基于RETINEX的图像增强方法探讨

Research on Image Enhancement Method Based on RETINEX

文章主要研究了几种基于视神经网络的图像增强方法。Retinex的基本概念是通过视网膜(人眼)和大脑皮层(大脑)的帮助,在观察一个物体的位置后,捕捉图像。根据视差理论,一个图像可以说是光照和物体反射率的产物。Retinex聚焦于图像的动态范围和颜色常性。迄今为止,研究者们提出了各种利用视神经网络增强图像对比度的方法。文章将讨论并比较了单尺度视Retinex(SSR)、多尺度Retinex(MSR)以及改进的Retinex图像增强(IRIE)和改进的MSR夜间增强(MSRINTE)。

This paper mainly studied several image enhancement methods based on optic neural network.The basic concept of Retinex was to capture an image with the help of the retina(the eye)and the cerebral cortex(the brain),after observing the position of an object.According to the parallax theory,we could say that an image was a product of light and the reflectivity of the object.Retinex focused on the dynamic range and color constancy of the image.So far,researchers had proposed various methods to enhance image contrast by using optic neural network.This article discussed single-scale Retinex(SSR),multi-scale Retinex(MSR),improved Retinex image enhancement(IRIE),and improved MSR night enhancement(MSRINTE).