基于改进的小波分解织物疵点检测

Textile defect detection based on improvement wavelet decomposes

采用小波分解的改进方法,运用二维离散小波变换进行分解,有效地从图像中提取信息,分析织物的纹理特征并进行相应处理,实现目标图像的特征提取和输入LMBP神经网络进行学习训练。实验结果表明,对油污、破洞、断经、断纬能比较准确地识别和定位,可快速有效地进行织物疵点检测。

The improved method was put forward, using the 2-D discrete wavelet decomposition and extracting the effective information from the pictures. Then characteristics were obtained by relevant processing, and the LMBP neural network was input to carry on the training. The experimental results indicated that it could recognize and localize accurately, carry on the textile defect detection effectively.