【摘要】 视频中包含的文字信息与视频的语义内容有很强的相关性,将视频中的文字信息提取出来进行分析处理可以有效地理解电视视频语义,从而实现对视频内容的安全监控。针对文字检测提出一种基于小波变换、角点特征图像和统计特征的有效方法,并运用基于彩色空间的文字提取方法获取二值图像,更有利于后面OCR的文字识别。更多还原

【Abstract】 Text in frames has a strong correlation with the semantic content of videos. Analysing text extracted from the video frames can effectively improve the semantic information understanding,which can be utilized in video security monitoring. In this paper,a text detection algorithm is proposed based on wavelet decomposition,corner point and statistical features. And a color space based text extraction method is employed to obtain binary image, which can promote the text recognition result of OCR.更多还原