摘要: This paper presents an automatic method for creating a collage from collection of photos scene taken different viewpoints. The is constructed by aligning the images (in terms their positions, rotations and scales) using least-squares formulation. We have developed graph-based optimization algorithm layering so as to minimize fragmentation collage. A can be displayed with opaque layers, transparent or blended image boundaries. viewed piece-wise perspective representation visible seams. has not only aesthetic value but also conveys structure camera motion in intuitive way. To capture live-action collages dynamic scenes we arrays that physically flexed user continuously vary composition scene. design our enables reconfigure them spatial arrangement cameras matter minutes. show several still examples demonstrate provide new interesting way experience scenes.