摘要
针对某些特定视频中,画面文字信息经常包含较为敏感文字信息,导致信息泄露,设计实现了一种视频画面中的文字的检测识别系统,对视频画面中的文字标语、文字条幅,新闻画面中的文字导语等信息进行识别与比对;采用基于双阈值的视频镜头分割算法,根据颜色直方图信息提取关键帧,采用最大稳定极值区域算法提取图像中稳定区域,通过聚类和级联分类器实现文字区域提取,最后将文字区域分割后进行OCR识别,实验表明系统针对复杂背景中的文字能够达到较高的检测识别率。
This article introduced a text extraction and recognition system ior video content, which aimed to the text banners, texts o! the news images in the videos. Presents a algorithms based on adaptive dual--threshold which could divide the video into segments by shots. Then the key frames are extracted based on the color histogram information of the frame. Use Maximally Stable Extremal Regions （MSER） for detecting text character candidates in natural scene images. Then a combination of two classifiers is applied to filter non--text meaningful groups. At last, We put the regions of text into OCR system for recognition. The system that design in this paper achieved high recognition rate.
出处
《计算机测量与控制》
2015年第5期1754-1757,共4页
Computer Measurement &Control
关键词
双阈值
颜色直方图
最大稳定极值区域
均值聚类
分类器
dual--threshold
color histogram information
MSER
perceptual organization clustering
classifiers
