We present a non-photorealistic algorithm for retargeting large images to small size displays, particularly on mobile devices. This method adapts large images so that important objects in the image are still recognizable when displayed at a lower target resolution. Existing image manipulation techniques such as cropping works well for images containing a single important object, and down-sampling works well for images containing low frequency information. However, when these techniques are automatically applied to images with multiple objects, the image quality degrades and important information may be lost. Our algorithm addresses the case of multiple important objects in an image. The retargeting algorithm segments an image into regions, identifies important regions, removes them, fills the resulting gaps, resizes the remaining image, and re-inserts the important regions. Our approach lies in constructing a topologically constrained epitome of an image based on a visual attention model that is both comprehensible and size varying, making the method suitable for display-critical applications.

@InProceedings{STRGG05,
  author       = "Setlur, Vidya and Takagi, Saeko and Raskar, Ramesh and Gleicher, Michael and Gooch, Bruce",
  title        = "Automatic Image Retargeting",
  booktitle    = "International Conference on Mobile and Ubiquitous Multimedia (MUM)",
  year         = "2005",
  ee           = "http://portal.acm.org/citation.cfm?id=1149499",
  url          = "http://graphics.cs.wisc.edu/Papers/2005/STRGG05"
}