|Table of Contents|

One-dimensional Range Profile Recognition Based on FSVM


Research Field:
Publishing date:


One-dimensional Range Profile Recognition Based on FSVM
LUAN Ying-hongLI Yue-hua
School of Electronic Engineering and Optoelectronic Technology,NUST,Nanjing 210094,China
vectors radars radar target recognition support vector machine membership functions affinity
Based on the discussion of membership functions computed by distance and affinity among samples,a new membership function is introduced which can reflect the distribution of radar echo signal in feature space,in order to avoid the influence of noise and isolated points in millimeter-wave radar one-dimensional range profile recognition.Super-ellipse is introduced to improve membership function and applied to fuzzy support vector machine(FSVM) to distinguish support vectors with outliers or noise.Compact super-ellipse is used to surround samples and its direction radius can measure the affinity between samples.The experimental results show that the recognition rate of support vector machine drops down rapidly with the decrease of signal-noise-ratio for echo waves,but only one percent of FSVM reduces,which reduces the influence of noise on recognition results of one-dimensional range profiles effectively.


[ 1] Cortes C, Vapn il V. Support-vector netw ork[ J]. M ach ine Learn ing, 1995, 20( 3) : 273- 297.
[ 2] 刘江华, 程君实, 陈佳品. 支持向量机训练算法综述[ J]. 信息与控制, 2002, 2( 1): 45- 50.
[ 3] 吴丹, 顾学迈. 一种新的基于支持向量机的自动调制识别方案[ J]. 南京理工大学学报( 自然科学版), 2006, 30( 5): 569- 573.
[ 4] L in Chunfu, W ang Shengde. Fuzzy suppo rt vectorm ach ines[ J]. IEEE Trans on Neura l Ne tw orks, 2002, 13( 2): 464- 471.
[ 5] 阎满富, 杨志民. 模糊支持向量机与模糊模拟[ J]. 系统工程, 2004, 22( 11): 12- 14.
[ 6] H uang H anpang, L iu Y ihung. Fuzzy support vector m achines for pa ttern recogn ition and data m ining[ J]. Journal o f Fuzzy System s, 2002, 4( 3): 826- 835.
[ 7] Takuya I, Shigeo A. Fuzzy support vecto r m ach ines for pattern classification[ A] . Interna tiona l Jo in t Conference on Neura l Netwo rks Proceed ing s[ C ]. W ashing ton, USA: Neural Netwo rks Counc il of IEEE, 2001. 1449- 1455.
[ 8] 张翔, 肖小玲, 徐光祐. 基于样本之间紧密度的模糊支持向量机方法[ J]. 软件学报, 2006, 17 ( 5): 951- 958.
[ 9] 汤光华, 王俐莉, 严榴香, 等. 基于支持向量机的雷达一维距离像识别[ J]. 仪器仪表学报, 2006, 27 ( 6) : 799- 800.
[ 10] 马永军, 方凯, 方廷健. 基于支持向量机和距离度量的纹理分类[ J]. 中国图像图形学报, 2002, 7 ( A11): 151- 155.
[ 11] 李明, 荆竹翠. 两种实现模式分类的椭球法[ J]. 山西大学学报(自然科学版), 2005, 28( 4) : 355- 357.
[ 12] Zhu Qium ing, Ca iYao, Liu Luzheng. A m ultiple hyperellipsoidal subclass model for an evolutionary c lassifier [ J]. Pa ttern Recognition, 2001, 34( 3): 547- 560.
[ 13] G lineru F. Pattern separation v ia ellipso ids and conic programm ing[ D ]. M ons, H ainaut, Belg ium: M athem atics and Opera tions Research Departm ent, Facu lte Po lytechn igue deM ons, 1998.


Last Update: 2012-11-19