Emotion-based image retrieval using multiple-queries and consistency feedback

In this paper, we propose the effective image retrieval method based on the userpsilas emotion. Existing Emotion/Contents-based image retrieval which used a single query image reflects the userpsilas retrieval intention by methods of moving the query points or updating the weights and so on. However, these methods have the limitation of retrieval by the complicated factors of emotion. In this paper, multiple-query images are added in every feedback process for fully expressing the userpsilas emotion. Also the images that the users want can consistently be searched through the proposed consistency feedback method. We got the retrieval satisfaction more than 90% within the 4-th retrieval through the experiments for given 6 representative emotion represented with adjectives and confirmed that the proposed method can be expended by any emotion query adjectives.