Chinese Optical Character Recognition for Information Extraction from Video Images

摘要

A number of research work on text extraction from videos is conducted in these few years, but not many focus on the Chinese language. Due to different characteristics between Chinese and English, new methods are urgently needed for Chinese character extraction. We are applying optical character recognition (OCR) techniques to video frames, and trying to extract character texts from videos. By doing this, we can automatically convert video contents to text, and then exploit the Chinese subtitles for indexing and searching in a digital video library. This paper covers ways we filter the heavy noise and segment out each Chinese character in video segments. We also describe how we perform the optical character recognition for Chinese and evaluate its performance. Some future work will also be described.