Histogram thresholding by minimizing graylevel fuzziness

Abstract

The problem of Histogram sharpening and thresholding by minimizing greyness ambiguity using the measures of fuzziness in a set is considered. This work provides a complete mathematical formulation of the said problem in order that both fuzzy and crisp segmentation result. The criteria regarding the choices of membership functions and the window sizes are established. They involve symmetry in ambiguity around the crossover point and bound functions for restricting the variation of membership function values. The segmentation thus obtained provides not only flexibility in methodology but also effective results. Experimental results further establish the same conclusions.