Image fusion‐based video deraining using sparse representation

Abstract

By disregarding the ‘rain component’ but keeping the ‘non-rain component’ only, results reconstructed by the conventional morphological component analysis decomposition-based rain removal algorithms lose a lot of detail information. On the basis of the limitation, an image fusion strategy is introduced to remove rain from a video. The final rain-free frame is recovered by employing the fused coefficients and the fused dictionary. Experimental results demonstrate that the proposed method can efficiently remove rain streaks, while at the same time preserve more detail information.