Vertical Displacement of Slab End Detection and Quantification Based on Multidimensional Information Fusion of Point Cloud

• Main Authors: Wang, Yu, Gan, Xiaodong, Qi, Hongtuo, Shan, Wenchen, Zhou, Jing, Wu, Zhou
• Format: Conference Proceeding
• Language: English
• Subjects: Accuracy; detection and quantification; Fitting; Fuses; Image segmentation; information fusion; multidimensional; point cloud; Point cloud compression; Quantization (signal); Safety; Slabs; Three-dimensional displays; Transforms; vertical displacement of slab end
• ISSN: 2831-3410
• DOI: 10.1109/ISPCE-ASIA64773.2024.10756250

The detection and quantification for vertical displacement of slab end (VDSE) is critical in infrastructure construction for ensuring project quality and construction safety, etc. Traditional reliance on manual measurement is inefficient, and using the method without multidimension to address this challenge is a significant limitation. To this end, we propose an end-to-end method for VDSE detection and quantification through point cloud with multidimensional information. Firstly, we transform the 3D point cloud into a 2D depth map to detect the location of the VDSE and record the its pixel coordinates. Secondly, we reverse-project the VDSE 2D coordinates to the space 3D coordinates, and split the raw point cloud into two subsets based on the spatial position of the VDSE. Finally, we fuse 2D and 3D information and propose a planar phase method to quantify the VDSE. Experimental results show that our method achieves 85.50/0 precision, 91.3% recall and a maximum 16.4% error for VDSE detection and quantification, providing valuable insights and strategies for addressing analogous challenges in infrastructure construction.