一种基于改进SURF的视频DNA生成算法研究

为有效提高视频内容管理的准确性和高效性，本文借鉴生物信息学的方法，提出了一种降维改进的SURF算法和K-means算法相结合的视频DNA提取算法。该算法通过提取关键帧的SURF特征得到特征点集合，并对其进行K-means聚类构建视觉词袋模型，并将SURF特征通过视觉词袋模型量化为视觉词汇，并最终编码生成视频DNA。实验结果表明，采用改进的SURF算法生成的视频DNA具有良好的准确性和鲁棒性，并能在时间开销方面得到一定的提高。

Study of a Video DNA Extraction Algorithm Based on Improved SURF

In order to improve the accuracy and efficiency of video content management, a combined method of improved SURF and K-means was proposed in this paper to extract the DNA of video, which referenced the Bioinformatics to identify and analysis videos. By extracting SURF features on video key frames, a feature set was formed. Then, the feature set was processed by K-means, and built a model of Bag of words. Finally, SURF features were quantized visual vocabulary by the model of Bag of words and encoded as video DNA. The experimental results showed that the proposed algorithm of extraction of video DNA has high accuracy and good robustness, and can make improvement in terms of time cost.