Exploiting forced alignment of time-reversed data for improving HMM-based handwriting segmentation

作者：

Highlights：

• Analysis of online handwriting segmentation boundaries obtained from HMM.

• Identification of early boundary problem in the Viterbi forced alignment.

• Exploration of time-reversed data for improving the segmentation boundaries.

• Incorporation of forward-reverse alignment in HMM based handwriting segmentation.

• Proposal is evaluated by improve in word recognition performance.

摘要

•Analysis of online handwriting segmentation boundaries obtained from HMM.•Identification of early boundary problem in the Viterbi forced alignment.•Exploration of time-reversed data for improving the segmentation boundaries.•Incorporation of forward-reverse alignment in HMM based handwriting segmentation.•Proposal is evaluated by improve in word recognition performance.

论文关键词：Online handwriting recognition,DNN,IAM,UNIPEN,Word recognition

论文评审过程：Received 3 January 2018, Revised 26 November 2018, Accepted 6 December 2018, Available online 7 December 2018, Version of Record 19 December 2018.

论文官网地址：https://doi.org/10.1016/j.eswa.2018.12.012