中国机械工程

• 智能制造 • 上一篇    下一篇

基于梯度方向信息熵的印刷电路板缺陷检测

李云峰1,2;李晟阳1,2   

  1. 1.河南科技大学机电工程学院,洛阳,471003
    2.机械装备先进制造河南省协同创新中心,洛阳,471003
  • 出版日期:2017-03-25 发布日期:2017-03-23

Defect Detection for PCBs Based on Gradient Direction Information Entropy

LI Yunfeng1,2;LI Shengyang1,2   

  1. 1.School of Mechatronics Engineering,Henan University of Science and Technology,Luoyang,Henan,471003
    2.Collaborative Innovation Center of Machinery Equipment Advanced Manufacturing of Henan Province,Luoyang,Henan,471003
  • Online:2017-03-25 Published:2017-03-23

摘要: 针对印刷电路板裸板缺陷在线视觉检测,提出了一种适用于电路板彩色图像的缺陷检测算法。该算法主要通过分析缺陷区域边界像素的梯度方向信息获得对应的典型图像特征,具体由滤波去噪、目标分割和特征提取三部分组成。为减弱环境光干扰同时保证边缘细节的清晰,首先在CIE Lab色彩空间对图像进行双边滤波,然后利用该色彩空间的均匀性分割出需要检测的目标区域,最后设计了邻域梯度方向信息熵这一描述子用于提取缺陷特征和构造特征向量,利用支持向量机对缺陷进行识别。实验结果表明:所提算法能够对印刷电路板裸板存在的短路、断路、孔洞、余铜、划痕等常见缺陷进行快速精确的定位,能够满足生产过程中的实时检测要求。

关键词: 印刷电路板, 缺陷检测, 双边滤波, 特征提取, 信息熵

Abstract: Aiming at real-time visual inspections of PCB defects, a defect detection algorithm was proposed herein for applying to PCB color image. The typical image features were obtained by analyzing the gradient direction information of the edge pixels in the defect regions. The algorithm consisted of three parts in detail, which were de-noising, target segmentation and feature extraction. Firstly, bilateral filtering in CIE Lab color space was used to reduce the interference of ambient light and ensure the clarity of the edge details; then the uniformity of this color space was used to segment the target areas which needed to be detected; lastly, the neighborhood gradient orientation information entropy was designed as a descriptor for defect feature extraction and feature vector construction, then support vector machine was used to identify defects. Experimental results show that: the new algorithm is able to locate the common defects existing in PCBs rapidly and accurately such as short circuit, open circuit, empty cavity, copper residue, scratch etc. which may meet the real-time detection requirements in production processes.

Key words: printed circuit board(PCB), defect detection, bilateral filtering, feature extraction, information entropy

中图分类号: