Image Semantic Information Retrieval Based on Parallel Computing
Yun Ling, Yi Ouyang · 2008
Most of the algorithms proposed in the literature deal with the problem of digital image retrieval. To interpret semantic of image, many researcher use keywords as textual annotation. Concept recognition is a key problem in semantic information searching. In order to be effective and efficient, we proposed a parallel algorithm for semantic concept mapping, which adopts two-stages concept searching method. The first stage is to implement image low-level feature extraction schema; the second step is to implement latent semantic concept model searching, and bridging relationship between image low- level feature and global sharable ontology. Through combining ontology and image SIFT feature, the images on web pages and semantic concept can be mapping together for semantic searching. Experiments on several web pages sets show that it can outperform other methods in terms of precision and recall.