A Grid Based Resizing Framework via Effectively Combining Cropping with Warping

IEEE Computer Society
Publication Type:
Conference Proceeding
2012 IEEE International Conference on Image Processing ICIP 2012, 2012, pp. 2997 - 3000
Issue Date:
Filename Description Size
2012004452OK.pdf Published version1.04 MB
Adobe PDF
Full metadata record
Image retargeting is a problem of adapting images to arbitrary aspect ratios in order to maximize users browsing experience. As two major solutions for image retargeting, warping and cropping have their own advantages and limitations respectively. In this paper, a grid based resizing framework is proposed for effectively combining warping with cropping. Firstly, warping preserves more important content within the cropping window through retaining the aspect ratios of salient grids and distorting the non-salient ones. Secondly, cropping provides extra space for warping to absorb the spatial deformation and assures the important content is retained in retargeted image simultaneously. Finally the objective function is formulated as two energy terms for warping and cropping respectively. And, a nonlinear optimization is applied to obtain the retargeting results. Our approach could make warping and cropping complement each other, and improve the quality of retargeted image effectively. Experiments and comparisons on the ReTargetMe dataset demonstrate the superiority of our approach.
Please use this identifier to cite or link to this item: