Characters Segmentation of Cursive Handwritten Words based on Contour Analysis and Neural Network Validation

Fajri Kurniawan (Department of Computer Graphics and Multimedia Universiti Teknologi Malaysia)
Mohd. Shafry Mohd. Rahim (Department of Computer Graphics and Multimedia Universiti Teknologi Malaysia)
Ni'matus Sholihah (Department of Computer Graphics and Multimedia Universiti Teknologi Malaysia)
Akmal Rakhmadi (Department of Computer Graphics and Multimedia Universiti Teknologi Malaysia)
Dzulkifli Mohamad (Department of Computer Graphics and Multimedia Universiti Teknologi Malaysia)

Journal of ICT Research and Applications
Vol. 5 No. 1 (2011)

This paper presents a robust algorithm to identify the letter boundaries in images of unconstrained handwritten word . The proposed algorithm is based on  vertical  contour  analysis.  Proposed  algorithm  is  performed  to  generate  presegmentation by analyzing the vertical contours from right to left. The unwanted segmentation  points  are  reduced  using  neural  network  validation  to  improve accuracy  of  segmentation.  The  neural  network  is  utilized  to  validate segmentation  points.  The  experiments  are  performed  on  the  IAM  benchmark database.  The  results  are  showing  that  the  proposed  algorithm  capable  to accurately locating the letter boundaries for unconstrained handwritten words.