A deep-learning-based secure video compressive sensing scheme

Abstract

With the advancement of technology and the increased use of video-based communication, we must deal with both massive volumes of information and highly sensitive data in the form of video. As a result, video data must be stored, accessed, and processed in a safe, efficient, and effective way. For that purpose, we first introduce the notion of compressive sensing in this study and then offer a deepnetwork- based compressed sensing approach that investigates both temporal and spatial correlation of video during signal restoration by employing compensation through multilayer deep features. We also offer a unique encryption approach for protected transmission of sampled video frames based on chaotic sequence and maximum distance separable (MDS) matrices.