Halo control for LHE based local adaptive tone mapping

Abstract

We propose a halo control method that is suitable for smoothed local histogram equalization (LHE). Smoothed LHE based tone mapping, e.g., Apical Iridix®, is effective and widely used for local adaptive image enhancement. However, halo artifacts occur at the edge, which causes unnaturalness. There are no detailed discussions on the principles of smoothed LHE halo-artifacts. In this paper, we analyze the cause of the artifacts and propose a halo reduction/control method that uses a weighted local histogram. In addition, we separate the artifacts into light and dark and evaluate the effectivity for the human visual sense. The adequate weighting function of our method can be set automatically at each pixel by using the difference between the input and mean luminance of the local window so as to enhance the natural local edge.