Date of Award


Degree Name

Master of Science


Industrial Engineering


Tzu L. Tseng


Recently more and more industrial applications use image acquisition to improve product manufacturing. The observed growth in the last few years is mainly due to the great advances in acquisition devices which are now affordable for more industrialists. Moreover, the increasing demands in the quality requirements of the products are a great stimulus to apply vision tools which allow a better understanding of the impact of the manufacturing process on the product quality. Vision tools have been used in many industrial fields. But the traditional 2D vision is not as reliable as 3D measurement due to the limitations of the technology and the structure of a part. In this study, a novel approach which integrates photometric stereo reconstruction and 3D measurement for classifying the parts into different categories is presented. The data extracted from several case studies demonstrates the proposed methodology. Results show that the new methodology yielded superior results compared to the traditional inspection approaches with very high classification accuracy. Moreover, the proposed approach is capable to archive 3D models of the parts and achieve rapid quality control. This paper forms the basis for solving many other similar problems that occur in many industries.




Received from ProQuest

File Size

78 pages

File Format


Rights Holder

Jun Zheng