登录    |    注册

您好,欢迎来到澳门新葡新京网赌!

澳门新葡新京首页>澳门新葡新京网赌期刊>本期导读>精确型傅里叶高度函数描述子的服装款式识别方法

精确型傅里叶高度函数描述子的服装款式识别方法

117    2019-08-27

¥0.00

全文售价

作者:孙国栋, 徐亮

作者单位:湖北工业大学机械工程学院, 湖北 武汉 430068


关键词:精确型傅里叶高度函数;支持向量机;服装款式识别


摘要:

针对服装款式自动识别时存在提取特征困难、识别率低和分类效率低等问题,该文在精确型傅里叶高度函数(accurate Fourier height functions 2,AFHF2)与线性核函数支持向量机(support vector machine,SVM)基础上提出一种新的服装款式识别方法。首先,利用AHFH2描述子对衣服轮廓进行特征提取,对服装轮廓全局信息和局部信息进行充分表征;然后,在不需要调整参数的情况下使用线性核函数SVM对AFHF2描述子特征进行快速训练与测试。通过自建的服装图形库验证该方法的有效性,实验结果表明该算法优于现有的算法,其中AFHF2描述子优于傅里叶描述子、高度函数(HF)和TCDs等算法,线性核函数SVM算法优于径向基函数SVM算法、K-近邻算法、概率神经网络以及反向传播神经网络算法,其平均识别率能达到97.91%。


Clothing style recognition approach based on the accurate Fourier height functions shape descriptor
SUN Guodong, XU Liang
School of Mechanical Engineering, Hubei University of Technology, Wuhan 430068, China
Abstract: In order to overcome the difficulty in extracting the suitable feature of the clothing and low recognition rate and low classification efficiency of the identification on clothing style, a new clothing style recognition approach based on the accurate Fourier height functions 2 (AFHF2) shape descriptor and the support vector machine (SVM) using linear kernel function is proposed. First, AFHF2 shape descriptor is applied to extracting feature that can fully characterize the global and local information of clothing contours. Then the SVM using linear kernel function, which does not require tuning parameters, is applied to quickly training and testing the extracted AFHF2 shape descriptor features. The effectiveness of the proposed approach is verified by a self-built clothing graphics library. The experimental results show that the proposed algorithm outperforms the existing algorithms, and the AFHF2 descriptor is superior to Fourier descriptor (FD), height function (HF) descriptor, and triangular centroid distances (TCDs) descriptor, and the SVM using the linear kernel function is superior to the SVM using radial basis function, K-nearest neighbor (KNN), probabilistic neural networks (PNN) and back propagation neural network (BPNN), and the average recognition rate of the proposed approach can reach 97.91%.
Keywords: accurate Fourier height functions;support vector machine;clothing style recognition
2019, 45(8):130-134  收稿日期: 2019-01-17;收到修改稿日期: 2019-03-29
基金项目: 国家自然科学基金项目(51775177,51675166)
作者简介: 孙国栋(1981-),男,湖北天门市人,教授,博士,主要从事图像处理和机器学习方面的研究
参考文献
[1] GUPTA M, BHATNAGAR C, JALAL A S. Clothing image retrieval based on multiple features for smarter shopping[J]. Procedia Computer Science, 2018, 125:143-148
[2] 燕平.服装款式设计[M].重庆:西南师范大学出版社, 2011:3-4.
[3] 李东,万贤福,汪军,等.基于轮廓曲率特征点的服装款式识别方法[J].东华大学学报(自然科学版), 2018, 44(1):87-92
[4] 李东,万贤福,汪军.采用傅里叶描述子和支持向量机的服装款式识别方法[J].纺织学报, 2017(5):122-127
[5] 纪娟,纪娟,秦珂,等.基于HOG和几何特征的服装细节要素识别与分类[J].图学学报, 2016, 37(1):84-90
[6] 于雨桐.基于多特征融合的服装图像款式识别的研究与应用[D].北京:北京工业大学, 2016.
[7] WU B, FENG S L, SUN G D, et al. Identification method of shaft orbit in rotating machines based on accurate fourier height functions descriptors[J]. Shock and Vibration, 2018(3737250):1-10
[8] 李霁蒲,赵荣珍.近邻概率距离在旋转机械故障集分类中的应用方法[J].振动与冲击, 2018, 37(11):48-54
[9] 刘朝阳,陈以,李少博.概率神经网络在手写汉字识别中的应用[J].电子设计工程, 2016, 24(2):32-34
[10] 张鹏林,徐旭,杨超,等.基于VMD能量熵和BP神经网络风电叶片缺陷研究[J].澳门新葡新京, 2018, 44(9):115-120
[11] WANG J W, BAI X, YOU X E, et al. Shape matching and classification using height functions[J]. Pattern Recognition Letters, 2012, 33(2):134-143
[12] ALAJLAN N, RUBE I E, KAMEL M S, et al. Shape retrieval using triangle-area representation and dynamic space warping[J]. Pattern Recognition, 2007, 40(7):1911-1920
[13] YANG C Z, WEI H, YU Q, et al. A novel method for 2d nonrigid partial shape matching[J]. Neurocomputing, 2018, 275(1):1160-1176
[14] CHANG C C, LIN C J. LIBSVM:a library for support vector machines[J]. ACM Transactions on Intelligent Systems and Technology, 2011, 2(3):1-27
[15] 刘祥楼,贾东旭,李辉,等.说话人识别中支持向量机核函数参数优化研究[J].科学技术与工程, 2010, 10(7):1669-1673

Xml地图