Intuitive Visualization for Online Image Retrieval

摘要

We proposed a efficient and intuitive method to show the Online image retrieval result for searching information. The method incorporates visualization techniques in content-based image retrieval to show the hidden information in the result. We incorporated image browsing into online image retrieval. We give users a initial display based on PageRank, then use the users' feedback to compute similarity function, then we compute the dissimilarity between images, get the position of images in the display space. If users are not satisfied with the display, they may feed back some more interested images to the system to improve the display. With the interactions provided by the system, users can browse a large number of images efficiently and find the exact images fast.