Automatic contrast enhancement by variational minimax optimization

Abstract

We present an automatic global contrast enhancement method using variational optimization. A novel energy functional is formulated as a convex combination of histogram equalization and quadratic distortion constraint regularization. The desired image gray-level transformation is obtained through minimax optimization. This method sets a generic framework for developing distortion regularized histogram equalization techniques for global or spatially adaptive contrast enhancement. We demonstrate the capability of our method using various types of low contrast images by visual inspection and quantitative comparison.