单张无人机影像水体高光自动检测与补偿方法

无人机获取地面影像时，水体反射会导致影像中存在较为明显的水体高光，给数据处理带来一定困难并对DOM（digital orthophoto map）质量产生显著影响。提出了一种单张无人机影像水体高光自动检测与补偿方法，先基于高光分量对高光区域进行多尺度检测，再使用Grabcut算法对其边界进行优化；然后根据地物反射的几何与光谱特性建立决策树剔除非水体高光；接着精化高光区域，将附近的细碎高光点纳入到区域内；最后，使用提出的改进Criminisi算法对高光区域进行补偿。使用无人机影像对该方法的有效性进行验证，综合主观目视与PNSR（peak signal to noise ratio）、SSIM（structural similarity index）指数的客观评价结果，本文方法优于Mallick、Shen、Yoon等人的方法；使用Pix4Dmapper对高光处理前后的影像生成DOM，本文方法可以明显提高DOM质量。

A Method for Automatic Water High Light Detection and Removal in Single UAV Image

UAV(unmanned aerial vehicle) is convenient and low cost, water reflection can cause high light in the image and has adverse effect on UAV data processing and the DOM(digital orthophoto map) quality. We propose a method for automatic water high light detection and removal in single UAV image. Firstly, we extracted initial candidate highlight regions using multi-scale threshold detection in a high light component proposed in this paper and used the Grabcut algorithm to optimize them. Next, a decision tree was built according to the characteristics of water high light to eliminate the error detection. Then, high light regions were refined by the high light points nearby. Finally, we modified Criminisi algorithm to remove high light regions. In actual UAV images, the experimental results show that our method can remove the water high light well and is superior to other methods proposed by Mallick, Shen and Yoon in terms of PNSR(peak signal to noise ratio) and SSIM(structural similarity index) parameters and visual evaluation, and improves the DOM quality as well.