Abstract：In this paper a novel method is presented to segment gray level image through maximizing the fuzzy partition entropy of two-dimensional histogram. After the concept of fuzzy partition is briefly introduced first, a new definition of fuzzy partition entropy based on condition probability and condition entropy is presented. Then, the multi-dimensional triangular-norm is applied to construct generalized Cartesian product of non-interactive fuzzy sets, and also an approach for generating multi-dimensional fuzzy partition is presented. Finally, a new method for segmenting gray level image through maximizing the fuzzy partition entropy of two-dimensional histogram is put forward. Experiments are conducted on real object pictures, and the results show that the approach presented herein performs better than some classical threshold selection methods do.
2002, Vol. 24 
