ABSTRACT :
The aim of our project is to propose a high-capacity CDMA based watermarking scheme based on orthogonal pseudorandom sequence subspace projection. We introduced a novel idea to eliminate the interference due to the correlation between the host image and the code sequences in the watermark extraction phase, and therefore, improve the robustness and message capacity of the watermarking scheme. We give the implementation steps of the proposed scheme and test its performance under different attack conditions by a series of experiments. Experimental results show that the proposed scheme shows higher robustness than the canonical scheme under different attack conditions.
Digital watermarking is a widely used technique for copyright protection and authentication of intellectual properties. A watermark is a piece of information such as a logo, a license number, or any other sign of copyright or ownership that is embedded into the digital products such as audios, photos, videos and other multimedia products. Imperceptibility, robustness and security are the main requirements in most watermarking applications. Code Division Multiple Access (CDMA) principles provided a robust and secure way for watermarking. In a CDMA system, different messages are hidden in the same noise-like signal using uncorrelated codes, i.e., low cross correlation value (orthogonal/near orthogonal) among codes. So it is highly robust to interference and noise. Moreover, third party cannot reconstruct the base-band signals without the key, so CDMA systems are secure against private attacks.
Joseph proposed the first CDMA-based watermarking scheme, which spreads out the watermark information to the m sequences in the form of string sequences. Silvestre et al embedded the watermark information into the frequency domain using orthogonal codes. Kohda et al using CDMA techniques to embed watermark information into the DCT domain of color images. They first transform the RGB image into YIQ signal and perform DCT transformation on the Y, I, Q components, then select the first 15 DCT coefficients of Y, 6 coefficients of I, and 3 coefficients of Q to form the separate CDMA channel with spreading sequences of variable-period to transmit YIQ signals. Vassaux et al. divided the host image into multiple layers and embedded the watermark messages in the 1,2,4,8 layers using CDMA scheme. Bijan stated that spreading sequence CDMA communication principles had natural applications in uncompressed digital video watermarking.