Date Added: Apr 2011
Content-aware image retargeting has attracted a lot of interests recently. The key and most challenging issue for this task is how to balance the tradeoff between preserving the important contents and minimizing the visual distortions on the consistency of the image structure. In this paper, the authors present a novel filtering-based technique to tackle this issue, called "Importance filtering". Specifically, they first filter the image saliency guided by the image itself to achieve a structure-consistent importance map. They then use the pixel importance as the key constraint to compute the gradient map of pixel shifts from the original resolution to the target.