A Fast Discrete Homotopy Solution Method for Two Problems in Image Deconvolution

Blur and noise are commonly found in real-life images. Blur occurs in the acquisition whereas noise mostly occurs due to the signal converted into a digital representation due to the sampling process. Severe blur and noise in a given image results in resolution loss and restricts accurate post-processing assessments. To improve image restoration and on the same time build a fast variational framework we propose the incorporation of restarted homotopy analysis method as an alternative compared with costly direct methods. Through experiments, we demonstrate the effectiveness of our new technique.