|
Abstract
|
Article Information:
An Efficient Character Segmentation Based on VNP Algorithm
S. Chitrakala, Srivardhini Mandipati, S. Preethi Raj and Gottumukkala Asisha
Corresponding Author: S. Chitrakala
Submitted: March 18, 2012
Accepted: April 14, 2012
Published: December 15, 2012 |
Abstract:
|
Character segmentation is an important preprocessing stage in image processing applications such
as OCR, License Plate Recognition, electronic processing of checks in banks, form processing and, label and
barcode recognition. It is essential to have an efficient character segmentation technique because it affects the
performance of all the processes that follow and hence, the overall system accuracy. Vertical projection profile
is the most common segmentation technique. However, the segmentation results are not always correct in cases
where pixels of adjacent characters fall on the same scan line and a minimum threshold is not observed in the
histogram to segment the respective adjacent characters. In this study, a character segmentation technique based
on Visited Neighbor Pixel (VNP) Algorithm is proposed, which is an improvement to the vertical projection
profile technique. VNP Algorithm performs segmentation based on the connectedness of the pixels on the scan
line with that of the previously visited pixels. Therefore, a clear line of separation is found even when the
threshold between two adjacent characters is not minimal. The segmentation results of the traditional vertical
projection profile and the proposed method are compared with respect to a few selected fonts and the latter,
with an average accuracy of approximately 94%, has shown encouraging results.
Key words: Dissection based segmentation, optical character recognition, segmentation, vertical projection, , ,
|
|
Abstract
|
PDF
|
HTML |
|
Cite this Reference:
S. Chitrakala, Srivardhini Mandipati, S. Preethi Raj and Gottumukkala Asisha, . An Efficient Character Segmentation Based on VNP Algorithm. Research Journal of Applied Sciences, Engineering and Technology, (24): 5438-5442.
|
|
|
|
 |
ISSN (Online): 2040-7467
ISSN (Print): 2040-7459 |
 |
| Information |
|
|
|
|
| Sales & Services |
|
|
|
|
