An efficient access method for multimodal video retrieval

This paper presents the Slim (2) -tree, an efficient and effective content-based video retrieval technique allowing the use of multiple modalities within a single index structure. Slim (2) -tree is capable of dealing with different distance measures for the modalities and can perform both multimodal and unimodal searches using the same tree structure. Experimental studies on a large real dataset show the video similarity search performance of the proposed technique. Additionally, we present experiments comparing our method against state-of-the-art of multimodal solutions. Comparative test results demonstrate that our technique improves the performance of video similarity queries.