Data Hiding Using Interpolation Technique

Lim, Sze Hwan (2018) Data Hiding Using Interpolation Technique. Final Year Project (Bachelor), Tunku Abdul Rahman University College.

Text Lim Sze Hwan_FULL TEXT.pdf Restricted to Registered users only Download (2MB)

Abstract

The purpose of this project involves the use of high resolution images to be embedded with secret data. Over the years high resolution images were already around but monitors could not take advantage due to cost of producing high resolution monitors are expensive. The challenge of this experiment is achieving a high embedding capacity on an image and the maintaining the image quality after embedding. The research questions for this project are how to maintain image quality after embedding and what is the embedding capacity of the proposed method. Objectives of this research are propose a novel data hiding technique via interpolation and trying to maintain the image quality afterwards. Throughout the experiments the images used are 4 low resolution images (512x512), 3 medium resolution(1920x1280) images and 3 high resolution images(more than 6000x4000). In our experiments, the images will be embedded with short and long messages and also be embedded to its maximum capacity. The proposed method is based on Neighbor Mean Interpolation (NMI) by Jung et al. The proposed will be developed using MATLAB from MathWorks. After embedding the secret data using the interpolation method on the test images they are assessed using Peak Signal to Noise Ratio (PSRN), Structure Similarity (SSIM) and visual observation to check for quality against the original. The results show that both high and low resolution images were low in quality when the entire image is fully embedded. Long messages however when embedded on high resolution images, results also show that it is also low in quality but were visually better than low and medium resolutions. However it achieved high embedding capacity on the image. Even more so high resolution images compared to low resolution images when embedded to the maximum is visually better although PSNR and SSIM tests said otherwise. For future improvements the method could apply some image quality maintaining techniques to help maintain the loss of quality through interpolation.