Font Size: a A A

Recognition Of The Special Characters, Mathematical Formula Based On The Vector Line

Posted on:2012-10-24Degree:MasterType:Thesis
Country:ChinaCandidate:G GaoFull Text:PDF
GTID:2208330335497791Subject:Computer software and theory
Abstract/Summary:PDF Full Text Request
With the development of computer science and network technology, it is more and more important to translate the printed documents into electronic format to store and manage in computer by optical character recognition technology (Optical Characters Recognition, OCR). Traditional OCR systems have been widely used in handwriting, character recognition and document printing with a high recognition rate, but in some specific areas:such as a mix of characters, graphics, formula mixed literature, because literature is not recognized mathematical formulas and diagrams, its efficiency and accuracy are significantly reduced. How to accurate extraction, analysis and reconstruction of mathematical formulas in the literature, has become a hot spot areas identified.This design can be adapted to contain special characters, fonts of different sizes, two-dimensional distribution of the special features of character recognition algorithms. This paper is aimed to find a new solution for character recognition to break the limitations. The basic idea of this paper is extracting straight lines from characters and comparing them to find the best match of the tested character. This paper introduces six different CR algorithms as long as noise remove algorithm. In addition, this paper builds a prototype that contains a wealth of character modules and highly scalable database which is used for character recognition matching.What is more, this paper implements a comprehensive test structure, which meets the need of all six different character recognition algorithms. With the test result, we further optimize the database and algorithm design, and finally proved by experiments that the algorithms show robust adaptability for the structure of mathematical expressions, as well as great accuracy of recognition.
Keywords/Search Tags:Character recognition, Feature extraction, Approximate polyline, Noise removing, Matcher
PDF Full Text Request
Related items