Critical Points Extraction from Building Façades by Analyzing Gradient Structure Tensor

Dong Chen,Lincheng Wan,Jiju Peethambaran,Guiqiu Xiang,Jing Li,Shaoning Di,Xianghong Li

doi:10.3390/rs13163146

Dong Chen, Lincheng Wan + Show 5 more

Open Access

https://doi.org/10.3390/rs13163146

Copy DOI

Abstract

This paper proposes a building façade contouring method from LiDAR (Light Detection and Ranging) scans and photogrammetric point clouds. To this end, we calculate the confidence property at multiple scales for an individual point cloud to measure the point cloud’s quality. The confidence property is utilized in the definition of the gradient for each point. We encode the individual point gradient structure tensor, whose eigenvalues reflect the gradient variations in the local neighborhood areas. The critical point clouds representing the building façade and rooftop (if, of course, such rooftops exist) contours are then extracted by jointly analyzing dual-thresholds of the gradient and gradient structure tensor. Based on the requirements of compact representation, the initial obtained critical points are finally downsampled, thereby achieving a tradeoff between the accurate geometry and abstract representation at a reasonable level. Various experiments using representative buildings in Semantic3D benchmark and other ubiquitous point clouds from ALS DublinCity and Dutch AHN3 datasets, MLS TerraMobilita/iQmulus 3D urban analysis benchmark, UAV-based photogrammetric dataset, and GeoSLAM ZEB-HORIZON scans have shown that the proposed method generates building contours that are accurate, lightweight, and robust to ubiquitous point clouds. Two comparison experiments also prove the superiority of the proposed method in terms of topological correctness, geometric accuracy, and representation compactness.

Highlights

Building façade contouring framework: We propose a generic framework for building façade contouring from LiDAR and photogrammetric point clouds
Since the purpose of this paper is to extract critical points from building façades, the confidence definition should be directly related to the characteristics of these critical points
We present a method for extracting building façade contours

Summary

Introduction

As the most basic and critical feature, building contours provide the foundations for scene understanding [1], semantic annotation [2], and 3D abstract perception [3]. In the initial stage of computer vision, a majority of contouring methods are image-based because the “linear” features of images are more accurate when the resolution is guaranteed. The contour of images has a clear definition: the image pixels at the discontinuities in gray level, color, texture, etc. The company ESRI (https://www.esri.com/en-us/home, last accessed on 6 August 2021) used the deep learning framework, i.e., MaskRCNN [9], to extract accurate building contours from high-resolution aerial and satellite imagery.

Objectives

Methods

Findings

Conclusion