运用直流抑制傅里叶功率谱特征总和鉴定平纹坯布织物疵点

织物疵点诸如断经、双纬、局部微小疵点,以及稀纬和松纬,都会出现在喷气织机或剑杆织机织造的平纹衬衫坯布中,因此,这些疵点对织物的质量控制至关重要。讨论了应用由傅里叶变换获得的直流抑制傅里叶功率谱(DCSFPS),依据目录频率意义和机织物的周期性来分析织物图像,从而鉴定织物疵点。分析基于由三种不同线密度纱线织造的平纹机织物而实施,计算多达20个处于DCSFPS边缘的特征,同时将DCSFPS输入基于LMBP算法的神经网络。神经网络分别在输入层、隐藏层和输出层最佳化地运用20、40和3个神经元,对坯布织物疵点进行鉴别和分类。神经网络的分析结果重复点不超过20个,且给出的分类精确性几乎达到100%。

Identification of plain grey fabric defects using DC suppressed Fourier power spectrum sum features

The fabric defects such as warp break, double pick and localized microstructure defects; namely loose weft and thick place occurring in grey shirting fabrics woven on looms such as air-jet and rapier looms are becoming crucial for quality control. This paper discusses the application of DC suppressed Fourier power spectrum （DCSFPS） obtained from Fourier transform （FT） for the analysis of fabric images in terms of significant frequency contents and the periodicity of the woven fabric in order to identify the fabric faults. The analysis was carried out on real plain weave grey fabric of 3 different yarn counts by computing as many as 20 features from the marginals of DCSFPS which were used as inputs to the Neural Network （NN） implementing Levenberg-Marquardt Back-propagation algorithm （LMBP）. The results of NN optimized with 20, 40 and 3 neurons in the input, hidden and output layer respectively for identification and classification of grey fabric defects are encouraging with NN converging in less than 20 iterations and giving classification accuracy of almost 100%.