结合多尺度和分数阶微分的单幅图像去雾算法

提出一种非下采样轮廓波变换(non-subsampled contourlet transform, NSCT)和分数阶微分相结合的图像去雾算法.该算法首先通过对低质有雾图像进行NSCT分解，得到一个低频子带与多尺度多方向的多个高频子带；然后采用分数阶微分算子对图像的低频子带进行增强，同时通过对各子带的高频系数进行非线性处理，实现高频子带的增强；最后进行NSCT重构，得到增强后的图像.对不同低质有雾图像进行实验比较，结果表明：本算法增强了主观视觉效果，使图像变清晰的同时，具有较高的对比度增益、清晰度增益、信息熵和平均梯度.

Single image defogging algorithm for combining multiscale and fractional differential

In a foggy environment, the image taken will be blurred, the visibility and contrast are greatly weakened, coupled with the image transmission and transformation, the image may be damaged, which is not conducive to the subsequent observation and processing of the image. We therefore propose an image demogging algorithm combining non - downsampling contour wave transform (Non-subsampled contourlet transform,NSCT) and fractional differential. The algorithm first obtained a low - frequency and multiscale high-frequency subband, and enhances the low-frequency sub - band; finally NSCT is reconstructed to obtain the enhanced image. Experimental comparison of different low - quality fog images shows that the results enhance the subjective visual effect and make the image clear, along with high contrast gain, clarity gain, information entropy and average gradient.