Automatic text location using cluster-based template matching
Eun Yi Kim, Jae Sik Chang, Hang Joon Kim · 2003
This paper proposes a method for automatic text location in images including multi-segment characters. Accordingly, we use shape information in addition to size and location information.This shape information is represented by cluster-based templates obtained from target characters using K -means clustering algorithm. Experimental results show the effectiveness of the proposed method.