A Framework for Building Point Cloud Cleaning, Plane Detection and Semantic Segmentation

TL;DR

This paper addresses building information modeling from 3D point clouds by proposing a three-stage framework for cleaning, plane detection, and semantic segmentation. It combines an adaptive z-score based outlier removal, robust RANSAC plane detection, and a PointNet-inspired semantic segmentation network whose features are classified by an SVM. A LOOCV evaluation on S3DIS and a private LEICA dataset demonstrates strong performance: effective outlier removal, accurate plane extraction, and high per-class segmentation accuracy (e.g., Floor $98.3\%$, Ceiling $93.2\%$, Wall $83.1\%$). The framework advances BIM workflows by delivering more accurate, automation-ready 3D building representations with improved efficiency; room-level segmentation remains manual, offering a direction for future automation.

Abstract

This paper presents a framework to address the challenges involved in building point cloud cleaning, plane detection, and semantic segmentation, with the ultimate goal of enhancing building modeling. We focus in the cleaning stage on removing outliers from the acquired point cloud data by employing an adaptive threshold technique based on z-score measure. Following the cleaning process, we perform plane detection using the robust RANSAC paradigm. The goal is to carry out multiple plane segmentations, and to classify segments into distinct categories, such as floors, ceilings, and walls. The resulting segments can generate accurate and detailed point clouds representing the building's architectural elements. Moreover, we address the problem of semantic segmentation, which plays a vital role in the identification and classification of different components within the building, such as walls, windows, doors, roofs, and objects. Inspired by the PointNet architecture, we propose a deep learning architecture for efficient semantic segmentation in buildings. The results demonstrate the effectiveness of the proposed framework in handling building modeling tasks, paving the way for improved accuracy and efficiency in the field of building modelization.

Figures (6)

• Figure 1: Overview of the proposed framework.
• Figure 2: Histograms of X, Y, Z axis
• Figure 3: Point cloud cleaning result by combining X and Y axis filtration
• Figure 4: visual representations of the scanned buildings.
• Figure 5: visual results and comparison of the outlier removal methods.