Scene text recognition using a Hough forest implicit shape model and semi-Markov conditional random fields

作者：

Highlights：

• Part-based character models are built in a Hough forest.

• The Hough forest robustly detects characters even if characters are highly degraded.

• Checking text line properties effectively remove false detections.

• Utilizing character models throughout the process raises end-to-end text recognition accuracy.

摘要

Highlights•Part-based character models are built in a Hough forest.•The Hough forest robustly detects characters even if characters are highly degraded.•Checking text line properties effectively remove false detections.•Utilizing character models throughout the process raises end-to-end text recognition accuracy.

论文关键词：Scene text recognition,Character recognition,ISM,Hough forests,Part-based character models,Semi-Markov CRFs

论文评审过程：Received 27 November 2013, Revised 30 March 2015, Accepted 3 May 2015, Available online 19 May 2015, Version of Record 16 July 2015.

论文官网地址：https://doi.org/10.1016/j.patcog.2015.05.004