论文部分内容阅读
I first discuss how to impose positivity constraint in the variational total variation model for restoration of images with noise and blur,highlighting our new method of implicitly imposing the constraint.Then,to simultaneously restore both the image and the kernel,I present our blind deconvolution work where positivity is crucial and such models will not work otherwise.Finally I show some work on fractional order derivatives and their advantage of preserving smoothness without staircasing.