Flexible printed circuit defective detection based on image registration

This paper studies the application of advanced computer image processing techniques for solving problem of automated defect detection for flexible printed circuit (FPC). Referent image and inspecting image of FPC are obtained, and align them in the same position through affine transformation which is also called image registration. According to the principle that mutual information entropy (MI) increases as image registration accuracy improves, genetic algorithm integrated with MI is applied to searching optimal image registration parameters to improve image registration accuracy. The experimental result prove that the registration result can be optimized effectively. On the basis of that, inspecting image is subtracted from referent image, the residual image shows possible defect. Through analyzing the square and the shape of possible defect, all defects can be detected finally. The inspecting result proves accurate defect detection. The inspecting accuracy can reach 93.5%. And 200 inspecting targets cost 3891ms, The method proposed in paper does not only improve inspecting accuracy, but also improve efficiency in practice.