基于颜色校正和明暗通道的沙尘降质图像增强算法

Dust Image Enhancement Algorithm Based on Color Correction and Bright and Dark Channel

针对沙尘天气下拍摄的图片中存在的颜色偏移、对比度低等问题,提出了一种沙尘降质图像增强算法.该算法由基于高斯模型的偏色校正和基于暗通道和亮通道的对比度增强2个部分组成.首先根据沙尘降质图像的RGB 3个通道直方图的分布特点,使用高斯模型分别对各通道进行建模,从而进行颜色调整,解决沙尘图像的偏色问题,其次采用暗通道先验去雾的原理,结合亮通道,计算像素级的大气光值,然后对透射图和大气光值进行优化,从而恢复出清晰的无沙尘图像,最后实验结果表明本文算法可以较好地提高沙尘降质图像的重建主观质量,并保持良好的色彩保真度和亮度.

In the report,in order to solve the problems of color deviation and low color contrast in the images taken in dust weather,an algorithm for dust quality image enhancement was proposed.The algorithm consists of two parts:color correction based on gaussian model and contrast enhancement based on dark channel and light channel.Firstly,according to the characteristics of the histogram of three channels of RGB in the dust quality image,gaussian model was used to model each channel,so as to conduct color adjustment,which solves the problem of color deviation of the dust image;secondly,dark channel prior and light channel prior were adopted to calculate the atmospheric light at pixel level,and then the transmission diagram and atmospheric light value were optimized to obtain a clear restored image.The results showed that the algorithm can improve the subjective quality of image reconstruction and maintain good color fidelity and brightness.