Seamlet carving for shape-aware image resizing

In this paper we propose an optimized seam-carving approach for anti-shearing image resizing. Current image/video seam-carving strategy only focuses on deleting the pixels along 8-connected seams, which may lead to the obvious information loss during resizing. Unlike the traditional seam-carving methods, our approach can construct a seamlet to handle the discontinuous pixels and further achieve an optimized image resizing results with shape-preservation. In particular, to suppress zigzag artifacts, we introduce the anti-shearing energy into the seamlet generation, which includes two steps: 1) the energy map is formalized by using Gabor filtering and the saliency map; 2) The energy computation is optimized in Gabor feature space to obtain the resizing results without the connectivity restriction. The experimental results and comparisons show the effectiveness of our method.