摘要
针对无人机航拍图像在雾霾天气下降质特点,提出一种基于暗通道先验原理的航拍图像去雾改进算法。根据大气物理模型和暗通道原理,对有雾图像进行自适应分块求取暗通道图;通过降采样得到初始透射率图,再应用导引滤波对投射率进行优化,同时利用最小值滤波和均值滤波求解大气光强;应用雾天模型恢复降质图像。通过实验对比验证,该算法简单且能够高效去雾,在客观评价指标上总体性能也优于其他算法。
UAV aerial images are seriously degraded in haze weather. To solve this problem,the paper proposes an aerial image dehazing algorithm based on dark channel prior principle.According to the physical model of atmosphere and dark channel principle,the hazed image is adaptively segmented to obtain the dark channel graph; the initial transmissivity diagram is obtained through down sampling; and the projection ratio is optimized by using guided filter. At the same time,the gas intensity is solved by using minimum filter and mean filter; finally,the dehazed image is recovered from the degraded image through the haze model.The experimental comparison and verification show that the image dehazing algorithm is simple and can be highly efficient,and the overall performance of the objective evaluation index is better than other algorithms.
出处
《无线电工程》
2016年第11期38-41,78,共5页
Radio Engineering
关键词
无人机侦察
图像去雾
大气物理模型
暗通道
UAV reconnaissance
image dehazing
physical model of atmosphere
dark channel
