摘要
为提高航空公路图像中不良天气中的清晰度,提出一种基于改进分数阶微分算子及景深尺度变化Retinex的图像增强算法。首先将图像进行一种新的分数阶微分操作,然后将雾图像作暗原色先验处理,得到图像的估计景深图,计算原图各处的Retinex尺度,再进行单尺度Retinex变换得到结果图像。实验结果表明,与现有的He算法及其他算法相比,对于场景深度差异较大的模糊航空图像,该算法具有更快的运行速度和图像增强效果。
In order to improve visibility in the poor quality weather for aerial road images, this paper proposed a changing scale Retinex algorithm based on fractional differential and depth map. After a new fractional differential operation, it required the image dark channel prior treatment to obtain the estimated depth map. Then according to the depth map, Retinex scales was under calculating in each part of the image. Finally the single scale Retinex transform was under performing to obtain the enhanced image. Experimental results show that after comparing with the He＇ s algorithm and others, the new algorithm has the faster speed and better image enhancement effect for the images that have greatly different scene depths.
出处
《计算机应用研究》
CSCD
北大核心
2015年第9期2844-2848,共5页
Application Research of Computers
基金
国家自然科学基金资助项目(61170147)
陕西省科学技术研究发展计划项目(国际合作2013KW03)
西安市科技计划项目(CX1252(8))
中央高校基本科研业务费专项资金项目(2013G2241019)
