A Bidirectional Flow Joint Sobolev Gradient for Image Interpolation

摘要

An energy functional with bidirectional flow is presented to sharpen imageby reducing its edge width, which performs a forward diffusion in brighter lateral on edgeramp and backward diffusion that proceeds in darker lateral. We first consider the diffusion equationsasL2gradient flows on integral functionals and then modify the inner product fromL2to a Sobolev inner product. The experimental results demonstrate that our model efficientlyreconstructs the real image, leading to a natural interpolation with reduced blurring, staircaseartifacts and preserving better the texture features of image.