|Table of Contents|

ISAR image target recognition based on B(2D)2PGNMF

《南京理工大学学报》(自然科学版)[ISSN:1005-9830/CN:32-1397/N]

Issue:
2013年06期
Page:
863-868
Research Field:
Publishing date:

Info

Title:
ISAR image target recognition based on B(2D)2PGNMF
Author(s):
Wang FangSheng WeixingMa XiaofengWang Hao
School of Electronic Engineering and Optoelectronic Technology,NUST,Nanjing 210094,China
Keywords:
inverse synthetic aperture radar block two-directional and two-dimensional non-negative matrix factorization with projected gradient target recognition
PACS:
TN957
DOI:
-
Abstract:
In order to make use of local spatial structure information and classification information of inverse synthetic aperture radar(ISAR)to realize target recognition,a novel ISAR image target recognition algorithm based on block two-directional and two-dimensional non-negative matrix factorization with projected gradient(B(2D)2 PGNMF)is proposed here.Target image is constructed by the form of non-negative weighted combination of basis vectors.The weighted vectors resolved by the B(2D)2PGNMF are regarded as target features,and five-type aircraft targets are classified using a nearest neighbor classifier.The numerical results show:with the same compress ratios or dimensions of base matrix,the two-dimensional non-negative matrix factorization with projected gradient(PGNMF)has higher distinguishing accuracy than one-dimensional PGNMF; the distinguishing result of block non-negative matrix factorization with projected gradient(BPGNMF)is better than the PGNMF; the distinguishing result of the B(2D)2PGNMF is better than the two-directional and two-dimensional non-negative matrix factorization with projected gradient((2D)2PGNMF).With the same dimensions of base matrix,the compress ratio of the two-dimensional PGNMF is higher than that of the one-dimensional PGNMF,the running time of the BPGNMF is the longest and the running time of the(2D)2PGNMF is the shortest.This method has better recognition performance,and has no effect on operation efficiency.

References:

[1] 许人灿,刘朝军,黄小红,等.基于超分辨ISAR成像的空中目标自动识别[J].系统工程与电子技术,2006,28(1):46-48.
Xu Rencan,Liu Chaojun,Huang Xiaohong,et al.Automatic target recognition based on super-resolution ISAR images[J].Aerospace Electronics Information Engineering and Control,2006,16(1):46-48.
[2]Toumi A,Hoeltzener B,Khenchaf A.Using watersheds segmentation on ISAR image for automatic target recognition[A].2nd International Conference on Digital Information Management[C].Lyon,France:IEEE,2007:285-290.
[3]高兴斌,刘永坦.ISAR目标象的特征提取和特征选择[J].哈尔滨工业大学学报,1994,26(5):77-81.
Gao Xingbin,Liu Yongtan.Feature extraction and feature selection of ISAR object image[J].Journal of Harbin Institute of Technology,1994,26(5):77-81.
[4]李大芳.基于ISAR图像目标识别算法的研究[D].哈尔滨:哈尔滨工业大学电子信息工程学院,2004.
[5]Lin Bo,Yan Fengxia,Zhu Jubo.Feature extraction of 2D radar profile via double-sides 2DPCA for target recognition[A].2nd International Congress on Image and Signal Processing[C].Tianjin,China:IEEE,2009:1-5.
[6]Lee D D,Seung H S.Learning the parts of objects by non-negative matrix factorization[J].Nature,1999,401(6755):788-791.
[7]Lin Chinjen.Projected gradient methods for non-negative matrix factorization[J].Neural Computation,2007,19(10):2756-2779.
[8]潘彬彬,陈文胜,徐晨.基于分块非负矩阵分解人脸识别增量学习[J].计算机应用研究,2009,26(1):117-120.
Pan Binbin,Chen Wensheng,Xu Chen.Incremental learning of face recognition based on block non-negative matrix factorization[J].Application Research of Computers,2009,26(1):117-120.
[9]宿韬,张强,魏小鹏,等.基于NMF分组策略的人脸识别[J].计算机工程,2009,35(4):199-203.
Su Tao,Zhang Qiang,Wei Xiaopeng,et al.Face recognition based on NMF group strategy[J].Computer Engineering,2009,35(4):199-203.
[10]Zhang Daoqiang,Chen Songcan,Zhou Zhihua.Two-dimensional non-negative matrix factorization for face representation and recognition[A].Proceeding of the ICCV'05 Workshop on Analysis and Modeling of Faces and Gestures[C].Beijing,China:Springer,2005:350-363.
[11]高宏娟,潘晨.基于(2D)2NMF及其改进算法的人脸识别[J].计算机应用,2007,27(7):1660-1666.
Gao Hongjuan,Pan Chen.Face recognition based on(2D)2NMF and its improvement[J].Computer Applications,2007,27(7):1660-1666.
[12]狄文羽,何明一,梅少辉.基于快速非负矩阵分解和RBF网络的高光谱图像分类算法[J].遥感技术与应用,2009,24(3):385-389.
Di Wenyu,He Mingyi,Mei Shaohui.A hyper-spectral image classification algorithms based on quick non-negative matrix factorization and RBF neural network[J].Remote Sensing Technology and Application,2009,24(3):385-389.
[13]高涛,何明一.改进投影梯度非负矩阵分解的单训练样本特征提取研究[J].电子与信息学报,2010,32(5):1121-1125.
Gao Tao,He Mingyi.Using improved non-negative matrix factorization with projected gradient for single-trial feature extraction[J].Journal of Electronics and Information Technology,2010,32(5):1121-1125.
[14]薛二娟,鲍长春,李如玮.基于二维非负矩阵分解的1kb/sWI语音编码算法[J].电子学报,2010,38(7):1574-1579.
Xue Erjuan,Bao Changchun,Li Ruwei.1kb/s waveform interpolative speech coding based on two-dimensional nonnegative matrix factorization[J].Acta Electronica Sinica,2010,38(7):1574-1579.
[15]Kim K T,Seo D K,Kim H T.Efficient classification of ISAR images[J].IEEE Trans Antennas and Propagation,2005,53(5):1611-1621.

Memo

Memo:
-
Last Update: 2013-12-31