明亮区域的暗原色先验算法

针对暗原色先验算法在处理含有大面积明亮区域的有雾图像时,复原图像会产生严重色彩偏移的不足,提出了一种基于像素点的透射率修正方案.在分析暗原色先验去雾原理以及复原图像色彩偏移成因的基础上,对去雾模型进行推导,使用暗通道亮度与大气光数值接近度及像素通道间数值接近度作为明亮区域判定机制,得出更具普适性的透射率求取方法.明亮区域采用容差机制纠正错误估计的透射率,非明亮区域仍采用原透射率求取方式,并首次将改进算法应用到雪天模糊图像清晰化中.实验结果表明,改进算法使去雾图像清晰自然,可有效修正色彩失真.

Dark channel prior algorithm for bright regions

In order to solve the problem that the restored images will exhibit serious color distortion when the haze images in the bright regions with large area are processed with dark channel prior algorithm, a transmission correction scheme based on pixels was proposed. Based on the analysis for both principle of dark channel prior haze removal and color distortion cause of restored images, the derivation was performed for the haze removal model. Through using the numerical proximity between dark channel and atmospheric light as well as the numerical proximity between channels as the decision mechanism of bright regions, a more universal method of calculating the transmission was obtained. For the bright regions, the error estimation of transmission was corrected with the tolerance mechanism, while the original transmission estimation method was still used in the non bright regions. The improved algorithm was firstly applied to the clearness of snow images. The experimental results show that the haze removal images are clear and nature with the improved algorithm, and the color distortion can be effectively corrected.