An overview of scalar quantization based data hiding methods

In Ref. [1], Costa presented a communications framework that provided useful insights into the study of data hiding. We present an alternate and equivalent framework with a more direct data hiding perspective. The difference between the two frameworks is in how channel dependent nature is reflected in optimal encoding and decoding operations. The connection between the suggested encoding/decoding scheme and practical embedding/detection techniques is examined. We analyze quantization based embedding/detection techniques in terms of the proposed framework based on three key aspects. The first aspect is the type of postprocessing utilized at the embedder (i.e. distortion compensation [2,3], thresholding [4], Gaussian mapping [5]). The second key aspect is the form of demodulation used at the extractor. The third is the criteria used to optimize the embedding/detection parameters. The embedding/detection techniques are compared in terms of probability of error, correlation, and mutual information (hiding rate) performance merits.