Handwritten Numeral Recognition Technology Research And The Automatic Grading System

Handwritten numeral recognition belongs to the field of pattern recognition, which is a hot field for a large number of researchers and also is a critical step in entry of information. It is widely used in public security, taxation, transportation, finance, education and other industries in the practical activities. At present, identification of a variety of ways, but can not identify the technology fully to achieve a correct recognition rate. In order to apply handwritten numeral recognition to real practice, we not only studied the handwriting recognition algorithms but also designed a handwritten numeral recognition system applied to a student test system.After doing more learning and researches on handwritten numeral recognition technology, we chose the BP neural network as classifier training and recognition algorithms. In fact, BP neural network implements a mapping from input to output. In theory, it is capable of the realization of any complex non-linear mapping. Thus, it is suitable for solving complex problems.In this paper, we have designed a number identification system. In this system, first, a scanner scanned students'test papers and saved them as digital images; then, some preprocess would be done for the digital images such as binarization, noise, etc; and then, training the neural network classification model; finally, identifying the number by using this model. The performance of this system is acceptable.