Accuracy assessment of 3-dimensional LiDAR building extraction

• Published in: 2018 IEEE 14th International Colloquium on Signal Processing & Its Applications (CSPA) pp. 261 - 266
• Main Authors: Saraf, Noraain Mohamed, Hamid, Juazer Rizal Abdul, Halim, Maisarah Abdul, Rasam, Abdul Rauf Abdul, Lin, Saharuddin
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.03.2018
• Subjects: 3-D LiDAR; accuracy assessment; building extraction; Buildings; Data mining; Data models; Feature extraction; Laser radar; Solid modeling; Three-dimensional displays
• DOI: 10.1109/CSPA.2018.8368723

This study intent to perform an accuracy assessment of building extraction from Light Detection and Ranging (LiDAR) derived Digital Surface Model (DSM). A normalized DSM was extracted to separate the buildings from other spatial features. Slope and aspect analyses were conducted based upon segmentation to extract building points. Furthermore, height of building was estimated from contour analysis. Hence, an accuracy assessment was done and findings were documented. As a result, the percentage of error assessment was less than 10 percent compared to building footprint. In addition, there was only a slight difference between building's height extracted from LiDAR point cloud and height measured on the ground. It certifies that the height of building estimated from 3-D LiDAR is within 1m tolerance. The segmentation applied based on contour, slope and aspect analyses indicated that the approach can derive reliable and accurate results in extracting building points and estimating building's height. The findings from this study demonstrated the capability and the effectiveness of LiDAR data.