中文摘要：

针对消除背景光照对图像的影响时出现的细节弱化、色彩失真的问题，提出了一种基于边缘检测的Retinex彩色图像增强算法。根据人类视觉特性从图像中提取亮度分量并检测边缘信息，在平滑点和边缘点处采用不同的模板估计背景光照，以避免强边缘处的光晕现象；通过调整图像中的背景光照比例提升局部对比度，并根据反射图像直方图自适应调整全局对比度；利用R、G、B通道及亮度分量的等价变换进行色彩恢复，以保证增强前后图像色调一致。实验结果表明，增强后的图像标准差提升了19．34％，信息熵增大了13．18％。

英文摘要：

In order to solve the problems of detail reduction and color distortion when eliminating the effect of illumination conditions, an edge-preserved Retinex algorithm for color image enhancement is proposed. According to the characteristics of human visual system, luminance is extracted from the original image. Halo effect at strong edge is avoided by estimating the background luminance with different filters at the edge and the smooth region after edge detection. Local contrast is increased by adjusting the proportion of background luminance, and global contrast is adjusted according to the histogram of reflection image. Hue of the images before and after enhancement is preserved because of the color recovery based on equivalence of direct proportion transformation in R, G, B and I channel. Experiment results show that contrast increases by 19.34% and entropy increases by 13.18% after enhancement.