Automated Registration of 3D Scans Using Geometric Features and Normalized Color Data

Abstract

Laser range sensors are playing an increased role in construction. These devices are used to collect a large number of points from different locations and then, those points are registered in a common framework. This article describes a new procedure for the registration of point clouds, especially suited to the fields of architecture and cultural heritage. Often, in these fields, the registration of point clouds is subject to errors due to the fact that an important number of points do not lie on particular geometric features. In this article, an accurate and efficient approach for 3D data registration based on Iterative Closest Point (ICP) algorithm is proposed, which takes advantage of the color data acquired along with range data. Points suitable for registration are selected according to their local geometry and/or color properties, thus a significant improvement on performance convergence and processing time is obtained. The algorithm performs an automatic, on‐the‐flight estimation of the overlapping region, taking into account possible color differences produced by lighting changes through the measurement process. The proposed approach has been tested on real scanned data from cultural heritage buildings and compared to other approaches, showing a better performance in terms of automation degree, accuracy, and speed.

A Review of Heritage Building Information Modeling (H-BIM)

“…In particular, Laser Scanner Technologies (TLS) work via a laser beam that travels towards the area being scanned and back, measuring angles and distances with accuracies from millimeters to centimeters [34] (Figure 2a). In this context, TLS obtain a precise and detailed geometric reproduction of three-dimensional objects in a short time [35], in the form of millions of points (a cloud of points), with geometric coordinates (X, Y, Z) ( Figure 2b) [9], in a digital environment, with both metric and radiometric information. In addition, three different types stand out among the TLS systems: triangulation, phase difference, and Time of Flight (TOF).…”

A Kinect‐Based System for 3D Reconstruction of Sewer Manholes

“…Improvement of the accuracy of alignments is an interesting area for future research (Gómez‐García‐Bermejo et al., ).…”

Color-Based Algorithm for Automatic Merging of Multiview 3D Point Clouds

“…The proposed correlation is based on the assumption that for every range image there is available a corresponding intensity image. Gómez- García-Bermejo et al [2013] extended the Iterative Closest Point (ICP) algorithm to deal with texture data using consideration of convergence and processing time improvement. However, none of those algorithms can be used for point clouds that lack geometrical features and are devoid of additional information about initial alignment or corresponding intensity images.…”