基于暗通道先验的改进图像去雾算法

Improved Image Dehazing Algorithm Based on Dark Channel Prior

针对传统暗通道先验去雾算法处理得到的去雾图片存在亮度偏暗,颜色失真等问题,提出一种基于暗通道先验的改进图像去雾算法。首先,将有雾的图像根据尺寸大小均匀分割成八份;然后,对分割后的图像进行暗通道先验去雾处理,并为每一个分割图像块找到一个最合适的w值;最后,将去雾后图像转为HSI颜色空间,对亮度进行限制对比度自适应直方图均衡化处理。实验结果显示:改进后的算法有效改善了去雾后图像亮度偏暗的问题,且极大程度避免了色彩失真现象。

Aiming at the problems of dim brightness and color distortion in the dehazing images obtained by the traditional dark channel prior dehazing algorithm, an improved image dehazing algorithm based on dark channel prior was proposed. Firstly, the image with fog is evenly divided into eight equal parts according to size. Then, the segmented image is dehazed by a dark channel prior, and an optimal w value is found for each segmented image block. Finally, the dehazed image is transformed into HSI color space, and the brightness limit contrast adaptive histogram equalization process is performed. Experimental results show that the algorithm can effectively improve the image brightness after dehazing, and avoid color distortion to a large extent.