新闻视频帧中的标题字幕探测  

Editing Caption Detection in News Video Frames

作　　者：徐峰[1] 梁学战[1] 

机构地区：[1]中国人民解放军63636部队

出　　处：《中国科技信息》2009年第23期117-118,120,共3页China Science and Technology Information

摘　　要：新闻视频中出现的字幕往往包含大量的信息,是视频分析的重要语义线索,探测并识别出来的文字可以为基于内容的视频检索提供索引。通过对新闻字幕的特点进行分析,文章提出了一种基于边缘特征的新闻字幕探测方法。算法首先对图像进行灰度变换,然后对灰度图进行边缘检测,最后进行字幕区域的探测与合并。通过对不同的新闻视频帧进行实验,获得了平均96.8%的探测准确率。实验结果表明,文章提出的方法能够较好的完成新闻字幕检测的任务。Captions in news video often contain the massive information, which is of the important sources for the realization of automatic retrieval, analysis and comprehension of videos since it implies lots of semantics eherently. Analyzing the characteristics of news captions, this thesis proposes a news captions detection method based on edge features. The algorithm first makes gray--scale transformation of images and carries out edge detection, finally, captions are extracted through regional explorasion and captions merger. On test with different news video frame in our experiment the accuracy rate 96,8% is obtained. The experimental results show that the approach can accomplish the task of detection news caption.

关 键 词：新闻视频 边缘检测 标题字幕 边缘特征 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]