Halftone Image Steganography Based on Reassigned Distortion Measurement

Abstract

AbstractMost state-of-the-art halftone image data hiding methods aim to preserve good visual quality when embedding messages, while ignoring the statistical security. This paper proposed a halftone steganographic scheme that improves the visual quality and the statistical security of the anti-steganalysis. First, a general distortion measurement for halftone images based on human visual system (HVS) model is proposed. Utilizing the Least-Mean-Square (LMS) method, halftone images can be converted to grayscale images and the objective image quality assessment is applied to evaluate the distortion caused by flipping pixels. Different distortion measurements can be derived from different image quality assessments. Then, to further measure the embedding distortions, we combine these distortion measurements to construct a reassigned distortion measurement based on the controversial pixels prior (CPP) rule. Finally, syndrome-trellis code (STC) is employed to minimize the number of flipping pixels. Experimental results have shown that the proposed steganographic scheme achieves strong statistical security with high capacity and visual quality.KeywordsHalftone image steganographyReassigned distortion measurementSyndrome-trellis code