Develoment of an adaptive discrete cosine transform (DCT)-based Mod 4 image steganographic tool / Nur Fatin Ibrahim

In developing a steganographic tool, the factor of imperceptibility is the most important factor to be highlighted. The stego-image must not leave any sign of modification in order to achieve higher level of imperceptibility. The proposed solution in this project is by using the adaptive Mod-4 steganographic technique in discrete cosine transform (DCT) domain. Groups of 2 x 2 quantized DCT coefficients (GQC) are selected as the valid embedding area from the selected standard steganography cover images. These images are grayscale JPEG images with size of 256 x 256 pixels. The modulo 4 arithmetic operation is further applied to all the valid GQCs to adaptively embed a pair of binary bits. The imperceptibility will be tested by using Peak-Singnal-Noise-Ratio (PSNR) The experimental results shows that all images achieved PSNR value greater than 30dB as well as high quality because the average value of PSNR for this tool is 40.526dB. Therefore, this demonstrates that the adaptive DCT-Mod-4 tool can successfully preserve the quality of the images and stays undetected.