基于支持向量机的珍珠多特征分类方法
|
摘要
珍珠企业在珍珠分类的过程中需要同时考虑珍珠的形状、纹理和色泽等特征信息,传统珍珠分类方法只针对单一特征对其进行分类,因此提取珍珠的多个特征对其进行分类有着现实意义.在利用单目多视角摄像装置直接获取5个不同视角的珍珠表面图像并进行预处理之后,参考实际人工分类的步骤,用珍珠边缘轮廓得到其傅里叶级数的系数作为形状特征,并用灰度共生矩阵得到珍珠的全局纹理特征,此外还设计了一种新的局部纹理特征提取方法;通过从珍珠的多个视图中提取珍珠的形状特征、全局纹理特征和局部纹理特征,进而构建支持向量机分类器,实现二分类.实验结果表明:所提出的特征提取方法有效,在1 100颗测试珍珠上分类精度达到85.73%.
Pearl companies need to consider the characteristics of pearls such as shape,texture,and color in the process of pearl classification.Traditional pearl classification methods only classify them by single feature,so extracting multiple features to classify pearls has important practical significance.The pearl's surface images from five different visual angles were obtained by a monocular multi-view imaging device and preprocessed.Then,according to the actual manual classification steps,the Fourier series coefficients was computed as the shape features of the pearl and its global texture features was extracted based on the gray-level co-occurrence matrix.Moreover,as the complement of the global texture features,a novel method was developed to extract the local texture features.The SVM model was constructed by utilizing the proposed shape features,the global texture features,and the local texture features.The binary classification of pearls was realized.The experimental results showed that the presented method is quite efficient,and the classification accuracy arrives 85.73%on 1 100 pearls with properly selected kernel function.
Pearl companies need to consider the characteristics of pearls such as shape,texture,and color in the process of pearl classification.Traditional pearl classification methods only classify them by single feature,so extracting multiple features to classify pearls has important practical significance.The pearl's surface images from five different visual angles were obtained by a monocular multi-view imaging device and preprocessed.Then,according to the actual manual classification steps,the Fourier series coefficients was computed as the shape features of the pearl and its global texture features was extracted based on the gray-level co-occurrence matrix.Moreover,as the complement of the global texture features,a novel method was developed to extract the local texture features.The SVM model was constructed by utilizing the proposed shape features,the global texture features,and the local texture features.The binary classification of pearls was realized.The experimental results showed that the presented method is quite efficient,and the classification accuracy arrives 85.73%on 1 100 pearls with properly selected kernel function.
引文
[1]汤一平,夏少杰,李陈荣,等.基于单目多视角视觉的珍珠品质检测[J].农业机械学报,2014,45(4):276-283.
[2]李革,李斌,王莹,等.珍珠形状的计算机视觉识别[J].农业机械学报,2008,39(7):129-132.
[3]夏少杰.基于单目多视角机器视觉的珍珠分级技术研究[D].杭州:浙江工业大学,2015.
[4]汤一平,夏少杰,冯亦军,等.基于单目多视角机器视觉的珍珠在线分类装置[J].农业机械学报,2014,45(1):288-292.
[5]LI G,GONG Y,YUAN W.Human gait recognition based on earth movers distance and zernike moments[J].Procedia CIRP,2016,56:461-464.
[6]郑华文,曹衍龙,杨将新.基于计算机视觉的珍珠形状分级识别技术研究[J].工程设计学报,2008,15(5):365-368.
[7]LI B,CHENG K,YU Z.Histogram of oriented gradient based gist feature for building recognition[J].Computational intelligence and neuroscience,2016(6):1-9.
[8]CHAKI J,PAREKH R,BHATTACHARYA S.Plant leaf recognition using texture and shape features with neural classifiers[J].Pattern recognition letters,2015,58:61-68.
[9]TAHIR M,KHAN A.Protein subcellular localization of fluorescence microscopy images:employing new statistical and texton based image features and SVM based ensemble classification[J].Information sciences,2016,345:65-80.
[10]HARALICK R M,SHANMUGAM K.Textural features for image classification[J].IEEE transactions on systems,man,and cybernetics,1973(6):610-621.
[11]LLOYD K,ROSIN P L,MARSHALL D,et al.Detecting violent and abnormal crowd activity using temporal analysis of grey level co-occurrence matrix(GLCM)-based texture measures[J].Machine vision and applications,2017,28(3/4):361-371.
[12]吴福理,鲁锦樑,胡同森.基于BF-WS的肝脏CT图像自动分割[J].浙江工业大学学报,2015,43(6):630-635.
[13]CORTES C,VAPNIK V.Support-vector networks[J].Machine learning,1995,20(3):273-297.
[14]叶永伟,任设东,陆俊杰,等.基于SVM的汽车涂装线设备故障诊断[J].浙江工业大学学报,2015,43(6):670-675.
[15]SYARIF I,PRUGEL-BENNETT A,WILLS G.SVM parameter optimization using grid search and genetic algorithm to improve classification performance[J].Telecommunication computing electronics and control,2016,14(4):1502-1509.
[16]兰秀菊,张丽霞,鲁建厦,等.基于小波分析和PSO-SVM的控制图混合模式识别[J].浙江工业大学学报,2012,40(5):532-536.
[2]李革,李斌,王莹,等.珍珠形状的计算机视觉识别[J].农业机械学报,2008,39(7):129-132.
[3]夏少杰.基于单目多视角机器视觉的珍珠分级技术研究[D].杭州:浙江工业大学,2015.
[4]汤一平,夏少杰,冯亦军,等.基于单目多视角机器视觉的珍珠在线分类装置[J].农业机械学报,2014,45(1):288-292.
[5]LI G,GONG Y,YUAN W.Human gait recognition based on earth movers distance and zernike moments[J].Procedia CIRP,2016,56:461-464.
[6]郑华文,曹衍龙,杨将新.基于计算机视觉的珍珠形状分级识别技术研究[J].工程设计学报,2008,15(5):365-368.
[7]LI B,CHENG K,YU Z.Histogram of oriented gradient based gist feature for building recognition[J].Computational intelligence and neuroscience,2016(6):1-9.
[8]CHAKI J,PAREKH R,BHATTACHARYA S.Plant leaf recognition using texture and shape features with neural classifiers[J].Pattern recognition letters,2015,58:61-68.
[9]TAHIR M,KHAN A.Protein subcellular localization of fluorescence microscopy images:employing new statistical and texton based image features and SVM based ensemble classification[J].Information sciences,2016,345:65-80.
[10]HARALICK R M,SHANMUGAM K.Textural features for image classification[J].IEEE transactions on systems,man,and cybernetics,1973(6):610-621.
[11]LLOYD K,ROSIN P L,MARSHALL D,et al.Detecting violent and abnormal crowd activity using temporal analysis of grey level co-occurrence matrix(GLCM)-based texture measures[J].Machine vision and applications,2017,28(3/4):361-371.
[12]吴福理,鲁锦樑,胡同森.基于BF-WS的肝脏CT图像自动分割[J].浙江工业大学学报,2015,43(6):630-635.
[13]CORTES C,VAPNIK V.Support-vector networks[J].Machine learning,1995,20(3):273-297.
[14]叶永伟,任设东,陆俊杰,等.基于SVM的汽车涂装线设备故障诊断[J].浙江工业大学学报,2015,43(6):670-675.
[15]SYARIF I,PRUGEL-BENNETT A,WILLS G.SVM parameter optimization using grid search and genetic algorithm to improve classification performance[J].Telecommunication computing electronics and control,2016,14(4):1502-1509.
[16]兰秀菊,张丽霞,鲁建厦,等.基于小波分析和PSO-SVM的控制图混合模式识别[J].浙江工业大学学报,2012,40(5):532-536.