| Author | Naing, Oke |
| Call Number | AIT Thesis no. CS-91-40 |
| Subject(s) | Pattern recognition systems
|
| Note | A thesis submitted in partial fulfillment of the requirements for the degree of Master of Engineering, School of Engineering and Technology |
| Publisher | Asian Institute of Technology |
| Abstract | This study is concerned with a neural network approach for detection of the orientation of Myanmar characters. The character orientation detection has many applications particularly in industrial automation. The model reported here uses a three layer perceptron with back- propagation training algorithm and was simulated on a Sun Workstation. Training is done based on multiple inputs, multiple outputs parallel distributed processing. The input to the system is 8x8 bit matrix representing Myanmar characters. Standard transformation and thinning algorithm are used to form these matrices. The network performs classification of characters and detects orientation. The success rate is high and is tolerant for considerable deformation and distortion of the input characters. The network seems to have, however, some difficulty with the input characters "Ka", "Kha" and "Hnit". This may be explained because of the resemblance of these characters with others. The system can be extended to other similar characters in other languages. |
| Year | 1991 |
| Type | Thesis |
| School | School of Engineering and Technology (SET) |
| Department | Other Field of Studies (No Department) |
| Academic Program/FoS | Computer Science (CS) |
| Chairperson(s) | Sadananda, Ramakoti |
| Examination Committee(s) | Zhao, Ming ;Hawkey, R. |
| Scholarship Donor(s) | German Academic Exchange Service (DAAD) |
| Degree | Thesis (M.Eng.) - Asian Institute of Technology, 1991 |