Abstract: This paper presents an algorithm regarding word group as a special symbol to detect video text automatically. It extracts an 18 dimension features vector according to strokes feature and two-dimension layout feature of video text, and uses a pre-trained Support Vector Machine(SVM) to partition frame regions into text and non-text regions. A multi-resolution model is used to detect texts of different font sizes, and a dilatation-shrink process is employed to adjust the text position. Experimental result shows that the detection rate of the method achieves 93.17% and the false detection is 0.73%.

Key words: video content analysis, video text detection, Support Vector Machine(SVM)

摘要： 提出一种以词组作为模式识别对象的中英文视频文本检测算法，其根据视频中文本的笔画结构特点和聚集特性构造一个18维的特征向量，利用支持向量机将视频帧分为文本和非文本区域，通过多分辨率模型检测不同尺寸的文本，采用扩张-收缩的后处理过程校准文本区域位置。实验结果表明，该算法的检测准确率达93.17%，误检率仅为0.73%。

关键词: 视频内容分析, 视频文本检测, 支持向量机

CLC Number: 

• TP391.41