Auto Recognition And Prototype System Design In Financial Bill

The auto recognition of the financial bill is one of the most important application fields in Pattern Recognition. It is a synthetic research work in both theory and application, which includes almost all branches of pattern recognition and image processing technical fields. Based on plenty of papers, technology reports and dissertations, the methods of document image understanding are discussed in this dissertation. It is focus on document image incline rectify, document analysis and document recognition. It realizes a prototype system, and by testing, this system is satisfying. The main work is as follows:At first, this paper's research is based on the preprocessing of document image. The document understanding and OCR is sensitive to the skewed document image. A new skew detection method based on directional projection was presented, using Bresenham line drawing algorithm. Through advancing the Hough Transaction algorithm to realize incline rectify. The experiments show that this method has high precession and adaptability.Secondly, this paper proposed a new table structure expression method and the table recognition algorithm in the table layout structure analysis part. The algorithm consists of two major parts. Firstly, a line detection algorithm applied in financial document images is presented according to the specific characters of financial documents' structure. Secondly, a financial document pattern recognition algorithm based on the fact that different financial documents have different geometrical structures is proposed.At last, this paper presented a new method for the logical structure extraction of form document in the table logical structure analysis part. The algorithm of it consists of three phases: global division of the whole document, local logical structure analysis and global-division of the whole document. This method emphasizes the synthetic analysis of the global and local layout structure of a form document. Compared with other local-layout-structure-analysis methods, it has higher accuracy and can treat with more complex form documents.