Fusion of Local Rank Transform and Tone Mapping for Contrast Enhancement: Application to Skin Lesion Image Analysis

Abstract

A novel approach is proposed for the contrast enhancement of skin lesion images. The colors of the enhanced image become naive and unrealistic; however this approach can successfully fulfill the purpose of contrast enhancement as a pre-processing step for segmentation. The algorithm highly emphasizes the details present in an image. The novelty lies in the idea of merging Local Rank Transform (LRT) with tone mapping that makes the details visible in low contrast image. Through LRT, artificial exposures are created from the image itself rather than capturing the scene multiple times through the camera at different exposure settings. After that, any kind of tone mapping can be applied over those artificially created exposures in order to make visible the existent details in an image.