摘要
为进一步提高织物瑕疵检测算法对瑕疵类型的通用性,提出一种采用非负字典学习的机织物瑕疵检测算法。首先对正常机织物图像进行窗口分割,将每个子窗口按列展开成列向量,所有列向量联合组成1个矩阵;然后对该矩阵进行非负字典学习,得到个数最佳的非负字典,即基向量;最后应用该字典对待检测样本在最小平方误差下进行近似,并在重构误差的基础上进行疵点检测。重点探讨了窗口大小和字典个数对检测效果的影响。对4 864个样本的实验结果表明,所提算法能在误检率小于10%情况下,取得90%的检出率。
In order to improve the versatility of detection algorithm on varying fabric defect types,an algorithm for woven fabric defect detection using non-negative dictionary learning（ NNDL） is proposed.Firstly,normal fabric image is divided into small image patches and unfolded into a column vector,then all the column vectors are combined into a matrix. Secondly,the matrix composed of column vectors is solved by NNDL,and then the non-negative dictionary with the optimal number（ basis vectors） is extracted. Finally the dictionary is applied to reconstruct testing samples in minimal least square error,where the reconstruction error between original and reconstruction image is used for defect detection. The influence of the patch size and the number of the dictionary are also investigated. Experiment results on4 864 samples show that the proposed method can achieve 90% of detection rate with false detection rate below 10%.
出处
《纺织学报》
EI
CAS
CSCD
北大核心
2016年第3期144-149,共6页
Journal of Textile Research
基金
国家自然科学基金项目(61379011)
关键词
非负字典学习
子窗口尺寸
字典大小
机织物瑕疵
检测
non-negative dictionary learning
patch size
dictionary size
fabric defect
detection
