摘要
针对现有检测方法不能很好识别混合纤维网中纤维分布情况的问题,提出采用随机共振和图像识别技术相结合的方法。先通过对原始图像做去噪和灰度化处理,再对灰度图像的特征进行分析提取出非目标纤维作为噪声,采用二维级联随机共振对目标纤维特征进行提取,最后通过相似度指标评价混合纤维网中纤维的分布情况。以4种不同质量配比的玄武岩纤维/涤纶混合纤维网和玻璃纤维/涤纶混合纤维网共8组作为检测对象,结果表明:混合纤维网上不同区域的纤维分布差别不大;混合纤维网中目标纤维质量分数越大,所得相似度越大;相同质量配比的混合纤维网,玄武岩纤维相似度高于玻纤纤维相似度。
In view of the problem that the existing detection methods could not identify the fiber distribution in the mixed fiber net very well,the method of stochastic resonance combined with image recognition technology was proposed.Firstly,the original picture was disposed by denoised and grayscale process.Secondly,the feature of gray scale image was analyzed to extract non-target fibers as noise,and then two-dimensional cascaded stochastic resonance was used to extract target fibers.Finally,the similarity index was used to evaluate the fibers distribution in the mixed fiber net.Taken eight groups of basalt fiber/polyester mixed fiber net and glass fiber/polyester mixed fiber net with four different quality ratios as test objects,the result showed that,the distribution of fibers at different areas of mixed fiber net was not significantly different;the greater the mass fraction of the target fibers in the mixed fiber net,the greater the similarity obtained;the similarity of the basalt fiber was higher than that of the glass fibers with the same quality ratio of mixed fiber net.
作者
梁振江
邓辉
王庆涛
朱珍军
Liang Zhenjiang;Deng Hui;Wang Qingtao;Zhu Zhenjun(School of Textiles,Tiangong University,Tianjin 300387,China;Key Laboratory of Advanced Textile Composites,Ministry of Education of China,Tiangong University,Tianjin 300387,China)
出处
《产业用纺织品》
北大核心
2020年第10期33-40,共8页
Technical Textiles
基金
天津市应用基础与前沿技术研究计划(重点项目)(15JCZDJC38500)
国家科技支撑计划项目(2014BAE09B00)。
关键词
混合纤维网
图像识别
二维级联随机共振
相似度
mixed fiber net
image identification
two-dimensional multistage stochastic resonance
similarity
