基于支持向量机的视频字幕自动提取方法

An Automatic Video Subtitles Extraction Method Based on Support Vector Machine

视频字幕的尺寸大小经常发生变化,原有提取方法只考虑字幕本身固有特征,在自动提取过程中难以保证字幕的运动状态,不能精准地进行视频中字幕与非字幕的分类,导致字幕提取的正确率较低.对此本研究基于支持向量机的视频字幕自动提取方法,以提高字幕提取正确率.以角点检测法为基础,提取视频字幕特征,基于支持向量机分割字幕特征集合,二值化区域理想阈值自动提取视频字幕,完成基于支持向量机的视频字幕自动提取方法设计.实验结果表明:以四组不同形式的视频为测试对象,对其中包含的字幕组进行提取,在本文方法应用下没有出现漏检情况.对提取完成的字幕进行正确率计算,本文方法的提取正确率平均为99.6%,而深度学习提取方法的正确率平均为94.4%,主要成分分析提取方法的正确率平均为96.5%,较之分别提高了5.2%和3.1%,具有实际应用意义.

The size of video subtitles often changes,and the original extraction methods only consider the inherent characteristics of the subtitles themselves.It is difficult to ensure the motion state of the subtitles during the automatic extraction process,and cannot accurately classify subtitles and non subtitles in the video,resulting in a low accuracy of subtitle extraction.Based on corner detection method,extract video subtitle features,segment the subtitle feature set using support vector machine,and automatically extract video subtitles using binary region ideal threshold.Complete the design of a support vector machine based automatic video subtitle extraction method.The experimental results show that using four different forms of videos as the test subjects,the subtitle groups included in them were extracted,and there were no missed detections in the application of this method.The accuracy calculation of the extracted subtitles shows that the average extraction accuracy of the method in this paper is 99.6%,while the average accuracy of the deep learning extraction method is 94.4%.The average accuracy of the main component analysis extraction method is 96.5%,which is 5.2%and 3.1%higher than that of the previous method,respectively,and has practical application significance.