Constellation, le dépôt institutionnel de l'Université du Québec à Chicoutimi

Résumé

With the increase in the automatization level of industrial production, the three-dimensional laser scanning system has been applied in industrial production more and more widely with its advantages of high identification rate, non-destructive capability, and so on. In order to increase the product quality, product testing becomes an important and indispensable link in the industrial production. The research aim of this thesis is, to develop an automatic on-line testing system, based on 3D laser scanner technology. Such a system can contribute to the realization of a complete automatic process from measuring the product to calculating its form error and further judging whether the product is satisfactory or not.

This thesis, starting from the data points obtained from a from 3D scanner, after filtering the noise points in the right and left CCD, makes use of NURBS surface to fit these point data, and reconstruct the surface of the measured object. Based on the principle of the Least Region to assess the form error of the surface, we obtain a mathematic model of the form error of the surface, and further making use an improved Genetic Algorithm, calculate the form error of the object. According to the given acceptable scale of the error, we can judge whether the product is satisfactory or not. Finally, the program design work of the whole system is realized through the use of VC++6.0, with a user interface to calculate form error being implemented.

This research can be applied in the industrial production in order to increase the product quality and the level of automatization. Other potential applications include artificial intelligence, medicine, etc.