Adaptive Image Contrast Enhancement by Computing Distances into a 4-Dimensional Fuzzy Unit Hypercube

A new fuzzy procedure for adaptive gray levelimage contrast enhancement is presented in this paper. Startingfrom the pixels belonging to a normalized gray level image, anappropriate smooth S-shaped fuzzy membership function (MF)is considered for gray-scale transformation and is adaptivelydeveloped through noise reduction and information loss minimization.Then, a set of fuzzy patches is extracted from the MF,and for each support of each patch, we compute four ascendingorder statistics that become points inside a 4-dimensional fuzzyunit hypercube after a suitable fuzzification step. Contrastenhancement is performed by computing the distances amongthe above points and the points of maximum darkness andmaximum brightness (special vertexes in the hypercube) and bydetermining the rotation of the tangent line to the MF arounda crucial point where fuzzy patches and the MF coexist. Theproposed procedure enables high contrast enhancement in allthe treated images with performance that is fully comparable tothat obtained by three more sophisticated fuzzy techniques andby standard histogram equalization.