2009-2015 ©
Journal Publication
Research Title การสกัดคุณลักษณะเด่นแบบผสมเพื่อรู้จำตัวเขียนอักษรธรรมอีสานด้วยโครงข่ายประสาทเทียม  
Date of Distribution 7 July 2012 
     Title of the Conference 2012 4th Conference Knowledge and Smart Technology (KST) 
     Organiser Burapha University 
     Conference Place 3rd floor of 50 years of Burapha University Celebration Building Burapha University 
     Province/State ชลบุรี 
     Conference Date 7 July 2012 
     To 8 July 2012 
Proceeding Paper
     Volume 2012 
     Page 45-52 
     Abstract This paper presented Isarn Dharma hand written alphabetical character recognition using back-propagation Neural Network. In the study, we proposed a hybrid of zone and histogram projection based for extracting the Isarn Dharma alphabet features. In the experiments, thirty-six Isarn alphabet datasets, each set contains 25 alphabets, were collected from palm leaf. In order to evaluate the efficiency of the proposed method, 900(36x25) difference characters were used for classification and recognition by using k-folds cross validation. The recognition rates are ที่ 62.22%, 52.24%, 63.87% and 71.22% for zone based, histogram projection, hybrid of zone base with horizontal histogram projection, and vertical histogram projection, respectively. The experimental results showed that the proposed method performed higher recognition accuracy compared to the conventional zone based and histogram projection. 
525020165-8 Mr. AMONTAP WIJITJAREON [Main Author]
Science Master's Degree

Peer Review Status มีผู้ประเมินอิสระ 
Level of Conference ชาติ 
Type of Proceeding Full paper 
Type of Presentation Oral 
Part of thesis true 
Presentation awarding false 
Attach file
Citation 0