首页 | 本学科首页   官方微博 | 高级检索  
     检索      

一种新的维吾尔文字幕关键帧提取方法
引用本文:闫轲,哈力旦·阿布都热依木,李敏强.一种新的维吾尔文字幕关键帧提取方法[J].安徽大学学报(自然科学版),2015(5):68-73.
作者姓名:闫轲  哈力旦·阿布都热依木  李敏强
作者单位:新疆大学电气工程学院,新疆乌鲁木齐,830047
基金项目:国家自然科学基金资助项目
摘    要:为了快速有效地提取视频中含有维吾尔文字幕的关键帧提出一种新方法.该方法首先运用多尺度Harris角点检测,结合维吾尔文字笔画特点,完成字幕关键帧的初步提取;然后利用最小二乘法拟合直线,根据两条拟合直线的角度偏离情况对初步提取的结果作去冗余处理,得到最终字幕关键帧.实验结果表明:该方法快速简单,提取结果的准确率达到93%,具有很好的适应性.

关 键 词:维吾尔文  关键帧提取  多尺度  Harris角点检测  直线拟合

A new method of uyghur subtitle key frame extraction
Abstract:The authors proposed a new method to extract the key frame of the video with uyghur subtitles quickly and efficiently in this paper . Firstly , subtitle key frames were extracted preliminary by using multi‐scale Harris corner detection combined with the characteristics of the uygur text strokes . Then the object of preliminary extraction was redundant and processed according to judging the deviation angle of two fitting straight lines by least square method .Finally ,it could get the subtitle key frames .Experimental results showed that this method was quick and simple ,with the extract accuracy 93% ,and had a good adaptation .
Keywords:uygur text  key frame extraction  multi-scale Harris corner detection  linear fitting
本文献已被 CNKI 万方数据 等数据库收录!
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号