基于匹配滤波和多特征融合的羊绒羊毛纤维识别  

Identification of cashmere and wool based on matched filter and multi-feature fusion

作　　者：朱耀麟[1,2] 穆婉婉 李云红[1] 武桐[1] ZHU Yaolin;MU Wanwan;LI Yunhong;WU Tong(School of Electronics and Information,Xi′an Polytechnic University,Xi′an,Shaanxi 710048,China;School of Electronics and Information,Northwestern Polytechnical University,Xi′an,Shaanxi 710072,China)

机构地区：[1]西安工程大学电子信息学院,陕西西安710048 [2]西北工业大学电子信息学院,陕西西安710072

出　　处：《毛纺科技》2021年第11期64-69,共6页Wool Textile Journal

基　　金：陕西省教育厅服务地方科学研究计划(18JC012);陕西省教育厅重点研究计划产业用纺织品协同创新中心项目(20JY026);陕西省科技厅重点研发计划一般项目(2019GY-098);榆林市科技局科创新城项目(2018-2-24);绍兴市柯桥区西纺纺织产业创新研究院项目(19KQYB10)。

摘　　要：针对传统纤维骨架提取方法鲁棒性较差的问题,提出一种基于匹配滤波和多特征融合的羊绒羊毛纤维识别方法。该方法利用高斯曲线来拟合纤维SEM图像鳞片边缘横截面处灰度值的分布,设计合适的滤波核函数以分割纤维骨架,弱化了光照和噪声的影响;依据纤维骨架图像鳞片密度、鳞片均匀度、形状轮廓的不同,提取其灰度直方图统计特征、GLCM特征和HOG特征并将三者进行融合,避免了鳞片密度计算的复杂性,且较为完善地获取了纤维信息。实验结果表明:提出的算法在羊绒和羊毛纤维的分类上准确率提升至97.1%,相比其他算法至少提高了3.3%,证实了该方法的有效性。Aiming at the problem that the robustness of traditional fiber skeleton extraction method is poor,a method for identifying cashmere and wool based on matched filter and multi-feature fusion was proposed in this paper.Gaussian curve was used to fit the distribution of gray value at the cross-section of the scale edges of the fiber SEM image,the appropriate filter kernel function was designed to segment the fiber skeleton,the influence of light and noise was weakened.The statistical features of gray histogram,gray-level co-occurrence matrix(GLCM)features and histogram of oriented gradient(HOG)features were extracted according to the difference in scale density,scale uniformity and shape contour of fiber skeleton image,then the three features were fused,the complexity of scale density calculation was avoided,and the fiber information was obtained comprehensively.The experimental results show that the accuracy of the proposed algorithm is 97.1%in the classification of cashmere and wool fibers,which at least 3.3%higher than that of other algorithms.The validity of this method is confirmed.

关 键 词：羊绒羊毛 匹配滤波 多特征融合 灰度直方图统计 GLCM HOG 

分 类 号：TS102.3[轻工技术与工程—纺织工程]