Performance analysis of freeware filtering algorithms for determining ground surface from airborne laser scanning data

摘要: Numerous filtering algorithms have been developed in order to distinguish the ground surface from nonground points acquired by airborne laser scanning. These automatically attempt determine using various features such as predefined parameters and statistical analysis. Their efficiency also depends on landscape characteristics. The aim of this contribution is test performance six common embedded three freeware programs. algorithms’ adaptive TIN, elevation threshold with expand window, maximum local slope, progressive morphology, multiscale curvature, linear prediction were tested four relatively large (4 8  km2) diverse areas, which included steep sloped hills, urban ridge-like eskers, a river valley. results show that areas each algorithm yields commission omission errors. It appears TIN suitable while curvature best suited wooded areas. yielded overall average root-mean-square error values 0.35 m.