改进多尺度Retinex的彩色图像增强

摘要

由于假设入射光变化平稳，基于Retinex理论的传统彩色图像增强算法也有一些局限性，增强效果有限。在明暗对比度较强处，引入颜色失真和出现光晕现象。考虑到这些限制，本文提出了一种基于多尺度Retinex的改进算法。利用双边滤波进行亮度分量估计和局部对比度增强，将增强后的亮度图像与原始图像HSV彩色空间的亮度图像I进行对比，进行彩色图像恢复。实验结果表明，该方法在增强图像时，能够有效地避免颜色失真及抑制光晕。%There are also some limitations on the effect of enhancement which traditional color image enhancement algorithm based on the Retinex theory due to its assumption which the incident light is changing smoothly.It brings color distortion and appears a halo phenomenon during the strong contrast between the light and the dark .Considered these limitations,it puts toward an improved algorithm based on the multi-scale Retinex in this paper.The brightness component is estimated by bilateral filtering and local contrast enhancement.We will compare the brightness of the image after increasing with the original image of brightness component I of HSV color space , for color image restoration.The experimental results prove that the method can efficiently avoid cross color and restrain the halo phenomenon when enhancing the image.