Synergy between aerial imagery and low density point cloud for automated image classification and point cloud densification

Abstract

In this paper a synergy scheme between aerial imagery and sparse LIDAR point clouds is proposed for an automated aerial image classification. In this scheme, a point cloud and an image are chosen for a certain urban area. The point cloud is automatically classified into buildings, vegetation and roads using PCA and intensity variation. Afterwards, a projection of the point cloud into an image is obtained, such that it is registered with the aerial image. The aerial image classifier is trained with the LIDAR classification result to generate an automated classifier for aerial images. The classifier is tested with another image to demonstrate its accuracy. Another benefit of the synergy proposed is to densify the planar patches of the low density point cloud using the segmented aerial image to help modelling applications achieve more precise boundaries.