一种解决视频检索中语义不一致性方法

近年来,领域本体作为实现知识共享的基础已应用于视频语义检索。然而,同一或不同应用或领域知识可能存在多个本体,这些本体之间缺乏对术语使用的约定。这种现状所产生的语义不一致性导致难以实现检索过程中的知识推理和互操作过程的自动化,因而严重影响视频检索的效率和结果。提出一个两层本体结构,把同一或不同领域共享或共用的概念与知识映射到一个上层本体,从而解决多本体间语义上的不一致性,实现本体间知识的共享。并通过应用于视频场景的检索实验,说明该方法是有效和可行的。

Novel Solution for Semantic Inconsistence in Video Retrieve

In recent years,domain ontology has been used to video semantic retrieve as the base of sharing of knowledge.However, one domain may have more than an ontology .The ontology which has no restriction about terms among them lead to semantic inconsistence that is a problem to improve the performance of video semantic retrieve.We propose a tow-tie ontology model that maps the knowledge of multi ontology to an upper ontology to solve the problem. And the experiment we do in video scene retrieve shows that the solution is practice and effective.