Literal Feature Extraction and Recognition In An English Document

碩士 === 國立臺灣海洋大學 === 電機工程學系 === 94 === The purpose of this thesis is to develop an optical character recognition system according to the theory of image processing for recognizing the English character in a document. For a start, we perform the operation of image preprocessing, so as to emphasize the...

Full description

Bibliographic Details
Main Authors: Ying-Liang Hsieh, 謝瑩亮
Other Authors: Ying-Te Chang
Format: Others
Language:en_US
Published: 2005
Online Access:http://ndltd.ncl.edu.tw/handle/35374221834014641694
Description
Summary:碩士 === 國立臺灣海洋大學 === 電機工程學系 === 94 === The purpose of this thesis is to develop an optical character recognition system according to the theory of image processing for recognizing the English character in a document. For a start, we perform the operation of image preprocessing, so as to emphasize the key feature of an original image, in which due to the shape of a test pattern may differ from the same of a sample pattern, thus normalization and closing are both essential operations in said image preprocessing stage. In the second stage, we divide the image into a number of regions using projection approach, wherein each said region is thought of as a character. Next, to extract the feature from each character and then to employ recursive least square algorithm to perform feature matching and output the result thereafter. As for the accuracy of said recognition system, that is 99% for standard character as well as for defected character.