Color image enhancement based on adaptive multi-scale morphological unsharpening filter

Abstract

In order to overcome the problem in Retinex algorithm that it is possible to cause image detail loss by using Gauss function to estimate illumination, in this paper, we propose a color image enhancement algorithm based on multi-scale morphology unsharp method combined with Retinex. Firstly, the color image is converted from RGB space to HSV space. Then, the S component is enhanced by adaptive logarithmic transformation to improve its visual characteristics and make it consistent with human visual system. The method of multi-scale morphological unsharpening is adopted to estimate the V component, and the multi-scale weights are adaptively selected according to the features of the image. The V component is enhanced according to the Retinex principle. Finally, the H and the enhanced S and V components are recombined and reflected into RGB space to achieve the purpose of image enhancement. The experimental results showed that the algorithm is superior to the traditional SSR, MSR and MSRCR algorithms in terms of entropy, average gradient and sharpness, and outperforms those based on Gauss function estimation.