融合颜色信息与特征点的镜头边界检测算法  

Shot boundary detection algorithm combining color information and feature points

作　　者：蔡轶珩[1] 胡朝阳 崔益泽 

机构地区：[1]北京工业大学信息与通信工程学院,北京100124

出　　处：《计算机应用》2017年第A02期95-98,111,共5页journal of Computer Applications

基　　金：国家自然科学基金资助项目(61201360);北京市教委科技项目(KM201710005028)

摘　　要：针对新闻视频镜头检测算法在高效和准确两个方面的要求很难平衡的问题,提出了一种融合全局特征与局部特征的融合算法。首先介绍了镜头边界检测的相关理论知识。接着,分别对基于H-S颜色直方图这一全局特征,以及尺度不变特征变换(SURF)特征点这一局部特征的镜头边界识别算法进行分析,并应用于新闻视频中进行相应的步骤设定;在此基础上,将两类典型的算法融合形成一种新的镜头边界检测算法。通过构建视频镜头检测软件平台,对其有效性进行验证。实验结果表明:融合算法能够充分发挥两类算法各自的优势,使新闻视频镜头的查准率和查全率指标得到平衡,同时大幅降低程序的运行时间,提高算法的效率和准确性。最后,对该算法进一步优化的方向作了展望。It is difficult to get a balance between efficiency and accuracy in news video shot detection algorithm. A fusion algorithm combining global feature and local feature was proposed. First, the relevant theoretical knowledge of shot boundary detection was introduced. Then, the shot boundary recognition algorithm based on the global feature of H-S color histogram and the local feature of SURF feature point were analyzed, and the corresponding steps were set and applied to news video. On this basis, two typical algorithms were merged to form a new boundary detection algorithm. By constructing a video shot detection software platform, the validity of the fusion algorithm was verified. The experimental results show that the fusion algorithm make full use of the advantages of the two algorithms, so that the precision and recall of the news video lens are balanced. Meanwhile, the running time of the program is greatly reduced and the efficiency and accuracy of the algorithm are improved. Finally, a prospect of further optimization of the algorithm is discussed.

关 键 词：镜头边界检测 新闻视频 H-S颜色直方图 加速稳健特征特征点 算法融合 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]