基于图像线灰度和BP神经网络的织物疵点检测  

Fabric Defect Detection Method Based on Image Line Grayscale and BP Neural Network

作　　者：徐晓峰[1] 

机构地区：[1]山东省威海职业学院

出　　处：《棉纺织技术》2012年第5期26-29,共4页Cotton Textile Technology

摘　　要：探讨织物疵点自动检测的方法。通过对4种常见织物疵点的图像进行线灰度曲线分析和处理,提取疵点图像的特征值,送入BP神经网络进行识别,从而实现织物疵点的检测。试验结果表明,该方法取得了较好的检测效果,织物疵点识别率达到93%以上。认为,此法能够有效识别出织物中的几类常见疵点,应进一步研究,以提高其识别准确率。Fabric defect automatic detection method was discussed. Images of four kinds of common fabric defects were analyzed and treated by line grayseale curve, characteristics of fabric defect images were abstracted and identified in BP neural network,then fabric defect detection can be realized. The test result shows that the detection result is better, fabric＇, defect recogntion rate can reach above 93%. It is considered that several types of defects can be recognized efffctively by the method,the reeogntion should be increased further.

关 键 词：线灰度 神经网络 织物图像 织物疵点 检测方法 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] TP274[自动化与计算机技术—计算机科学与技术]