Please use this identifier to cite or link to this item:
https://scholarbank.nus.edu.sg/handle/10635/42111
DC Field | Value | |
---|---|---|
dc.title | A new run length based method for scene text detection | |
dc.contributor.author | Basavanna, M. | |
dc.contributor.author | Shivakumara, P. | |
dc.contributor.author | Srivatsa, S.K. | |
dc.contributor.author | Hemantha Kumar, G. | |
dc.date.accessioned | 2013-07-04T08:43:37Z | |
dc.date.available | 2013-07-04T08:43:37Z | |
dc.date.issued | 2011 | |
dc.identifier.citation | Basavanna, M.,Shivakumara, P.,Srivatsa, S.K.,Hemantha Kumar, G. (2011). A new run length based method for scene text detection. Proceedings of the 5th Indian International Conference on Artificial Intelligence, IICAI 2011 : 1730-1736. ScholarBank@NUS Repository. | |
dc.identifier.isbn | 9780972741286 | |
dc.identifier.uri | http://scholarbank.nus.edu.sg/handle/10635/42111 | |
dc.description.abstract | Scene text detection and extraction from images is not as easy as detection of text in plain background document images because scene text images usually have complex background, different fonts, font size, colors, orientation etc. Hence it is an emerging and challenging research area in document analysis. Besides, it is essential for bridging gap between high and low level features to retrieve images accurately and efficiently. To achieve this, we present a new method based on run length algorithm for scene text detection in complex background images. The run lengths are calculated for the Sobel edge map of input image. The run length helps in detecting text because of the fact that when text appears in the image, characters and words have regular spacing between characters and words. This feature is explored to detect text in scene images by introducing run length algorithm in this work. The benchmark database ICDAR 2003 competition is used for evaluating the performance of the proposed method. The result of the proposed method is compared with the result of existing methods to show its superiority over existing methods in terms of f-measure. The robustness of the proposal is also tested by conducting experiments on our own data, captured by high resolution (camera) and low resolution (mobile). | |
dc.source | Scopus | |
dc.subject | Complex scene text image | |
dc.subject | Max-Min clustering | |
dc.subject | Run length | |
dc.subject | Scene text detection | |
dc.subject | Scene text recognition | |
dc.subject | Sobel edge map | |
dc.type | Conference Paper | |
dc.contributor.department | COMPUTER SCIENCE | |
dc.description.sourcetitle | Proceedings of the 5th Indian International Conference on Artificial Intelligence, IICAI 2011 | |
dc.description.page | 1730-1736 | |
dc.identifier.isiut | NOT_IN_WOS | |
Appears in Collections: | Staff Publications |
Show simple item record
Files in This Item:
There are no files associated with this item.
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.