... Text embedded in video frames often carries the most important information, such as time, place, name or topics, etc. This information may do great help to video indexing and video content ... understanding. To extract text information from video frames, which is often referred as video OCR, the first essential step is to detect the text area in video frames. Many methods have been introduced ... and background in video images. To detect the text efficiently, those methods usually defined a lot of rules that are largely dependent of the content of video. Because the video background...