Printed Text and Handwriting Identification in Noisy Document Images
Mohammed Jassim
Dr. Mohamed Jassim, Assistant Professor, College of Science and Information Technology, Iraq.
Manuscript received on 20 November 2015 | Revised Manuscript received on 30 November 2015 | Manuscript published on 30 November 2015 | PP: 10-12 | Volume-4 Issue-5, November 2015 | Retrieval Number: E1502114515©BEIESP
Open Access | Ethics and Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: In this project, we address the problem of segmenting and identifying text in noisy document images. The identification of machine printed text and handwriting is important because: (1) recognition techniques available for machine printed text and handwriting are significantly different (2) Handwriting in a document indicates corrections, additions that should be treated differently from the main content. Instead of using simple noise filtering techniques, our approach treat noise itself is treated as a class. Thus our approach becomes a three-class identification problem (Machine Printed Text, Handwriting and Noise). After performing text identification, post processing is performed to refine classification accuracy.
Keyword: Document Analysis, Printed Text, Segmentation, Markov Random Field, Hidden Markov Model.
Scope of the Article: Classification