Nonparametric Steganalysis of QIM Steganography Using Approximate Entropy

Date Added: Jul 2011
Format: PDF

This paper proposes an active steganalysis method for Quantization Index Modulation (QIM) based steganography. The proposed nonparametric steganalysis method uses irregularity (or randomness) in the test-image to distinguish between the cover-image and the stego-image. The authors have shown that plain quantization (quantization without message embedding) induces regularity in the resulting quantized-object, whereas message embedding using QIM increases irregularity in the resulting QIM-stego. Approximate entropy, an algorithmic entropy measure, is used to quantify irregularity in the test-image. The QIM-stego image is then analyzed to estimate secret message length. To this end, the QIM codebook is estimated from the QIM-stego image using first-order statistics of the image coefficients in the embedding domain.