摘要
视频字幕提取是建立视频信息索引的有效手段,字幕区的检测和定位是新闻视频字幕提取过程中影响识别效率的关键问题。利用新闻视频字幕区在视频帧序列中的变化规律,通过计算视频字幕帧之间的时间相关性能够实现新闻视频字幕场景的快速分割。对经过场景分割同主题新闻视频帧序列中的字幕,以像素的帧序列方差描述其字幕区时间不变性,提出了一种基于帧序列颜色分析的视频字幕区定位提取方法,并通过实验验证了方法的有效性。
The extraction of video captions is an effective method for constructing the index of video information.The sticking point for the process of captions extraction is identification and location of the area with captions in image frames and affects efficiency of recognizing text.A method is used to divide the video into scenes by calculating time correlation between video frames according to the change of frame background.Then a novel method is presented to locate and extract the area with captions in video frames sequence based on chromatic analysis.The method makes up of the variance of the pixel with the same location in video frames to describe time inflexibility after dividing video into scenes.Finally,the experimental result shows that the method is valid.
出处
《安徽电子信息职业技术学院学报》
2013年第3期6-9,共4页
Journal of Anhui Vocational College of Electronics & Information Technology
基金
湖北省教育厅科学技术研究项目视频字幕的提取设计与实现(B20129203)
关键词
图像处理
字幕检测
时间相关性
色度分析
image processing
captions extra ction
time correlation
chromatic analysis