Estimating a Small Signal in the Presence of Large Noise

Video magnification techniques are useful for visualizing small changes in videos. For instance, Eulerian video magnification has been used to visualize the flow of blood in the human face. Such visualizations have possible applications in remote monitoring or screening for diseases. However, when visualizing blood flow, the signal of interest may be similar in amplitude to the noise in the video. This raises the question of what one is actually seeing in a magnified video: signal or noise? We seek to understand these signal and noise characteristics with the goal of producing informative and accurate visualizations. We present a preliminary algorithm for estimating the signal amplitude in the presence of relatively high noise. We demonstrate that the algorithm can be used to accurately estimate the signal amplitude in an uncompressed simulated video, but is susceptible to compression noise and motion.