Character-Position-Free On-Line Handwritten Japanese Text Recognition by Two Segmentation Methods

DOI Web Site 16 References
  • LIANG Jianjuan
    Department of Computer and Information Sciences, Tokyo University of Agriculture and Technology
  • ZHU Bilan
    Department of Computer and Information Sciences, Tokyo University of Agriculture and Technology
  • KUMAGAI Taro
    Automotive Electronics Development Center, Mitsubishi Electric Corporation
  • NAKAGAWA Masaki
    Department of Computer and Information Sciences, Tokyo University of Agriculture and Technology

Abstract

The paper presents a recognition method of character-position-free on-line handwritten Japanese text patterns to allow a user to overlay characters freely without confirming previously written characters. To develop this method, we first collected text patterns written without wrist or elbow support and without visual feedback and then prepared large sets of character-position-free handwritten Japanese text patterns artificially from normally handwritten text patterns. The proposed method sets each off-stroke between real strokes as undecided and evaluates the segmentation probability by SVM model. Then, the optimal segmentation-recognition path can be effectively found by Viterbi search in the candidate lattice, combining the scores of character recognition, geometric features, linguistic context, as well as the segmentation scores by SVM classification. We test this method on variously overlaid sample patterns, as well as on the above-mentioned collected handwritten patterns, and verify that its recognition rates match those of the latest recognizer for normally handwritten horizontal Japanese text with no serious speed restriction in practical applications.

Journal

References(16)*help

See more

Keywords

Details 詳細情報について

Report a problem

Back to top