基于神经网络的织物疵点识别技术

因织物组织繁多,表面特征各异,很难建立一个统一的织物疵点识别模型。为了解决这一问题,实现自动验布,提出采用双层神经网络和小波变换来识别织物疵点的方法。先对正常织物进行训练,得到织物的特征,应用第1层简单BP网络来分辨正常织物和疵点。然后对疵点图像进行二维离散小波变换,并去除织物本身的特征,利用已训练的BP网络进行具体疵点识别。试验证明,这种方法的准确性较高,速度快,基本接近自动验布系统的要求。

Fabric defect detection technique based on neural network

Owing to numerous kinds of fabric weaves and varied characteristics of fabric surface, it is very difficult to establish a universal fabric defect detection model. In order to solve this problem and realize automatic inspection of fabric, a method for identification of the defects on the fabric by using the double layer neural network and wavelet analysis is proposed. To be specific, the first layer BP neural network is utilized to tell the defects of a normal fabric and acquire its characteristics through repeated training, and then, the twodimensional discrete wavelet transformation based on the image of the defects is conducted, wiping off the inherent characteristics of the fabric and identifying the defects by means of the trained BP neural network. The experiments demonstrated that this method is of high accuracy and high speed, satisfying the elementary requirements of automatic cloth inspection.