基于多特征融合图像分析技术的羊毛与羊绒鉴别
|
摘要
为快速准确地鉴别羊毛与羊绒,提出一种基于多特征融合的鉴别方法。首先利用光学显微镜及数码相机对羊毛与羊绒纤维进行图像采集,然后分别采用2种类型的预处理操作得到单根纤维图像与去除背景的纤维二值图像;其次通过灰度共生矩阵算法提取第1类预处理后羊毛与羊绒纤维图像的纹理特征参数,基于中轴线算法提取第2类预处理后纤维图像的直径形态特征参数;最后将纹理及形态特征参数融合成多维数组并通过K均值算法进行聚类识别。实验结果显示,与传统利用单一纤维特征提取算法进行识别的方法相比,该算法平均识别率可达到95.25%,识别率较高,可用于羊毛与羊绒纤维的自动分类识别。
For rapid identification of wool and cashmere, a method based on the multi-feature fusion for the fiber identification was proposed. Firstly, the images of wool and cashmere fibers were captured by an optical microscope and a digital camera. Secondly, two kinds of preprocessing operations were carried out respectively, and the binary images of single fiber image and background free fiber were obtained. Then, the texture parameters of the first kind of cashmere and wool fiber images were extracted by the gray level co-occurrence matrix algorithm and the diameter parameters of the second kinds of fiber images were extracted based on the central axis algorithm. Finally, the texture and morphological feature parameters were fused into multidimensional array and the clustering analysis was carried out by the K-means algorithm. The experimental results show that the average identification rate of the algorithm proposed can reach 95.25%. Compared with the conventional single fiber feature extraction algorithm, the recognition rate is high, which confirmed that this method can be used for automatic classification and identification of cashmere and wool fibers.
For rapid identification of wool and cashmere, a method based on the multi-feature fusion for the fiber identification was proposed. Firstly, the images of wool and cashmere fibers were captured by an optical microscope and a digital camera. Secondly, two kinds of preprocessing operations were carried out respectively, and the binary images of single fiber image and background free fiber were obtained. Then, the texture parameters of the first kind of cashmere and wool fiber images were extracted by the gray level co-occurrence matrix algorithm and the diameter parameters of the second kinds of fiber images were extracted based on the central axis algorithm. Finally, the texture and morphological feature parameters were fused into multidimensional array and the clustering analysis was carried out by the K-means algorithm. The experimental results show that the average identification rate of the algorithm proposed can reach 95.25%. Compared with the conventional single fiber feature extraction algorithm, the recognition rate is high, which confirmed that this method can be used for automatic classification and identification of cashmere and wool fibers.
引文
[1] 蒋高平, 钟跃崎, 王荣武, 等. 基于谱线特征的羊绒与羊毛的鉴别[J]. 纺织学报, 2010, 31(4): 15-19.JIANG Gaoping, ZHONG Yueqi, WANG Rongwu, et al. Identification of wool and cashmere based on characteristics of spectral line [J]. Journal of Textile Research, 2010, 31(4): 15-19.
[2] 唐杰, 赵世海, 银海燕, 等. 羊毛、羊绒纤维鉴别方法综述[J]. 毛纺科技, 2014, 42(7): 48-50.TANG Jie, ZHAO Shihai, YIN Haiyan, et al. Identification methods description of wool and cashmere [J]. Wool Textile Journal, 2014, 42(7): 48-50.
[3] 张荣娜, 秦士忠, 朱林平. 山羊绒与其他动物纤维CAT扫描电子显微镜系统研究[J]. 检验检疫科学, 2001, 11(3): 1-4,11.ZHANG Rongna, QIN Shizhong, ZHU Linping. Study on CAT scanning electron microscope system of cashmere and other animal fibers [J]. Inspection and Quarantine Science, 2001, 11(3): 1-4,11.
[4] 陈国华, 王金泉. 羊绒与羊毛的碱溶度差异[J]. 毛纺科技, 2000, 28(2): 18-20.CHEN Guohua, WANG Jinquan. Difference in alkali solubility of cashmere and wool [J]. Wool Textile Journal, 2000, 28(2): 18-20.
[5] TANG M, ZHANG W, ZHOU H, et al. A real-time PCR method for quantifying mixed cashmere and wool based on hair mitochondrial DNA [J]. Textile Research Journal, 2014, 84(15): 1612-1621.
[6] HARALICK R M. Statistical and structural approaches to texture [J]. Proceeding of IEEE, 1975, 67(5): 786-504.
[7] 汪黎明, 陈健敏, 王锐, 等. 织物折皱纹理灰度共生矩阵分析[J]. 青岛大学学报(工程技术版), 2003, 18(4): 5-8.WANG Liming, CHEN Jianmin, WANG Rui, et al. The analysis of grain of fabric winkle by concurrence matrix of gray degree[J]. Journal of Qingdao University (Engineering Technology Edition), 2003, 18(4): 5-8.
[8] 白雪冰, 王克奇, 王辉, 等. 基于灰度共生矩阵的木材纹理分类方法的研究[J]. 哈尔滨工业大学学报, 2005, 37(12): 1667-1670.BAI Xuebing, WANG Keqi, WANG Hui, et al. Research on the classification of wood texture based on gray level co-occurrence matrix [J]. Journal of Harbin Institute of Technology, 2005, 37(12): 1667-1670.
[9] PORTILLO-GARCIA J, TRUEBA-SANTANDER I, MIGUEL-VELA G D, et al. Efficient multi-spectral texture segmentation using multivariate statistics [J]. IEE Proceeding: Vision Image, and Signal Processing, 1998, 145(5): 257-364.
[10] 刘玮, 李发源, 熊礼阳, 等. 基于区域生长的黄土地貌沟沿线提取方法与实验[J]. 地球信息科学学报, 2016, 18(2): 220-226.LIU Wei, LI Fayuan, XIONG Liyang, et al. Shoulder line extraction in the loess plateau based on region growing algorithm [J]. Journal of Geo-Information Science, 2016, 18(2): 220-226.
[11] 王蒙, 吕建平. 基于边缘检测和自动种子区域生长的图像分割算法[J]. 西安邮电学院学报, 2011, 16(6): 16-19. WANG Meng, Lü Jianping. An image segmentation algorithm based on edge extraction with automatic seeded region growing [J]. Journal of Xi′an Institute of Posts and Telecommunications, 2011, 16(6): 16-19.
[12] 辛栋. 心脏造影图像中血管直径测量技术研究[D]. 郑州:郑州大学, 2012: 22-26.XIN Dong. Study on measurement of blood vessel diameter in cardiac contrast images [D]. Zhengzhou: Zhengzhou University, 2012: 22-26.
[13] OSHER S, FEDKIW R. Level Set Methods and Dynamic Implicit Surfaces [M]. New York: Springer-Verlag, 2002:15-30.
[14] 朱俊平, 路凯, 柴新玉, 等. 羊绒与羊毛直径的水平集中轴线法测量[J]. 纺织学报, 2017, 38(9): 14-18.ZHU Junping, LU Kai, CHAI Xinyu, et al. Level set of central axis method of cashmere and wool diameter [J]. Journal of Textile Research, 2017, 38(9): 14-18.
[15] 孙可, 刘杰, 王学颖. K均值聚类算法初始质心选择的改进[J]. 沈阳师范大学学报(自然科学版), 2009, 27(4): 448-450.SUN Ke, LIU Jie, WANG Xueying. K Mean cluster algorithm with refined initial center point [J]. Journal of Shenyang University (Natural Science Edition), 2009, 27(4): 448-450.
[16] 焦明艳. 一种基于灰度共生矩阵的羊绒与羊毛识别方法[J]. 成都纺织高等专科学校学报, 2017 (3): 126-129.JIAO Mingyan. A recognition method of cashmere and wool based on gray level co-occurrence matrix [J]. Journal of Chengdu Textile College, 2017(3): 126-129.
[2] 唐杰, 赵世海, 银海燕, 等. 羊毛、羊绒纤维鉴别方法综述[J]. 毛纺科技, 2014, 42(7): 48-50.TANG Jie, ZHAO Shihai, YIN Haiyan, et al. Identification methods description of wool and cashmere [J]. Wool Textile Journal, 2014, 42(7): 48-50.
[3] 张荣娜, 秦士忠, 朱林平. 山羊绒与其他动物纤维CAT扫描电子显微镜系统研究[J]. 检验检疫科学, 2001, 11(3): 1-4,11.ZHANG Rongna, QIN Shizhong, ZHU Linping. Study on CAT scanning electron microscope system of cashmere and other animal fibers [J]. Inspection and Quarantine Science, 2001, 11(3): 1-4,11.
[4] 陈国华, 王金泉. 羊绒与羊毛的碱溶度差异[J]. 毛纺科技, 2000, 28(2): 18-20.CHEN Guohua, WANG Jinquan. Difference in alkali solubility of cashmere and wool [J]. Wool Textile Journal, 2000, 28(2): 18-20.
[5] TANG M, ZHANG W, ZHOU H, et al. A real-time PCR method for quantifying mixed cashmere and wool based on hair mitochondrial DNA [J]. Textile Research Journal, 2014, 84(15): 1612-1621.
[6] HARALICK R M. Statistical and structural approaches to texture [J]. Proceeding of IEEE, 1975, 67(5): 786-504.
[7] 汪黎明, 陈健敏, 王锐, 等. 织物折皱纹理灰度共生矩阵分析[J]. 青岛大学学报(工程技术版), 2003, 18(4): 5-8.WANG Liming, CHEN Jianmin, WANG Rui, et al. The analysis of grain of fabric winkle by concurrence matrix of gray degree[J]. Journal of Qingdao University (Engineering Technology Edition), 2003, 18(4): 5-8.
[8] 白雪冰, 王克奇, 王辉, 等. 基于灰度共生矩阵的木材纹理分类方法的研究[J]. 哈尔滨工业大学学报, 2005, 37(12): 1667-1670.BAI Xuebing, WANG Keqi, WANG Hui, et al. Research on the classification of wood texture based on gray level co-occurrence matrix [J]. Journal of Harbin Institute of Technology, 2005, 37(12): 1667-1670.
[9] PORTILLO-GARCIA J, TRUEBA-SANTANDER I, MIGUEL-VELA G D, et al. Efficient multi-spectral texture segmentation using multivariate statistics [J]. IEE Proceeding: Vision Image, and Signal Processing, 1998, 145(5): 257-364.
[10] 刘玮, 李发源, 熊礼阳, 等. 基于区域生长的黄土地貌沟沿线提取方法与实验[J]. 地球信息科学学报, 2016, 18(2): 220-226.LIU Wei, LI Fayuan, XIONG Liyang, et al. Shoulder line extraction in the loess plateau based on region growing algorithm [J]. Journal of Geo-Information Science, 2016, 18(2): 220-226.
[11] 王蒙, 吕建平. 基于边缘检测和自动种子区域生长的图像分割算法[J]. 西安邮电学院学报, 2011, 16(6): 16-19. WANG Meng, Lü Jianping. An image segmentation algorithm based on edge extraction with automatic seeded region growing [J]. Journal of Xi′an Institute of Posts and Telecommunications, 2011, 16(6): 16-19.
[12] 辛栋. 心脏造影图像中血管直径测量技术研究[D]. 郑州:郑州大学, 2012: 22-26.XIN Dong. Study on measurement of blood vessel diameter in cardiac contrast images [D]. Zhengzhou: Zhengzhou University, 2012: 22-26.
[13] OSHER S, FEDKIW R. Level Set Methods and Dynamic Implicit Surfaces [M]. New York: Springer-Verlag, 2002:15-30.
[14] 朱俊平, 路凯, 柴新玉, 等. 羊绒与羊毛直径的水平集中轴线法测量[J]. 纺织学报, 2017, 38(9): 14-18.ZHU Junping, LU Kai, CHAI Xinyu, et al. Level set of central axis method of cashmere and wool diameter [J]. Journal of Textile Research, 2017, 38(9): 14-18.
[15] 孙可, 刘杰, 王学颖. K均值聚类算法初始质心选择的改进[J]. 沈阳师范大学学报(自然科学版), 2009, 27(4): 448-450.SUN Ke, LIU Jie, WANG Xueying. K Mean cluster algorithm with refined initial center point [J]. Journal of Shenyang University (Natural Science Edition), 2009, 27(4): 448-450.
[16] 焦明艳. 一种基于灰度共生矩阵的羊绒与羊毛识别方法[J]. 成都纺织高等专科学校学报, 2017 (3): 126-129.JIAO Mingyan. A recognition method of cashmere and wool based on gray level co-occurrence matrix [J]. Journal of Chengdu Textile College, 2017(3): 126-129.