Lossless Watermarking Using Bandelet and Optimal Matrix Norms

Abstract

In this paper, a multi-frequency lossless watermarking algorithm that the original image need not been destroyed is proposed by using bandelet and optimal matrix norms. The wavelet transform is first performed to the original image, the bandelet transform is then applied to the high and middle frequency parts of the transformed image, and the texture and edge traced by using the geometric flow via feature of the image to yield the high frequency lossless watermarking. For low frequency part of wavelet coefficients, by selecting optimal matrix norm, a low frequency watermarking scheme is also achieved. A lossless watermarking approach composed of multi-frequency information is then proposed. The experimental tests show the proposed watermarking system is evaluated against standard watermarking attacks.