Infrared image enhancement algorithm using local entropy mapping histogram adaptive segmentation

Abstract

The traditional histogram equalization algorithm may cause problems such as local over-enhancement and noise amplification while enhancing the image. To solve these problems, this paper proposes an infrared image enhancement method using local entropy mapping histogram adaptive segmentation. Firstly, a local entropy mapping histogram is built through the modified ‘sigmoid’ function to describe the detailed distribution of the infrared image. Then the LOESS algorithm and local minimum examination are used to adaptively segment the local entropy mapping histogram into multiple sub-histograms. And follow, the double plateau constraint of shape preservation is adopted, and the double plateau thresholds of each interval of the local entropy mapping histogram are adaptively optimized according to the genetic algorithm. Finally, multiple sub-histograms are equalized to obtain an enhanced image. Comparative experiments on real infrared images show that our method is ahead of other superior methods in qualitative and quantitative evaluation.