A visible image watermark embeds a clearly visible mark onto a gray or color photographic image. Using a watermark in this way identifies the ownership of materials and reminds viewers of their limited copying rights.
Reversible visible watermarking is a form of visible image watermarking developed at the IBM Tokyo Research Laboratory. Reversible visible watermarking is used for online content distribution. The image is marked with a reversible visible watermark before distribution, and the watermarked image content serves as a "preview" that users may view or download for free. The watermark can be removed to recreate the unmarked image by using a "vaccine" program that is available for an additional fee.
Fragile image watermarking is used to determine if an image has been altered since it was watermarked. IBM is working on several techniques for exploiting this. This method of "image authentication" could be used to detect alterations or replacements in image libraries or to create a "secure" digital camera.
One of the techniques IBM is working with is used to detect the presence of tampering in an image. A robust watermark is embedded into the image simply to identify that the image is authenticated, and then another layer of fragile watermark is embedded on top; this watermark is designed to be extremely sensitive to the alteration of the image. The first layer tells the user to check the second layer; the second layer acts as an "alarm" if the image has been tampered with.
Another technique uses error diffusion to preserve the color content of an image as it undergoes watermarking. This can be important if the images are to be used in a color-critical application. This technique also retains a partial watermark; if the watermarked image is cropped, the extraction process can determine whether the remaining portion has otherwise been altered.
Both techniques require an image-specific authentication key to extract the watermark, making it more difficult for a malicious party to detect or estimate the watermark in a watermarked image. Both techniques will display the extracted watermark as an image for visual authentication, and both permit automatic authentication. Both can also localize any changes that have taken place in an altered image.
IBM is also investigating multiple techniques for robust image watermarking that would help identify the owner or recipient of a watermarked image. We will briefly discuss two such techniques.
The first technique uses a suite of technologies called DataHidingTM, and was developed at the IBM Lab to allow users to embed invisible digital data into the digital content. The target media range from still image to video, and to audio data. This suite of technologies is very flexible when it comes to data payload and level of robustness; also, the data can be automatically extracted and detected without human observation or interaction.
In still image and video DataHiding, the data to be embedded is converted to a binary bitstream and embedded into the image by altering the luminance level of the pixels following a set of predefined rules.
The second variation is based on a technique that modulates the brightness of the image pixels with a random noise field to embed the watermark. The color of the pixels is not altered in the watermarking, so that image color is preserved. When a watermark is extracted from an image, the watermark will be displayed even if it has been damaged by processing applied to the watermarked image. A visual inspection will quickly verify the resemblance of the inserted and extracted watermarks. In limited experiments, this technique has produced watermarks that survive printing, rescanning, reduction, and JPEG compression.
Both techniques described here require an image-specific watermark key to extract the watermark from the watermarked image. This makes it more difficult for a malicious party to detect or estimate the watermark. Both techniques insert the watermark data many times; this redundancy permits the watermark extraction to work more reliably. Neither technique requires that an unmarked image is present in order to extract a watermark.