Variational exposure fusion with optimal local contrast

Abstract

In this paper, we present a variational method for exposure fusion. In particular, we combine differently exposed images to a single composite that offers optimal exposedness, saturation, and local contrast. To this end, we formulate the output image as a convex combination of the input, and design an energy functional that implements important perceptually inspired concepts from contrast enhancement such as a local and nonlinear response. Several experiments demonstrate the quality of our technique and show improvements w.r.t. state-of-the-art methods.