| 一种有效的视频文字提取方法 |
| |
| 引用本文： | 田破荒,李弼程,彭天强.一种有效的视频文字提取方法[J].电视技术,2007,31(10):87-89. |
| |
| 作者姓名： | 田破荒 李弼程 彭天强 |
| |
| 作者单位： | 信息工程大学,信息工程学院,河南,郑州,450002 |
| |
| 摘 要： | 根据文字区域内集中分布着短小的水平垂直边缘,而背景区的边缘则比较粗、比较长的特征,提出一种视频文字提取方法.它使用相关熵二值化和形态学膨胀获得候选文字区,然后提取候选区域的变异直方图,根据该直方图获得精确的文字区.选用不同视频截图实验,结果表明该方法简单,具有较高的正确率、精度和鲁棒性.
|
| 关 键 词： | 视频文字提取 Sobel算子 相关熵二值化 形态学膨胀 变异直方图 |
| 文章编号： | 1002-8692（2007）10-0087-03 |
| 修稿时间： | 2007-07-16 |
Efficient Approach of Video Text Retrieval |
| |
| TIAN Po-huang,LI Bi-cheng,PENG Tian-qiang.Efficient Approach of Video Text Retrieval[J].Tv Engineering,2007,31(10):87-89. |
| |
| Authors: | TIAN Po-huang LI Bi-cheng PENG Tian-qiang |
| |
| Affiliation: | Department of Information Science, School of Information Engineering, Information Engineering University, Zhengzhou 450002, China |
| |
| Abstract: | According to characters that short vertical and horizontal edges mix together in text area while edges in background are thick and long, a segmentation algorithm for segmenting text from video frames is presented. The method gets potential text areas using relative entropy-based threshold and expansion in mathematical morphology. The gained varied gray-scale histograms of potential areas are involved to refine the text areas. Experiments on different Video segmentations prove the simplicity, high-accuracy, and robustness of this method. |
| |
| Keywords: | video text retrieval Sobel operator relative entropy-based threshold morphological dilation varied gray-scale histogram |
| 本文献已被 维普 万方数据 等数据库收录！ |
|
