Clustering-Based Navigation of Image Search Results on Mobile Devices

With the Internet information explosion as well as the rapid increasing of mobile users, web image search and browsing on mobile devices will become a big application in foreseeable future. Since mobile devices generally suffer from limited resources, like small screen factor and interaction facilities, etc, the interface provided by current image search engine is not suitable for search results navigation on mobile devices. In this paper, we propose a new method to increase user's experience on search results browsing and navigation. By clustering result images based on visual and semantic cues, user can catch the overview of the search results with only a few clicks; by using an optimal layout algorithm, the summary of each cluster can be presented on small screen with maximum scaling ratio; by employing the smart thumbnail of the presentation image, more information can be shown to users on limited screen. The navigation operations are designed to make the new interface easy to learn and simple to use. Experiment results demonstrate the effectiveness of the new presentation method.