Title :
Aircraft image recognition using back-propagation
Author :
Somaie, A.A. ; Badr, A. ; Salah, T.
Author_Institution :
R&D Center, Ain Shams Univ., Cairo, Egypt
Abstract :
The back-propagation neural network (BNN) is presented to recognize the aircraft images. Different training algorithms were used with different number of hidden neurons. The presented training algorithms satisfied the high recognition performance when the number of hidden neurons was less than or equal to ten times the number of classes. The effect of the activation function on the recognition performance and error convergence were studied and it was concluded that the log sigmoid function is better than the tanch sigmoid function. It was found that the network recognizes the test images correctly even the noisy and the incomplete images
Keywords :
Gaussian noise; aircraft; backpropagation; convergence of numerical methods; error analysis; image recognition; neural nets; Gaussian noise; activation function; aircraft image recognition; back-propagation neural network; error convergence; hidden neurons; incomplete images; log sigmoid function; noisy images; recognition performance; tanch sigmoid function; test images; training algorithms; Aircraft; Convergence; Electronic mail; Image processing; Image recognition; Neural networks; Neurons; Pattern recognition; Shape; Testing;
Conference_Titel :
Radar, 2001 CIE International Conference on, Proceedings
Conference_Location :
Beijing
Print_ISBN :
0-7803-7000-7
DOI :
10.1109/ICR.2001.984755
