基于机器视觉的FPC补强片缺陷智能检测仿真

Simulation of Intelligent Defect Detection of FPC Reinforcement Based on Machine Vision

针对当前电路板缺陷检测方法存在召回率低和复杂度高的问题,提出基于机器视觉的FPC补强片缺陷智能检测方法。通过空域或者时域上的连续图像转换为离散采样点实现柔性印制电路补强片图像采样,将采样得到的柔性印制电路补强片图像函数连续数值转换成其数字等价量,实现图像量化。将量化结果代入中值滤波,利用数据排序方式将图像中没有被污染的点与噪声点替换,完成图像噪声滤除处理。基于处理后的补强片图像,将FPC补强片缺陷检测划分成全局检测与局部检测。利用直方图配准与八连通域面积对全局缺陷进行识别,实现补强片缺陷初步检测,通过投影配准与相关系数对局部缺陷进行检测。实验结果表明,上述方法可有效提升补强片缺陷检测召回率,计算复杂度低于当前相关研究成果。所提方法性能优越,具有合理性与鲁棒性。

Due to low recall rate and high complexity of current detection methods,this article focused on a method for intelligently detecting the defect of FPC reinforce patch based on machine vision.By converting continuous images in spatial domain or time domain to the discrete sampling points,we realized the sampling of reinforce patch image of flexible printed circuit.And then,we converted the continuous function value of image of flexible printed circuit reinforce patch to its equivalent,so as to achieve the image quantization.Moreover,we plugged the quantization results into median filter,and replaced the uncontaminated points and noise points in image by data arrangement,and the image noise filtering was completed.Based on the processed image of reinforce patch,we divided the defect detection of FPC reinforcing patch into the global detection and the local detection.In addition,we used histogram registration and the area of eight connected domains to detect the global defects,so that the preliminary defect detection of reinforce patch was realized.Finally,we detected the local defects by projection registration and correlation coefficient.Simulation results show that the proposed method can effectively improve the recall rate of reinforces panel defect detection and the computational complexity is lower than that of the current research results.The proposed method has superior performance,rationality and robustness.