A Novel Fuzzy Entropy Definition and Its Application in Image Enhancement

Abstract

In order to improve the image enhancement quality and to reduce the processing time, a novel fuzzy entropy definition for image self-adaptive enhancement is proposed based on the exponential behavior of information-gain and a fuzzy domain partition method M. The proposed fuzzy entropy definition can avoid the defect of logarithmic one and makes the definition much reasonable and makes the physical meaning of the definition much evident due to exponential definition. And the partition method M enables the optimal enhancement for different images. The self-adaptive fuzzy parameters are gotten by enumeration method and classic genetic algorithm (GA) based on maximum entropy principle respectively. The experiment results show that processing time based on the new entropy definition is cut down a little on condition that the image enhancement quality is better or unchanged compared to that based on the existed entropy definition. And parameters optimization of GA costs less time than that of enumeration method for the simple optimization problem which the fuzzy domain partition method M is given for different images. The automatic acquisition of the partition method M is the next research.