摘要
为保证织物疵点自动检测技术在尽可能低的误检和漏检率条件下,达到尽可能高的检测速度,本文基于开放计算机视觉(OpenCV),给出了判断疵点存在性和利用小波变换提取疵点纹理特征的方法。在Visual C++6.0环境下,开发了单色织物疵点检测和特征值提取程序,并对刮线、色污、长残、脏污、断纱、白杠、反丝、飞花、锈斑和掉扣等疵点样片进行检测实验。实验结果表明,经过小波变换处理后的子图,对于线状疵点织物,其特征值在疵点走向方向上(经向子图或者纬向子图)的变化明显,而对于面状疵点,其特征值在经、纬方向上均变化明显。该研究对实际检测系统的设计开发具有应用价值。
In order to ensure that the technique for automatically detecting fabric defects can achieve a fast defection speed under a lower mistake rate and omission rate, this paper presents the methods to determine the existence of defects and to extract defects~ texture characters using wavelet transform. In Visuall C＋＋6.0, the program to detect monochromatic fabric defects and extract eigenvalue has been developed, and detection experiments on kerf, unclean color, long residual, smear, weft-lacking, white streak, anti-silk, loom fly, rusty spot and buckle-off has been done. The results showed that after the wavelet transform, the texture characters extracted from the subgraph change obviously in the direction of detects （the merid- ional subgraph or zonal subgraph） for the linear fabric, but for the planer defects, the texture characters are changed obviously in the warp and weft directions. The study has the actual application value for the design and development of actual detection system.
出处
《青岛大学学报（工程技术版）》
CAS
2014年第2期75-80,共6页
Journal of Qingdao University(Engineering & Technology Edition)
基金
青岛市科技局基础研究项目(12-1-4-2-(9)-jch)
关键词
疵点检测
开放计算机视觉
小波变换
特征值提取
defect defection
open source computer vision （OpenCV）
wavelet transform
feature extraction
作者简介
时峰（1988-）.男,山东济南人,硕士研究生,主要研究方向为测控技术与智能仪器.
