Document Analysis Of The Financial Bill And Recognition Of Handwritten Digit

The auto recognition of the financial bill is one of the most important application fields. It is a synthetic research work in both theory and application, which includes almost all branches of pattern recognition and image processing technical fields. This paper is based on the practical work. In this paper, I summarize the detailed work in the bill document analysis and recognition, study some advanced methods and ideas in these fields, bring up an complete bill document analysis system model and give some suggestions to improve the present system. After giving a complete bill document analysis system model, I discuss every module in the bill document analysis procedures, and furthermore give solving methods in detail. The first step is to measure document features. It mainly includes two parts, preprocessing and producing a table of information about bill documents. And then, the second step is type recognition of the document. And I provide two vavs in this paper. The third step is dynamic area抯 document analysis, taking off the frame without loss and cutting apart the string. I only discuss numbers in dynamic area. In fact, this procedure is the preceding one of the handwritten digit recognition. 7he handwritten digit recognition is one of my main topics. First, I probe into the handwritten numeral recognition method called method based on topological structure in the present system, and supply the detailed measures and the partial experiment results. Moreover, I study the method based segment outlines for the hand~ritten digit recognition as well, which is compared with the method based topological structure, from the feature extracting, the storage and expression of the feature, the style of sorting and the disposing of the refused character. From the present method in our system, for offsetting disadvantages, I propose an improvement model on it. Nowadays. there are a lot of research achievements in the handwritten digit recognition. For example, multiple classifier and neural network has attracted the concern of the mass. At last, I offer a model to improve the present method. All in all. This paper gives a complete bill document analysis system model, discussing every module in the bill document analysis procedures; probes into the handwritten digit recognition method called method based on topological structure and supplies the detailed measures and the partial experiment results; offers a model to improve the present method.