Acta IMEKO, Title ACTA IMEKO ISSN: 2221-870X June 2018, Volume 7, Number 2, 39-44 ACTA IMEKO | www.imeko.org June 2018 | Volume 7 | Number 2| 39 Point cloud processing techniques and image analysis comparisons for boat shapes measurements Sofia Catalucci1,2, Roberto Marsili1, Michele Moretti1, Gianluca Rossi1 1Università degli Studi di Perugia, Dipartimento d’Ingegneria, Via Goffredo Duranti 93, 06125 Perugia, Italy 2Present address: University of Nottingham, Advanced Manufacturing Building, Jubilee Campus, NG81BB Nottingham, United Kingdom Section: RESEARCH PAPER Keywords: photomodelling; metrology; point clouds; Iterative Closest Point algorithm Citation: Sofia Catalucci, Roberto Marsili, Michele Moretti, Gianluca Rossi, Point cloud processing techniques and image analysis comparisons for boat shapes measurements, Acta IMEKO, vol. 7, no. 2, article 7, June 2018, identifier: IMEKO-ACTA-07 (2018)-02-07 Section Editor: Fabio Leccese, Università degli Studi di Roma Tre, Italy Received January 17, 2018; In final form March 25, 2018; Published June 2018 Copyright: © 2018 IMEKO. This is an open-access article distributed under the terms of the Creative Commons Attribution 3.0 License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited Funding: This work was supported by the Engineering Department, Università degli Studi di Perugia, Italy Corresponding author: Sofia Catalucci, e-mail: sofia.catalucci@gmail.com 1. INTRODUCTION The design and modelling of a boat involve complex free- form geometric shapes, difficult to measure and survey with traditional metrology methods. Scan data allow creating precise 3D models that can be used by naval designers and engineers to ensure the quality of interior and exterior construction, as well as for simulation and inspection purposes. For this purpose, the paper describes photomodelling technique, a recent and fast image processing and alignment method that leads to the reconstruction of three-dimensional models, starting from the simple acquisition of photographic images. Close to photogrammetry, the result obtained is a 3D point cloud, a set of x,y,z space coordinates, first form of the object surveyed [1]. A point cloud can be identified as a pixel cloud, because of the direct relationship between photomodelling and photography: each pixel of an image corresponds to a point of the cloud, thus preserving the chromatic characteristics of the object surveyed [2]-[6]. 2. TEST PERFORMED Object of the survey is a Beneteau First 456/s boat (Figure 1), 1984, with 3 cabin all with toilets, engine Yanmar 55 cv, 14 m length. The measurements have been performed on the driving seat area, by using photomodelling technique and the Creaform 3D scanner, placed on the same surface (Figure 2). At first, the survey has been performed using the 3D scanning system (Figure 3); target stickers have been applied randomly to facilitate the capture process, due to the auto similarity of the surface pattern. Figure 1. The Boat. ABSTRACT Photomodelling is a new and fast solution for 3D modelling, based on the same principles of photogrammetry. The comparison between photomodelling and the metrological technique of structured light 3D scanning, provided by the Creaform Go Scan 50 with metrological certification, is the aim of this paper, defining performances and verifying the potential of this innovative, simple and economical technique. ACTA IMEKO | www.imeko.org June 2018 | Volume 7 | Number 2| 40 The survey carried out by the photomodelling technique included a totality of 20 images, loaded in three different dedicated software: Agisoft Photoscan, Visual SfM and Autodesk Remake. Each instrument is different because of its time data processing, difficulty of use, accuracy and precision of results [7]. Furthermore, the proposed applications include both open-source and commercial software. 3. DATA PROCESSING The 3D data processing is the same for each software: the first operation is the manipulation of the 3D point cloud. The next step is the creation of three-dimensional model, called “triangulation”: starting from the input data vertices, edges and faces are generated. The result obtained is a set of coordinates, which is converted into a polygonal surface [8]-[10]. With editing software as MeshLab and Geomagic Studio, it is possible to perform manual editing of data, merging, scaling, aligning of different surfaces, creating a three-dimensional surface with the aid of different algorithms [11]-[13] (Figure 4, Figure 5, Figure 6). 4. COMPARISON After the alignment process, the research concerned the comparison between the results of photomodelling technique, identified as TEST, and a REFERENCE model. The surface reconstructed by the scanning system has been chosen as reference [14]-[16]. Figure 4. Surface 3D model and point cloud by Agisoft Photoscan. Figure 5. Surface 3D model and point cloud by Visual SfM. Figure 2. Driving seat area. Figure 3. Boat surface scanning by Creaform ScanGo 50. ACTA IMEKO | www.imeko.org June 2018 | Volume 7 | Number 2| 41 The software chosen for the comparison is Geomagic Qualify. The comparison has been made using algorithms that provide variances and deviations between geometric entities in the space and it had as output 2D and 3D maps of such deviations. It was possible to derive a matrix in .CSV format of the test and reference points spatial coordinates x,y,z and of the values of the mutual distances for each pair of points. After the manual and automatic alignment process, a spectrum of 15 intervals of values has been set: -10 and +10 mm deviation is the range of acceptability; the range between -1 and +1 mm is the optimum correspondence between the test and reference points. As it is clear from the various elaborate graphs, the best result is the comparison with the TEST model of Agisoft Photoscan (Table 1, Figure 7). The results obtained with Visual SfM are close to the previous ones (Table 1, Figure 8). On the contrary, Autodesk Remake highlights the differences in terms of precision and accuracy of both the previously mentioned software programs, especially the second one. Autodesk Remake is a simple software, which can be used by every kind Figure 6. Surface 3D model by Autodesk Remake. Table 1. Deviation distributions. Deviation distribution range Agisoft Photoscan Number of points Visual SfM Number of points Autodesk Remake Number of points n>10 1536 1550 10136 8.510 1536 1550 10136 8.5