Watermarks for customer identification

Details Watermarks for customer identification Watermarks for customer identification

1999-11-30

2004-10-19

Sheikh, Ayaz (Department: 2131)

Electrical computers and digital processing systems: support

Multiple computer communication using cryptography

Particular communication authentication technique

C380S054000

Reexamination Certificate

active

06807634

ABSTRACT:

BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention generally relates to imperceptible watermarking of images, and more particularly to a digital watermarking method that performs positive identification of customers who purchase and publish images.
2. Background Description
An imperceptible watermark is defined as an alteration of a data set which is, for the most part, imperceptible to a human i.e., the watermark should be invisible or almost invisible, but can be recognized by a machine such as a computer. The general principle of developing a watermark is disclosed, for example, by M. M. Yeung et al. in “Digital Watermarking for High-Quality Imaging”,
Proceedings of the IEEE Signal Processing Multimedia Workshop
, Princeton, N.J. (1997).
A robust watermark is one conceived to survive modifications of the image, and more precisely, to still be detectable when the image has been modified to some reasonable extent. Robust watermarks are usually required to be non-removable by an adverse party, at least not removable without visible distortion to the image. They are usually designed, for instance, to establish ownership or to help protect copyrights. In several cases, one expects them to be detectable when the image is transferred to the analog world, such as when printed or transformed into analog signals such as required for display on a TV screen.
Watermarks cannot be replaced by digital signatures in this context, as known signature schemes with no visible trace on the image do not resist transfer from the digital to analog world. Because of the dual constraint of invisibility and resistance to minor changes in the image, robust watermark detection is therefore based on statistical analysis: any individual components of a mark can be altered by attack and/or by digital/analog (or D/A) conversion so that one is reduced to check that traces of the overall mark persist.
In general, a robust watermark consists of an array M(K) of elements, M(K)(h,v) defined by a cryptographically designed key K. Here, h and v stand respectively for horizontal and vertical coordinates. According to one technique for generating a robust watermark, M(K) is a matrix of “0's” and “1's”, where a zero means a pixel gets brighter, and a 1 in the matrix means a pixel gets darker after watermarking (one can have a richer set of possibilities where, for instance, a “2” in the matrix means that no change is made.) The watermark M(K) (referred to as a “mask”) is used to modify some pixels attributes according to some algorithm, defining some function F, which generates the modified pixel attribute m(h,v) at location (h,v) as a function of the mask M, the pixel's attribute before marking A(h,v), and the attributes before marking An(h,v) of neighboring pixels, i.e.,
m
(
h,v
)=
F
(
M
(
K
)(
h,v
),
A
(
h,v
),
An
(
h,v
)).
The amount by which watermarking modifies the luminosity at a given pixel is not necessarily uniform for better visual performance and/or for better resistance to attacks. Because the image and the accompanying watermark are expected to suffer some modifications, the key K, as well as the function F, are usually chosen as secret keys: a description of cryptographic techniques, with directions on how to use several of their implementations, can be found in
Handbook of Applied Cryptography
by Alfred J. Menezes, Paul C. van Oorschot and Scott A. Vanstone, CRC Press, 1997.
An example of robust invisible robust watermark has been disclosed, for instance, in U.S. Pat. No. 5,825,892 issued to Mintzer and Braudaway.
Problem to be Solved
An aim of the present invention is to address the following business case. Suppose an image vendor (whom we will call The Ven Company in this hypothetical) sells digital images to a pool of customers C
1
,C
2
, . . . , C
N
. In fact not all images are necessarily sold to all customers. Each customer indeed only buys the right to print these images in some publications. No customer has the right to circulate or sell digital form of the images.
The Ven Company wants to prevent unauthorized resale of these images. In particular, if one of the images is found in some publication not listed by any customer, The Ven Company will legitimately suspect that one of its customers has broken his contract. It would therefore be advantageous for The Ven Company to identify the customer in its pool which has acted in this way, so that The Ven Company can pursue legal remedies including suing the customer for injunctive relief to prevent the unauthorized publication and/or damages, terminate its business relationship with this customer, or both.
In principle, this problem may be addressed using a robust invisible watermark scheme as defined above. However, given an image carrying a watermark, W
i
, belonging to a large collection W
1
, W
2
, . . . , W
m
, recognizing which one of these marks has been used, if any, is difficult and time consuming. To segregate among different customers, the number, m, of watermarks needs to be as large as the maximum number of customers that The Ven Company expects for its best selling digital images, Also, because of the statistical nature of robust watermark detection, detection can be unreliable for large m, and careless use of watermarks could lead to a false positive identification, leading to potentially embarrassing problems with non-offending customers. On the other hand, the problem of a false negative is, in principle, solved by the intrinsic qualities of the watermarking scheme used, and therefore reduces the problem To selecting which mark is used out of a large collection.
A well knows kind of attack on watermarks in a situation as described above is collusion of several customers. This can be described as follows: assume customers Ci
1
, Ci
2
, . . . , Ci
n
all buy the same image I, which has been watermarked with different marks Wi
1
, Wi
2
, . . . , Wi
n
, so that they receive image files Ii
1
, Ii
2
, . . . , Ii
n
, respectively. These customers can together create a new image file Ic (where c stands for collusion) where each pixel (h,v) is defined by a combination of averaging and random choices among the (h,v) pixels of all images Ii
1
, Ii
2
, . . . , Ii
n
. It has been recognized that arbitrary collusions will defeat watermarking. Prior work can be found on protection against collusion attacks, such as “Collusion secure fingerprinting for digital data” by D. Boneh, and J. Shaw,
IEEE Transactions on Information Theory
, Vol 44, No. 5, pp. 1897-1905,(1998).
SUMMARY OF THE INVENTION
It is an object of the present invention to provide an improved method for encoding a watermark into a digital image, so that a person who engaged in an unauthorized use of th digital image may be identified with greater accuracy compared with conventional methods.
It is another object of the present invention to provide an improved method for detecting when an unauthorized digital image has been formed as a result of collusion between or among cuts of a digital image vendor, and then to identify the colluding parties with greater accuracy compared with conventional methods.
It is another object of the present invention to provide a method of the aforementioned types which significantly reduces the number of processing steps required to determine the customer or customers who engaged in an unauthorized use of a digital image.
It is a further object of the present invention to provide a method for performing collusion detection based on an analysis of several images sold to one or more customers.
These and other objects of the invention are achieved by providing a digital watermarking method which encodes two watermarks, or masks, into a digital image offered for use by a vendor. The method may advantageously be applied for watermark detection alone, or may also be used to perform collusion detection among two or more customers. The type of watermarks and the manner in which the watermarks are encoded are determined accordingly.
In accordance wit a first embodiment of the method for performing watermark

Affiliated with

Also associated with

Rating

Watermarks for customer identification does not yet have a rating. At this time, there are no reviews or comments for this patent.

If you have personal experience with Watermarks for customer identification, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Watermarks for customer identification will most certainly appreciate the feedback.

Rate now

Comments { 0 }

Profile ID: LFUS-PAI-O-3286076