Accurate crack detection and characterization on concrete are essential for the maintenance, safety, and serviceability of various infrastructures. In this paper, an innovative approach was developed to automatically measure the cracks from 3D point clouds collected by a phase-shift terrestrial laser scanner (TLS) (FARO Focus3D S120). The approach integrates several techniques to characterize the cracks, which include the deviation on point normal determined using k-nearest neighbor (kNN) and principal components analysis (PCA) algorithms to identify the cracks, and principal axes and curve skeletons of cracks to determine the projected and real dimensions of cracks, respectively. The coordinate transformation was then performed to estimate the projected dimensions of cracks. Curve skeletons and cross sections of cracks were extracted to represent the real dimensions. Two cases of surface cracks were used to validate the developed approach. Because of the differences in definitions of the crack dimension in the three methods and due to the curve shape of the crack, the width and depth of cracks obtained from the cross-section method and manual measurement were close but slightly smaller than those measured by the projection algorithm; whereas the length of cracks determined by the curve-skeletons method was slightly larger than those obtained by the manual measurement and projection method. The real dimension of a crack has good agreements with real situations when compared with the results of the manual measurement and projection method.


Civil, Architectural and Environmental Engineering


Missouri University of Science and Technology, Grant 2017YFC1501303

Keywords and Phrases

Automated Characterization; Automated Detection; Cracks; Laser Scanning; Point Clouds; Point Normal Variance

International Standard Serial Number (ISSN)

1943-555X; 1076-0342

Document Type

Article - Journal

Document Version

Final Version

File Type





© 2023 American Society of Civil Engineers, All rights reserved.

Publication Date

01 Jun 2023