Cross-Media Retrieval Method Based on Temporal-spatial Clustering and Multimodal Fusion

Abstract

Aiming at the problem of the "semantic gap" and the "dimensionality curse", this paper discussed the model of cross-media retrieval. The methods of feature extraction and fusion of multimedia were given for processing high-dimensional data, and a nonlinear hybrid classifier based on support vector hidden Markov models was design for implementation semantic mapping and learning. According to Shannon information theory, calculation methods of similarity and correlation were given to implement temporal-spatial clustering. Typhoon and other multimedia disaster data are selected for experiments and comparisons. Experimental results show that this method improves the performance of cross-media retrieval.