The Recovery of High Speed Vibration Video Based on Compressive Sensing

The compressive sensing is a technique which can reduce the sampling rate at the time of sampling the signal. Most high speed video cameras have relatively low spatial resolution. It is hard to overcome the tradeoff between temporal and spatial resolution without incurring a remarkable increase in hardware costs. In this paper, in order to overcome this tradeoff, we adopt compressive sensing techniques for time sampling of video, representing and reconstructing the high speed video from low frame rate image sample which is compressed by the space-time coded exposure measurement matrix, and accomplish sparse representation of a video by learning an over-complete dictionary on video patches. The experimental results prove the effectiveness of the method, and realize the recovery of the high speed vibration video from low frame rate compressed sampling video.