A few years ago three dimensional printers and technologies that are able to scan 3D objects appeared, and they are becoming more and more popular. The 3D scanners can mostly generate point clouds from which – for 3D printing, for example – a triangle mesh generation might be required.
In this paper, a triangle mesh is created from a point cloud – generated by a special 3D scanner – on which error detection and error correction can be efficiently performed. This allows each deficient part of the previous scan – because of the shadow effect, for example – to be partially scanned again. The settings for each following scan can be calculated from the corresponding deficient section of the surface.
In this paper, the automatic generation of a closed mesh, its analysis, error detection and organization is presented. The algorithms of the described process has optimal complexity, so an average computer can also carry out the mentioned tasks on a model consisting of a few million points in a few seconds. Furthermore, the created triangle mesh is still closed, and the borders of the calculated error groups are continuous. In this paper, I provide a detailed complexity analysis and corresponding measurement results that prove the theory and which serve as evidence of the suitability of the process.
Further scanning setups can automatically be calculated from the results of the described algorithms, and methods can be created to correct the error groups.
Depending on the properties of the scanner, the most optimal line of sight of the object can be obtained by different settings of the scanner, which is required for rescanning the faulty parts. The scanned geometry gradually gets more precise and finally it will reach the point when it cannot be further improved by the given hardware.
In this paper, a concrete method is also provided for error correction, which can be used for inserting a proper rescan into the previous triangle mesh.