A robust against rotation attack watermarking technique based on wavelet transform

Abstract

Watermark robustness is one of the major characteristics that influence the performance and applications of digital image watermarks. The main issue being focused relates to tradeoff between three major components that covers embedding capacity, imperceptibility and robustness. Discrete Wavelet Transform (DWT) is a technique in frequency domain to transform image into four sub-images. Each sub-image of DWT has its own ability and advantage of good embedding to get better robustness and imperceptibility. Robustness is one of the areas to be improved when an attack is implied to watermarked image. One of the main issues is rotation attack when the watermarked image is rotated with high angle of rotation. The purpose of this research is to improve the robustness of rotation attack and increase the quality of watermarked image. This research proposed a new scheme called RA_SPARSE which determine embedding phase and new model called R_E_Z_A for rotation attacks detection. The effect of embedding on different sizes of standard messages and rotation attack on all angles is identified and analyzed to observe the performance of the developed scheme. One of the findings is the discovery about the nature of the images give the different results for the effectiveness on imperceptibility and robustness of watermarked image. The results of the research for image with much smooth area produced the highest visual quality of PSNR which is 59.37 decibel or equivalent to 8.3 percent improvement at level two of diagonal sub-image (HH) when compared to other available current schemes. Proposed scheme also had the utmost robustness of NCC with 0.9998 or equivalent to 6.8 percent perfection at level three of vertical sub-image (LH) when compared to former existing recent schemes. The research proves that the nature of image with many smooth areas presents more robustness and imperceptibility.