SVD Based Performance Improvement in Hiding a Message Behind an Image

Abstract

The paper describes an effective method to embed grayscale images or watermarks in the original cover images. To do so, we used a famous linear algebra technique called Singular Value Decomposition. To carry out the test work results and analysis, we observed the two most common metrics namely Peak Signal to Noise Ratio (PSNR) and Mean Square Error (MSE) during the research work. Correlation was found between the original image and the recovered image during extraction of watermark. This method was found fully robust against various attacks which were carried out throughout the research.