Caption extraction method and device

Abstract

The invention discloses a caption extraction method and device. The method comprises the following steps: a video is decoded, video frames are obtained, pixels in the video frames are subjected to connection operation in a caption arrangement direction, connected domains in the video frames are obtained, video frames containing the same caption are determined based on the connected domains in the video frames, caption zones in the video frames containing the same caption are determined based on distribution positions of the connected domains in the video frames containing the same caption, an assembly tree is correspondingly constructed according to a plurality of channels in the caption zones, a contrast ratio extremum area corresponding to each channel is extracted via use of the constructed assembly tree, contrast ratio extremum areas of the plurality of channels are subjected to color enhancement processing operation, color enhancement contrast ratio extremum areas are formed, and the color enhancement contrast ratio extremum areas of the plurality of channels are integrated. Via implementation of the caption extraction method and device, strong interference backgrounds such as light illumination, clothing and the like in images of the caption zones can be effectively eliminated, the backgrounds can be well separated from the caption, and improvement of subsequent character identification efficiency and precision can be facilitated.

Description

technical field [0001] The communication video technology of the present invention particularly relates to a subtitle extraction method and device. Background technique [0002] Currently, subtitles are recorded in various ways in video files, for example, embedded subtitles, built-in subtitles, and external subtitles. Among them, the embedded subtitle is when the subtitle is overlaid on the video frame of the video, and integrated with the video frame of the video. No matter how the video format changes, the size, position and effect of the subtitle are unchanged. Embedded subtitles are to encapsulate video files and subtitle files into multiple audio tracks and multiple subtitle tracks, and dubbing and subtitles can be selected during playback. The subtitle file corresponding to the external subtitle is independent of the video file. When the video needs to be played, the subtitle file to be used by the video player is loaded on the video. [0003] The video player of th...

AI Technical Summary

Problems solved by technology

[0003] The video player of the terminal supports various forms of subtitles, but it is difficult to extract and recognize all forms of subtitles. In this way, users cannot extract subtitles in text form for sharing or recording during the process of watching videos

Images