【摘要】 视频总结是提取视频中运动目标并在时间和空间上对其重新展现的过程。在提取运动目标特征点时,针对传统的SURF算法会进行重复性计算而消耗较多时间的特点,对SURF算法进行改进,避免对特征点领域Haar小波响应的二次计算,并增大扇形窗口滑动步长,从而减少了许多交叉区域的重复计算,提高了算法的实时性,最后,利用改进的SURF算法进行视频总结,形成了包含很多重要信息的总结视频。实验证明了该方法的有效性。更多还原

【Abstract】 Video summary is a process of extracting the moving objects in the video and show them again in the time and space.When extracting the feature points of the moving objects,traditional SURF algorithm may consume more time.This paper improves the SURF algorithm to avoid secondary calculation of the Haar wavelet response,and increases the sliding step of the window.Then it reduces the calculation of the intersectional region to improve the real-time of the algorithm.Finally use the improved SURF algorithm to summarize the video,and form the video summary containing a lot of important information.Experiments show the effectiveness of the method.更多还原