Versatile and resilient holographic sensing on images

Holographic representations of data have two main objectives. The sensing process generates and then distributes multiple descriptions of information in packets that enable progressive recovery. The packets are designed to have as equal importance as possible to guarantee smoothness in the quality of the recovered information, independent of the order of the packet’s arrival. We recently developed a least-squares approach to the design of holographic representation for stochastic data vectors. While it relies on the framework widely used in modeling images, it had not been rigorously tested on actual images. This paper describes the results of such tests on various images under noisy sensing environment. We report that holographic sensing is indeed versatile and resilient in actual deployment.