基于脉冲神经网络的织物瑕点检测算法

Fabric Defect Detection Algorithm Based on Spiking Neural Network

针对纺织业中织物瑕点检测尚未达到自动化、智能化,提出一种新的织物检测算法——基于脉冲神经网络(SNN)的织物瑕点检测算法.该算法根据生物信息处理机制,把网络模型分成3层,分别为:接收层、中间层、输出层.接收层负责将图像信息转为神经元的输入序列;中间层根据四个不同权重矩阵的输出信息判断是否属于四个方向上的边缘像素点;输出层将汇总中间的输出信息并判断是否为边缘图像点;最后根据特征值经mallat算法定位瑕点区域.通过实验分析和对比,其结果表明该算法具有较好的检测率.

In view of the lack of automation and intelligence in fabric defect detection,a new fabric defect detection algorithm based on spiking neural network(SNN)is proposed in this paper.According to the biological information processing mechanism,the network model is divided into three layers,namely,receiving layer,middle layer and output layer.The receiving layer is responsible for converting the image information into the neuron’s input sequence;The middle layer determines whether the edge pixel points belong to four directions according to the output information of four different weight matrices.The output layer will summarize the intermediate output information and determine whether it is edge image point.Finally,the defect area is located by Mallat algorithm based on eigenvalues.Experimental analysis and comparison show that the algorithm has good detection rate.