Intelligent crack extraction and analysis for tunnel structures with terrestrial laser scanning measurement

Abstract

An automatic and intelligent method for crack detection is significantly important, considering the popularity of large constructions. How to identify the cracks intelligently from massive point cloud data has become increasingly crucial. Terrestrial laser scanning is a measurement technique for three-dimensional information acquisition which can obtain coordinates and intensity values of the laser reflectivity of a dense point cloud quickly and accurately. In this article, we focus on the optimal parameter of Gaussian filtering to balance the efficiency of crack identification and the accuracy of crack analysis. The innovation of this article is that we propose a novel view of the signal-to-noise ratio gradient for Gaussian filtering to identify and extract the cracks automatically from the point cloud data of the terrestrial laser scanning measurement.