The automatic classification 3D point clouds based associative markov network using context information

Abstract

Many applications of mobile mapping want to automatic classification point clouds into different classes for further processing. In this chapter we present a new approach for labeling 3D point clouds with using a novel feature descriptor-the four directions scan line gradient, and context classification models-associative Markov network (AMN). To build informative and robust 3D feature point representations, our descriptors encode the underlying surface geometry around a point using multi-scanlines gradients. It is more stable and reliable than normal vectors in urban environments with wide variety of natural and manmade objects. By defining objects models of 3D geometric surfaces and making use of contextual information of AMN, our system is able to successfully segment and label 3D point clouds. We use FC09 datasets to evaluate the proposed algorithm. © 2012 Springer Science+Business Media B.V.