基于内容的项目视频关键帧识别技术研究

Video Key Frame Recognition Technology Research Based on the Content

摘要: 关键帧的提取对于视频的自动切分至关重要，提取的好坏直接影响最终的结果，本文通过对项目视频的分析研究，提出了模板匹配，直方图比较，关键帧检测与关键帧识别的算法，并比较了模板匹配与直方图对比的识别准确率和识别速率，其中在关键帧识别中，采用了新的字符分割的算法。最后，总结了本文方法的优越性，并且取得了很好的效果。   The key frame extraction is vital for automatic segmentation of video; the standard of extracted key frame directly affects the final result. Based on the analysis and research about project video, the paper proposes template matching, histogram comparison, the key frame detection and key frame recognition algorithm, and for the recognition accuracy and recognition rate of template matching and histogram comparison, the paper also does the comparison. In the recognition, the key frames adopted a new character segmentation algorithm. Finally, the paper summarizes the advantages of this method and good results have been achieved.