UAV-Based Terrain Modeling in Low-Vegetation Areas: A Framework Based on Multiscale Elevation Variation Coefficients

The removal of low vegetation is still challenging in UAV photogrammetry. According to the different topographic features expressed by point-cloud data at scales, a vegetation-filtering method based on multiscale elevation-variation coefficients proposed for terrain modeling. First, virtual grids are constructed and average elevation values corresponding point clouds obtained. Second, amount change any two scales each grid calculated obtain difference surface characteristics (degree change) scales. Third, variation coefficient that corresponds largest degree calculated, threshold segmentation performed relation vegetated regions much larger than those regions. Finally, optimal calculation neighborhood radius analyzed, discussed. experimental results show can accurately remove points reserve ground low- densely areas. type I error, II total error study areas range from 1.93 9.20%, 5.83 5.84%, 2.28 7.68%, respectively. 2.43–2.54% lower CSF, TIN, PMF algorithms This provides foundation rapid establishment high-precision DEMs