Abstract: Image enhancement provides good dynamic range compression in each red，green and blue（RGB） channel of real-color image，but it fails in color rendition.This paper proposes to transform a real-color image from RGB space to HSV space at first. Then based on illumination-reflectance model，value is enhanced by homomorphic decomposition and wavelet transform.At last，the color purity of saturation is adjusted based on spectral sensitivity of human visual system.Experiments show the compression of real-color image dynamic range is obviously effective，it is successful in color rendition and the real color images enhanced have adapted well to human vision.Real color image enhanced by the algorithm of this paper is better than the multi-scale Retinex’s with color restoration.

摘要： 在RGB空间中分别对真彩图像的三个通道进行增强能够很好地压缩图像动态范围，但往往会产生色彩偏离。因此，先将真彩图像由RGB空间变换到HSV空间，进行色度、饱和度和亮度分离；然后基于入射-反射模型，对亮度通道进行同态分解和小波变换的图像增强；再根据人类视觉系统光谱感觉特性，调整色饱和度分量的色彩纯度。实验证实，增强后的真彩图像色彩基本无偏差，图像动态范围压缩良好，能够很好地适应人类视觉。该方法对真彩图像的增强优于带色彩恢复的多尺度Retinex方法（msrcr）。