The research of neural network study system in vehicle identification

This dissertation also used BP network to recognize the characters of the vehicle plates. However, BP network has inherent defects and, by improving it with network learning method, this dissertation proved the possibility of an increased network learning efficiency, effectively solving the low-speed and local minimum problems of the neural network constringency. Furthermore, it introduced the variable differentiations of NN learning methods, greatly enhancing the performance of the whole system.