Video compression in signal-dependent noise

This work investigates the performance of video compression techniques in the presence of signal-dependent noise. The signal-dependent noise sources most commonly encountered are film-grain noise and speckle. Film-grain noise degradation occurs when a photographic film is scanned for the purpose of digitization [6]. All types of coherent imaging techniques, such as synthetic aperture radar (SAR) imagery, laser illuminated imagery, astronomical imagery and ultrasonic medical imagery are affected by speckle. Noise in the video not only affects the quality of the video, but also the compression scheme for the video. It is of utmost importance to improve the quality of video and also the achievable compression, for the sake of archiving, in applications such as medical imagery. This work aims to investigate techniques to improve the quality and achievable compression of moving pictures (video), keeping in mind such applications. There is no real consensus yet on the "best" quality measure to use for determining the quality of the output video, so we will use the standard mean square error, log mean square error, signal-to-noise ratio and perceptual mean square error (which is modeled on the human visual system) in this work.