jorge Salvador Marques, Arnaldo J. Abrantes · 2002
This paper extends a class of constrained clustering methods for shape estimation by using the concept of extended features. The extended features consist of edge points and associated image properties, e.g., gradient, texture and color. Experimental results show that the use of extended features improves the performance of the algorithm in the presence of cluttered background.