Adaptive to pixel value and pixel value difference irreversible spatial data hiding method using modified LSB for grayscale images

Abstract

A new irreversible data hiding spatial domain method, A2MLSB, adaptive to pixel value (PV) and pixel value difference (PVD) using variable threshold on PV and PVD ranges, and modified least-significant bit (MLSB) substitution method for grayscale images is proposed. Correctness of A2MLSB is proved. Theoretical estimates of the mean squared error (MSE), peak signal-to-noise ratio (PSNR, dB) and embedding capacity (EC, bpp) for A2MLSB method matching experimental results are obtained. Theoretical MSE estimates are obtained for MLSB and LSB as functions of EC; they can be used for assessing performance of data hiding method based on MLSB and LSB, and serve as benchmarks. A2MLSB has, on average, better PSNR (by 1.33 dB) and EC (by 0.14 bpp) than 37 variants of nineteen known data hiding methods. It resists RS, PVD histograms, SPAM, and SRM steganalysis.