中文摘要：

针对基于Retinex理论的经典图像增强算法中存在的灰化现象、噪声扩大现象和光晕现象，提出一种基于亮度的自适应单尺度Retinex图像增强算法。算法中引入了基于边缘信息的高斯滤波器，’主要使用同类性质的像素进行平滑处理用以改善光晕现象和抑制噪声；在提取的反射分量中加入像素的原始亮度以改善衣化现象和减小噪声。对比仿真实验结果显示，基于亮度的自适应单尺度Retinex图像增强算法可以有效增强细节、减小光晕和控制噪声，同时处理后的图像具有良好的色彩保真度。

英文摘要：

To deal with the ＇graying out＇, ＇noise enlargement＇ and ＇halo＇ effects existing in classical Retinex image enhancement algorithms, an adaptive single scale Retinex scheme based on luminance is proposed. The edge information is incorporated into the Gaussian filter to restrain ＇halo＇ and ＇noise enlargement＇ effects, in which the filtering is carried out with adjacent pixels possessing the same characteristics. The original luminance is added to the obtained reflectance to restrain ＇graying out＇ and ＇noise enlargement＇ effects. Compared with other algorithms, the simulation results indicate that the proposed adaptive single scale Retinex scheme based on luminance can acquire more details, less ＇halo＇, less noise and better color fidelity.