A Photo Visualization System Considering Region of Interest

Cheap and highly-functional digital cameras are now readily available to the public. In contrast to traditional film-based cameras, tasks of refining, classification and clustering images is burdensome to camera users. Therefore, we need an automated procedure to assist manual management of digital photos. One way to overcome this problem is to provide a summarized view covering the entire set of photos taken. This paper proposes a new photo summarization system for the sets of temporarily sequential photos taken by a user in exploiting EXIF (EXchangeable Image Format) kept for each photo. First, we present a one clustering method using event-based classification for the temporarily dense set of photos. Second, we propose a method to select a representative photo from each cluster to construct a summarized view sequence photo set. Finally, we present a new layout method allowing the overlapped configuration of photos to increase the space efficiency compared to the plain grid-based layout method. In experiment, we compared our system with that of previous work (traditional grid layout), in terms of space efficiency and user-friendliness empirically, to evaluate our system.