Abstract: Under the condition of rainy day, because of the rain has the bad influence on imaging
system, it will lead the decrease of the quality of the image, and does not help to further processing
image information. To deal with this problem, a kind of automatic image rain removal approach is
proposed. Firstly, the rainy image is decomposed into rough sub-graph and detail sub-graph based on
internal generative mechanism theory. And then, the sparse decomposition based on morphological
component analysis method was used to divide the detail sub-graph into rainy component and
non-rain component. Finally, the rough sub-graph was combined with non-rain component so as to
reconstruct the rain-free image. The experimental results show that the proposed method can
effectively remove the rain streaks in image and keep the state-of-the-art detail information of
the scene.

Key words: image processing, raindrop, sparse decomposition, morphological component, oriented
gradients