基于暗通道先验的图像去雾改进

Improvement of Image Dehazing Based on Dark Channel Prior

雾天环境一般会导致相机拍摄的图像对比度下降、色彩失真。传统暗通道先验去雾算法虽然可以估算雾天透射率并获得较好的去雾效果,但仍然存在算法复杂度高、容易出现块效应,以及明亮区域色彩偏离问题。针对上述问题提出一种改进的去雾算法,该算法通过引入容差机制校正明亮区域的透射率,然后利用引导滤波进一步细化透射率。实验结果表明,使用改进后的算法可有效改善色彩偏离问题,提高运算速度。

In a foggy environment,images captured by the camera may cause a drop in contrast and color distortion.Although the traditional dark channel prior defogging algorithm can estimate the fog day transmittance and obtain better defogging effect,there is still a problem that the algorithm is high in complexity,prone to blockiness,and of bright area color deviation.In view of the above problems,this paper proposes an improved defogging algorithm,which corrects the transmittance of the bright area by introducing a tolerance mechanism,and then uses guide filtering to refine the transmittance.Experimental results show that the improved algorithm can effectively improve the color distortion and improve the speed of operation.