Handwritten Devanagari word recognition using robust invariant feature transforms

Prathima Guruprasad, Guruprasad

Research output: Chapter in Book/Report/Conference proceedingConference contribution

1 Citation (Scopus)

Abstract

Local features in Images can be described by the various algorithms in Computer Vision. Handwritten Devanagari word scripts are usually present in different illumination, sizes, orientation and occlusion. These are recognized by finding the points of Interest followed by the extraction of features around these interest points. In this paper we discuss extraction of Invariant Features from Handwritten Devanagari words of various forms using the Scale Invariant Feature Transform (SIFT) and Speeded Up Robust Features (SURF) techniques. On analysis of the patterns of word images and their recognition capability, the criteria for robust detection is derived. A dataset of over 1000 Devanagari words of various sizes and forms is created. The features extracted from the query image is subjected to Image matching by comparison to those features in the database using Random sample consensus. Comparison of Results indicate that SURF is computationally efficient while SIFT is more apt for detecting deformed images.

Original languageEnglish
Title of host publicationProceedings of the 2015 International Conference on Applied and Theoretical Computing and Communication Technology, iCATccT 2015
EditorsManjunath Aradhya, S.K. Niranjan
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages327-330
Number of pages4
ISBN (Electronic)9781467392228
DOIs
Publication statusPublished - 19-04-2016
Event1st International Conference on Applied and Theoretical Computing and Communication Technology, iCATccT 2015 - Davangere, Karnataka, India
Duration: 29-10-201531-10-2015

Publication series

NameProceedings of the 2015 International Conference on Applied and Theoretical Computing and Communication Technology, iCATccT 2015

Conference

Conference1st International Conference on Applied and Theoretical Computing and Communication Technology, iCATccT 2015
Country/TerritoryIndia
CityDavangere, Karnataka
Period29-10-1531-10-15

All Science Journal Classification (ASJC) codes

  • Computational Theory and Mathematics
  • Computer Science Applications
  • Computer Networks and Communications

Fingerprint

Dive into the research topics of 'Handwritten Devanagari word recognition using robust invariant feature transforms'. Together they form a unique fingerprint.

Cite this