Content based photograph slide show with incidental music

In this paper, we proposed a new scheme to generate photograph slide show with incidental music based on the content of the photographs and music. Unlike existing photograph slide show software, which show the photographs one by one (perhaps with some fancy transition effects) with music in the background, our algorithm utilizes content analysis of the photographs and the incidental music. Firstly, low-quality photographs, such as blurred or under exposed, are filtered out by a photograph quality detector; and the rest of the photographs are grouped into a three-layer construction, namely, day, scene, and group of similar photographs (GoS). Incidental music is segmented based on its beats, and then aligned with photographs, which makes the photograph transitions occur at the beat position of the music. Various transition effects and transformation effects are applied based on the structure and content of the photographs. According to different users' preferences, several editing styles are also supported in our algorithm. The experimental results are very impressive according to the user study.