Research on Image Fusion Algorithm Based on Fuzzy Clustering and Semantics
Jingxiu Ni, Qian Xu, Guoying Zhang, Xinkai Xu · 2017
Image fusion aims at integrating complementary or ambiguous information of different images. In order to bridge the semantic gap during the image fusion, a hierarchical semantic labeling model of images is proposed and an image fusion algorithm based on semantic labeling and fuzzy clustering is presented. Images are segmented into regions which are classified into different types using low-level features. Afterwards the labeling model learns to map region types with keywords. The fusion mechanism is based on the calculation of semantic similarity. Experimental results demonstrate that the image fusion algorithm proposed in the paper is more precise than the state-of-the-art methods, especially when noisy labels appear in semantic fusion.