Construction quality detection based on point cloud nonuniform thinning method

Abstract

This study proposes a construction building quality inspection method based on 3D laser scanning and the associated point-cloud data thinning to overcome the low efficiency of traditional methods owing to subjective factors. First, 3D laser scanning was used to collect, process, and stitch the data of a target building to obtain high-precision 3D point-cloud data. Second, to solve the problems associated with large amounts of scanning data and low processing efficiencies, a nonuniform thinning method was designed based on the characteristics of building quality inspection indicators for point-cloud data preprocessing that preserves the concave and convex features of the wall without loss of detection information. Thereafter, automatic feature extraction and point-cloud data fitting were achieved by combining the random sampling consistency (RANSAC) algorithm and the eigenvalue method, after which the corresponding geometric parameters were obtained. Finally, based on the extracted spatial geometric features and the construction quality detection index principle, a spatial geometric operation index detection method based on feature fitting was designed for construction quality detection. The experimental results showed that the proposed uneven thinning method could effectively preserve the concave and convex characteristics of the wall; its flatness value is only reduced by 0.37 % and the data-thinning ratio reached 54.7 %. The proposed construction quality detection method based on 3D laser scanning technology can effectively detect the wall size, flatness, and verticality index. Compared with the traditional detection method, the dimensional accuracy was improved by 49.84 % and the efficiency by 31.67 %, which is conducive to the standardization, processing, and automation of construction detection.