Nontarget-Based Displacement Measurement Using Computer Vision and 3-D Point Cloud

Abstract

The computer vision-based displacement measurement method, which is a noncontact method, has recently drawn significant attention because of its measurement convenience and great performance. However, computer vision-based methods generally require the installation of an artificial target on the structural surface to measure the displacement. Although the artificial target provides clear traceable patterns for robust pattern tracking and metric information for coordinate transformation, target-based methods have practical limitations in field owing to the need for target installation and restrictions on camera positioning. To overcome these limitations, a nontarget-based approach, which uses structural features instead of artificial targets, has been developed. Existing nontarget-based methods still require calibration using artificial targets or the known dimensions of the structure. Moreover, most existing methods are limited to one- or two-degrees of freedom (DOF) in displacement measurement, which limits the reliability of structural health monitoring (SHM) for structures requiring out-of-plane displacement measurement. This paper proposes nontarget-based 6-DOF and in-plane structural displacement measurement approaches using a three-dimensional (3-D) point cloud. The 6-DOF displacement measurement method utilizes the combined red, green, and blue (RGB) and depth (D) information obtained using an RGB-D camera. This method provides 3-D information of the structure and suggests a coordinate transform scheme to estimate the physical 6-DOF displacement without using a target. The proposed method is validated in two laboratory experiments using a shear building model and a pan-tilt with a predefined measurement plate, in which both translational and rotational displacements are measured. In the proposed nontarget-based, in-plane displacement measurement method, the laser scanning 3-D point cloud combined with high-resolution RGB information of structure is used to determine the precise physical displacement tailored to the field measurement. The proposed method also utilizes light detection and ranging (LiDAR) and a camera that provides the combined 3-D point cloud and RGB information, presenting the relationship between the image and physical displacement for the coordinate transform. The proposed method is validated in a laboratory-scale experiment using a concrete specimen and a field test with a full-scale bridge.