Research On Chinese And English Mixed Bussiness Card Recognition System

Business card as a convenient carrier of personal information is getting more and more popular,but the management of business card information becomes a troublesome problem. Manualmanagement is trouble and error-prone, so research of automatic recognition and storage for businesscard has practical significance. This paper intends to do some research and exploration on Chineseand English mixed business card recognition system. It mainly studies the business card layoutanalysis algorithm, character segmentation and information classification problem about charactersmixed with Chinese and English. At the same time, this paper tries to improve and optimize theexisting algorithm to make the business card recognition system more perfect and efficient.The main work of this paper is as follows.1. This paper analyzes the existing image preprocessing method (especially on two value imageproblems and tilt correction), compares and summarizes the advantages and disadvantages of variousmethods and puts forward their solutions for business card recognition applications.2. In this paper, compared with the existing layout analysis algorithm, a connected element andrecursive projection profile integrated layout analysis algorithm is presented. By this method, we cando complex layout analysis and solve the judgement problem of division of section’s property better.3. The existing method for character segmentation in Chinese and English mixed environmentsand different font text mixed cases is difficult to carry on the accurate segmentation. On the basis ofthe present algorithm, we put forward the improved methods based on global features extraction ofmixed Chinese/English character. Different language region will be determined and separated firstly,and then different segmentation algorithm is used for different language areas. Through this, we cansolve the character adhesion and incorrect segmentation problems and improve the segmentationaccuracy rate.4. This paper presents the working process of traditional business card recognition system, andanalyzes the existing shortages, then proposes a new information classification approach withfeedback mechanism for business cards. Layout analysis is considered as an important factor based ontraditional semantic concept, and the result of classification affects layout analysis in turn. By doingthis, the system is able to correct errors automatically, while recognition rate is increased.The experiment proves, this method achieves good results with the Chinese/English mixedbusiness card recognition problem.