Fractional means based method for multi-oriented keyword spotting in video/scene/license plate images

作者：

Highlights：

• A novel approach for keyword spotting in video, scene and license plate images.

• Anew model based on fractional means for reducing background complexity.

• The combination of Radon and Fourier coefficients to extract context features.

• Minimum cost path based ring growing to restore missing characters.

摘要

•A novel approach for keyword spotting in video, scene and license plate images.•Anew model based on fractional means for reducing background complexity.•The combination of Radon and Fourier coefficients to extract context features.•Minimum cost path based ring growing to restore missing characters.

论文关键词：Text candidates,Fractional mean,Radon coefficient,Fourier coefficient,Context features,Minimum cost path estimation,Ring growing,Word detection,Keyword spotting

论文评审过程：Received 18 August 2017, Revised 10 July 2018, Accepted 10 August 2018, Available online 11 August 2018, Version of Record 5 October 2018.

论文官网地址：https://doi.org/10.1016/j.eswa.2018.08.015