Abstract

This paper addresses the problem of rain streak removalfrom a single image. Rain streaks impair visibility of an im-age and introduce undesirable interference that can severe-ly affect the performance of computer vision algorithms.Rain streak removal can be formulated as a layer decomposition problem, with a rain streak layer superimposed on a background layer containing the true scene content. Exist-ing decomposition methods that address this problem em-ploy either dictionary learning methods or impose a lowrank structure on the appearance of the rain streaks. Whilethese methods can improve the overall visibility, they tendto leave too many rain streaks in the background imageor over-smooth the background image. In this paper, wepropose an effective method that uses simple patch-based priors for both the background and rain layers. These pri-ors are based on Gaussian mixture models and can accommodate multiple orientations and scales of the rain streaks. This simple approach removes rain streaks better than the existing methods qualitatively and quantitatively. We overview our method and demonstrate its effectiveness over prior work on a number of examples.