On improved spread spectrum watermark detection under compressive sampling

Abstract

This paper studies performance of correlation based energy detection (ED) and generalized likelihood ratio test (GLRT) to resolve binary hypothesis on spread spectrum (SS) watermark in digital images under compressive sampling (CS) paradigm. Watermark information in the form of independent and identically distributed (i.i.d) Gaussian pattern is embedded during image acquisition at low measurement space i.e. CS platform. Diversity technique used in communication receiver is then applied to improve watermark detector performance. Simulation results highlight that at low (watermark) signal-to-noise ratio (WNR/SNR), GLRT based detector offers high probability of detection (PD) while ED performs almost at par with GLRT at high SNR i.e. at high measurement space for a given watermark power. Simulation results also show the improved detector performance compared to the conventional correlator-detector and existing CS based watermarking methods.