Image Upsampling Based on Moving-Constrained Spring-Mass Model
Gong Yong · Chinese Journal of Computers · 2011
Image upsampling is an important area in visual media technologies.In this paper the authors propose a new method for upsampling images by considering the image upsampling as the problems of mechanics balance based on moving-constrained spring-mass model and upsampling based on deformed grid.According to the characters of image upsampling,the authors modify the classical definition of stretching force in spring-mass model to reduce the stretch disturbing on those masses outside the neighborhood of the mass from the mass's moving.A centripetal force is also introduced to restrict the mass's moving scope.The external force is designed based on the gradient field of the module of the gradient of source image,which will point to the direction of edges for masses in the neighbor of edge,and will nearly be zero for the other masses.Also,a regular term based on Helmholtz-Hodge theory is used to guarantee the smoothness of grid deformation.The authors offer a series of experimental results to show the usefulness of the method.