面向圖像識(shí)別的測地局部典型相關(guān)分析方法

許歡; 蘇樹智; 顏文婧; 鄧瀛灝; 謝軍

doi:10.11999/JEIT200123

面向圖像識(shí)別的測地局部典型相關(guān)分析方法

doi: 10.11999/JEIT200123 cstr: 32379.14.JEIT200123

1.
安徽理工大學(xué)計(jì)算機(jī)科學(xué)與工程學(xué)院淮南 232001
2.
北京工商大學(xué)計(jì)算機(jī)與信息工程學(xué)院北京 100037

基金項(xiàng)目: 國家自然科學(xué)基金(61806006)，安徽省高等學(xué)校自然科學(xué)研究基金(KJ2018A0083)，中國博士后科學(xué)基金(2019M660149)

詳細(xì)信息

作者簡介:
許歡：女，1982年生，助教，研究方向?yàn)闄C(jī)器學(xué)習(xí)、圖像處理、模式識(shí)別

蘇樹智：男，1987年生，副教授，研究方向?yàn)槎嗄B(tài)模式識(shí)別、特征學(xué)習(xí)、子空間融合、圖像處理

顏文婧：女，1984年生，講師，研究方向?yàn)闄C(jī)器學(xué)習(xí)、模式識(shí)別、信號(hào)處理

通訊作者:
蘇樹智　sushuzhi@foxmail.com

中圖分類號(hào): TN911.73; TP391.4
計(jì)量
- 文章訪問數(shù): 1533
- HTML全文瀏覽量: 499
- PDF下載量: 63
- 被引次數(shù): 0
出版歷程
- 收稿日期: 2020-02-21
- 修回日期: 2020-07-23
- 網(wǎng)絡(luò)出版日期: 2020-07-23
- 刊出日期: 2020-11-16

A Geodesic Locality Canonical Correlation Analysis Method for Image Recognition

1.
College of Computer Science and Engineering, Anhui University of Science & Technology, Huainan 232001, China
2.
School of Computer and Information Engineering, Beijing Technology and Business University, Beijing 100037, China

Funds: The National Natural Science Foundation of China (61806006), The Anhui Province Natural Science Research Foundation of Institutions of Higher Learning (KJ2018A0083), The China Postdoctoral Science Foundation (2019M660149)

摘要

摘要: 典型相關(guān)分析(CCA)是一種經(jīng)典的多模態(tài)特征學(xué)習(xí)方法，能夠從不同模態(tài)同時(shí)學(xué)習(xí)相關(guān)性最大的低維特征，然而難以發(fā)現(xiàn)隱藏在樣本空間中的非線性流形結(jié)構(gòu)。該文提出一種基于測地流形的多模態(tài)特征學(xué)習(xí)方法，即測地局部典型相關(guān)分析(GeoLCCA)。該方法利用測地距離構(gòu)建了低維相關(guān)特征的測地散布，并進(jìn)一步通過最大化模態(tài)間的相關(guān)性和最小化模態(tài)內(nèi)的測地散布學(xué)習(xí)更具鑒別力的非線性相關(guān)特征。該文不僅在理論上對提出的方法進(jìn)行了分析，而且在真實(shí)的圖像數(shù)據(jù)集上驗(yàn)證了方法的有效性。
- 圖像識(shí)別 /
- 典型相關(guān)分析 /
- 多模態(tài)特征學(xué)習(xí) /
- 流形學(xué)習(xí)
Abstract: Canonical Correlation Analysis (CCA) is a classic multi-modal feature learning method, which can learn low-dimensional features with the maximum correlation from different modalities. However, it is difficult for CCA to find the nonlinear manifold structures hidden in the sample spaces. This paper proposes a multi-modal feature learning method based on geodesic manifolds, namely Geodesic Locality Canonical Correlation Analysis (GeoLCCA).The geodesic distances are used to construct the geodesic scatters of low-dimensional correlation features, and the nonlinear correlation features with better discriminative power are learned by maximizing the between-modal correlation and minimizing the within-modal geodesic scatters. This paper not only analyzes the proposed method in theory, but also verifies the effective of the proposed method on the real-world image datasets.
- Image recognition /
- Canonical Correlation Analysis (CCA) /
- Multi-modal feature learning /
- Manifold learning

HTML全文

表 1 在GT圖像數(shù)據(jù)集上的識(shí)別率(%)及標(biāo)準(zhǔn)差

	訓(xùn)練樣本數(shù)5	訓(xùn)練樣本數(shù)6	訓(xùn)練樣本數(shù)7	訓(xùn)練樣本數(shù)8
GeoLCCA	67.26±2.01	71.36±1.83	76.10±1.28	78.20±1.31
GMCCA	65.22±1.64	66.64±1.56	69.70±1.75	72.06±1.66
LPCCA	44.84±1.73	50.09±3.79	54.15±1.74	57.46±2.56
DMCCA	63.56±2.77	67.80±1.29	73.67±1.71	75.80±1.99
CCA	59.08±1.81	61.78±1.35	66.22±1.66	68.14±2.01
A±B: A表示平均識(shí)別率(%)，B表示對應(yīng)的識(shí)別率標(biāo)準(zhǔn)差

下載: 導(dǎo)出CSV

表 2 在ORL圖像數(shù)據(jù)集上的識(shí)別率(%)及標(biāo)準(zhǔn)差

	訓(xùn)練樣本數(shù)5	訓(xùn)練樣本數(shù)6	訓(xùn)練樣本數(shù)7	訓(xùn)練樣本數(shù)8
GeoLCCA	95.15±1.58	97.19±1.33	98.25±0.83	99.50±0.65
GMCCA	93.90±2.04	95.19±0.89	97.00±1.53	98.50±1.42
LPCCA	84.70±3.00	87.81±2.40	89.17±2.00	94.25±2.58
DMCCA	93.80±1.53	95.50±1.74	96.75±1.49	99.38±0.66
CCA	90.35±1.94	93.19±1.94	93.83±1.68	97.25±1.15
A±B: A表示平均識(shí)別率(%)，B表示對應(yīng)的識(shí)別率標(biāo)準(zhǔn)差

下載: 導(dǎo)出CSV

參考文獻(xiàn)(25)

劉政怡, 段群濤, 石松, 等. 基于多模態(tài)特征融合監(jiān)督的RGB-D圖像顯著性檢測[J]. 電子與信息學(xué)報(bào), 2020, 42(4): 997–1004. doi: 10.11999/JEIT190297

LIU Zhengyi, DUAN Quntao, SHI Song, et al. RGB-D image saliency detection based on multi-modal feature-fused supervision[J]. Journal of Electronics &Information Technology, 2020, 42(4): 997–1004. doi: 10.11999/JEIT190297

YE Qiaolin, FU Liyong, ZHANG Zhao, et al. Lp- and Ls-norm distance based robust linear discriminant analysis[J]. Neural Networks, 2018, 105: 393–404. doi: 10.1016/j.neunet.2018.05.020

王肖鋒, 孫明月, 葛為民. 基于圖像協(xié)方差無關(guān)的增量特征提取方法研究[J]. 電子與信息學(xué)報(bào), 2019, 41(11): 2768–2776. doi: 10.11999/JEIT181138

WANG Xiaofeng, SUN Mingyue, and GE Weimin. An incremental feature extraction method without estimating image covariance matrix[J]. Journal of Electronics &Information Technology, 2019, 41(11): 2768–2776. doi: 10.11999/JEIT181138

YUAN Sen and MAO Xia. Exponential elastic preserving projections for facial expression recognition[J]. Neurocomputing, 2018, 275: 711–724. doi: 10.1016/j.neucom.2017.08.067

WANG Rong, NIE Feiping, HONG Richang, et al. Fast and orthogonal locality preserving projections for dimensionality reduction[J]. IEEE Transactions on Image Processing, 2017, 26(10): 5019–5030. doi: 10.1109/TIP.2017.2726188

ZHU Yani, ZHU Chaoyang, and LI Xiaoxin. Improved principal component analysis and linear regression classification for face recognition[J]. Signal Processing, 2018, 145: 175–182. doi: 10.1016/j.sigpro.2017.11.018

KUMAR S, BHUYAN M K, LOVELL B C, et al. Hierarchical uncorrelated multiview discriminant locality preserving projection for multiview facial expression recognition[J]. Journal of Visual Communication and Image Representation, 2018, 54: 171–181. doi: 10.1016/j.jvcir.2018.04.013

GAJJAR S, KULAHCI M, and PALAZOGLU A. Real-time fault detection and diagnosis using sparse principal component analysis[J]. Journal of Process Control, 2018, 67: 112–128. doi: 10.1016/j.jprocont.2017.03.005

WANG Hao, FAN Yuanyuan, FANG Baofu, et al. Generalized linear discriminant analysis based on Euclidean norm for gait recognition[J]. International Journal of Machine Learning and Cybernetics, 2018, 9(4): 569–576. doi: 10.1007/s13042-016-0540-0

董書琴, 張斌. 基于深度特征學(xué)習(xí)的網(wǎng)絡(luò)流量異常檢測方法[J]. 電子與信息學(xué)報(bào), 2020, 42(3): 695–703. doi: 10.11999/JEIT190266

DONG Shuqin and ZHANG Bin. Network traffic anomaly detection method based on deep features learning[J]. Journal of Electronics &Information Technology, 2020, 42(3): 695–703. doi: 10.11999/JEIT190266

SUN Quansen, ZENG Shenggen, LIU Yan, et al. A new method of feature fusion and its application in image recognition[J]. Pattern Recognition, 2005, 38(12): 2437–2448. doi: 10.1016/j.patcog.2004.12.013

CHEN Jia, WANG Gang, and GIANNAKIS G B. Graph multiview canonical correlation analysis[J]. IEEE Transactions on Signal Processing, 2019, 67(11): 2826–2838. doi: 10.1109/TSP.2019.2910475

LIU Yiqi, LIU Bin, ZHAO Xiujie, et al. A mixture of variational canonical correlation analysis for nonlinear and quality-relevant process monitoring[J]. IEEE Transactions on Industrial Electronics, 2018, 65(8): 6478–6486. doi: 10.1109/TIE.2017.2786253

HONG Kan, LIU Guodong, CHEN Wentao, et al. Classification of the emotional stress and physical stress using signal magnification and canonical correlation analysis[J]. Pattern Recognition, 2018, 77: 140–149. doi: 10.1016/j.patcog.2017.12.013

SAFO S E, AHN J, JEON Y, et al. Sparse generalized eigenvalue problem with application to canonical correlation analysis for integrative analysis of methylation and gene expression data[J]. Biometrics, 2018, 74(4): 1362–1371. doi: 10.1111/biom.12886

GAO Lei, QI Lin, CHEN Enqing, et al. Discriminative multiple canonical correlation analysis for information fusion[J]. IEEE Transactions on Image Processing, 2018, 27(4): 1951–1965. doi: 10.1109/TIP.2017.2765820

GENG Fazhan and QIAN Suping. An optimal reproducing kernel method for linear nonlocal boundary value problems[J]. Applied Mathematics Letters, 2018, 77: 49–56. doi: 10.1016/j.aml.2017.10.002

MELZER T, REITER M, and BISCHOF H. Appearance models based on kernel canonical correlation analysis[J]. Pattern Recognition, 2003, 36(9): 1961–1971. doi: 10.1016/s0031-3203(03)00058-x

ALAM M A, FUKUMIZU K, and WANG Yuping. Influence function and robust variant of kernel canonical correlation analysis[J]. Neurocomputing, 2018, 304: 12–29. doi: 10.1016/j.neucom.2018.04.008

SUN Tingkai and CHEN Songcan. Locality preserving CCA with applications to data visualization and pose estimation[J]. Image and Vision Computing, 2007, 25(5): 531–543. doi: 10.1016/j.imavis.2006.04.014

CHEN Jia, WANG Gang, SHEN Yanning, et al. Canonical correlation analysis of datasets with a common source graph[J]. IEEE Transactions on Signal Processing, 2018, 66(16): 4398–4408. doi: 10.1109/TSP.2018.2853130

BALASUBRAMANIAN M, SCHWARTZ E L, TENENBAUM J B, et al. The Isomap algorithm and topological stability[J]. Science, 2002, 295(5552): 7. doi: 10.1126/science.295.5552.7a

ZHANG Guiying, ZOU Wenbin, ZHANG Xianjie, et al. Singular value decomposition based virtual representation for face recognition[J]. Multimedia Tools and Applications, 2018, 77(6): 7171–7186. doi: 10.1007/s11042-017-4627-8

SU Shuzhi, GE Hongwei, YUAN Yunhao, et al. A label embedding kernel method for multi-view canonical correlation analysis[J]. Multimedia Tools and Applications, 2017, 76(12): 13785–13803. doi: 10.1007/s11042-016-3786-3

SU Shuzhi, FANG Xianjin, YANG Gaoming, et al. Self-balanced multi-view orthogonality correlation analysis for image feature learning[J]. Infrared Physics & Technology, 2019, 100: 44–51. doi: 10.1016/j.infrared.2019.05.008

相關(guān)文章

施引文獻(xiàn)

資源附件(0)

訪問統(tǒng)計(jì)