当前位置: X-MOL 学术Proc. IEEE › 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.)
Front Cover
Proceedings of the IEEE ( IF 23.2 ) Pub Date : 2020-12-18 , DOI: 10.1109/jproc.2020.3040564


Motivated by the need to monitor mountain glaciers, this article addresses the estimation of a mountain glacier’s flowline, along which the movement of the glacier’s terminus can be captured. We propose a method for estimating glacier flowlines through local linear regression gradient descent (gd-flow) on digital elevation model (DEM) images. We additionally propose an aggregation method (median gd-flow) that increases the accuracy of gd-flow by combining flowlines corresponding to different gd-flow window sizes. We compare gd-flow and median gd-flow to a standard hydrological flow model (D8) and a variant of D8 (D8-smooth), which applies D8 to a smooth DEM, across 25 glaciers using manually drawn flowlines as a reference standard. We assess the performance via simulations considering variable terrain and image quantization. Our proposed algorithm (gd-flow) outperforms D8 and D8-smooth in both applied and simulated settings, while median gd-flow further improves gd-flow by leveraging the accuracy of multiple window sizes.

中文翻译:

 封面


出于监测山地冰川的需要,本文讨论了山地冰川流线的估计,沿着该流线可以捕获冰川终点的运动。我们提出了一种通过数字高程模型(DEM)图像上的局部线性回归梯度下降(gd-flow)来估计冰川流线的方法。我们还提出了一种聚合方法(中值 gd-flow),通过组合对应于不同 gd-flow 窗口大小的流程来提高 gd-flow 的准确性。我们将 gd-flow 和中值 gd-flow 与标准水文流量模型 (D8) 和 D8 的变体 (D8-smooth) 进行比较,后者将 D8 应用于平滑 DEM,跨越 25 个冰川,使用手动绘制的流线作为参考标准。我们通过考虑可变地形和图像量化的模拟来评估性能。我们提出的算法 (gd-flow) 在应用和模拟设置中均优于 D8 和 D8-smooth,而中值 gd-flow 通过利用多个窗口大小的准确性进一步提高了 gd-flow。
更新日期:2020-12-18
down
wechat
bug