Auto scene text detection based on edge and color features
Xiaodong Huang, Kehua Liu, Lishang Zhu · 2012
In this paper we present a novel approach to detecting scene text based on the edge and color features. Firstly, because the character edge feature is not sensitive to the luminance changes, we extract the edge features to locate the candidate text region coarsely. Secondly, according to the text row character will keep similar color, we use the K-means clustering to extract color feature and locate the candidate text regions accurately. Finally, we use a trained SVM classifier to distinguish the text region from non-text region in these candidate regions. Experimental results show that our algorithm performs well for detecting scene text with various color, font-size and text alignment.