A method of retrieving text data from a broadcast image

摘要

Text data is retrieved from a broadcast image by capturing a sequence of frames at predetermined intervals. Each captured frame image is processed to emphasise text content and de-emphasise changing backgrounds materials. The presences or absence of text in the frame is detected by measuring a luminance ratio of the brightest part of the image (the potential text) and the background part. Selected frames are then passed to OCR software for text retrieval.