Khmer printed characters recognition by using wavelet descriptors

Name: Chanoeurn Chey

Organization : King Mongkut's University of Technology Thonburi. Faculty of Engineering. Electrical and Information Engineering

keyword: Optical character recognition devices

; อุปกรณ์รู้จำอักขระด้วยวิธีการทางแสง

Abstract: In Khmer printed characters, same character has various shapes according to the fonts andsome characters are very similar in shape. Although these two problems are similar to otherlanguages such as Thai, there is need to explore the Khmer character recognition. Currently no research has been done on the topic. This thesis proposes a method to solve the problemof Khmer printed character recognition by using Wavelet Descriptors. In the recognition, firstly the Khmer printed character images are converted to skeleton forms; then, skeletonof Khmer characters are converted to temporal domain. The templates are obtained by wavelet coefficients from the character training set. To match the input characters withtemplates, the character recognition method using deformable wavelet descriptor is adapted by using fixed template and Euclidean distance classifier for matching instead. Thesmallest distance is the recognition result of the proposed method. As a result, the deformation can be skipped because it might get low recognition rate of similarcharacters. The experiment consists of two parts. The first part is to evaluate the overall recognition rate of input characters with three different sizes (22-point, 18-point and 12-point) from 10 different fonts of Khmer printed character. Twenty styles of characters are used as the training set. The results show 92.85 percent recognition rate for the 22-point, 9 1.66 percent recognition rate for the 18-point and 89.27 percent recognition rate for the 12-point.The second part is to specifically evaluate the system, testing with one document that has 21 pages of Khmer printed character with different resolutions from a scanner andfacsimile (fax). The document is initially printed with 300 dpi (dots per inch), then scanned with three different resolution, hoodpi, 300dpi and 1 5 0 g ~ othre documentthat received from Fax machine is scanned with 300dpi. The results show 92.99 percent recognition rate for 600 dpi resolution, 88.61 percent recognition rate for 300 dpiresolution, 80.05 percent recognition rate for 150 dpi resolution and 71.68 percent recognition rate for the input from a fax machine.

King Mongkut's University Technology Thonburi. Library

Address: กรุงเทพมหานคร

Email: info.lib@mail.kmutt.ac.th

Name: Pinit Kumhom

Role: Advisor

Name: Kosin Chamnongthai

Role: Advisor

Created: 2004

Issued: 2007-07-20

Modified: 2010-12-19

วิทยานิพนธ์/Thesis

ISBN: 9741836023

CallNumber: ENE462

eng

DegreeName: Master of Engineering

Level: Master's Degree

Descipline: Electrical and Information Engineering

Grantor: King Mongkut's University of Technology Thonburi

©copyrights King Mongkut's University of Technology Thonburi

ลำดับที่.	ชื่อแฟ้มข้อมูล	ขนาดแฟ้มข้อมูล	จำนวนเข้าถึง	วัน-เวลาเข้าถึงล่าสุด
1	ENE462.pdf	3.52 MB	20	2024-10-21 09:57:45
2	ENE462ab.pdf	14.58 KB	4	2018-05-10 18:10:17
3	ENE462.txt	2.54 KB	4	2018-05-10 18:10:17

ใช้เวลา

0.046245 วินาที