Polygonal Element Image Warping

With the ever-increasing update of terminal display devices,there are more and more types of commonly used display devices which may have different resolutions and aspect ratios appear,especially for mobile display devices.In order to improve the user experience,we need to warp the output picture or videoadaptively to meet the different resolution requirements of the display screen.The purpose of image warping is to preserve visually prominent regions when resizing the image,and large deformation of unimportant background is allowed,this research is one of the current hot topic in computer graphics.We present an image warping method based on polygonal elements.Firstly,we combine the technique of image saliency detection to generate an adaptive polygon mesh which make the polygons of the salient areas denser,and that of background sparser.Secondly,we define the polygonal element shape function space on the generated polygon mesh and the warping function is expressed as a linear combination of the polygonal element shape functions,and then we represent the discrete warping problem as a continuous problem.Finally,we use the deformation gradient to measure the deformation degree and introduce different image warping energy functions.We can obtain the optimal warping function under different metrics by optimizing these energy functions.We analyze the image warping method proposed in this paper.The adaptive mesh in this paper performs better in preserving details of the image than the random mesh.The more polygons generated,the better warping image can be obtained.In addition,comparing with the previous methods,our method can guarantee the continuity on polygon mesh boundaries in image deformation.Our method can better preserve the salient regions than previous methods by using an equal number of cells,and we can achieve warping results similar to or ever better than other methods using fewer cells.