A photo summarization technique based on location and orientation with geographic information

Abstract

With the growing popularity of digital cameras and camera phones, more and more people take and view pictures more frequently. This trend of recent movement brings out dramatically increasing number of photos and research challenges for dealing with the huge quantity of photographs, such as searching, analyzing, summarizing, browsing, and sharing photos efficiently. This thesis presents a new photo summarization method that utilizes the photographer``s location and the orientation information, in order to select representative pictures among large collections. In particular, we implemented a system, which performs automatically selecting the representative photos for subjects and landscapes from a particular region, by analyzing, clustering, and summarizing methods based on the location and the orientation information with the geographic information. As a result, the representative photographs can provide various and distinctive views of subjects or landscapes, helping people understand them easily. For acquiring digital photos with location and orientation information, a simple capture system containing GPS and digital compass sensor was developed, and approximately 900 photos were taken in Gyeongbok Palace for the experiment. The photos were classified and summarized by location and orientation information. Also, the representative photos were selected successfully, which were verified by the user preference test. The proposed method can be one of the efficient ways of summarizing photos and it will be useful in various applications, such as map-based photo sharing services, online tour assistants, photo recommendation services, and etc.