Image Retargeting Method Based on Grid Deformation and Content-Awareness

This paper proposes an image retargeting method based on grid deformation and content awareness. The method adjusts the image to different sizes while preserving its prominent and structural features. It starts with a grid partitioning technique, where the image deformation is controlled by the coordinates of the grid vertices. This transforms the image retargeting problem into an optimization problem of the grid vertex coordinates. Simultaneously, saliency detection is performed, aiming to minimize the deformation of visually prominent objects while allowing more deformation of less noticeable elements, thus preserving the original structure. To protect linear features in the image, line segments need to be detected and these features incorporated into the energy equation to optimize the grid vertex coordinates. Experimental results demonstrate that the method provides visually pleasing retargeted images while retaining the original structural features.