US6834344B1 — Semi-fragile watermarks

Claim 1

Independent Claim 133 words

A method for watermarking an image file where benign modifications to the file still render the file authenticatible, comprising the steps of: constructing a digest N from stable contents of the file which survives benign modifications but which is changed by modifications to significant content of the file; embedding said digest N into the file with a watermark algorithm; and wherein said stable contents of the image file is determined by the steps: constructing a class of images having statistical properties which are a representative subset of possible benign modifications to the image; and decomposing the image with said statistical properties into two components, a first of said two components comprising said benign modifications to said image and a second of said two components, being orthogonal to the first, comprising said stable contents.

Claim 2

Independent Claim 133 words

A method for watermarking an image file where benign modifications to the file still render the file authenticatible, comprising the steps of: constructing a digest N from stable contents of the file which survives benign modifications but which is changed by modifications to significant content of the file; embedding said digest N into the file with a watermark algorithm; and wherein said stable contents of the image file is determined by the steps: constructing a class of images having statistical properties which are a representative subset of possible benign modifications to the image; and decomposing the image with said statistical properties into two components, a first of said two components comprising said benign modifications to said image and a second of said two components, being orthogonal to the first, comprising said stable contents.

Claim 3

Dependent Claim 22 words

A method for watermarking an image file as recited in claim 1 wherein said watermark algorithm is an invisible robust watermark algorithm.

Claim 4

Dependent Claim 22 words

A method for watermarking an image file as recited in claim 1 wherein said watermark algorithm is a visible robust watermark algorithm.

Claim 5

Dependent Claim 25 words

A method for watermarking an image file as recited in claim 1 wherein said benign modifications comprise printing the file and subsequently scanning the file.

Claim 6

Dependent Claim 22 words

A method for watermarking an image file as recited in claim 1 wherein said benign modifications comprise compressing and decompressing the file.

Claim 7

Dependent Claim 27 words

A method for watermarking an image file as recited in claim 1 wherein said digest is a hash of said second orthogonal component comprising and stable contents.

Claim 8

Dependent Claim 24 words

A method for watermarking an image file as recited in claim 1 wherein said step of decomposing the image comprises principle component analysis (PCA).

Claim 9

Dependent Claim 22 words

A method for watermarking an image file as recited in claim 1 wherein said step of decomposing the image comprises Wold's decomposition.

Claim 10

Independent Claim 120 words

A method for watermarking an image where benign modifications to image file still render the file authenticatible, comprising the steps of: cutting an image into squares of n×n pixels; computing transform coefficients for each of said squares; quantizing said transform coefficients according to a quantization table to produce quantized coefficients; encrypting said quantized coefficients with a secret key; hashing said encrypted quantized coefficients to obtain a digest N; embedding said digest N in said image; and wherein authenticating said file comprises the steps of extracting said digest N from said image; reconstructing a new digest N′ from said from said image; and comparing said digest N to said new digest N′, wherein a match indicates that said image is authentic.

Claim 11

Independent Claim 120 words

A method for watermarking an image where benign modifications to image file still render the file authenticatible, comprising the steps of: cutting an image into squares of n×n pixels; computing transform coefficients for each of said squares; quantizing said transform coefficients according to a quantization table to produce quantized coefficients; encrypting said quantized coefficients with a secret key; hashing said encrypted quantized coefficients to obtain a digest N; embedding said digest N in said image; and wherein authenticating said file comprises the steps of extracting said digest N from said image; reconstructing a new digest N′ from said from said image; and comparing said digest N to said new digest N′, wherein a match indicates that said image is authentic.

Claim 12

Dependent Claim 21 words

A method for watermarking an image as recited in claim 9 wherein said transform coefficients comprise discrete cosine transform (DCT) coefficients.

Claim 13

Dependent Claim 32 words

A method for watermarking an image as recited in claim 9 wherein said match comprises at least 50% of bits in said digest N and said new digest N′ being the same.

Claim 14

Dependent Claim 30 words

A method for watermarking an image as recited in claim 9 wherein reconstructing said new digest N′ comprises the step of: extracting and decoding the embedded watermark in said image.

Claim 15

Dependent Claim 101 words

A method for watermarking an image as recited in claim 9 wherein authenticating said file comprises the steps of: extracting said digest N from said image; reconstructing a new digest N′ from said from said image; decomposing said digest N into a collection of DCT coefficients N(1), N(2) . . . N(I); decomposing said new digest N′ into a collection of DCT coefficients N′(1), N′(2) . . . N′(I); authenticating said image when N(I) matches N′(I) if the distance between N′(I) and a quantization interval Q(I) is small with the property that every member of Q(I) will be mapped into N(I).

Claim 16

Dependent Claim 30 words

A method for watermarking an image as recited in claim 13 wherein reconstructing said new digest N′ comprises the step of: extracting and decoding the embedded watermark in said image.

Claim 17

Dependent Claim 53 words

A method for watermarking an image as recited in claim 9 further comprising the steps: considering said digest N as a q-ary word; letting C be a q-ary error correcting code having a same length as said q-ary word; changing said digest N be a codeword in C closest to said q-ary word.

Claim 18

Independent Claim 142 words

A method for marking a file for authentication so that benign modifications to the file still render the file authenticatible, comprising the steps of: constructing a digest N from stable contents of the file which survives benign modifications but which is changed by modifications to significant content of the file; embedding encrypted second number computed out of said digest N into the file with a mark on the file; and wherein said stable contents of the image is determined by the steps: constructing a class of images having statistical properties which are a representative subset of possible benign modifications to the image; and decomposing the image with said statistical properties into two components, a first of said two components comprising said benign modifications to said image and a second of said two components, being orthogonal to the first, comprising said stable contents.

Claim 19

Independent Claim 142 words

A method for marking a file for authentication so that benign modifications to the file still render the file authenticatible, comprising the steps of: constructing a digest N from stable contents of the file which survives benign modifications but which is changed by modifications to significant content of the file; embedding encrypted second number computed out of said digest N into the file with a mark on the file; and wherein said stable contents of the image is determined by the steps: constructing a class of images having statistical properties which are a representative subset of possible benign modifications to the image; and decomposing the image with said statistical properties into two components, a first of said two components comprising said benign modifications to said image and a second of said two components, being orthogonal to the first, comprising said stable contents.

Claim 20

Dependent Claim 26 words

A method for marking a file as recited in claim 16 wherein said encrypted second number is computed using the private part of a digital signature.

Claim 21

Dependent Claim 24 words

A method for marking a file as recited in claim 16 wherein said benign modifications comprise printing the file and subsequently scanning the file.

Claim 22

Dependent Claim 21 words

A method for marking a file as recited in claim 16 wherein said benign modifications comprise compressing and decompressing the file.

Claim 23

Dependent Claim 17 words

A method for marking a file as recited in claim 16 wherein said file is an image.

Claim 24

Dependent Claim 23 words

A method for marking a file as recited in claim 16 wherein said step of decomposing the image comprises principle component analysis (PCA).

Claim 25

Dependent Claim 21 words

A method for marking a file as recited in claim 16 wherein said step of decomposing the image comprises Wold's decomposition.

Claim 26

Dependent Claim 53 words

A method for marking an image as recited in claim 16 wherein authenticating said file comprises the steps of: extracting said digest N from said image; reconstructing a new digest N′ from said image; and comparing said digest N to said new digest N′, wherein a match indicates that said image is authentic.

Claim 27

Dependent Claim 32 words

A method for marking an image as recited in claim 23 wherein said match comprises at least 90% of bits in said digest N and said new digest N′ being the same.

Claim 28

Dependent Claim 99 words

A method for marking an image as recited in claim 23 wherein authenticating said file comprises the steps of: extracting said digest N from said image; reconstructing a new digest N′ from said image; decomposing said digest N into a collection of DCT coefficients N(1), N(2) . . . N(I); decomposing said new digest N′ into a collection of DCT coefficients N′(1), N′(2) . . . N′(I); authenticating said image when N(I) matches N′(I) if the distance between N(I) and a quantization interval Q(I) is small with the property that every member of Q(I) will be mapped into N′(I).

Claim 29

Dependent Claim 53 words

A method for watermarking an image as recited in claim 16 further comprising the steps: considering said digest N as a q-ary word; letting C be a q-ary error correcting code having a same length as said q-ary word; changing said digest N be a codeword in C closest to said q-ary word.

Claim 30

Independent Claim 81 words

A method for marking an image where benign modifications to the file still render the file authenticatible, comprising the steps of: cutting an image into squares of n by n pixels; computing transform coefficients for each of said squares; quantizing said transform coefficients according to a quantization table to produce quantized coefficients; encrypting said quantized coefficients with a secret key; hashing said encrypted quantized coefficients to obtain a digest N; and embedding encrypted version of said digest N in said image.

Claim 31

Independent Claim 81 words

A method for marking an image where benign modifications to the file still render the file authenticatible, comprising the steps of: cutting an image into squares of n by n pixels; computing transform coefficients for each of said squares; quantizing said transform coefficients according to a quantization table to produce quantized coefficients; encrypting said quantized coefficients with a secret key; hashing said encrypted quantized coefficients to obtain a digest N; and embedding encrypted version of said digest N in said image.

Claim 32

Dependent Claim 21 words

A method for marking an image as recited in claim 27 wherein said transform coefficients comprise discrete cosine transform (DCT) coefficients.

Annuity Stage	Window Opens	Payment Deadline	Surcharge Period Until	Estimated Fee	Status
3.5 Years (Year 4)	2007-12-21	2008-06-21	2008-12-21	$2000.00	Window Closed / Lapsed
7.5 Years (Year 8)	2011-12-21	2012-06-21	2012-12-21	$3760.00	Window Closed / Lapsed
11.5 Years (Year 12)	2015-12-21	2016-06-21	2016-12-21	$7700.00	Window Closed / Lapsed

Semi-fragile watermarks

Abstract

Inventors

Bibliographic Details

Claims Analysis

References Cited (Backward Citations)

USPTO Annuity & Maintenance Windows

Assigned Categories & Tags

Assign Category Tag