Tensor representation learning based image patch analysis for text identification and recognition

作者：

Highlights：

• A novel model, tensor representation learning based image patch analysis (TRL-IPA), is proposed for document understanding.

• TRL-IPA is built on a general formulation of the convergent tensor representation learning (CTRL) algorithms.

• The CTRL algorithms are theoretically guaranteed to converge to a local optimal solution of the learning problem.

• Extensive experiments demonstrate the superiority of TRL-IPA over related vector and tensor representation based approaches.

摘要

Highlights•A novel model, tensor representation learning based image patch analysis (TRL-IPA), is proposed for document understanding.•TRL-IPA is built on a general formulation of the convergent tensor representation learning (CTRL) algorithms.•The CTRL algorithms are theoretically guaranteed to converge to a local optimal solution of the learning problem.•Extensive experiments demonstrate the superiority of TRL-IPA over related vector and tensor representation based approaches.

论文关键词：Tensor representation learning,Convergence,Ancient document understanding,Text identification,Text recognition

论文评审过程：Received 1 August 2013, Revised 23 September 2014, Accepted 27 September 2014, Available online 24 October 2014.

论文官网地址：https://doi.org/10.1016/j.patcog.2014.09.025