Automatic point cloud registration for large outdoor scenes using a priori semantic information

Abstract

As an important and fundamental step in 3D reconstruction, point cloud registration aims to find rigid transformation that register two point sets. The major challenge in point cloud registration techniques is finding correct correspondences in the scenes which may contain many repetitive structures and noise. This paper is primarily concerned with improving registration using a priori semantic information in the search for correspondences. In particular, we present a new point cloud registration pipeline for large outdoor scenes that takes advantage of semantic segmentation. Our method consists of extracting semantic segments from point clouds uses an efficient deep neural network; then, detecting the key points of the point cloud and using a feature descriptor to get the initial correspondence set; finally, applying a Random Sample Consensus (RANSAC) strategy to estimate the transformations that align segments with the same labels. Instead of using all points to estimate a global alignment, our method aligns two point clouds using transformations calculated by each segment with the highest inlier ratio. We evaluate our method on the publicly available Whu-TLS registration dataset. These experiments demonstrate how a priori semantic information the improves registration in terms of precision and speed.