Video Subtitle Extraction Method Research Of Radio And Television Based On Triple-play

In the convergence process of the telecommunications network, cable network andInternet network, will be facing a more serious security challenges. Traditional cablenetworks are relatively independent and closed, data transfer mode is a single top-downbroadcast-style. With new technologies and new business, data transmission mode willalso incude terminal to the service side, between the terminal. Television is influential, andaffects a lot of audiences, So video content regulation has important implications. Videosubtitles contains a large number of intuitive information, its extraction and analysis is animportant part of the entire video content regulation.Here is a method for the extraction and recognition of video subtitles based edgedetection. Video subtitles added later, with a clear outline and stroke characteristics.According to this feature, first make edge detection for picture, to remove the irrelevantinformation in the picture, using OTSU binarization, further removing the backgroundinformation, and then remove noise through the morphological operations, to give thesubtitle area. Then divide the pictures into simple background and complex backgroundboth cases. For subtitle picture with simple background, make binarization directly andOCR recognition to give subtitle text content. For complex background, first make edgedetection, then binarization and morphological operation, finally, OCR recognition.Test subtitle picture in the case of two kinds of simple background and complexbackground, the subtitles area location can be accurately positioned, effectively removingthe background information, then use OCR software identification, has a high recognitionrate.