TY - GEN
T1 - S3DA
T2 - 1st International Symposium on Software Fault Prevention, Verification, and Validation, SFPVV 2024
AU - Xu, Yilongfei
AU - Wang, Zhewei
AU - Liang, Jinhao
AU - Zhang, Yueling
AU - Feng, Jincao
AU - Miao, Weikai
AU - Wang, Jiangtao
AU - Pu, Geguang
N1 - Publisher Copyright:
© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2025.
PY - 2025
Y1 - 2025
N2 - Surface Mount Technology (SMT) is prevalent in Printed Circuit Board (PCB) assembly, mainly using solder printing to connect the components and the board. During the process of solder printing, solder defects due to machine failure and environmental factors are widespread. Existing defect detection methods mainly use computer vision to detect solder defects. The main idea of this type of method is to obtain the image information and defect features of the PCB and use the machine learning model to identify the solder defects of the PCB. In actual industrial PCB assembly, the lack of illumination and the occlusion caused by other workpieces leads to incomplete input images for machine learning models, which makes existing methods unable to detect such occluded defects. In order to solve the above problems, this paper proposes a new algorithm for solder defect detection using 3D point cloud data. First, the point cloud data is obtained by scanning the 3D point cloud camera. Next, the point cloud data is denoised and filtered, and the area of interest is further screened to obtain the solder area to be calculated. Finally, using the idea based on integral summation, solder defects are identified by calculating the solder volume. This algorithm can automatically assist manual judgment and effectively identify possible defects in solder processing.
AB - Surface Mount Technology (SMT) is prevalent in Printed Circuit Board (PCB) assembly, mainly using solder printing to connect the components and the board. During the process of solder printing, solder defects due to machine failure and environmental factors are widespread. Existing defect detection methods mainly use computer vision to detect solder defects. The main idea of this type of method is to obtain the image information and defect features of the PCB and use the machine learning model to identify the solder defects of the PCB. In actual industrial PCB assembly, the lack of illumination and the occlusion caused by other workpieces leads to incomplete input images for machine learning models, which makes existing methods unable to detect such occluded defects. In order to solve the above problems, this paper proposes a new algorithm for solder defect detection using 3D point cloud data. First, the point cloud data is obtained by scanning the 3D point cloud camera. Next, the point cloud data is denoised and filtered, and the area of interest is further screened to obtain the solder area to be calculated. Finally, using the idea based on integral summation, solder defects are identified by calculating the solder volume. This algorithm can automatically assist manual judgment and effectively identify possible defects in solder processing.
KW - Automatic Optical Inspection (AOI)
KW - Defect detection
KW - Point cloud
KW - Solder joint
UR - https://www.scopus.com/pages/publications/86000457996
U2 - 10.1007/978-981-96-1621-3_8
DO - 10.1007/978-981-96-1621-3_8
M3 - 会议稿件
AN - SCOPUS:86000457996
SN - 9789819616206
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 115
EP - 130
BT - Software Fault Prevention, Verification, and Validation - 1st International Symposium, SFPVV 2024, Proceedings
A2 - Liu, Shaoying
PB - Springer Science and Business Media Deutschland GmbH
Y2 - 2 December 2024 through 3 December 2024
ER -