A multiscale contrast enhancement method

This paper contributes a novel approach to contrast enhancement. The proposed approach measures local contrast within the context of a nonlinear scale-space representation. The original image is locally probed at multiple resolutions generated through anisotropic diffusion. Once local contrast has been estimated across an optimal range of scales, its value is used to enhance the initial image. Due to the choice of the anisotropic scale-space, the method also accounts for nonlinear edge contributions. Properties of local contrast within scale-space are introduced and discussed.