Multiobject detection of targets with fine details, scale and translation variations

被引:17
作者
Diab, SL [1 ]
Karim, MA
Iftekharuddin, KM
机构
[1] Univ Dayton, Dept Elect & Comp Engn, Dayton, OH 45469 USA
[2] Timken Co, Timken Res, Canton, OH 44706 USA
关键词
automatic target recognition; feature extraction; logarithmic polar transform; clustering; learning vector quantization neural network;
D O I
10.1117/1.601921
中图分类号
O43 [光学];
学科分类号
070207 ; 0803 ;
摘要
We propose an enabling algorithm for automatic target detection wherein the targets are almost similar but differ in fine details. Multiple translated and scaled target images are processed in the Mellin transform domain and subsequently detected using a learning vector quantization (LVQ) neural network. (C) 1998 Society of Photo-Optical Instrumentation Engineers. [S0091-3286(98)02203-X].
引用
收藏
页码:876 / 883
页数:8
相关论文
共 8 条
[1]   REAL-TIME DEFORMATION INVARIANT OPTICAL-PATTERN RECOGNITION USING COORDINATE TRANSFORMATIONS [J].
CASASENT, D ;
XIA, SF ;
LEE, AJ ;
SONG, JZ .
APPLIED OPTICS, 1987, 26 (05) :938-942
[2]   Feature evaluation and selection based on an entropy measure with data clustering [J].
Chi, ZR ;
Yan, H .
OPTICAL ENGINEERING, 1995, 34 (12) :3514-3519
[3]  
Goodman W., 2005, INTRO FOURIER OPTICS, V3rd
[4]   Wavelet preprocessed amplitude-modulated phase-only filter for recognition applications [J].
Iftekharuddin, KM ;
Tuthill, TA ;
Karim, MA ;
Jemili, K .
OPTICAL ENGINEERING, 1996, 35 (07) :2088-2093
[5]   Rotation-invariant target recognition using an amplitude-coupled minimum-average correlation-energy filter [J].
Iftekharuddin, KM ;
Ahmed, F ;
Karim, MA .
OPTICAL ENGINEERING, 1996, 35 (04) :1009-1014
[6]  
IFTEKHARUDDIN KM, 1995, OPT ENG, V35, P3143
[7]   TRANSFORMATION OF IMAGE POSITIONS, ROTATIONS, AND SIZES INTO SHIFT PARAMETERS [J].
JENSEN, AS ;
LINDVOLD, L ;
RASMUSSEN, E .
APPLIED OPTICS, 1987, 26 (09) :1775-1781
[8]  
Zurada J., 1992, INTRO ARTIFICIAL NEU