Low-rank matrix recovery of dynamic events

Low-rank matrix recovery problems arise in a variety of scientific and engineering applications. For instance, blind de-convolution in signal processing and communication, phase retrieval in computational imaging, and recommendation systems in machine learning. In this paper, we present an algorithm for reconstructing a time-varying low-rank matrix from sequential measurements. In particular, we present an algorithm for estimating video frames from dynamic measurements by exploiting inter-frame motion. We discuss two applications of our proposed model and algorithm: (1) Auto-calibration in dynamic MRI. (2) Phase retrieval of a video signal. We demonstrate the performance of our method on real and synthetic data.