当前位置: X-MOL 学术J. Indian Soc. Remote Sens. › 论文详情
Our official English website, www.x-mol.net, welcomes your feedback! (Note: you will need to create a separate account there.)
An Adaptive Technique to Detect and Remove Shadow from Drone Data
Journal of the Indian Society of Remote Sensing ( IF 2.2 ) Pub Date : 2020-11-05 , DOI: 10.1007/s12524-020-01227-z
Ankush Agarwal , Sandeep Kumar , Dharmendra Singh

Unmanned aerial vehicles (UAVs)/drones are used to capture and monitor real-time data for various applications. Different sensors may be mounted on the UAV like 4K RGB camera, RedEdge-M and many others. 4K camera is used to capture an image in RGB bands, and RedEdge-M captures an image in five bands namely Blue, Green, Red, RedEdge and Near Infrared having a center wavelength of 475, 560, 668, 717 and 840 nm, respectively. Quality of an image can be judged with its basic two properties, i.e., spatial and spectral information. The images obtained by these sensors are useful for monitoring various targets. One of the most important challenges for any UAV sensor image is a shadow that affects a lot the quality of an image. A shadow can hinder the identification of the target and also affect the vegetation parameters which highly depend on the band values. Huge change in the class labels has been identified during classification because of shadow. Band values can highly suffer from shadow, and thus, it is required to minimize the shadow effect without compromising the image quality. Therefore, in this paper, an attempt has been made to minimize the shadow effect by proposing an algorithm that is based on spatial distribution of neighboring pixels and is compared with other known techniques. The study area chosen is an agriculture field nearby Roorkee region, which lies in the northern part of India having central latitude–longitude as 29.9457°N, 78.1642°E. The images were captured using DJI Phantom 3 pro at an altitude of 100 m which provides a spatial of 0.05 m. It is observed that the proposed method shows better results as compared to others.

中文翻译:

一种从无人机数据中检测和去除阴影的自适应技术

无人机 (UAV)/无人机用于捕获和监控各种应用的实时数据。无人机上可以安装不同的传感器,如 4K RGB 相机、RedEdge-M 等。4K相机用于拍摄RGB波段的图像,RedEdge-M拍摄五个波段的图像,分别是Blue、Green、Red、RedEdge和Near Infrared,中心波长分别为475、560、668、717和840 nm . 一幅图像的质量可以通过它的两个基本属性来判断,即空间信息和光谱信息。这些传感器获得的图像可用于监控各种目标。对于任何无人机传感器图像而言,最重要的挑战之一是阴影,它会极大地影响图像质量。阴影会阻碍目标的识别,也会影响高度依赖于波段值的植被参数。由于阴影,在分类过程中已经识别出类标签的巨大变化。Band 值可能会受到阴影的影响,因此需要在不影响图像质量的情况下最小化阴影效果。因此,在本文中,尝试通过提出一种基于相邻像素的空间分布并与其他已知技术进行比较的算法来最小化阴影效应。选择的研究区域是位于印度北部的 Roorkee 地区附近的农田,中经纬度为 29.9457°N,78.1642°E。这些图像是使用 DJI Phantom 3 pro 在 100 m 的高度拍摄的,提供了 0.05 m 的空间。
更新日期:2020-11-05
down
wechat
bug