A hand gesture recognition system based on canonical superpixel-graph

作者：

Highlights：

•

• A new distance metric, namely canonical superpixel-graph earth mover’s distance (CSG-EMD), is proposed.

• The hand gesture is represented by a canonical superpixel-graph.

• The structural stress of the superpixel-graph is utilized to formulate the proposed CSG-EMD.

• High mean accuracies with a small number of templates are achieved on five public datasets.

• The recognition performance is not sensitive to the parameters.

摘要

•A new distance metric, namely canonical superpixel-graph earth mover’s distance (CSG-EMD), is proposed.•The hand gesture is represented by a canonical superpixel-graph.•The structural stress of the superpixel-graph is utilized to formulate the proposed CSG-EMD.•High mean accuracies with a small number of templates are achieved on five public datasets.•The recognition performance is not sensitive to the parameters.

论文关键词：Hand gesture recognition,Kinect,Superpixel earth mover’s distance,Canonical forms

论文评审过程：Received 4 November 2016, Revised 27 June 2017, Accepted 27 June 2017, Available online 6 July 2017, Version of Record 20 July 2017.

论文官网地址：https://doi.org/10.1016/j.image.2017.06.015