Title :
MACE prefiltering for neural network based automatic target recognition
Author :
Gader, Paul ; Keller, James M. ; Jones, Thomas ; Miramonti, Joseph ; Hobson, Greg
Author_Institution :
Dept. of Electr. & Comput. Eng., Missouri Univ., Columbia, MO, USA
fDate :
27 Jun- 2 Jul 1994
Abstract :
A method for achieving improved target recognition rates with low false alarm rates is presented. The method combines frequency domain advantages of the minimum average correlation energy (MACE) filter with the superior classification abilities of multi-layer, feedforward networks. Results are provided on forward looking infrared (FLIR) images of real targets. A new evaluation metric is proposed that measures the quality of the results
Keywords :
correlation methods; feedforward neural nets; filtering theory; image classification; multilayer perceptrons; object recognition; MACE prefiltering; classification abilities; evaluation metric; false alarm rates; forward looking infrared images; frequency domain advantages; minimum average correlation energy filter; multilayer feedforward networks; neural network based automatic target recognition; target recognition rates; Feedforward neural networks; Frequency domain analysis; Infrared imaging; Matched filters; Mean square error methods; Multi-layer neural network; Neural networks; Nonlinear filters; Power engineering and energy; Target recognition;
Conference_Titel :
Neural Networks, 1994. IEEE World Congress on Computational Intelligence., 1994 IEEE International Conference on
Conference_Location :
Orlando, FL
Print_ISBN :
0-7803-1901-X
DOI :
10.1109/ICNN.1994.374854
