A survey of document image word spotting techniques

作者:

Highlights:

• This work reviews the word spotting methods for document indexing.

• The nature of texts addressed by word spotting techniques is analyzed.

• The core steps that compose a word spotting system are thoroughly explored.

• Several boosting mechanisms which enhance the retrieved results are examined.

• Results achieved by the state of the art imply that there are still goals to be reached.

摘要

•This work reviews the word spotting methods for document indexing.•The nature of texts addressed by word spotting techniques is analyzed.•The core steps that compose a word spotting system are thoroughly explored.•Several boosting mechanisms which enhance the retrieved results are examined.•Results achieved by the state of the art imply that there are still goals to be reached.

论文关键词:Word spotting,Retrieval,Document indexing,Features,Representation,Relevance feedback

论文评审过程:Received 26 February 2016, Revised 5 November 2016, Accepted 21 February 2017, Available online 27 February 2017, Version of Record 22 April 2017.

论文官网地址:https://doi.org/10.1016/j.patcog.2017.02.023