Videotext Detection for Content-based Indexing

E.Y. Kim, J.S. Chang, S.H. Park, and H.J. Kim (Korea)


Videotexts, text location, region segmentation,morphological operators, video indexing


This paper presents a method that can detect texts from video frames, regardless of text size, alignment and orientation. The proposed method consists of three major modules: video decomposition, region segmentation, and videotext detection. An input video is decomposed into several element frames, according to their color prototypes. Then, each element frame is segmented into character components and non-character. Finally, morphological operators are carried out on character components that allow extracting videotexts with various alignments, and then the videotexts obtained from each frames are combined. Experimental results show the effectiveness of the proposed method.

Important Links:

Go Back