|Table of Contents|

Object Confirmation Method Based on Principle Component Analysis and Its Application in Vehicle-logo Location and Recognition


Research Field:
Publishing date:


Object Confirmation Method Based on Principle Component Analysis and Its Application in Vehicle-logo Location and Recognition
WANG Mei12WANG Guo-hong1
1.Department of Electronic and Information Engineering,Naval Aeronautical Engineering Institute,Yantai 264001,China;2.Laboratory of Image Processing,Yantai Vocational College,Yantai 264670,China
principle component analysis plausibility function object confirmation vehicle-logo location and recognition
An object confirmation method based on principle component analysis(PCA)is presented to improve the accuracy of small object location and recognition.The object samples are regarded as a set of random vectors.A set of eigen-objects is given by PCA.The object reconstruction is executed by reflecting the located actual object to the eigen-space.The plausibility function is defined between original located object and reconstruction object and the function is used to confirm the true object or to delete the false object in order to reduce the false-alarm rate.Actual vehicle-logo images are used to test the method and the experiment shows that the accuracy of the object location is improved by 16.5 percent than the result without object confirmation.The invariant moment minimum distance classifier is used to complete the vehicle-logo recognition and the experiment result shows that the accuracy of the object recognition is improved by 20 percent.


[ 1] 刘兆磊, 邵锡军, 张光义, 等. 雷达组网目标定位精度分析和评估[ J]. 南京理工大学学报( 自然科学版), 2005, 29( 2): 182- 185.

[ 2] Zheng Da-nian, Zhao Yan-nan, W ang Jia-x in. An e-ffic ient me thod o f license plate location [ J]. PatternRecogn ition Letters, 2005, 26( 15): 2431- 2438.
[ 3] Shigueo N, Ke ijiY, O sam u K. A novel adaptive mo rpholog ical approach fo r deg raded charac ter im ag e segmen tation[ J]. Patte rn Recognition, 2005, 38 ( 11 ):1961- 1975.
[ 4] Ba iH ong- liang, Liu Chang-p ing. A hybr id license p la teex traction me thod based on edge statistics and m orpho-logy [ A ]. IEEE ICPR-2004 [ C ]. Cambridge, UK:IEEE Press, 2004. 831- 834.
[ 5] Liu Zh-i fang, You Zh-i sheng, Zhang J-i p ing. Veh ic le realtmi e location based on v isual perception model[ J].Sem -i conductor Photonics and Techno logy, 2003, 9( 1):55- 61.
[ 6] Guldogan E, Gu ldogan O, K iranya z S, et a.l Compressione ffects on co lo r and texture based m ultim ed iaindex ing and retr ieva l[ A ]. IEEE IC IP-2003 [ C ].Barce lona, Spa in: IEEE Press, 2003. 9- 12.
[ 7] Tsai L W, H sieh JW, Fan K C. V ehic le detection usingno rma lized color and edgemap[ J]. IEEE T ransactionson Image Processing, 2007, 16( 16): 850- 864.
[ 8] 李贵俊, 刘正熙, 游志胜, 等. 基于能量增强和形态学滤波的车标定位方法[ J]. 光电子# 激光, 2005,16( 1): 76- 79.
[ 9] 庄永, 杨红雨, 游志胜, 等. 一种快速车标定位方法[ J] . 四川大学学报( 自然科学版), 2004, 41( 6):1167- 1171.
[ 10] Ra fael C G, R ichard EW, S teven L E. Dig ital im ag eprocessing using MATLAB [M ]. Be ijing: Pub lish ingH ouse o f E lectronics Industry, 2005. 379- 407.
[ 11] 徐涛, 王祁. PCA 在火箭发动机试车台传感器故障诊断中的应用[ J]. 南京理工大学学报( 自然科学版), 2006, 30( 6): 669- 672.
[ 12] 王枚, 王国宏. 椭圆配准方法及其在车标归一化中的应用[ J]. 激光与红外, 2007, 37( 6): 94- 97.[ 13] H uM K. V isua l patte rn recognition by mom ent invar-iants [ J ]. IRE Transactions on Inform ation Theory,1962, 8: 179- 187


Last Update: 2012-11-19