Intelligent Defect Review of Printed Circuit Board (PCBs) is a crucial step in the circuit semiconductor manufacturing process for making a second and accurate judgment on the suspected defects of PCBs detected initially. It uses intelligent means to screen out real defects and eliminate misjudgments, ensuring the quality of PCBs before leaving the factory. As the core component of modern electronic products, PCBs have complex manufacturing processes and more than a hundred types of defects. Due to different application scenarios, the inspection standards vary. Moreover, with complex backgrounds, strong process attributes, and diverse size specifications, it is highly difficult to adapt algorithms for PCBs.
In the early stage of PCB production, AOI/AVI equipment based on traditional algorithms was used to detect PCB defects. The equipment debugging was complicated, and the false alarm rate was high, resulting in a large number of “suspected defects”. Manual re - judgment required checking each PCB one by one, which was inefficient and prone to misjudgments. There are various types of defects (such as short circuits, open circuits, etc.). Some defect features are vague, and traditional algorithms have difficulty accurately distinguishing between real and false defects. This not only restricts the production line yield but also makes it difficult to control quality due to the high cost and poor consistency of manual re - judgment.
