Contour-based image alignment for automatic optical inspection in industry

• Main Authors: Chen-Fu Liao, 廖振甫
• Other Authors: Yen-Lin Chen
• Format: Others
• Language: zh-TW
• Published: 2017
• Online Access: http://ndltd.ncl.edu.tw/handle/3ama3p

碩士 === 國立臺北科技大學 === 資訊工程系所 === 105 === In this paper, we propose a new process which integrated by new algorithms and revise some odd methods to establish an automated optical inspection system. The system could inspect the rotation, row scalability, column scalability, translation and evaluate similarity score to achieve defect detection between component of production line, wafer, IC and its golden samples. The proposed system extracted border of image in edge detection of Gaussian image pyramid as feature points and kept important edges by revised topological structural analysis of digitized binary images by border following algorithm.The system speed up performance by revised search strategy of image pyramid, the final result could rivaled commercial software by finding critical value which using sub-pixel estimation to fit transformation into second order polynomial.