基于图像处理的棉亚麻纤维的自动检测

Automatic Detection of Cotton Flax Fiber Based on Image Processing

探讨基于图像处理的棉亚麻纤维自动检测方法。通过显微镜采集棉纤维和亚麻纤维的图像样本,然后采用图像预处理技术对纤维进行特征强化,提取出纤维图像的直径比、直径标准差、平均扭曲度、最大扭曲度、整体充满度和充满度标准差等6个特征值参数,并对特征值参数进行归一化处理和相关性分析,最后通过BP神经网络建立纤维数据库。试验结果表明:基于图像处理的6特征值6阈值与神经网络的组合识别模式自动检测效率较高。认为:该棉亚麻纤维自动检测系统得到的数据值更为客观和准确。

The automatic detection method of cotton flax fiber based on image processing was discussed.The image samples of cotton fiber and flax fiber were collected by microscope.Then,the characteristics of fibers were strengthened by adopting image pre-processing technology.Six characteristic value parameters including diameter ratio,diameter standard deviation,mean torsion resistance,maximum torsion resistance,whole fullness and fullness standard deviation were extracted from the fiber images.Moreover,the normalization processing and correlation analyses were analyzed for the characteristic value parameters.In the end,fiber database was established through BP neural network.The experimental results showed that the automatic detection efficiency of the combination recognition mode for six characteristic values&six threshold values of image processing and neural network was higher.It is considered that the data obtained by the automatic detection system for cotton flax fibers is more objective and accurate.