Scene Text Detection via Stroke Width
Li Yao, Huchuan Lu · 2016
In this paper, we propose a novel text detec-tion approach based on stroke width. Firstly, a u-nique contrast-enhanced Maximally Stable Extremal Region(MSER) algorithm is designed to extract charac-ter candidates. Secondly, simple geometric constrains are applied to remove non-text regions. Then by inte-grating stroke width generated from skeletons of those candidates, we reject remained false positives. Final-ly, MSERs are clustered into text regions. Experimental results on the ICDAR competition datasets demonstrate that our algorithm performs favorably against several state-of-the-art methods. 1.