A high-efficiency edge-preserving Bayesian method for image interpolation

Common interpolation methods such as bilinear and bicubic tend to smooth edges in the interpolated image. We propose a high-efficiency edge-preserving Bayesian method for image interpolation. In the proposed method, the problem of image interpolation can be expressed as a constrained optimization problem, where two keys of solving speed are how to obtain search direction and step length. We find a descent direction that has an analytic form and identify the step length using the Armijo rule, resulting in high-speed implementations. Experiment results show that this method guarantees the convergence rate of the solution and good visual quality of the interpolated image.