Combined similarity measure based approach to image retrieval

In this paper, a new combined similarity measure based approach is proposed to retrieve images by examples. Visual similarity and semantic similarity are combined as a double measure method to compute the similarity between the query image and the target image in the database. According to semantic similarity measure, the query image's candidate similar image set is formed; the images in the set are similar to the query image in semantics. Then visual similarity measure is used to compute similarity between the query image and the image in the candidate similar image set. Experimental results on Corel database show the effectiveness of our proposed approach.