Impact of denoising on watermarking: a perspective for information retrieval

This article presents a watermark technique for embedding a watermark in the frequency domain by applying a Fast Fourier Transform (FFT) algorithm. The frequency domain analysis takes advantages of FFT shift to embed the watermark, bringing low frequency components to the center of the cover image in order to avoid user's attention. With heterogeneous radial calculation, watermark is embedded in magnitude components, after separation of phase and magnitude components of the image. Moreover, protection of watermark is achieved by error correction coding and Advanced Encryption Standard (AES-128) as a double level of security. Extraction of the watermark is done by inverting the steps taken during embedding process. The impact of multiple noise elimination techniques is analyzed for the watermark technique proposed under additive white Gaussian noise (AWGN). The performance of the impact of multiple noise elimination techniques for the proposed watermark technique is evaluated in terms of image quality evaluation metrics (IQA) and the number of bits that are retrieved.