Integrating multiple character proposals for robust scene text extraction
作者：
Highlights：
• Proposed system separates text regions from images under unconstrained environment.
• Generalized clustering utilizes properties of scene text to detect text boundaries.
• Multiple image segmentations provide various interpretations on text regions.
• Two-step CRF approach models properties and relationship of text in graph structure.
• Character proposals are generated and integrated to find proper character regions.
摘要
•Proposed system separates text regions from images under unconstrained environment.•Generalized clustering utilizes properties of scene text to detect text boundaries.•Multiple image segmentations provide various interpretations on text regions.•Two-step CRF approach models properties and relationship of text in graph structure.•Character proposals are generated and integrated to find proper character regions.
论文关键词：Scene text extraction,Two-stage CRF models,Multiple image segmentations,Component,Character proposal
论文评审过程：Received 9 October 2012, Revised 1 August 2013, Accepted 29 August 2013, Available online 9 September 2013.
论文官网地址：https://doi.org/10.1016/j.imavis.2013.08.007
