基于独立分量分析的纺织品缺陷快速检测系统

A rapid detection system for textile defects based on independent component analysis

由于传统方法在进行纺织品缺陷检测过程中无法精准获取源信号分布情况,因此设计一种基于独立分量分析的纺织品缺陷快速检测系统。系统硬件包含织物传动结构、镜头、图像收集与处理卡板;软件为Linux操作系统,运用交叉研发模式完成软件程序研发,设计纺织品缺陷快速检测算法,即根据统计独立原则将测量信号分解成若干互相独立的元素,采用独立分量分析算法分析源信号的非高斯分布,以此提取纺织品图像的高、低阶矩特征,检测纺织品缺陷。实验结果表明,该系统对于纺织品缺陷检测的精度远高于传统方法,实际应用效果更好。

Since the traditional method can not accurately obtain the source signal distribution in the process of textile defect detection,a rapid textile defect detection system based on independent component analysis is designed.The system hardware includes fabric transmission structure,lens,image collection and processing board.The software is Linux operating system,using intersectant research and development mode to complete software application development.Besides,it also designs a fast detection algorithm for textile defects,namely decomposing the measurement signal into independent elements according to statistical independence principle,and uses independent component analysis algorithm to analyze the non-Gaussian distribution of the source signal to extract high and low moments of textile image feature,and detection of textile defects.The experimental results show that the accuracy of the system for textile defect detection is much higher than the traditional method,and the practical application effect is good.