Joint of Multi-feature Fusion and Mutual Information-Based Image Semantic Annotation

In order to solve the problem of semantic gap in annotation, we propose an automatic semantic annotation framework, Combination of Multi-feature fusion and Mutual information model (JMFMI), which consists of image processing, semantic learning, and semantic annotation. The improvement is reflected in two aspects: combining global features and local features; and forming incidence matrix between low-level features and high-level semantic. To reduce dimension of correlation matrix, we propose the method of mutual information based on experiential entropy. Experimental results show that the proposed method is a better solution to improve annotation accuracy and slow speed.