3D digitalization becomes a real need in many domains. It allows capturing and reproducing real-world objects or environments in order to understand and preserve them. It can be achieved by using either photogrammetry or laser scanner. Hence, a 3D point cloud is generated by merging several scans from several scanner positions. Nonetheless, the difficult challenge of this digitalization is to guarantee the quality of the cloud. This quality is mainly measured by completeness, resolution and accuracy. In this paper, we propose a criterion to estimate the completeness and a solution improving the accuracy of the point cloud and withdrawing needless points. Indeed, boolean operations on polygons for calculating the completeness of a predetermined convex area from data of the previous scanner positions is proposed. This approach does not need any sampling of the supposed environment. Furthermore, nearest neighborhood of each point of the cloud and neighbors coming from multiple scans are compared in order to remove useless points and improve the accuracy of digitalization. For evaluation purposes, we consider model of the Hassan mosque in Rabat (Morocco). Thus, the quality of this model was quantified by the three above considered measures.
[1]
William F. Lutz.
Sebastian
,
1923,
The Psychological Clinic.
[2]
Jin Bae Park,et al.
Complete coverage navigation of cleaning robots using triangular-cell-based map
,
2004,
IEEE Transactions on Industrial Electronics.
[3]
Michael M. Kazhdan,et al.
Screened poisson surface reconstruction
,
2013,
TOGS.
[4]
Guillermo Sapiro,et al.
Comparing point clouds
,
2004,
SGP '04.
[5]
Alexander Zelinsky,et al.
Planning Paths of Complete Coverage of an Unstructured Environment by a Mobile Robot
,
2007
.
[6]
Guillaume Caron,et al.
3D POINT CLOUD MODEL COLORIZATION BY DENSE REGISTRATION OF DIGITAL IMAGES
,
2015
.