Enhancement of digital grayscale image watermarking using sparse matrix

Abstract

Watermarking is a form of steganography that proved its worth in successfully protecting copyright information. It is the process of embedding data inside an audio or video or image message such that the embedded data is possible to be detected or extracted later. The core focus in watermarking techniques is their performance which is determined by imperceptibility along with robustness and capacity. These properties are often conflicting, which needs to accept some trade-offs between them. Despite the successes recorder in the area of digital watermarking, several challenges continue to persist particularly in the Areas of balancing these factors. This research aims to enhance the the processes in the watermarking technique for archieving imperceptibility with an acceptable balancing and enhance the security. The research proposed a new scheme using sparse matrix for improving the effectiveness of watermarked image using digital wavelet transform and inverse discrete wavelet transform to locate the best place and level in the image to embed the watermark. Sparse matrix is used to enhance the embedding process by selecting the proper coefficient. For more secure watermarking, additional encryption layer is utilized to increase the difficulty towards unauthorized extraction. The proposed technique generated the proper message size for each sub image based on the PSNR, which is used as an indicator for selecting the suitable level of embedding and for detecting the possibility of attacks. The proposed scheme improves watermarking quality by using the sparse matrix to select the appropriate coefficient for embedding. The experiments showed that the proposed scheme enhances 2.8479 dB of quality (PSNR) or equivalent to 5.3 % of improvements. The research proposed scheme achieved better PSNR in comparison with other research.