摘要
针对现有算法对机织物纹理适应性和实时性不佳的问题,提出一种基于奇异值分解(SVD)的机织物瑕疵检测算法。首先将正常织物图像的灰度值沿纵横方向进行投影,并将投影所得的序列组成联合投影序列;然后对联合投影序列组成的矩阵进行奇异值分解,并提取基向量;最后应用所提取的基向量对待检测样本进行重构,并通过重构误差区分瑕疵和正常纹理。重点探讨了基向量个数和子窗口大小对检测效果的影响。经过4 693个样本的实验,结果表明,在误检率小于10%情况下,本文算法的检出率可达90%。经比较,本文算法在检测精度和实时性上都优于AR模型算法。
Aiming at texture adaptability and real time issues challenging for most existing algorithms,a method for woven fabric defect detection using singular value decomposition（ SVD） is proposed. Firstly,the gray values of the normal fabric image are projected along the horizontal and vertical directions,and the resulting two sequences are combined into a joint sequence. Secondly,the matrix composed of the joint sequences is solved by SVD,and then the basis vectors are extracted. Finally the extracted basis vectors are used to reconstruct the sample to be tested and its reconstruction error can be used to discriminate defect from normal texture. The effect of the number of vectors and patch size on the detection result is investigated. The experimental results of 4 693 samples show that the proposed method can achieve a detection rate of 90% with a false detection rate below 10%. Comparison shows that the proposed method outperforms AR method in terms of both detection accuracy and real time.
出处
《纺织学报》
EI
CAS
CSCD
北大核心
2014年第7期61-66,共6页
Journal of Textile Research
基金
国家自然科学基金资助项目(61379011
61271006)
中央高校基本科研业务费专项资金资助项目(13D110115)
关键词
机织物
瑕疵检测
奇异值分解
基向量
重构误差
AR模型
woven fabric
defect detection
singular value decomposition
basis vector
reconstruction error
AR model
