A boundary-aware point clustering approach in Euclidean and embedding spaces for roof plane segmentation

TL;DR

A boundary-aware point clustering approach in Euclidean and embedding spaces constructed by a multi-task deep network for roof plane segmentation and results show that the proposed approach significantly outperforms the existing state-of-the-art approaches.

Abstract

Roof plane segmentation from airborne LiDAR point clouds is an important technology for 3D building model reconstruction. One of the key issues of plane segmentation is how to design powerful features that can exactly distinguish adjacent planar patches. The quality of point feature directly determines the accuracy of roof plane segmentation. Most of existing approaches use handcrafted features to extract roof planes. However, the abilities of these features are relatively low, especially in boundary area. To solve this problem, we propose a boundary-aware point clustering approach in Euclidean and embedding spaces constructed by a multi-task deep network for roof plane segmentation. We design a three-branch network to predict semantic labels, point offsets and extract deep embedding features. In the first branch, we classify the input data as non-roof, boundary and plane points. In the second branch, we predict point offsets for shifting each point toward its respective instance center. In the third branch, we constrain that points of the same plane instance should have the similar embeddings. We aim to ensure that points of the same plane instance are close as much as possible in both Euclidean and embedding spaces. However, although deep network has strong feature representative ability, it is still hard to accurately distinguish points near plane instance boundary. Therefore, we first group plane points into many clusters in the two spaces, and then we assign the rest boundary points to their closest clusters to generate final complete roof planes. In this way, we can effectively reduce the influence of unreliable boundary points. In addition, we prepare a synthetic dataset and two real datasets to train and evaluate our approach. The experiments results show that the proposed approach significantly outperforms the existing state-of-the-art approaches.

Figures (11)

• Figure 1: The whole framework of the proposed roof plane segmentation approach.
• Figure 2: The brief architecture of the proposed roof plane segmentation network.
• Figure 3: Illustration of the point classification branch. This figure presents the (a) input points with roof and non-roof points, the (b) shifted points, the (c) embedding features of points and the (d) final point classification result. In (a), (b) and (d), the white, green, blue and red dots represent the input, roof plane, boundary and non-roof points. In (c), the red dots labeled with $-1$ represent boundary points, other gray dots represent plane points.
• Figure 4: An example of plane point clustering and boundary point refinement. (a) is the input point clouds with two roof planes. (b) presents the result of point classification, where red, blue and green colors represent non-roof, boundary and plane points. (c) shows boundary points and the roof planar patches generated by plane point clustering . (d) is the final roof plane segmentation result. In (c) and (d), we apply yellow and cyan dots to denote two planar patches.
• Figure 5: Several examples of the synthetic dataset. Different colors represent different plane instances.