Simulation of 3D laser scanner for objects dimensional inspection

number: 
1904
English
Degree: 
Author: 
Omar Mobasher Salih
Supervisor: 
Dr. Mohammed Z. Al-Faiz
year: 
2008
Abstract:

The 3D laser scanner is an imaging device that provides the user with 3D models. 3D laser scanner has been used in most of the laser popular applications since it saves time in addition to the good quality results it provides. In this work an algorithm is proposed to simulate the processes of 3D LASER scanning and product dimensional inspection. The scanning operation is divided into two major algorithms, Searching Mode which searches for the sample limits in the range of the scanner with large step size, and Scanning Mode that scans the sample as accurate as possible to measure its dimensions. Secondary algorithms are also used to complete the scanning process. These algorithms are termed “scanned point validation” and “plane fitting” algorithms, these algorithms have been written in matlab 7 environment. The first algorithm was used instead of the laser receiver in the actual systems. With the aid of this algorithm the simulation program is able to recognize whether the scanned point belongs to the body or not. On the other hand, the plane fitting algorithm is used to find the best plane which contains the points that belong to certain groups. As an application of 3D laser scanner in industry, the resulted sample dimensions obtained are exposed to an algorithm of dimension inspection, which checks the sample for shape and dimensions validity with respect to the standard data saved in the database The inspected part has a shape of parallelepiped with square base. The algorithms of 3D laser scanning and dimension inspection are successfully applied on different cases. The first case has sample of an exact dimensions to that of the standard ones and other six cases differed from the standards in length, width, and height one at a time. The results of these cases show that the program accepts only the first case that has the same or near-by dimensions to the standards, and rejects any other cases since it is out of the accepted limits. The results show that the scanning accuracy depends only on the azimuth scanning angle step size. The algorithm is operated more reliable in three different cases (fixed elevation angle á=1o/step and variable azimuth angle 1o, 0.5o, 0.25o/step), where for the three tested cases the errors in measurement are 0.05 unit, 0.02 unit and 0.01 unit respectively. The required computation time is also found to be scanning angles step size dependant. The time needed for scanning is found to be inversely proportional to the step size of both vertical and horizontal scanning angles, where this time was 82s, 268s, and 610s respectively.