基于BP网络的字符模式识别

Pattern Recognition Based on Bp Network

神经网络理论已经成为解决某些问题的重要手段的方法。但利用神经网络进行解决问题和设计的时候,必定会涉及到大量的有关数值计算等问题,所以利用计算机对神经网络模型进行仿真和辅助设计时,仍是件很麻烦的事情。所以MATLAB的便利受到了青睐,BP网络在人工神经网络中应用最为广泛,而且在理论上十分完善,网络结构也比较直观。在BP网络中,模式识别是应用比较广泛的一个方面。该系统使用MATLAB中神经网络工具箱对英文表中的26个字母进行识别。通过建立网络,训练网络,测试网络,最后进行仿真,完成了正确识别26个英文字母的功能。该系统的操作界面简洁、友好、使用简单方便。

Neural network has already become an important method for resolving some problems, but when making use of the neural network to solve problems, large quantity of calculation must be involved, so it is still a piece of troublesome affair. Therefore MATLAB gets a very good reputation in this area. BP network is one of the widest application areas in the artificial neural network, its theory is very perfect, and its structure is very simple. Pattern recognition is very popular in BP network application. 26 letters of alphabets are identified by using MATLAB and its neural network tool box in this system. The steps of the program are creating the network, training the network, testing the network and the last is to carry out the emulation. In the end the system can identify the 26 letter of alphabets successfully. A brief and simple interface, easy to use are the features of the system.