Video Restoration Using Multichannel-morphological Component Analysis Inpainting

Morphological component analysis (MCA)[1, 2] is a popular image processing algorithm that extracts degrading patterns or textures from images and simultaneously performs inpainting (estimation of lost pixels). MCA has a wide range of uses, including MRI image enhancement and restoration of old photographs. However, in these authors’ opinions, an application that has been widely overlooked is the use of MCA in restoring video footage. We present a novel implementation of an augmented version of MCA known as multichannel morphological component analysis (mMCA), and use it to reconstruct heavily degraded video. Our algorithm examines consecutive frames of video footage in order to identify and remove distortions. Results in this paper indicate that this approach has some significant advantages over the use of conventional MCA on individual video frames.