Image analysis – Image compression or coding – Transform coding
Reexamination Certificate
2001-06-13
2004-09-07
Patel, Jayanti K. (Department: 2622)
Image analysis
Image compression or coding
Transform coding
C375S240180
Reexamination Certificate
active
06788821
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to a digital information embedding/extracting apparatus and method, and a medium including a program for executing the method recorded thereon. More particularly, the present invention relates to an apparatus and method for embedding digital data such as copyright information (hereinafter, generically called digital information) in a digital image signal and extracting the digital information from the digital image signal, and a medium including a program for executing the method recorded thereon.
2. Description of the Background Art
In recent years, it has become increasingly popular to provide information via the Internet. In particular, World-Wide Web (WWW) has been frequently used as an information send/receive service system that integrates images, sound, and the like. Once digital information such as images is made public on a network of the Internet, however, it can be easily copied by an indefinite number of users. This raises troubles such as that a third party illegally copies a copyrighted image without permission of the copyright holder and enjoys secondary use. Illegal copying is also a problem for the deployment of business on the Internet using image-base contents. To take measures against illegal copying, it is demanded to establish technology for protecting copyrights of image signals.
As one of the measures against illegal copying, a digital watermark technique is conventionally known, in which digital information is embedded in image data in a form unperceivable to humans. Using this digital watermark technique, if copyrighted digital information is illegally used, the copyright holder can claim the copyright by extracting embedded digital information and verifying the right.
An example of the conventional digital watermark embedding and extracting method is disclosed in Japanese Laid-Open Patent Publication No. 10-191330 (1998-191330) which will be briefly described as follows.
First, the digital watermark embedding method will be described with reference to FIG.
12
.
Data of an original image
1101
is picked up every block
1102
of 8×8 pixels based on normal JPEG or MPEG compression. A discrete cosine transform (DCT) operator
1103
a performs DCT operation for the retrieved data so as to transform the data into frequency components. A quantizer
1104
quantizes the data outputted from the DCT operator
1103
by referring to a quantization table
1105
. A watermark data embedding unit
1107
embeds watermark data
1106
in the data that is outputted from the quantizer
1104
. Specifically, the watermark data that is outputted from the quantizer
1104
. Specifically, the watermark data embedding unit
1107
performs processing which is represented by expression (1) below so as to obtain a new frequency component F(i) including the watermark data
1106
embedded therein.
F
(
i
)=
f
(
i
)+avg(
f
(
i
))×
w
(
i
)×constant (1)
wherein i denotes the element number of a frequency component of the 8×8 pixel block
1002
, w(i) denotes the value of each element of the watermark data
1106
, and avg(f(i)) denotes the average of the absolutes (absolutes values) of three neighboring points of each element, that is, avg(f(i))=(|f(i−1)|+|f(i)|+|f(i+1)|)/3.
The watermark data embedding unit
1107
includes: a partial average calculator
1108
for calculating the average avg(f(i)) of the absolutes (absolute values) of three neighboring points of the data outputted from the quantizer
1104
; a multiplier
1110
for multiplying the output of the partial average calculator
1108
by the watermark data
1106
for each element; a multiplier
1111
for multiplying the output of the multiplier
1110
by a constant &agr; (scaling parameter)
1109
that is used for changing the size of the watermark data
1106
to be embedded for each element; and an adder
1112
for adding the output of the multiplier
1111
to the output of the quantizer
1104
for each element.
An encoder
1113
encodes the frequency component F(i) outputted from the watermark data embedding unit
1107
based on JPEG or MPEG to generate compressed data
1114
including the watermark data embedded therein.
Next, the watermark extracting method will be described with reference to FIG.
13
.
Compressed data
1201
is decoded by a decoder
1202
. The decoded data is subjected to inverse-quantization by an inverse-quantizer
1203
and then inverse-DCT operation by an inverse-DCT operator
1204
according to normal JPEG or MPEG decoding, to provide reproduced image data
1205
.
A watermark data extractor
1206
retrieves each element from the data that is decoded by the decoder
1202
, and calculates F(i)/avg(F(i)) so as t extract data that is presumed to be embedded watermark data, wherein F(i) denotes a decoded frequency component of an 8×8 pixel block and avg(F(i)) denotes the partial average of three neighboring points of the data of each 8×8 pixel block outputted from the decoder
1202
, that is, avg(F(i))=(|f(i−1)|+|f(i)|+|f(i+1)|)/3.
The watermark data extractor
1206
includes: a partial average calculator
1207
for calculating the partial average avg(F(i)) of three neighboring points of the data; and a divider
1208
for dividing the frequency component F(i) outputted from the decoder
1202
by the partial average avg(F(i)) outputted from the partial average calculator
1207
.
An adder
1209
accumulates the extracted data for each 8×8 pixel block outputted from the watermark data extractor
1206
to obtain the sum of extracted data of one screen. Once the sum of extracted data of one screen is obtained, an inner product calculator
1210
calculates an inner product between watermark data
1211
expected to be detected and the sum of extracted data, and outputs a statistical similarity
1212
.
The statistical similarity
1212
is calculated in the following manner. First, watermark data W(i) is calculated from W(i)=F(i)/avg(F(i)), and then the sum WF(i) of the watermark data W(i) of one screen is calculated for each element i. The statistical similarity C between the watermark data w(i) expected to be detected and the sum WF(i) is calculated from C=WF×w/(WFD×wD) using an inner product of vectors, wherein W=(WF(
1
), WF(
2
), . . . , WF(n)), w=(w(
1
), w(
2
), . . . , w(n)), WFD denotes the absolute of vector WF, and wD denotes the absolute of vector w.
When the statistical similarity
1212
exceeds a predetermined value, it is determined that watermark data equivalent to the watermark data
1211
has been embedded in the compressed data
1201
. In this case, if the watermark data
1211
is data indicating “duplication prohibited”, for example, a reproduction machine that incorporates this watermark data extracting scheme can take some measures against duplication of the reproduced image data
1205
, for example.
The above conventional technique has the following problems. During the extraction of watermark data, the frequency component F(i) must be divided by the partial average avg(F(i)) for each element of an 8×8 pixel block. This complicates the operation.
Moreover, during the embedding of watermark data, since the constant &agr; used for changing the size of the watermark data is fixed, embedding depending on each frequency component of the 8×8 pixel block is not possible. Therefore, if the value of the constant &agr; is increased to improve the rate of detection of watermark data, the image quality is deteriorated.
SUMMARY OF THE INVENTION
Therefore, a main object of the present invention is to provide a digital information embedding/extracting apparatus and method where the construction and operation are simpler, the calculation cost is lower, and moreover deterioration in image quality is less observable, compared with conventional ones, by using a frequency compo
Inoue Hisashi
Katsura Takashi
Patel Jayanti K.
Wenderoth , Lind & Ponack, L.L.P.
LandOfFree
Digital information embedding/extracting apparatus and... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Digital information embedding/extracting apparatus and..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Digital information embedding/extracting apparatus and... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3200941