Abstract: Gray level transform will enhance contrast of color image if the technique straightly applies to R，G and B components，but color distortion will be introduced.MSRCR based on Retinex theory can improve visual effect of color image and not produce color distortion.However MSRCR must specify control parameters，namely its adaptivity is not well.And worse disadvantage of MSRCR is huge computation due to convolution，so processing time is much long.A novel adaptive color image enhancement algorithm is proposed.Original RGB image and the image with implementing gray level transform are all converted to nonlinear luminance/chroma color space.Consequently chroma component of the former and luminance component of the latter are recomposed，so the original color is preserved.This algorithm fully inherits advantages of gray level transform，such as extending effective gray levels，advancing whole contrast，improving image detail，and simple computation complexity，and parameters adaptability.So this algorithm can be applied to adaptive real time color image processing.

Key words: color image, gray level transform, color space, luminance/chroma

摘要： 灰度变换技术直接用于彩色图像的RGB各分量可以提高其对比度，但却会引起彩色失真；而基于Retinex理论的多尺度彩色复原算法（MSRCR）可以改善彩色图像的视觉效果，也不会引起彩色失真，但存在控制参数的取值问题，自适应性不是很好，更大的缺陷在于卷积运算巨大的运算量导致处理时间偏长。提出了一种基于灰度变换的自适应彩色图像增强算法，将原始RGB图像和RGB三分量灰度变换后的图像转换到非线性亮度/色度彩色空间，重组两者的色度分量和亮度分量，从而保持了原始图像的色彩。在继承灰度变换算法优点的基础上，扩展了原始图像的有效灰度范围，提高了整体对比度，增强了细节。此外，算法的计算量极小，不需要设定任何参数，可用于自适应实时彩色图像处理。

关键词: 彩色图像, 灰度变换, 彩色空间, 亮度/色度