A Novel Rapid Method for Viewshed Computation on DEM through Max-Pooling and Min-Expected Height

Abstract

Viewshed computation of a digital elevation model (DEM) plays an important role in a geographic information system, but the required high computational time is a serious problem for a practical application. Hitherto, the mainstream methods of viewshed computing include line-of-sight method, reference planes method, etc. Based on these classical algorithms, a new algorithm for viewshed computation is proposed in this paper: the Matryoshka doll algorithm. Through a pooling operation, the minimum expected height of the DEM is introduced as max-pooling with minimum expected height in the viewshed computing optimization. This is to increase the efficiency and adaptability of the computation of the visibility range. The experimental results demonstrate that the algorithm has obvious advantages in computing speed, but with the accuracy only slightly reduced.