基于模板匹配的肥料外包装印刷缺陷检测研究

Research on Defect Detection of Fertilizer Packaging Based on Template Matching

针对包装印刷品生产中人工缺陷检测成本高、效率低且漏检和误检率较高等问题,本研究提出了一种基于机器视觉与模板匹配算法的肥料外包装印刷品缺陷检测方案。该方案以某企业肥料外包装产品作为检测对象,首先使用灰度特征图方法对采集的外包装图像进行二值化处理,并通过形态学操作建立了感兴趣区域(Region of Interest,ROI)与形状模型。然后使用Frei-Chen算子提取图像边缘信息创建差异模型,利用形状模型与ROI对待测图像进行借用相似度量的定位、仿射变换以及灰度值的放缩等方法的处理。最后将得到的图像输入差异模型进行最终的缺陷检测。实验通过对大量具有不同缺陷的待测图像进行检测,验证了方案的有效性。

To sovle the problem of high cost,low efficiency,high rate of missed inspection and false inspection of manual defect detection in the industrial production of packaging and printing,a method of detecting printing defects in fertilizer packaging based on machine vision and template matching algorithm was proposed in this study.This method took an enterprise’s fertilizer packaging product as the detection object.First,the gray-level feature map method was used to binarize the collected packaging image.The region of interest(ROI)and shape model were established by morphological operation.Then,the Frei-Chen operator was used to extract the edge information of the image to create a different model.The shape model and ROI were used to process the image to be measured,such as the location with similarity measures,affine transformation and scaling of gray values.Finally,the obtained image was input into the different models for final defect detection.The experimental results verified the method effectiveness by detecting many printing images with various defects.