构造自适应模板的多目标缺陷检测方法

Method of Multi-Objective Defect Detection by Constructing Adaptive Template

模板匹配是机器视觉中一种常用的检测方法,通常的做法是在检测前预设标准模板,当模板种类繁多时,应用不便。文章提出了在多目标缺陷检测中构造自适应的模板的方法。首先在多个子目标图像中选取满足某些条件的子图像,然后进行图像配准后加权求和,自适应地生成标准模板,极大减少预设模板的工作量,并提高模板的适用性。基于自适应的模板利用模板匹配、定位、复制及区域差分运算,可简单有效地检测目标的缺陷位置和类型。

Template matching is a common detection method in machine vision. A common practice is to preset standard templates before detection. It is inconvenient for application when the variety of templates. A method of constructing adaptive template in multi-objective defect detection was proposed in this paper. Firstly, certain sub-image that meet some criteria in multiple target sub-image was selected. Then alignment and weighted summation was applied, an adaptive template was generated. This method greatly reduce the workload of preset templates, and improve the applicability of the template. Based on adaptive template, the defect location and type of the target is detected easily and effectively by matching, positioning and copying the template, and region differential operation.