Abstract: The digital recognitions have extensive application in a lot of important fields. The artificial network was studied in the paper. The figure specimens were filtered through the process of unconstrained handwritten numeral, that it was recognized by the improved BP neural network. The result of experiment demonstrates that the method has very high noise immunity capacity and overcomes the limitation of traditional BP algorithm. The recognition rate and precision rate are greatly improved at the same time.

Key words: Neural network, Digital recognition, Improve BP network