Title of Invention

METHOD FOR INSERTING A WATERMARK INTO DATA REPRESENTING A CONTENT TO BE PROTECTED AND DEVICE FOR INSERTING A WATERMARK INTO DATA.

Abstract Method for inserting a watermark into data (x) representing a content to be protected, comprising the steps consisting in: a) generating, from said data (x), a modulation sequence (m) representing the maximum amount of noise which can be added to said data; b) supplying a pseudo random noise sequence (v) to the input of a filter with predefined impulse response (h); c) multiplying said filtered pseudo noise sequence (w) with said modulation sequence; and
Full Text Digital data watermarking system using novel watermark
insertion and detection methods
The present invention relates to the field of
watermarking of digital data. It relates more1
particularly to a system for watermarking data using
novel watermark insertion and detection methods as well
as to devices for implementing these methods.
Recent methods for protecting against the
illicit copying of digital data use the principle of
data watermarking which consists in inserting a marking
item, commonly referred to as a "watermark", into a
multimedia content (still image, video, sound, etc.) in
an imperceptible manner. The watermark can for example
be a signal indicating that the content may not be
copied or any other item allowing the supplier of the
multimedia content to detect illegal copies.
In order to play its role perfectly, the
watermark must be robust to transformations of the.
watermarked content, whether these transformations be
made intentionally by a pirate who wishes to erase the
watermark, or whether they result from distortions
which occurred during the transmission of the signal
containing the watermarked data.
Various data watermarking techniques are known
from the prior art. Reference may in particular be made
to the documents EP-A-0 828 372, EP-A-0 840 513, WO-A-
98/03014 or WO-A-98/54897 which describe methods for
inserting watermarks into data to be protected and
methods for detecting the presence of such watermarks
in the data,
A scheme which is generally used to describe
the principle of data watermarking is that of Figure 1.
A first part 1 relates to the insertion of a hidden
item W (the watermark) into a content to be protected
C. This results in a watermarked content CT. Part 2
relates to the detection of the presence of the item W
in the content received CT. An additional datum K is
also necessary in the process for inserting and
detecting the watermark. This datum K which must be
shared in a secret manner by the device for inserting
and detecting the watermark is referred to as the key
by analogy with so-called symmetric or private-key
cryptography systems.
For example, a known watermarking technique
consists in adding a pseudo random noise sequence to
data which are to be watermarked. The detection process
is carried out, in this case, by performing a
correlation calculation: the data received are declared
watermarked if the correlation with the reference
pseudo noise sequence (used for the insertion of the
watermark) is greater than a given threshold. In this
example, the reference pseudo noise sequence
constitutes the key K of the data watermarking scheme
of Figure 1.
The problem with this scheme is that each
entity capable of detecting the watermark must share
the same key K as the entity which inserted the
watermark. In this case, the entity capable of
detecting the watermark can furthermore delete it or
modify it, thereby doing away with all the benefit of
the initial watermarking of the data. Consequently, a
supplier of content protected by watermarking should
not communicate his key K, which served for the
insertion of the watermark, other than in a secret
manner to trusted entities. This considerably limits
the possibilities of using data watermarking in
numerous fields.
In particular, in the field of consumer
electronic appliances, it is well known that it is
almost impossible, at any event at reasonable cost, to
store secret parameters in an appliance or in software
contained in such an appliance. Smart cards, which are
regarded as the only pieces of equipment allowing the
secure storage of a secret parameter, are not
themselves powerful enough to perform the calculations
connected with a watermark detection process.
In the example described above where the
watermarking is carried out by adding a pseudo random
noise sequence to the data which are to be watermarked,
even if the reference pseudo noise sequence is stored
secretly in the watermark detection device, it has been
demonstrated that a pirate can theoretically discover
the reference sequence and thus delete the watermark
from the data by observing the output from the detector
as a function of a large number of different input
signals.
The invention aims to solve the aforesaid
problems.
To this end, the invention relates to a method
for inserting a watermark into data representing a
content to be protected. According to the invention,
the method comprises the steps consisting in:
a) supplying a pseudo random noise sequence to
the input of a filter with predefined impulse response;
and
b) adding the filtered pseudo noise sequence to
the data.
According to a preferred aspect of the.
invention, the method furthermore comprises the steps
consisting in:
c) performing a pseudo random interleaving of
the data before step b); and
d) performing an inverse interleaving after
step b) so as to obtain the watermarked data.
The invention also relates to a method for
detecting a watermark in data representing a content
received, characterized in that it comprises the steps
consisting in:
i) performing a spectral analysis of the data;
and
ii) deducing therefrom whether the data include
a pseudo noise sequence which has been filtered by a
filter with predefined spectral response.
According to another preferred aspect of the
invention, a pseudo random interleaving of the data
received, which is identical to the interleaving-
performed in step c) above, is performed before step
i)
The invention also relates to a system for
watermarking data using a watermark insertion method
and a watermark detection method as those above.
According to the invention, a first series of
parametersr the private keyr is used for the insertion
of the watermark and a second series of parameters, the
public key, is used for the detection of the watermark,
so that:-
- knowledge of the public key does not make it
possible to know the private key; and
- knowledge of the method of detection and of
the public key does not make it possible to delete or
modify the watermark.
The invention also relates to a device for
inserting a watermark into data representing a content
to be protected. According to the invention, the device
comprises:
~ means for generating a pseudo random noise
sequence;
- filtering means having a predefined impulse
response and which are adapted for receiving the pseudo
noise sequence and for supplying a filtered pseudo
noise sequencer and
- means for adding the filtered pseudo noise
sequence to the data.
According to a preferred embodiment of the
invention, the device furthermore comprises:
- first means of pseudo random interleaving of
the data representative of the content to be protected,
so as to supply interleaved data, the interleaved data
being supplied to the addition means so as to be added
to the filtered pseudo noise sequence; and
- means of inverse interleaving of the first
interleaving means, linked to the output of the said
addition means so as to supply the watermarked data.
According to a particular embodiment of the
invention, the device comprises:
- means for transforming the content to be
protected into data representative of the content;
- means for generatinq a modulation sequence
indicative of the maximum amount of noise which can be
added to the data;
- first means of pseudo random interleaving of
the data representative of the content to be protected
so as to supply interleaved data;
- second means of pseudo random interleaving,
which are identical to the first adapted for receiving
the modulation sequence so as to supply an interleaved
modulation sequence;
- multiplication means adapted for receiving,
on the one hand the interleaved modulation sequence,
and on the other hand the filtered pseudo noise
sequence, so as to supply the watermark;
- means of addition of the interleaved data and
of the watermark, the output of the addition means
being linked to:
- means of inverse interleaving of the first
and second interleaving means so as to supply the
watermarked data; and
- means of inverse transformation of the.
watermarked data into a marked content.
The invention also relates to a device for
detecting a watermark in data representing a content
received-According to the invention, the device
comprises:
- means for estimating the power spectral
density of the data; and
- means of likelihood testing of hypotheses so
as to estimate whether the data include a pseudo noise
sequence which has been filtered by a filter with
predefined spectral response.
According to a particular embodiment, the
device furthermore comprises:
- means of pseudo random interleaving of the
data representing the content received, which are
adapted for performing the game interleaving as the.
first interleaving means of the insertion device, the
interleaved data being supplied to the means for
estimating the power spectral density.
According to another particular embodiment, the
device furthermore comprises:
- means for transforming the content received
into data representative of the content, the
transforming means being adapted for performing the
same transformation as the transforming means of the
insertion device.
Other characteristics and advantages of the
invention will become apparent on reading the following
description of a particular embodiment, which is non-
limiting, of the invention given with reference to the
appended figures, among which:
- Figure 1, described previously, illustrates a
known scheme for watermarking digital data;
- Figure 2 schematically represents a watermark
insertion device according to the invention;
- Figure 3 schematically represents a watermark
detection device according to the invention;
- Figure 4 illustrates a novel scheme for
watermarking digital data according to the invention.
Represented schematically in Figure 2 is a
device according to the invention for inserting a
watermark into a signal representative of a content to
be protected. This signal can in particular be a
digital video or audio signal or else a signal
representing a still image such as a photograph or a
computer-calculated synthetic image, or more generally,
any signal representing a multimedia content.
Firstly, the content to be protected is
transformed by a transformation module 10 into a
sequence of digital data x = {xn}, n lying between 1
and N. For example, if the content to be protected is
an image comprising N pixels, the coefficients xn can
correspond to the luminance of each pixel of the image.
These may also be coefficients of a Discrete Fourier
Transform of the signal representing the content to be
protected, or else coefficients of a Fourier-Mellin
Transform or coefficients of a wavelet decomposition
when the content to be protected is a still image.
The data sequence x representing the content to
be protected is transmitted on the one hand to a module
HPM 12 which outputs a modulation sequence m = {mn}, Vn
e [1-N], The module HPM calculates this modulation
sequence as a function of algorithms based on human
perception models, such as Sarnoff's model of the eye.
This sequence m = {mn} represents the maximum amount of
noise which can be added to each coefficient xn without
perceptible loss of quality.
According to one aspect of the invention, the.
data sequence x is transmitted moreover to an
interleaver 20/ which performs a random permutation p
of the coefficients xn so as to supply a sequence of
interleaved coefficients x = {xP(n)}. The purpose of
this interleaving of the data sequence x will be
explained subsequently.
The modulation sequence m is also transmitted
to an interleaver 14 which performs the same
permutation p of the coefficients mn as that performed
by the interleaver 20 so as to output an interleaved
modulation sequence m = {mp(n))}.
In order to constitute the watermark which will
be inserted into the data sequence x representing the
content to be protected, a pseudo random noise
generator (not represented} firstly supplies a pseudo
noise sequence v = {vn}, "n Î [1..N], with Gaussian
distribution. This pseudo noise sequence v is
transmitted to the input of a filter 16/ of Linear Time
Invariant (LTD type, whose impulse response is:
h = {hn}, "n Î [1..L] where L is an integer
corresponding to the length of the filter;
and whose spectral response is H(f), H(f) being
the Fourier Transform of h;
At the output of the filter 16 one obtains a
filtered pseudo noise sequence w = {wn}, Vn e [1..N]
satisfying the following equation (1):
4.In which Ä represents the convolution product.
From this may be deduced, from the interference
theorem, the following two equations (2) and (3):
in which jw(r) and jw(r) respectively represent
the auto-correlation functions of w and of v; and
in which jw(f) and jw(f) respectively
represent the power spectral densities of jw (r) and
jw(r) , that is to say their Fourier Transforms.
Since v is a pseudo random noise sequence with
Gaussian distribution, its spectrum, that is to say the
function jw(f), has a substantially flat shape. On the
other hand, once this sequence v is filtered by the
filter 16, the resulting sequence w exhibits a spectrum
jw (f} which is no longer flat on account of the term
|H(f) l2. It is also important to note, so as to
comprehend the rest of the invention, that knowledge of
|H{f) l2 (and by the same token, knowledge of the
modulus of H(f):lH(f)|) does not make it possible to
retrieve H(f) (and hence h) since there is an
uncertainty with regard to the phase of H(f).
Returning to Figure 2, the filtered pseudo
noise sequence w is multiplied (multiplier 18) by the
interleaved modulation sequence m and the resulting
sequence, which constitutes the watermark, is added
(adder 22} to the sequence of interleaved data x.
The output sequence from the adder 22 is
denoted y = {yP(n)} and satisfies the following
equations (4) and (5):
The power spectral density of the sequence of
watermarked interleaved data y is qiven by the
following equations (6) and (7):
In equation (7), md and aj respectively
represent the mean and the standard deviation of the
sequence j = {jn) with j e {x, m, v}r d(f) corresponds
Co the Dirac pulse and the expression is
equal to a constant.
The sequence of watermarked interleaved data y
is then transmitted to an inverse interleaver 24 which
performs the operation inverse to the permutation p
performed by the interleavers 20 and 14 so as to supply
a sequence of watermarked data y = {yn} whose
coefficients are in the same order as the initial order
of the data x = {xn}.
A transformation inverse to that performed by
the transformation module 10 is then performed by the
module 26 so as to obtain the marked content (or
watermarked content) which is thus protected against
illicit copying without the watermark being perceptible
within the content.
We shall now describe, in conjunction with
Figure 3, a device tor detecting a watermark in a
received content when this watermark has been inserted
into a content to be protected by a device such as that
of Figure 2.
The principle of the detection is based on the
spectral analysis of the signal received.
The signal received is representative of the
received content for which one will seek to determine
whether or not it is watermarked. This content is of
the same type as the content to be protected described
previously. In the example which follows, it will be
assumed that the content received is an image
containing N pixels.
The content received is firstly transmitted to
a transformation module 30 which performs the same
transformation operation as the module 10 of the
watermark insertion device of Figure 2 so as to supply
a data sequence r = {rn}, Vn e [1..N] representing the
content received. In our example, it is assumed that
the luminances rn of the pixels of the image received
are obtained as output from the transformation module
30.
If the content received were to correspond
exactly to the watermarked content emanating from the
device of Figure 2, that is to say if no transformation
or distortion of the signal had taken place during
transmission between the watermark insertion device and
the detection device, then one would have:
r = {rn} = y = {yn}
In practice, this is not always the case since
the signal sometimes undergoes transformations during
its. transmission.
Since the watermark has been inserted, in the
device of Figure 2, into a sequence of interleaved data.
x, the data sequence r will, in order to detect the
possible presence of a watermark in the content
received, be transmitted to an interleaver 32
performing the same permutation p of the coefficients
ra as that performed by the interleavers 20 and 14 of
Figure 2.
A sequence of interleaved data r = {rp(n)} is
obtained as output from the interleaver 32.
It was seen previously that when the watermark
inserted is a pseudo noise sequence filtered by a
filter with impulse response h and with spectral
response H(f), the power spectral density of the
(interleaved) data obtained y is expressed by relations
(6) and (7).
The purpose of the interleaving of the data
sequence x and of the modulation sequence m will now be
apparent. Indeed, if the data sequence x represents the
pixels of an image/ its spectral density has a very
Structured shape with very large amplitude differences.
The role of the interleaving of the data is to sever
the statistical coherence of this sequence so that the
spectral density of the sequence of interleaved data x
has a substantially flat shape, such as that of a
pseudo noise sequence with Gaussian distribution.
Thus, if a watermark consisting of a pseudo
noise sequence filtered by a filter with spectral
response H(f) is added to this interleaved sequence, a
data sequence is obtained whose power spectral density
can be expressed by relation (7) in which the
significant term |H(f)I2 can be detected.
The principle of the detection will therefore
be based on the spectral analysis of the sequence r and
on a Maximum Likelihood Ratio Hypothesis test (MLR
Hypothesis test), the hypothesis tested being the
following: if the sequence of interleaved data r
contains noise, is it noise which has been filtered by
a filter whose spectral response has a modulus similar
to lH(f)(? If the response is yes, one will deduce
from this that the noise present in the sequence r is a
watermark and, in the contrary case, one will conclude
from this that the content received was not
watermarked.
In practice, this analysis is based on
calculations relating to spectral analysis and the
likelihood testing of hypotheses which are described in
detail in the work by K. Dzhaparidze, "Parameter
Estimation and Hypothesis Testing in Spectral Analysis
of Stationary Time Series"/ Springer Series in
Statistics, Springer-Verlag, 1986, to which reference
may be made for further details.
Returning to Figure 3, the sequence of received
interleaved data r is transmitted to a module 34
performinq a Periodogram calculus. This calculus is
aimed at estimating the power spectral density of the
sequence r. A quantity IN(f) given by the following
relation (8)
is obtained at output.
This quantity is then transmitted to a module
36 performing a MLR Hypothesis test so as to determine
whether the content received is watermarked (output
response "Y") or not (output response "N").
The module 36 tests the likelihood of two
hypotheses:
- according to the first hypothesis Go, the
content received is not watermarked/ hence the spectral
density of the sequence r is substantially flat and can
be estimated via the following relation (9):
- according to the second hypothesis G1, the
content received is watermarked and the spectral
density of the sequence r can be estimated via the
following relation (10):
in which C is a constant and av is equal to 1
(one preferably chooses the pseudo noise sequence v at
the level of the insertion device so that av is equal
to 1, but one may equally choose other values),
Furthermore, fim is normed at the level of the insertion
device and equals for example 3,
To estimate the likelihood of the hypotheses GQ
and G1, the module 36 calculates two numbers UN,0(r) and
UN,1(r) representing the likelihoods of the hypotheses
Go and Gx according to the following relation (11) :
By then comparing these two numbers, the module
36 deduces from this:
if UN,i(r )>UN,o(r), then the response of the
detector is "Y" signifying that the content received is
watermarked; and
- if UK,i(r) detector is "N" signifying that the content received is
not watermarked.
It is also possible, in a preferential manner,
to calculate the difference (UN,l ( r } -UN,0 (r }) and to
perform the above comparisons only if this difference
is greater than a predetermined threshold, this being
so as to guarantee better exactness of detection.
The watermark insertion and detection methods
just described with reference to Figures 2 and 3 make
it possible to produce a novel watermarking system
which is illustrated by Figure 4. In this novel system
and according to a preferred aspect of the invention, a
parameter which is referred to as the "private key" KPRi
is used for the insertion (100) of a watermark W into a
content C, whereas another parameter which is referred
to as the "public key" KPUB is used for the detection
(200) of a watermark in a content received CT. The
terms "private key" and "public key" are used by
analogy with public key crytographic systems; It will
be noted that here the watermark W is binary, that is
to say that, either the content C is watermarked, or it
is not, but W does not contain any item of its own.
In the embodiment described above, the private
key Kpm is formed by the pseudo random noise sequence v
as well as by the impulse response h of the filter 16
(Fig. 2) . The sequences v = {vn} and h = {hn} are in
effect indispensable to the calculation of the sequence
w = {wn} which is itself, after having been multiplied
fey the interleaved modulation sequence m, inserted
into the data representing the content to be protected.
The public key used to detect the watermark in
the content received is for its part formed from the
modulus of the spectral response of the filter 16
IH(£)' Indeed/ in the spectral analysis calculations
performed (modules 34 and 36 of Figure 3) to detect the
presence of a watermark in a content received CT, only
the knowledge of lH(f) I is necessary. In particular, it
is not necessary to know v and h (the private key) to
perform the detection of the watermark. In actual fact,
as was seen earlier in the description, the knowledge
of IH(f)) does not suffice to know H(f) and hence h.
A system is therefore obtained in which
knowledge of the public key does not make it possible
to deduce the private key from this. Also, not knowing
the private key, it is impossible for the device
performing the detection of the watermark to delete it
or to modify it. The detection can therefore be
performed in a non-secure environment with no risk of
the watermark being erased.
1. Method for inserting a watermark into data (x) representing a
content to be protected, comprising the steps consisting in:
a) generating, from said data (x), a modulation sequence (m)
representing the maximum amount of noise which can be added
to said data;
b) supplying a pseudo random noise sequence (v) to the input of a
filter with predefined impulse response (h);
c) multiplying said filtered pseudo noise sequence (w) with said
modulation sequence; and
d) adding the filtered pseudo noise sequence (w) multiplied by the
modulation sequence to said data.
2. Method according to Claim 1 furthermore comprising the steps
consisting in:
e) performing a pseudo-random interleaving (p) of the modulation
sequence (m) before step c);
f) performing the same pseudo random interleaving (p) of the data
(x) before step d); and
g) performing an inverse interleaving after step d) so as to obtain
the watermarked data.
3. Method for detecting a watermark in data ( R ) representing a
content received, wherein it comprises the steps consisting in:
i) performing a spectral analysis of said data;
ii) estimating therefrom whether said data include a pseudo noise
sequence which has been filtered by a filter with predefined
spectral response (H(f)); and
iii) deducing from said estimation the presence of said watermark.
4. Method according to Claim 3 for detecting a watermark in data
( r ) representing a content received, the watermark being adapted to be
inserted in data to be protected (x) after performing a pseudo random
interleaving (p) of said data to be protected, wherein it furthermore
comprises a step consisting in:
iv) performing, before step i), the same pseudo random
interleaving (p) of the data(r) received.
5. Device for inserting a watermark into data (x) representing a
content to be protected, comprising:
- means for generating a pseudo random noise sequence (v);
- means (12) for generating, from said data (x), a modulation
sequence (m) indicative of the maximum amount of noise which
can be added to said data;
wherein it furthermore comprises
filtering means (16) having a predefined impulse response (h)
and which are adapted for receiving said pseudo noise sequence
(v) and for supplying a filtered pseudo noise sequence (w);
- multiplication means (18) for multiplying said filtered pseudo
noise sequence (w) with said modulation sequence; and
- means (22) for adding the filtered pseudo noise sequence
(w) multiplied by the modulation sequence (m) to said data (x).
6. Device according to Claim 5 wherein it furthermore comprises:
first means (20) of pseudo random interleaving of the data (x)
representative of the content to be protected so as to supply
interleaved data ( x ),
- second means (14) of pseudo random interleaving, which are
identical to the first means (20) and adapted for receiving said
modulation sequence (m) so as to supply an interleaved modula-
tion sequence ( m ), said interleaved modulation sequence being
supplied to the multiplication means (18) for multiplication with
said filtered pseudo noise sequence (w);
- wherein the interleaved data ( x ) are supplied to the addition
means (22) so as to be added to the filtered pseudo noise
sequence (w) multiplied by the interleaved modulation sequence
(m); and
means (24) of inverse interleaving of said first (20) interleaving
means, linked to the output of said addition means (22) so as to
supply the watermarked data.
7. Device according to Claim 6 further comprising:
- means (10) for transforming the content to be protected into data
(x) representative of said content; and
- means (26) of inverse transformation of the watermarked data
into a marked content.
8. Device for detecting a watermark in data ( r ) representing a
content received comprising:
- means (34) for estimating the power spectral density of said data;
means (36) of likelihood testing of hypotheses so as to estimate
whether said data include a pseudo noise sequence which has
been filtered by a filter with predefined spectral response (H(f));
and
means for deducing from said estimation the presence of said
watermark.
9. Device according to Claim 8 adapted for detecting a watermark
inserted by an insertion device, said insertion device comprising means
(20) of pseudo random interleaving of the data (x) to be protected so as
to supply interleaved data ( x") and means for inserting said watermark
in said interleaved data ( x ),
wherein the detection device furthermore comprises means (32)
of pseudo random interleaving of the date ( r ) representing the content
received, which are adapted for performing the same interleaving (p) as
said interleaving means (20) of the insertion device, said interleaved
data (!) being supplied to said means (34) for estimating the power
spectral density.
10. Device according to Claim 9 adapted for detecting a watermark
inserted by an insertion device further comprising means (10) for
transforming the content to be protected into data (x) representative of
said content,
wherein said detection device furthermore comprises means (30)
for transforming the content received into data ( r ) representative of
said content, said transforming means being adapted for performing the
same transformation as the transforming means (10) of the insertion
device.
Method for inserting a watermark into data (x) representing a
content to be protected, comprising the steps consisting in:
a) generating, from said data (x), a modulation sequence (m)
representing the maximum amount of noise which can be added
to said data;
b) supplying a pseudo random noise sequence (v) to the input of a
filter with predefined impulse response (h);
c) multiplying said filtered pseudo noise sequence (w) with said
modulation sequence; and

Documents:

in-pct-2001-1263-kol-granted-abstract.pdf

in-pct-2001-1263-kol-granted-claims.pdf

in-pct-2001-1263-kol-granted-correspondence.pdf

in-pct-2001-1263-kol-granted-description (complete).pdf

in-pct-2001-1263-kol-granted-examination report.pdf

in-pct-2001-1263-kol-granted-form 1.pdf

in-pct-2001-1263-kol-granted-form 18.pdf

in-pct-2001-1263-kol-granted-form 2.pdf

in-pct-2001-1263-kol-granted-form 26.pdf

in-pct-2001-1263-kol-granted-form 3.pdf

in-pct-2001-1263-kol-granted-form 5.pdf

in-pct-2001-1263-kol-granted-priority document.pdf

in-pct-2001-1263-kol-granted-reply to examination report.pdf

in-pct-2001-1263-kol-granted-specification.pdf

in-pct-2001-1263-kol-granted-translated copy of priority document.pdf


Patent Number 224160
Indian Patent Application Number IN/PCT/2001/1263/KOL
PG Journal Number 40/2008
Publication Date 03-Oct-2008
Grant Date 01-Oct-2008
Date of Filing 29-Nov-2001
Name of Patentee THOMSON LICENSING S.A.
Applicant Address 46 QUAI A LE GALLO, F-92100 BOULOGNE-BILLANCOURT
Inventors:
# Inventor's Name Inventor's Address
1 FURON TEDDY 13 RUE DE LA SANTE, F-35000 RENNES
2 DUHAMEL PIERRE 46 RUE BARRAULT, F-75013 PARIS
PCT International Classification Number H04N 1/32
PCT International Application Number PCT/EP00/04053
PCT International Filing date 2000-05-05
PCT Conventions:
# PCT Application Number Date of Convention Priority Country
1 99/07139 1999-06-01 France