Digit Recognition System Base On ARM Platform

With the rapid development of science and technology, The application of intelligent robot is more and more widely. Digit recognition has became an important area of machine vision because digit playing increasingly important role in our life. Now mostly digit recognition system has been based on PC- Windos system, because of the mobile robot in power, cost and hardware resource constraints, digit recognition system based on PC-Windos isn’t applied in intelligent robots. It is required to develop a digit identification system based on embedded system.Due to the limited resources of embedded system, the image acquisition, image processing, digit recognition are more stringent to select. In this paper, a digit recognition algorithm based on ARM- Linux system which is suit for mobile robot. In this algorithm, artificial bee colony algorithm has been improved to improve the traditional BP neural network convergence speed and precision, in order to improve the training speed and recognition rate. The algorithm was tested in MATLAB, the results show that the improved algorithm has to improve on the convergence speed and accuracy. Finally, Open CV and QT has been installed and ported in ARM11 to complete digit recognition system, including image acquisition, image preprocessing, feature extraction, recognition, and human-machine interface. Test results show that the digital identification system is well in real-time and recognition rate. It has a certain reference value for the development of digit recognition based on embedded system. At the same time, this digit recognition based on embedded system could be applied in portable or miniature digit recognition system.