A Content-Aware Scaling Method for Stereoscopic Images

Content-aware image scaling methods have received great attentions in the past 10 years. Most of them are proposed only for monocular 2D images. With the development of 3D display techniques, it is very desirable to extend these methods on stereoscopic images. The existing stereo methods fail in preserving objects with straight lines. In this paper, we introduce straight line preservation in the content-aware image scaling and apply content-aware random matching to the stereoscopic constraints. We first represent each image as a grid mesh and model the energy distortion function considering quad mesh deformation, line bending, boundary condition, horizontal alignment and disparity consistency. Then, we scale each quad mesh to the target size by minimizing the total energy function simultaneously preserve the straight lines. Finally, the target images are obtained by interpolating and mapping the scaled mesh. Experimental results demonstrate its effectiveness.