一种多模态信息融合的视频检索模型

针对包含复杂语义信息的视频检索的需要，提出了一种基于关系代数的多模态信息融合视频检索模型，该模型充分利用视频包含的文本、图像、高层语义概念等多模态特征，构造了对应于多个视频特征的查询模块，并创新地使用关系代数表达式对查询得到的多模态信息进行融合。实验表明，该模型能够充分发挥多模型视频检索及基于关系代数表达式的融合策略在复杂语义视频检索中的优势，得到较好的查询结果。

Video retrieval model based on multimodal information fusion

In allusion to the complex requirement of query, a new video retrieval model based on multimodal information fusion was brought forward in this paper. It included multi-models like text retrieval, image query, semantic features extraction, and used relational algebra expression to fuse these multimodal information. Experimental results demonstrate that our method could fully utilize the advantages of multimodal information fusion based on relational expression in video retrieval, and achieve good performance on complex semantic video retrieval.