Continuous Information from Cameras-Laser Scanners to Extract Geometric Parameters of Shapes with Accuracy; Validation of Methods

摘要

Up to last century, measures for land were made point by point with geodesic instruments, reaching accuracies better than one centimetre; at nineties, coordinate measuring machines reach accuracies better than one micrometer, but constrained to touch one point at once. In this century new instrumentation has arrived, which is able to catch thousands to millions of 3D points. The scanners with coupled cameras and laser curtains are one of them, permitting now to register at the time: shapes, assemblies, buildings, hills, and whole scenarios, giving out files of clouds of points. But these new technologies have however their costs, limitations, and risks. Many applications require measures in site, where coordinate machines do not arrive, nor require so high accuracy, nor heavy files with millions of points. The question is: are the scanners of coupled cameras-curtains of laser, and the files they provide out, able to support geometric analysis with accuracy? Our answer is yes if adequate pre-processing, post processing and tests are made. This paper will deal with principles of 3D accuracy, vectors arrangement, vectors processing in shapes, and principles and methods of validation, to give them public recognition, for recognized metrology purposes.