航拍图像逐步细化的云检测方法

A Progressive Refinement Scheme for Cloud Detection of Aerial Images

在现有航拍图像云检测算法研究的基础上,对航拍图像所获云层数据进行大量统计和分析,提出了一种改进的云检测新方法,即逐步细化的云检测方法.首先基于航拍图像的HSI模型构造云区显著图;再根据亮度和色调,设置最优阈值获得初步云检测结果;之后将另外通过滤波获得的细节图像与初步云检测结果相融合;再通过引导滤波获得最终检测结果.对包含薄云、水域、雪域、半透明云的典型图像进行处理,实验结果表明,与传统算法相比,该算法错误率更低,且检测结果更接近人眼实际观测结果.

Based on a large number of statistics and analysis of the obtained cloud data of aerial images,a novel progressive refinement scheme is proposed to detect clouds in the color aerial photographs under the framework of the current cloud detection algorithm.First,a cloud salient map is constructed under HSI model.Then,based on the cloud intensity and hue threshold statistics and Otsu algorithm,a coarse cloud detection result is obtained from the input aerial photograph.Then we construct a robust detail map derived from a filter operator,and blend it with the coarse cloud map.Finally,we perform a guided filter to achieve our final cloud detection result.The result of the experiment on some typical images containing thin clouds,water,snow,and translucent cloud shows that the error rate of this scheme is lower than that of the traditional algorithm,and the detection results are more close to what people actually observed.