Robust color image watermarking using discrete wavelet transform, discrete cosine transform and cat face transform

Abstract

The primary concern in color image watermarking is to have an effective watermarking method that can be robust against common image processing attacks such as JPEG compression, rotation, sharpening, blurring, and salt and pepper attacks for copyright protection purposes. This research examined the existing color image watermarking methods to identify their strengths and weaknesses, and then proposed a new method and the best embedding place in the host image to enhance and overcome the existing gap in the color image watermarking methods. This research proposed a new robust color image watermarking method using Discrete Wavelet Transform (DWT), Discrete Cosine Transform (DCT), and Cat Face Transform. In this method, both host and watermark images decomposed into three color channels: red, green, and blue. The second level DWT was applied to each color channel of the host image. DWT decomposed the image into four sub-band coefficients: Low-pass filter in the row, Low-pass filter in the column (LL) signifies approximation coefficient, High-pass filter in the row, Low-pass filter in the column (HL) signifies horizontal coefficient, Low-pass filter in the row, High-pass filter in the column (LH) signifies vertical coefficient, and High-pass filter in the row, High-pass filter in the column (HH) signifies diagonal coefficient. Then, HL2 and LH2 were chosen as the embedding places to improve the robustness and security, and they were divided into 4×4 non-overlapping blocks, then DCT was applied on each block. DCT turned a signal into the frequency domain, which is effective in image processing, specifically in JPEG compression due to good performance. On the other hand, the Cat Face Transform method with a private key was used to enhance the robustness of the proposed method by scrambling the watermark image before embedding. Finally, the second private key was used to embed the watermark in the host image. The results show enhanced robustness against common image processing attacks: JPEG compression (3.37%), applied 2% salt and pepper (0.4%), applied 10% salt and pepper (2%), applied 1.0 radius sharpening (0.01%), applied 1.0 radius blurring (8.1%), and can withstand rotation attack. In sum, the proposed color image watermarking method indicates better robustness against common image processing attacks compared to other reviewed methods.