PROJECT TITLE :

Compressive Sensing Forensics

ABSTRACT:

Identifying a proof’s origin and the way it absolutely was acquired is a vital forensic problem. Whereas forensic techniques currently exist to work out a symptom’s acquisition history, these techniques do not account for the chance that a signal may be compressively sensed. This can be an important drawback since compressive sensing techniques have seen increased popularity in recent times. During this paper, we tend to propose a collection of forensic techniques to spot signals acquired by compressive sensing. We tend to try this by initial identifying the fingerprints left during a signal by compressive sensing. We tend to then propose two compressive sensing detection techniques that may operate on a broad class of signals. Since compressive sensing fingerprints can be confused with fingerprints left by ancient image compression techniques, we propose a forensic technique specifically designed to identify compressive sensing in digital images. In addition, we have a tendency to propose a method to forensically estimate the number of compressive measurements used to acquire an indication. Through a series of experiments, we have a tendency to demonstrate that every of our proposed techniques can perform reliably below realistic conditions. Simulation results show that both our zero ratio detector and distribution-based mostly detector yield perfect detections for all cheap conditions that compressive sensing is employed in applications, and the precise 2-step detector for images will a minimum of achieve probability of detection of 90% for probability of false alarm <10%. Furthermore, our estimator for the number of compressive measurements can well replicate the $64000 variety.