| With the reduction in package size and the increase in installation density,the detection of irregular components on PCBs(Printed Circuit Board)has become more complex.Manual inspection not only incurs significant labor costs but also poses issues of missed or erroneous detections.Automated optical inspection(AOI)equipment,being efficient,accurate,and reliable,can effectively enhance product quality and reduce production costs.The paper focuses on the research and development of a machine vision-based AOI(Auto Optical Inspection)system for the detection of irregular components on PCBs,aiming to achieve objectives such as recognition,positioning,and defect detection of such components.The paper introduces the current research status of PCB inspection systems and machine vision-based PCB defect detection techniques both domestically and internationally.It designs a machine vision-based AOI solution for the detection of irregular components on PCBs and selects appropriate hardware devices such as cameras,lenses,and light sources.A hardware system is designed,and corresponding software is developed.As PCB components may undergo displacement and rotation during feeding,the captured images need to be corrected and registered.The paper proposes a scale-invariant feature extraction algorithm in combination with affine transformation to achieve precise positioning of marker points.To address the issues of recognition and positioning of irregular components,the paper presents an object recognition algorithm based on combined features and an improved template matching algorithm based on shape.Combined feature selection is performed on the regions of interest,and the minimum bounding rectangle method is used to generate separate rectangular regions for each region of interest,followed by template matching.Conventional shape-based template matching algorithms involve extensive normalization calculations,resulting in long computation time.The paper improves the conventional algorithm by employing the fast Fourier transform algorithm for normalization,significantly enhancing computational efficiency.For defecting detection of irregular components,the paper creates regions of interest(ROIs)for various defects such as missing components and polarity reversals based on the characteristics of irregular components.Feature templates are further generated for each defect and then combined for defect detection.Based on the above algorithms,the paper designs a machine vision-based AOI system for the detection of irregular components on PCBs and conducts experimental tests.The experimental results demonstrate a positioning error within 0.3mm for the center coordinates of marker points,a recognition rate of 95% for six types of irregular components on PCBs,and an accuracy rate of 96% for defect detection. |
