Automated tolerance inspection of free form objects

In quality control applications it is an important task to determine if a manufactured object deviates from the design requirements. Inspection involves with measurement and tolerance check of geometric dimensions of manufactured products. There had been presented many inspection solutions for products with regular features, such as quadrangular and circular shapes. However for products with free-form surfaces manufactured by CNC machines, turn benches or plastic injection, the inspection process is still a big problem. By being end-product or sub-product, free-form surfaces are widely used in many fields such as automotive, aerospace, biomedical and machining industries. In this thesis, a vision-based inspection system is proposed for the purpose of improving the speed and the accuracy of the process. We will develop a vision based inspection system which will inspect geometrical and also free form objects within given tolerances. The system will alert if the inspected object dimensions are not in given tolerances, thus the object is faulty. The system can be used in all manufacturing environments, where tolerances of objects are important. Several matching methods were investigated and some of them were tested to major on single method. The Hausdorff distance method is found to be robust and works stable with different shape features, but the processing speed of the method is still a big problem. By considering robustness and promising features, Hausdorff Distance method was studied to improve the speed of the matching process with using different auxiliary techniques. Finally, the results of the applied techniques will be discussed to present the best solution.