Combining Digital Watermarking and Fingerprinting Techniques to Identify Copyrights for Color Images

Algorithm 1

Feature extraction.

Input:   A color image H (N × N).

Output:   A feature image FT (N/8 × N/8).

Convert H to the YCbCr color space

Partition each of the Y, Cb, and Cr channels into N/8 × N/8 non-overlapping blocks of size 8 × 8

For each corresponding block of the three channels

    Take the first 4 samples from the Y channel, the next 2 from the Cb, and last 2 from the

    Cr as depicted in Figure 4 to form a packed block of size 8 × 8

End For

Apply 2D-DWT to each packed block to obtain N/8 × N/8 LL₂ blocks of size 2 × 2

For each of the LL₂ blocks

    Compute the average A of the four coefficients

    Obtain the feature type T according to (1)

    Determine the FT-share according to T, A, and Table 1

End For

Assemble the FT-shares to form the feature image FT