| Author | Hathai Tanta-ngai |
| Call Number | AIT Thesis no. CS-99-7 |
| Subject(s) | Optical character recognition devices
|
| 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 | In this thesis, we propose the use of Binary Decision Diagram (BDD) to represent 2-D blackand-
white images for optical character recognition (OCR). Traditionally in bit-mapped
representation, black-and-white images are represented as an array of black or white pixels,
such that and nxm image requires nxm bits of storage and for any binary image operation
require nxm operations. With BDD, an image can be viewed a group of rectangular blocks of
black pixels, thus a BDD with only the 'one' terminal is sufficient to represent a character
image. Redundant white pixels can be left out from our OCR problem. An optical character is
recognized by superimposing an image BDD cin a pattern BDD. Any discrepancy found is
taken as a dissimilarity measure of the two in1ages. We show that with BDD representation,
performance of our character recognition varies with the noise level while a point-by-point
global comparison is insensitive to noise. The performance of the algorithm suffers when noise
appears largely to be a set of isolated black pixels and in1proves again when the noisy pixels
appears to be grouped together forming larger rectangular black blocks. |
| Year | 1999 |
| Type | Thesis |
| School | School of Engineering and Technology (SET) |
| Department | Department of Information and Communications Technologies (DICT) |
| Academic Program/FoS | Computer Science (CS) |
| Chairperson(s) | Kanchana Kanchanasut |
| Examination Committee(s) | Sadananda, R.;Chidchanok Luahsinsap |
| Scholarship Donor(s) | Royal Thai Government (RTG) |
| Degree | Thesis (M.Eng.) - Asian Institute of Technology, 1999 |