基于PCA的可变框架模型Retinex图像增强算法

摘要

针对恶劣天气下拍摄图像的退化现象,提出了一种基于主成分分析(principal component analysis,PCA)的可变框架模型Retinex图像增强算法.PCA变换提供通道间良好的正交性,可避免由于亮度调整带来的色度失真.该算法通过PCA变换得到图像亮度分量、色度分量,对得到的亮度分量使用改进的可变框架模型Retinex进行处理,适当调整色度分量,最后对处理得到的RGB图像进行去相关拉伸.实验结果表明,该方法能有效改善恶劣天气造成的图像退化现象,提高图像的清晰度.%For the degradation of images taken under bad weather, this paper proposed a variation modal Retinex image enhancement algorithm based on PCA. PCA transformation provided orthogonality between channels and avoided producing incorrect colors despite the modification of luminance. Firstly, it utilized the PCA transformation to get the luminance component and the chrominance components. Secondly, processed the luminance component with improved variation modal Retinex image enhancement algorithm, and adjusted the chrominance components appropriately. Finally, processed the RGB image with decorrelation stretch. Experimental results show that the method can effectively ameliorate the image degradation taken under bad weather and improve image clarity.