AMULTISCALEFRAMEWORKFORCOMPRESSIVESENSINGOFVIDEO(视频压缩传感的多尺度框架).pdfVIP

A MULTISCALE FRAMEWORK FOR COMPRESSIVE SENSING OF VIDEO Jae Young Park Michael B. Wakin∗ Department of EECS, University of Michigan Division of Engineering, Colorado School of Mines Ann Arbor, MI 48109 Golden, CO 80401 ABSTRACT which point a compression algorithm may be applied to ex- ploit spatial and temporal redundancy. In some applications, Compressive Sensing (CS) allows the highly efficient acqui- such as imaging at non-visible (e.g., infrared) wavelengths, it sition of many signals that could be difficult to capture or may be difficult or expensive to obtain these raw samples. In encode using conventional methods. From a relatively small other applications, such as multi-image capture in camera net- number of random measurements, a high-dimensional signal works, implementing a compression algorithm may itself be a can be recovered if it has a sparse or near-sparse represen- challenge. We argue that these burdens may be reduced by us- tation in a basis known to the decoder. In this paper, we ing compressive imaging hardware (such as the “single-pixel consider the application of CS to video signals in order to camera” [3]) where random measurements are collected in- lessen the sensing and compression burdens in single- and dependently from each frame and no additional compression multi-camera imaging systems.