基于多对象匹配与融合字符特征的印刷质量检验方法

Printing Quality Inspection Method Based on Multi-object Matching and Fusion Character Features

中文关键词: 印刷质量检验  融合字符特征  自适应的合格范围  多对象匹配  缺陷类型分析

英文关键词: print quality inspection  fused character features  adaptive qualification range  multi-object matching  defect type analysis

中文摘要:

      目的 为了增强字符配准对字符位姿变化的鲁棒性和识别能力，以及印刷质量检验精度和缺陷类型分析对不同字符产品的自适应性，提出一种基于多对象匹配与融合字符特征的印刷质量检验方法。方法 采用多张合格字符样品图像进行模板构建；借助多对象匹配来配准多个待检验的字符，消除字符位姿的变化对字符配准的影响；进行逐像素的比对，检验字符区域的质量；利用灰度阈值分割以及Sobel边缘检测，将字符区域分成3个待检验的局部特征区域：边缘、前景、后景；进而获取边缘完整性，前景面积和灰度，背景面积和灰度这些显著的字符特征，由多张字符样品训练每个特征的自适应的合格范围；将其组合，形成融合字符特征，分析缺陷的类型。结果 测试数据表明，针对不同种类、不同精度要求的字符产品，所提方法对于字符质量的判断准确率达到100%，对缺陷类型的分类准确率保持在84.2%以上。结论 所提字符质量检验方法拥有良好的鲁棒性与自适应性，在包装、印刷等行业具备较高的应用价值。

英文摘要:

      This work aims to propose a character printing quality inspection method based on multi-object matching and fusion character features, to enhance the robustness and recognition of character registration to changes in character pose, as well as print quality inspection accuracy and defect type analysis for different character products. Firstly, multiple qualified character sample images were applied to template construction. Then the multi-object matching method was introduced to match the characters to be inspected to eliminate the effect of changes in character pose on character registration. Then a pixel-by-pixel comparison was performed to verify the quality of the character area. Last, with gray threshold segmentation and Sobel edge detection, the character region was divided into three local feature regions to be tested: edge, foreground, and background. Significant character features such as edge integrity, area and gray scale of foreground and background were collected. The adaptive qualifying range for each feature was trained from multiple character samples. The type of defect was given with the fused character features. Experimental results show that, when it came to different types of character products with different precision requirements, the accuracy of the proposed method for character quality was 100%, and the accuracy of classification of defect types remained above 84.2%. Featured by good robustness and adaptability, the proposed character quality inspection method has good application value in packaging, printing and other industries.