Watermarking Techniques: Sample Images

The table below presents sample images marked with the VW2D algorithm, with the appropriate watermark and key file. A full description of the VW2D technique can be found here. In our implementation only the green plane of an image is marked. The VW2D algorithm can accept either TIFF or JPEG images as input, and produce either TIFF or JPEG images as output. The full-size images behind the thumbnails below are the acutal JPEG images used as input, and the actual output JPEG images. The marked JPEG images are the undisturbed, marked images to which test images are compared. (Technical Note: the marked pixel data is first JPEG compressed, then decompressed. The inner product to be stored as part of the VW2D algorithm is then computed between the decompressed data and the watermark.)

Variable 2-D Watermark (VW2D) Samples

VW2D These images have been watermarked using the VW2D technique. (Please click on any thumbnail to see the actual original or watermarked image. The thumbnails themselves are not marked.) Each of these images have the same watermark embedded in them. Watermark A JPEG-compressed version of the watermark was produced by first converting the actual watermark to a binary image (a green pixel signifies a one and a black pixel signifies a zero in the mark), and then compressing this image to JPEG format. The actual mark itself in TIFF format, and the user-id file used to generate the mark is also available.	Original	Watermarked
	Original	Watermarked
	Original	Watermarked
	Original	Watermarked
	Original	Watermarked
	Original	Watermarked

Image Adaptive Watermark Samples

The following samples illustrate the image-adaptive watermarking technique applied in the DCT and Wavelet domains. A description of the IA-DCT and IA-Wavelet techniques can be found here.

Click on each thumbnail to see each image. Note that these images are JPEG encoded versions of the actual TIFF images used in these techniques, and are not the original images themselves. The watermark images below are difference images, which have been enhanced for visibility. Copies of the TIFF versions of all these images are available for download.