A novel semi-fragile image watermarking scheme based on wavelet

Abstract

A novel semi-fragile image watermarking scheme based on wavelet is presented in this paper. The original image is first transformed into wavelet domain. The image features are subsequently generated from the lowest-frequency subband as embedded watermark by the condition judgment based on the mean value. Lastly, the watermark is embedded into the predetermined bit-plane in the given lower-frequency subbands by adjusting the corresponding values of wavelet coefficients in the given subband. In the process of watermarking authentication, we compare the image features generated in the lowest-frequency wavelet subband of the received image by the condition judgment based on the mean value with the embedded watermarking information (the image features of the original image) extracted from the pre-determined bitplane in the given lower-frequency subband of the received image to decide whether the image is attacked maliciously or processed acceptably (embedded compression based on wavelet and communication noise). On the one hand, the proposed watermarking scheme is robust to the embedded image compression methods based on wavelet such as EZW, SPIHT and EBCOT in JPEG2000 to a pre-determined bit-plane into which the watermark is embedded. On the other hand, it is tolerant to the communication noise (e.g. Gaussian noise and Salt & Pepper noise) to some extent due to the condition opinion based on the mean value is adopted and the lower-frequency subbands are chosen to be embedded by watermark in the proposed watermarking scheme. Experimental results show the validity of our watermarking scheme in terms of both watermarking invisibility and watermarking robustness.