Project: ROBUST WATERMARKING OF IMAGES

Team:   Haitham Boshnak   hboshnak@stanford.edu
             Khaled Salama       knsalama@stanford.edu

Project Proposal:

    Several algorithms for robust watermarking will be studied,
including Spread Spectrum Watermarking and Sub-band techniques for
watermarking (DCT/wavelets). Based on some or all of the following
factors: SNR, Robustness, simplicity and salability an algorithm will be
chosen. The goal is to investigate the performance of the chosen
algorithm against various attacks.

    The chosen algorithm should be immune to common signal processing
operations such as re-sampling, re-quantization, dithering and image
enhancement techniques. It should be immune to various image
transformations such as scaling and compression. The watermark should
also be perceptually invisible or its presence should not interfere with
the image being protected. This will be tested using the Matlab toolbox
SCIELAB, specially designed for human eye perception of images. We will
consider various attacks including geometric image operations such as
rotation, translation and cropping. We will consider various filtering
operations, noise corruption, lossy compression and contrast change.

    Optimally, each of us will implement an algorithm and an attack
against the other. Finally we hope we would be able to justify our
selection and propose some recommendations for further improvements.

References:
[1] R. Dugad, K. Ratakonda and N. Ahuja, ^ÓA NEW WAVELET- BASED SCHEME
FOR WATERMARKING IMAGES^Ô,  ICIP 98 Chicago

[2] I. Cox, J. Kilian, T. Leighton anmd T. Shamoon, ^ÓSecure Spread
Spectrum watermarking for multimedia^Ô, IEEE trans Image Processing,
1997.