基于三边滤波的HSV色彩空间Retinex图像增强算法

Image Enhancement Algorithm Based on Retinex of Trilateral Filter in HSV Color Space

图像增强技术可以从对比度低、分辨率低等暗淡图中恢复原有图像信息并增强低动态范围图像的视觉效果,能从很大程度上改善图像清晰化程度.提出在HSV色彩空间下基于三边滤波的Retinex图像增强算法.该算法首先将传统的RGB空间上的处理转换到HSV空间.然后将三边滤波技术引入到亮度图像的提取阶段,可有效地降低图像高斯噪声、滤除脉冲噪声,同时又有效地保持图像边缘细节的特性.通过对三边滤波的滤波窗口进行角度倾斜,计算出像素"最佳"邻域范围且自动修改.引入了一个二值限制函数,进行"边缘限制性平滑"处理,从而进一步提高了图像边缘和特征的清晰度.将提取的亮度图像进行多尺度Retinex图像增强,使得图像信息更加准确恢复.最后将HSV空间转换到RGB空间,使得增强后的图像颜色得以保持.实验结果表明,该算法应用到图像增强处理中,处理后的图像在色彩得以保持的同时,细节信息也得到有效地突出,更有利于针对各种环境图像的更深入的理解和应用.

The image enhancement technique is commonly applied to recover the information from the dim images with some prob-lems, such as little visual contrast, poor resolution and so on, enhance the visual effect of low dynamic images. It can improve the im-age charity. In this paper an improved Retinex algorithm based on HSV color space with the trilateral filter is proposed, and it includesfollowing steps：Firstly, an input image is converted from traditional RBG space to HSV space. Then the trilateral filter technology istaken into the extraction stage of the image brightness, which can effectively reduce the Gaussian noise, remove the impulse noise,maintain the features and edge details of image. By inclining an angle of the filtering window we can calculate the best neighborhoodrange of pixel, and can modified automatically. With introducing a binary limiting function into the trilateral filter to limited edgesmoothing, so that it can improve the clarity of the features and edges of image. The multi-scale Retinex algorithm is applied to en-hance the extracted luminance image, so that the image information is more accurate recovered. Finally by transforming HSV space in-to RGB space, pixels color of the enhanced image is well maintained. Experimental results show that, using the algorithm to image en-hancing, the enhanced image effectively maintains the color and feature details, which will be more conducive to a better understandingand application of images from different environment.