ISBN: 978-981-18-5852-9 DOI: 10.18178/wcse.2022.04.194
Research on Inspection Method for Assembly Integrity of Electronic Components
Abstract— In order to improve the efficiency of assembly integrity inspection of electronic components and reduce the workload of manual inspection, an automatic inspection method of assembly integrity of electronic components based on vector drawings and digital image matching recognition is proposed. The method first abstracts the circuit symbols in the circuit sheet in vector format into a weighted undirected diagram, and then uses the isomorphic algorithm of the diagram to identify the type and number of symbols in the drawing. Secondly, the Mask R-CNN instance segmentation model is used to identify the type and quantity of electronic components in the circuit board. Finally, the recognition results of vector drawings and digital images are compared to determine whether the assembly is complete. On the experimental data set containing a total of 321 electronic components on 15 circuit boards, 14 pieces of this method were completely matched, and 1 piece of small number of electronic components were identified as errors, but it also can assist inspectors in judging the integrity of the assembly and provides ideas for automated inspections.
Index Terms— electronic components, assembly integrity inspection, vector drawing recognition, digital image recognition, isomorphism of graphs, mask R-CNN.
Yikang Li
School of Computer, Xi'an University of Posts & Telecommunications
Cite: Yikang Li, " Research on Inspection Method for Assembly Integrity of Electronic Components, " WCSE 2022 Spring Event: 2022 9th International Conference on Industrial Engineering and Applications, pp. 1696-1702, Sanya, China, April 15-18, 2022.