Kernel-Based Image Retrieval with Ontology

Due to the huge increase in the amount of digital images available in the explosive Internet era,making efficient content based image retrieval (CBIR)systems has become one of the major endeavors. In this paper, the authors study the integration of subsequence kernel function based on ontology. Using the VSM to create subsequence kernels, The kernel methodology described here not only overcome the VSM ignoring any semantic relation between words, but also result both in functional similarity and in sequence/words similarity by gap-weighted subsequences kernels, and semantic character is also taken into account. Experiments show that the method has more exact retrieval results, and its cost is under the accepted tolerance.