Development of a Non-Separable Integer-to-Integer Wavelet Transform-Based Digital Image Steganography System

Abstract

This paper develops a high payload and imperceptible digital image steganography system using a non-separable integer wavelet transform and diamond encoding. This became necessary to effectively utilize all the redundant bits in an image without any distortion. The developed non-separable integer-to-integer transform is responsible for decomposing the digital image into the frequency domain and has an advantage of resiliency to attacks, 75% lower rounding operations, and 50% lifting steps when compared with existing integer wavelet transform. The Haar was adopted as analyzing and synthesizing filter owing to its higher visual quality of 68.73 dB, 0.0087 and 0.9994 as peak signal-to-noise ratio (PSNR), structural similarity index, and mean square error (MSE) respectively when compared with Daubechies (59.11 dB, 0.0799, 0.9955), Symlet (56.15 dB, 0.0759,0.9955) and Coiflet (51.61 dB, 0.4978, 0.9758) under the same embedding conditions. The diamond encoding technique, on the other hand, is responsible for the evaluation of all the embeddable regions in the coefficients. The developed system maintained a high payload and imperceptibility with an average PSNR of 65.46 dB and MSE of 0.0244 which are 25.88% and 94.73% improvements over state of the art.

Keywords

• imperceptibility
• integer wavelet transform
• payload
• redundant bits
• sensitive
• steganography