Smoothing of HMM parameters for efficient recognition of online handwriting

作者：

Highlights：

• An HMM based unconstrained online handwriting recognition is presented.

• A novel approach to smoothing of HMM parameters has been considered.

• A circular distribution is used to exploit the directional nature of input data.

• A fully connected non-homogeneous HMM is used.

• It is used for limited vocabulary recognition of unconstrained Bangla handwriting.

摘要

Highlights•An HMM based unconstrained online handwriting recognition is presented.•A novel approach to smoothing of HMM parameters has been considered.•A circular distribution is used to exploit the directional nature of input data.•A fully connected non-homogeneous HMM is used.•It is used for limited vocabulary recognition of unconstrained Bangla handwriting.

论文关键词：Handwriting recognition,Online handwriting,Hidden Markov model,von Mises distribution

论文评审过程：Received 17 July 2013, Revised 14 March 2014, Accepted 18 April 2014, Available online 6 May 2014.

论文官网地址：https://doi.org/10.1016/j.patcog.2014.04.019