一种新的基于模糊聚类的镜头检索方法

镜头检索是基于内容的视频检索的重要内容,而镜头内部内容往往是变化的,这给镜头检索带来了很大的困难。为了描述这种镜头内部的不一致性,这里首次使用模糊聚类的方法,把镜头分为多个等价类,等价类内部内容是一致的,这些等价类客观全面地描述了镜头内部内容的变化。把这些等价类用于镜头检索,获得了良好的检索结果。实验对比结果也证实了该方法的有效性。

A New Approach for Content-based Shot Retrieval by Fuzzy Classification

Shot retr ie val plays a critical role in content-based video retrieval,but it is very dif ficult because some changes often exist in a shot's content.So we firstly pro pose a novel approach based on Fuzzy Classification to describe the inconsistenc ies.By the method,a shot is divided into many equivalence classes in each of which the content is coincident ,and the classes show the changes in the shot in objectivity and in the round.Then we apply the result above to shot retriev al,and experimental results indicate that our proposed approach achieves super ior performance than some existing methods.