BRDF-based correction of colorized aerial LiDAR point clouds

Abstract

The BRDF (Bi-directional Reflectance Distribution Function) describes the scattering properties of the surface as a function of the incidence direction, viewing geometry and wave length, and typically depends on the land-cover type being observed. In our work a laser scanning point cloud should be assigned multispectral values from aerial images. Since each laser point is covered by up to nine images, i.e. in general by nine different colour vectors, a unique set of colour vectors is sought for through radiometric correction by applying an optimal BRDF. As BRDF the widely used linear semi-empirical modified Walthall model has been chosen, for which the parameters should be determined by least squares adjustment. For the evaluation mostly uniform areas of four different object classes, such as road surfaces, lawn and grass area, trees and high vegetation, and roof tops, were used. It could be shown that the simple Walthall model yields acceptable results.