Robust image watermarking based on the wavelet contour detection

Robustness is one of the fundamental requirements for image watermarking. In this paper, we present a new approach to robust image watermarking, based on the contour of a host image, which is detected according to the Lipschitz singularity of the image. The watermark is embedded into the coefficients representing contours in the wavelet domain. Unlike other wavelet-based watermarking techniques which embed data into middle frequency bands and leave the low frequency coefficients untouched to avoid significant distortion, the new method embeds a watermark into the low pass frequency band after wavelet transform of the host image to provide more robustness, while the distortion is still invisible. Numerous simulations demonstrate that the presented approach is highly robust to JPEG compression, white noise and linear filtering.