LoG Based Watermarking Over Region Magnitude Using Chinese Remainder Theorem
Prajanto Wahyu Adi, Pramudi Arsiwi · 2018 International Seminar on Application for Technology of Information and Communication · 2018
An embedding model is proposed to hide the watermark in region magnitudes of the image using the Laplacian of Gaussian filter that is developed from the two previous researches that using block based and edge based watermarking through CRT algorithms. The block based CRT can cause the significant degradation of image when it lies on the homogeneous area. Meanwhile, the preceding study of edge based CRT detector is generated the embedding location as same magnitude in binary representation. The watermark bits are embedding on any edges location. The edge based CRT is unable to determine the most significant area for the embedding. The proposed method embeds the watermark bits from the highest region magnitudes in descending order of number of watermark bits. This is aimed to improve the previous researches that used block and edge based CRT algorithms. The experiments show that this model has higher structural similarity than the block based CRT which has average similarity value of 0.9996 and 0.9985 respectively. It also improved the robustness against compression the higher average correlation value of 0.6709 than previous CRT scheme which results in 0.6681 and 0.6616. The proposed model has better image quality of watermarked image in term of SSIM measurement and increase the robustness of watermark image in term of compression.