Performance Analysis of Point Cloud Preprocessing Algorithms Suitable for Construction Progress Analysis

Abstract

As the scale and complexity of construction projects increase, the importance of progress management becomes even more significant. The current construction management systems have limitations in accurately analyzing project delays and cost overruns. To achieve precise analysis of construction site conditions, various studies are being conducted to collect and analyze point clouds from construction sites. The point cloud data collected from construction sites are composed of a vast amount of information. For effective analysis of this data, it is essential to use appropriate preprocessing algorithms that can extract meaningful key data. To extract significant data from the entire dataset, various algorithms can be utilized. This study aims to compare and analyze the performance of three different point cloud preprocessing algorithms and propose the most suitable one for understanding the current status of construction sites. In the study, data on the exterior of the building was collected using drones, and point cloud coordinates were extracted. These coordinates were then processed using preprocessing algorithms, namely sor (statistical outlier removal), ror (radius outlier removal), and cor (conditional outlier removal). These algorithms were compared and analyzed based on accuracy, data processing time, and data preservation rate. The results of the study indicate that the sor algorithm exhibited the highest outlier removal rate in terms of point cloud accuracy. Additionally, the data processing took 2 seconds. In terms of data preservation, 99.6% of the key wall data was retained. The performance of preprocessing algorithms is a critical factor in the development of a progress management system utilizing point clouds; therefore, the sor algorithm is determined to be the most suitable point cloud preprocessing algorithm. © 2024, BPAS Publications. All rights reserved.