Rule-based trajectory segmentation for modeling hand motion trajectory

作者：

Highlights：

• Circular statistics based inference for hand gesture recognition.

• Derivation of parameter estimation formulae for the von Mises HMM.

• An automated process to determine a hyperparameter (# of states) of hand gesture HMM.

• Hand gesture trajectory segmentation based on sequential log-likelihood ratio test.

• Performance improvements by mitigating over-fitted parameter estimation.

摘要

•Circular statistics based inference for hand gesture recognition.•Derivation of parameter estimation formulae for the von Mises HMM.•An automated process to determine a hyperparameter (# of states) of hand gesture HMM.•Hand gesture trajectory segmentation based on sequential log-likelihood ratio test.•Performance improvements by mitigating over-fitted parameter estimation.

论文关键词：Trajectory segmentation,Hand gesture recognition,Hidden Markov model,HMM initialization

论文评审过程：Received 17 October 2012, Revised 11 August 2013, Accepted 8 November 2013, Available online 25 November 2013.

论文官网地址：https://doi.org/10.1016/j.patcog.2013.11.010