ImageHive: Interactive content-aware image summarization

Abstract

The ImageHive tool communicates facts, ideas, and stories from an image collection by generating a summary image that preserves the relationships between images and avoids occluding their salient parts. The tool applies a constrained graph layout algorithm that preserves image similarities and ensures that salient parts remain visible. Then, a fast, constrained Voronoi tessellation refines the layout locally and tiles the image plane. Users can interactively adjust the generated result. Use cases demonstrate that ImageHive is versatile enough for use in different devices and applications. Also, the fast layout algorithm is up to 100 times more efficient than the traditional state-of-the-art approach and has significantly better results than standard alternatives, including principal component analysis and Isomap. The article includes a video demonstration of ImageHive, available as a Web Extra on the IEEE Computer Society Digital Library.