FR2820928A1 - Digital word sequence watermarking authentication having first random sequence correlation measurement criteria calculated/second sequence generated and insertion second signal capacity estimated. - Google Patents

Abstract

The digital signal capacity measurement technique measures the length of the signal as a function of a predetermined error probability. A first random sequence is calculated (E1) as a function of the correlation measurement criteria. A second random sequence (E20) is then generated. The capacity of the digital signal is then estimated (E5) to allow addition of a second sequence.

Description

<Desc / Clms Page number 1>

The present invention relates generally to the insertion of a message such as a secret mark in a digital signal.

 The digital signal considered in the following will more particularly be a digital image signal.

 The insertion of a message envisaged in the context of the invention is part of the technical field of marking (watermarking in English) of digital data which can be interpreted as the insertion of a seal in digital data, allowing by example of authenticating the content of a digital data file. This marking is also called digital tattooing.

 The marking generally involves the modification of coefficients representative of the digital image. This modification is imperceptible to the eye, but can be decoded by an appropriate decoder.

 We are particularly interested in the capacity of the digital image, that is to say the number of marking bits that it is possible to insert into it. The capacity is determined according to the invisibility of the message and the robustness of the insertion. Robustness is defined by the possibility of finding the inserted message even after subsequent processing, such as compression and decompression for example.

 To estimate the capacity of an image, French patent application No. 99 04462 filed by the applicant proposes a method according to which a message is inserted into an image. An attack, such as compression followed by decompression, is then performed. Then we do, so

<Desc / Clms Page number 2>

 recursive, assumptions about capacity. Each hypothesis corresponds to a division into a quaternary tree of the image. For each leaf of the quaternary tree, a correlation measure is compared to a predetermined threshold.

 The number of sheets for which the correlation measure exceeds the threshold gives the capacity of the image.

 After estimating the capacity, a message is inserted in the image, according to the decomposition into a quaternary tree corresponding to the capacity.

 To later retrieve the message, repeat the correlation measurements. It is therefore necessary to know the threshold used.

 The article Capacity of the Watermark-channel: How many bits can be hidden within a digital image? published in Proceeding of SPIE vol. 3657, Security and watermarking of multimedia content, Electronic imaging'99, San Jose, CA, January 1999, offers a theoretical method for estimating the capacity of an image.

 This method gives good results, provided that the theoretical model of the image is reliable. In addition, if one wishes to take into account an attack on this image, it is then necessary to model this attack.

This type of model is very complex to implement.

 The present invention aims to remedy the drawbacks of the prior art, by providing a method and a device for estimating the capacity of a digital signal which make it possible to obtain this estimation more simply than according to the prior art.

 To this end, the invention proposes a method for estimating the capacity of a digital signal for the insertion of a message, as a function of a predetermined probability of error on the detection of any message of equal length. to the capacity, characterized in that it comprises the steps of: - calculation of a first pseudo-random sequence according to a criterion on a measure of correlation of this first sequence with data representative of the signal,

<Desc / Clms Page number 3>

 - calculation of a second pseudo-random sequence as a function of the first sequence and the probability of error, - insertion of the second sequence in the digital signal, - estimation of the capacity of the digital signal in which the second sequence was inserted.

 The invention also relates to a method for estimating the capacity of a digital signal for the insertion of a message, as a function of a predetermined probability of error on the detection of any message of length equal to the capacity. , characterized in that it comprises the steps of: - calculation of a first pseudo-random sequence according to a criterion on a measure of correlation of this first sequence with data representative of the signal, - calculation of a second pseudo-random sequence as a function of the first sequence and the probability of error, - insertion of the second sequence into the digital signal, - application of distortion on the digital signal in which the second sequence was inserted, - estimation the capacity of the digital signal into which the second sequence was inserted.

 Correlatively, the invention proposes a device for estimating the capacity of a digital signal for the insertion of a message, as a function of a predetermined probability of error on the detection of any message of length equal to the capacity, characterized in that it comprises: - means for calculating a first pseudo-random sequence as a function of a criterion on a measure of correlation of this first sequence with data representative of the signal, - means for calculating of a second pseudo-random sequence as a function of the first sequence and of the probability of error, - means for inserting the second sequence into the digital signal, - means for estimating the capacity of the digital signal in which the second sequence has been inserted.

<Desc / Clms Page number 4>

 The invention also relates to a device for estimating the capacity of a digital signal for inserting a message, as a function of a predetermined probability of error on the detection of any message of length equal to the capacity. , characterized in that it comprises: - means for calculating a first pseudo-random sequence as a function of a criterion on a measure of correlation of this first sequence with data representative of the signal, - means for calculating d '' a second pseudo-random sequence as a function of the first sequence and the probability of error, - means for inserting the second sequence into the digital signal, - means for applying distortion to the digital signal in which the second sequence has been inserted, means for estimating the capacity of the digital signal in which the second sequence has been inserted.

 Thanks to the invention, it is possible to estimate the capacity of a digital signal in a simpler way than according to the prior art.

 According to a preferred characteristic, the criterion on a correlation measurement is that the value of the correlation of the first sequence with the data representative of the signal is negative and of absolute value as large as possible. Alternatively, this criterion is that the value of the correlation of the first sequence with the data representative of the signal is positive and of absolute value as large as possible.

 The first sequence is thus the worst or alternatively the best sequence according to a criterion on its correlation with the signal.

 According to a preferred characteristic, the calculation of the first sequence comprises, for each coefficient of the first sequence, the steps of: - comparison of the value of a coefficient representative of the digital signal with a comparison value,

<Desc / Clms Page number 5>

 - determination of the binary value of the coefficient of the first sequence according to the result of the comparison.

 Thus, the first sequence is determined from simple calculations.

 According to a preferred characteristic, for each coefficient representative of the digital signal, the comparison value is an average calculated over a region containing said coefficient.

 According to a preferred characteristic, at least part of the signal is filtered by high-pass filtering before the calculation of the first sequence.

 This filtering is analogous to a filtering which can be used during the subsequent extraction of a message in order to improve the detection of the message.

 According to a preferred characteristic, the calculation of the second sequence comprises the steps of: - determining a number of coefficients to be inverted in the first sequence, as a function of the probability of error, - inverting the previously determined number of coefficients of the first sequence to form the second sequence.

 The second sequence is thus determined from simple and quick calculations to be implemented.

 According to a preferred characteristic, the capacity estimation comprises the steps of: - calculation of a correlation measure between the second sequence and the digital signal into which the second sequence has been inserted according to a predetermined insertion law, - determination of the maximum length of a message corresponding to a correct detection of this message, inserted into the digital signal by modulation of the second sequence and according to the predetermined insertion law, according to a criterion on the correlation.

<Desc / Clms Page number 6>

 These calculations make it possible to quickly obtain the capacity of the digital signal.

 According to a preferred characteristic, the criterion on the correlation is that it corresponds to a correct detection of a predetermined proportion of the coefficients of the inserted message.

 This criterion makes it possible to adjust a desired level of reliability of detection.

 According to a preferred characteristic, the steps of the capacity estimation are repeated from a maximum capacity to an estimated capacity. Alternatively, the steps of the capacity estimation are repeated from a maximum capacity to an estimated capacity, and in that the maximum capacity is modified until convergence between the maximum capacity and the estimated capacity.

 According to this variant, the capacity estimate is finer.

 The device according to the invention comprises means for implementing the above characteristics and has similar advantages.

 The invention also relates to a digital device including the device according to the invention, or means for implementing the method according to the invention. This digital device is for example a digital camera, a digital camcorder or a scanner. The advantages of the device and of the digital apparatus are identical to those previously exposed.

The invention can be implemented by a computer program.
An information storage means, readable by a computer or by a microprocessor, integrated or not in the device, possibly removable, stores the program implementing the method according to the invention.

 The characteristics and advantages of the present invention will appear more clearly on reading a preferred embodiment illustrated by the attached drawings, in which:

<Desc / Clms Page number 7>

- la figure 1 représente un mode de réalisation de dispositif d'estimation de capacité selon l'invention, - la figure 2 représente un mode de réalisation de dispositif selon l'invention, -la figure 3 est un mode de réalisation d'insertion de message dans des données, - la figure 4 représente une image numérique segmentée dans laquelle des bits d'un message sont à insérer, - la figure 5 représente un mode de réalisation de procédé d'estimation de capacité selon l'invention, - la figure 6 est un mode de réalisation de calcul de première séquence, inclus dans le procédé de la figure 5, - la figure 7 est un mode de réalisation de calcul de seconde séquence, inclus dans le procédé de la figure 5, - la figure 8 est un premier mode de réalisation d'estimation de capacité, inclus dans le procédé de la figure 5,

- la figure 9 est un second mode de réalisation d'estimation de capacité, inclus dans le procédé de la figure 5.

- Figure 1 shows an embodiment of capacity estimation device according to the invention, - Figure 2 shows an embodiment of device according to the invention, - Figure 3 is an embodiment of insertion of message in data, - Figure 4 represents a segmented digital image into which bits of a message are to be inserted, - Figure 5 represents an embodiment of capacity estimation method according to the invention, - Figure 6 is an embodiment of first sequence calculation, included in the method of FIG. 5, - FIG. 7 is an embodiment of second sequence calculation, included in the method of FIG. 5, - FIG. 8 is a first embodiment of capacity estimation, included in the method of FIG. 5,

FIG. 9 is a second embodiment of capacity estimation, included in the method of FIG. 5.

 According to the embodiment chosen and shown in FIG. 1, a device for estimating the capacity of a digital signal comprises a memory 1 adapted to store the digital signal whose capacity is to be estimated. The digital signal considered is more particularly a digital image IM.

 The memory 1 is connected to a circuit 2 for calculating the first pseudo-random sequence Ww.

 The circuit 2 is connected to a circuit 3 for calculating a second pseudo-random sequence W. The second pseudo-random sequence is calculated as a function of the first sequence and of a probability of error Pe defined by a user.

 Circuit 3 is connected to circuit 4 for inserting the second sequence W into the image under consideration.

<Desc / Clms Page number 8>

 Circuit 4 is connected to a drive circuit 5 which performs processing on the image supplied by circuit 4. The drive is for example compression followed by decompression of the image. It should be noted that the attack is optional.

The circuit 5 is connected to a circuit 6 for estimating the capacity of the image.
The operation of the device according to the invention will be detailed below.

 As shown in FIG. 2, a device implementing the invention is for example a microcomputer 10 connected to different peripherals, for example a digital camera 107 (or a scanner, or any means of acquisition or storage of image) connected to a graphics card and providing information to be processed according to the invention.

 The device 10 includes a communication interface 112 connected to a network 113 capable of transmitting digital data to be processed or, conversely, of transmitting data processed by the device. The device 10 also includes a storage means 108 such as for example a hard disk. It also includes a disc drive 109. This disc 110 can be a floppy disk, a CD-ROM, or a DVD-ROM, for example. The disk 110 like the disk 108 can contain data processed according to the invention as well as the program or programs implementing the invention which, once read by the device 10, will be stored in the hard disk 108. According to a variant, the program allowing the device to implement the invention may be stored in read-only memory 102 (called ROM in the drawing). In the second variant, the program can be received to be stored in an identical manner to that described previously via the communication network 113.

 The device 10 is connected to a microphone 111. The data to be processed according to the invention will in this case be an audio signal.

 This same device has a screen 104 making it possible to view the data to be processed or to serve as an interface with the user who can thus

<Desc / Clms Page number 9>

 configure certain processing modes, using the keyboard 114 or any other means (mouse for example).

 The central unit 100 (called CPU in the drawing) executes the instructions relating to the implementation of the invention, instructions stored in the read-only memory 102 or in the other storage elements. During power-up, the processing programs stored in a non-volatile memory, for example the ROM 102, are transferred to the RAM RAM 103 which will then contain the executable code of the invention as well as registers for storing the variables necessary for the implementation of the invention.

 More generally, an information storage means, readable by a computer or by a microprocessor, integrated or not in the device, possibly removable, stores a program implementing the method according to the invention.

 The communication bus 101 allows communication between the different elements included in the microcomputer 10 or connected to it. The representation of the bus 101 is not limiting and in particular the central unit 100 is capable of communicating instructions to any element of the microcomputer 10 directly or through another element of the microcomputer 10.

 The operation of the device according to the invention will now be described by means of algorithms.

 The insertion of a message into the image is detailed with reference to FIG. 3 in the form of an algorithm comprising steps E100 to E112.

 Note that the capacity of the image depends on the insertion algorithm used. Consequently, the estimation of capacity also depends on the insertion algorithm.

 However, thanks to the invention, the capacity and its estimation are independent of the insertion parameters, which are here the pseudo-random sequence and the message. Thus, it will not be necessary to recalculate the capacity for each pseudo-random sequence used and for each

<Desc / Clms Page number 10>

 message to insert. It should be noted that the pseudo-random sequences used for the insertion have no link with the sequences used during the capacity estimation.

 Step E100 is the segmentation of the data into regions, for example into adjacent blocks. The number and / or size of the regions can be predetermined or adjustable by a user. An example of segmentation is given in Figure 4.

 Generally, the message M to be inserted comprises L symbols, where L is an integer. Each symbol M, with the integer i varying from 1 to L is associated with at least one region in the next step E101. A given region is associated with a single symbol to insert. For the association of symbols with regions, the latter are traversed in a predetermined order.

en huit régions rectangulaires Ri à R8 dans le cadre de l'invention. For example, Figure 4 shows an image that has been cut out.

in eight rectangular regions Ri to R8 in the context of the invention.

A message to insert has three symbols bo, b1 and b2. Each region is assigned a message symbol.

 Referring again to FIG. 3, the next step E102 is an initialization to consider the first symbol M1 to be inserted, as well as the first region in which this symbol is to be inserted.

 In the next step E103 a variable C1 representing the rank of the current symbol is set to the value 1 and a variable C2 is set to the value 0. The variable C2 represents the number of times the current symbol has already been inserted. The variables C1 and C2 are linked to the length of the message M.

The next step E104 is the generation of a key K as a function of an initial key Knit and of the variables C1 and C2. For example,
K = Kin, t + N. C2 + C1
Where N is an integer equal to Lmax + 1 and where Lima is the maximum value that the length of the message can take.

According to a second example,:
K = Kin, t + C2.

 The next step E105 is the generation of a pseudo-random sequence as a function of the key K previously generated.

<Desc / Clms Page number 11>

 The next step E106 is the modulation of the symbol M, by the pseudo-random sequence previously generated, which results in a second pseudo-random sequence.

 The next step E107 is a psycho-visual weighting of the second pseudo-random sequence to ensure its invisibility in the image.

 The pseudo-random sequence thus modified is then added to the current region in the next step E108.

 The next step E109 is a test to determine if the current region is the last for the current symbol. If the answer is negative, it means that at least one region remains in which the current symbol must be inserted. Step E109 is then followed by step E110. In step E110, the following region is considered in which the symbol M is to be inserted, and the variable C2 is incremented by one.

 Step E110 is followed by step E104 previously described.

 When the answer is positive in step E109, this means that the current symbol has been inserted in all the regions associated with it.

 Step E109 is then followed by step E111 which is a test to determine if the current symbol is the last symbol to be inserted. If the answer is negative, this means that at least one symbol remains to be inserted, and this step is followed by step E112 in which the parameter i is incremented by one to consider the next symbol M + 1 and the first region associated with it.

 Step E112 is followed by step E103 previously described.

 When the answer is positive in step E111, this means that all the symbols have been inserted in the image.

 It should be noted that the insertion which has been described is carried out on the pixels of an image. It is however possible to transform this image beforehand, by a reversible transformation. This transformation is for example a transformation into image odelettes or a transformation into discrete cosine, called DCT, by blocks. The transformation performs a decomposition of the digital image, and provides a set of coefficients. If this transformation is a transformation into odelettes, these

<Desc / Clms Page number 12>

 coefficients carry spatio-temporal information. If this transformation is a block DCT transformation, the coefficients are spectral coefficients.

 The algorithm of FIG. 5 represents the general operation of the capacity estimation device according to the invention and includes steps E1 to E5.

 This algorithm can be stored in whole or in part in any information storage means capable of cooperating with the microprocessor.

This storage means can be read by a computer or by a microprocessor.

This storage means is integrated or not to the device, and can be removable.

For example, it may include a magnetic tape, a floppy disk or a CD-ROM (compact disk with frozen memory).

 The method aims to estimate the capacity of a digital signal, here a digital image, according to a given distortion and a given probability of error.

 By distortion, we consider here two types of distortion. The first type is caused by the image itself, which can be considered as noise in relation to the message inserted therein, which constitutes a useful signal.

The energy of the message being much weaker than that of the image, this influences the capacity of the image when we want to ensure good detection of the inserted message.

 The second type of distortion is due to an attack carried out after insertion of the message into the image, for example lossy compression.

This type of processing leads to a deterioration of the image and consequently of the message inserted in it.

 The probability of error is the probability of not correctly extracting a message whose size is equal to the capacity and which has been inserted into the digital image.

 Step E1 is the calculation of a first pseudo-random binary sequence from an image. The first sequence can be either the worst sequence Ww or the best sequence. We will consider more

<Desc / Clms Page number 13>

 especially the first case. The calculation of this sequence will be explained below.

 The next step E2 is the calculation of a second pseudo-random sequence W, from the first sequence and of a probability of error Pe.

The W sequence is the same size as the original image. The sequence W can be considered as the result of the modulation of this sequence with a message of any size of which all the bits are equal to one. Step E2 will be detailed below.

 The next step E3 is the insertion of the second sequence W into the image. For this, the sequence W is weighted to ensure its invisibility then each of its coefficients is added to one of the pixels of the image. It should be noted that a region of the image corresponds to a region of the sequence W.

 The next step E4 is an attack on the image into which the sequence W has been inserted. The attack is for example a compression of the JPEG type (from the English Joint Photographic Expert Group) followed by a decompression.

 The attack is a simulation of any processing that can be done on the image. Thus, the degradations brought by these treatments are taken into account in the estimation of the capacity. The greater the degradations, for example the greater the compression ratio, the lower the capacity.

 Of course, it is possible not to carry out an attack.

 The next step E5 is an estimation of the capacity of the image.

This step will be detailed below.

 The step E1 of calculating the first pseudo-random sequence is detailed in FIG. 6 in the form of an algorithm comprising steps E10 to E18.

 This algorithm calculates the worst sequence. A bad sequence is a sequence which, when inserted into the image as in step E3, necessarily leads to an erroneous result when it is extracted. This erroneous result is noted by a correlation measurement between the

<Desc / Clms Page number 14>

 inserted sequence and image. The worst sequence is the one whose correlation measure provides a negative value with the largest absolute value.

 It should be noted that the image here may have been previously processed. This preprocessing is for example a high pass filtering, which consists in subtracting from each pixel the average of the region to which it belongs. The preprocessing is applied to the so-called non-centered areas of the image, that is to say those whose pixel values are not centered on zero.

 If the insertion is carried out in the spatial domain (unprocessed image), then the pre-processing is carried out on the entire image. If the insertion is carried out on an image transformed by transformation into odelettes, then the preprocessing is carried out on the low-frequency sub-band.

 This pre-processing is due to the fact that the image can be pre-processed by high-pass filtering during the subsequent extraction of the message, so as to improve the detection of the message.

 Alternatively, the best sequence can be used, which is the one whose correlation measure provides a positive value with the largest absolute value. The best sequence has all of its coefficients which are reversed from the worst sequence.

 Step E10 is the segmentation into regions of the image. The regions are those which are used for the insertion as previously described.

 The next step EU is the selection of a region of the image. The regions will be considered successively.

 The next step E12 is the calculation of the average of the pixels of the current region.

 The next step E13 is the selection of a pixel pin from the region, where n is an integer varying between 1 and the total number of pixels from the region. All the pixels of the current region will be considered successively.

 The next step E14 is a comparison of the value of the current pixel Pn with the average determined in step E12, or alternatively with zero if the average is considered to be zero.

<Desc / Clms Page number 15>

 If the value of the current pixel is greater than the average, then step E14 is followed by step E15 at which the value of the coefficient Wn of the worst sequence corresponding to this bit is-1.

 If the value of the current pixel is lower than the average, then step E14 is followed by step E16 at which the value of the coefficient Wn of the worst sequence corresponding to this bit is +1.

 Step E15 or E16 is followed by step E17 which is a test to determine whether at least one pixel remains to be processed, and, if so, to consider a next pixel. Step E17 is then followed by step E14 previously described.

 When all the pixels of the current region have been processed, then step E17 is followed by step E18 which is a test to determine if there remains at least one region to be treated, and, if so, to consider a next region. Step E18 is then followed by step E12 previously described.

 When all the regions have been treated, the worst sequence is completely determined.

 The step E2 of calculating the second sequence W is now detailed with reference to FIG. 7, in the form of an algorithm comprising steps E20 to E25. The second sequence is the one used to estimate the capacity of the image. The second sequence is determined from the first sequence and the user-defined probability of error Pe.

 Step E20 is the calculation of the probability of correct decision during the extraction of the message. This probability Pc is equal to: 1-Pe. Ette is also equal to the product: Pi. P2 ..... PCmax, where Cmax is a maximum number of bits of the message, called maximum capacity, and P, is the probability of correct detection of a bit b ,, with i between 1 and Cmax.

The next step E21 is the calculation of the probability of correct detection of a bit. As we suppose that the probability of error on the different bits of the message is the same, the probability of correct detection of a bit is equal to:

<Desc / Clms Page number 16>

The next step E22 is the determination of a number Xb of coefficients of the worst sequence Ww which must be inverted for a bit. The number Xb is determined so that the probability that any pseudo-random sequence is better than the sequence being created W is equal to Pi.

The number Xb is the value that best verifies the following formula:

In which 2Nb represents the total number of possible binary pseudo-random sequences, CmNb represents the number of sequences in which m coefficients have been inverted from the sequence Ww on a support of Nb coefficients.

 Step E23 determines the number of regions associated with a bit, as a function of the maximum capacity Cmax and of the segmentation of the image.

 Steps E22 and E23 are followed by step E24 which determines the number of coefficients to be inverted by region. Assuming that a bit is inserted in N regions, then the number of coefficients to be inverted by region is X = Xb / N.

 The next step E25 is the inversion of X coefficients by region of the first sequence Ww. These X coefficients are randomly selected.

 As a variant, the selection of the coefficients to be inverted is not random, but is deterministic and depends for example on the value of the pixels of the image.

 The result of step E25 is the second sequence W.

 The capacity estimation step E5 is detailed with reference to FIG. 8 in the form of an algorithm comprising steps E50 to E57.

<Desc / Clms Page number 17>

 Step E50 is an initialization to consider the maximum capacity Cmax as the initial capacity. The capacities will be considered successively, in descending order, up to the estimated capacity of the image.

 The next step E51 is the selection of a first bit inserted, in a message of current length C. The bits will be considered successively. It should be noted that the message considered is composed of 1, since the image in which the second sequence W has been inserted in step E3 is considered below.

 The current bit corresponds to regions in the image. Indeed, as we saw during insertion (Figures 3 and 4), a given bit is inserted on regions which depend on an insertion law. For each capacity tested, the insertion law is applied to the second pseudo-random sequence.

 Indeed, the current bit also corresponds to regions in the second pseudo-random sequence. This second sequence has the same size as the image, and a region of the image corresponds to a region of the sequence.

 The next step E52 is the selection of the regions of the image and of the second pseudo-random sequence corresponding to the current bit.

 The next step E53 is a correlation calculation between the regions of the image into which the second sequence has been inserted and the regions of the second pseudo-random sequence selected in the previous step.

 The next step E54 is a test on the sign of the result of the previous correlation.

 If the correlation is negative in step E54, this means that the detection of the bit is not correct. In this case, there is no need to test other bits. Step E54 is then followed by step E55 to consider a capacity reduced by one unit. Step E55 is followed by step E51 previously described.

 If the correlation is positive in step E54, this means that the detection of the bit is correct. In this case, step E54 is followed by step E56 which is a test to determine whether the C bits have been tested.

<Desc / Clms Page number 18>

 If the answer is negative, then this step is followed by step E57 in which a next bit is considered. Step E57 is followed by step E51 previously described.

 If the response is positive in step E56, this means that the current capacity C is the largest for which a message is correctly detected. The capacity of the image was thus estimated.

 It has been considered here that a message is correctly detected if all of its bits are detected. As a variant, it is possible to consider that a message is correctly detected if only a predetermined proportion of this message is, for example 90% of the bits.

 A second embodiment of the capacity estimation step E5 is detailed with reference to FIG. 9 in the form of an algorithm comprising steps E500 to E502.

 This embodiment makes it possible to better approach the capacity than in the previous embodiment, which underestimates the capacity C in the case where it is less than the Coax value. it should be noted that the computation times here are greater than those of the first embodiment.

 Step E500 is an initialization at which the value Cmax is initialized to the value CI nit.

 The next step is step E50 previously described. It is followed by steps E51 to E57 previously described.

 When the response is positive in step E56, then this step is followed by step E501 which is a convergence test. Step E501 tests whether the value of C is close to the value of Cmax, while being different from this value.

valeur de Cmax est : Cmax = (Cmax + Climax)/2, avec C1max qui est la dernière valeur de Cmax pour laquelle la capacité calculée C était inférieure à Cmax. Ainsi, les calculs convergent par dichotomie vers la valeur optimale de Cmax. If the answer is negative, then this step is followed by step E502 at which a new value of Cmax is calculated. For example, the new value of Cmax is: (Cmax + C) / 2 if the value of C is different from that of Cmax. In the case where the values of C and Cmax are equal, then the new

value of Cmax is: Cmax = (Cmax + Climax) / 2, with C1max which is the last value of Cmax for which the calculated capacity C was less than Cmax. Thus, the calculations converge by dichotomy towards the optimal value of Cmax.

<Desc / Clms Page number 19>

 Step E502 is followed by step E50 previously described.

 When the response is positive in step E501, then the capacity C is determined.

 This embodiment makes it possible to find a capacity C which is close to the value Cmax, which reduces the underestimation of the capacity.

 Of course, the present invention is not limited to the embodiments described and shown, but encompasses, quite the contrary, any variant within the reach of ordinary skill in the art.

 In particular, the worst sequence was considered. Similarly, we can consider the best sequence.

Claims (27)

 1. Method for estimating the capacity of a digital signal for the insertion of a message, as a function of a predetermined probability of error on the detection of any message of length equal to the capacity, characterized in that that it comprises the steps of: - calculation (E1) of a first pseudo-random sequence (Ww) as a function of a criterion on a measure of correlation of this first sequence with data representative of the signal, - calculation (E2 ) a second pseudo-random sequence (W) according to the first sequence and the probability of error, - insertion (E3) of the second sequence (W) in the digital signal, - estimation (E5) of the capacity of the digital signal in which the second sequence was inserted.  2. Method for estimating the capacity of a digital signal for inserting a message, as a function of a predetermined probability of error on the detection of any message of length equal to the capacity, characterized in that that it comprises the steps of: - calculation (E1) of a first pseudo-random sequence (Ww) as a function of a criterion on a measure of correlation of this first sequence with data representative of the signal, - calculation (E2 ) a second pseudo-random sequence (W) as a function of the first sequence and the probability of error, - insertion (E3) of the second sequence in the digital signal, - application (E4) of a distortion on the digital signal in which the second sequence has been inserted, - estimation (E5) of the capacity of the digital signal in which the second sequence has been inserted. <Desc / Clms Page number 21>  3. Method according to claim 1 or 2, characterized in that the criterion on a correlation measurement is that the value of the correlation of the first sequence (Ww) with the data representative of the signal is negative and of greatest absolute value possible.  4. Method according to claim 1 or 2, characterized in that the criterion on a correlation measurement is that the value of the correlation of the first sequence with the data representative of the signal is positive and of absolute value as large as possible.  5. Method according to any one of claims 1 to 4, characterized in that the calculation of the first sequence comprises, for each coefficient of the first sequence, the steps of: - comparison (E14) of the value of a coefficient representative of the digital signal with a comparison value, - determination (E15, E16) of the binary value of the coefficient of the first sequence as a function of the result of the comparison.  6. Method according to claim 5, characterized in that, for each coefficient representative of the digital signal, the comparison value is an average calculated over a region containing said coefficient.  7. Method according to any one of claims 1 to 6, characterized in that at least part of the signal is filtered by high-pass filtering before the calculation of the first sequence.  8. Method according to any one of claims 1 to 7, characterized in that the calculation of the second sequence comprises the steps of: - determination (E22) of a number (Xb) of coefficients to be inverted in the first sequence, depending on the probability of error, <Desc / Clms Page number 22>  - inversion of the previously determined number of coefficients of the first sequence (Ww) to form the second sequence (seen).

 9. Method according to any one of claims 1 to 8, characterized in that the estimation of capacity (C) comprises the steps of: - calculation (E53) of a correlation measure between the second sequence (W) and the digital signal into which the second sequence has been inserted according to a predetermined insertion law, - determination of the maximum length (C) of a message corresponding to a correct detection of this message, inserted into the digital signal by modulation of the second sequence and according to the predetermined insertion law, according to a criterion on the correlation.  10. Method according to claim 9, characterized in that the criterion on the correlation is that it corresponds to a correct detection (E54) of a predetermined proportion of the coefficients of the inserted message.  11. Method according to claim 9 or 10, characterized in that the steps of estimating capacity are repeated from a maximum capacity (Cmax) to an estimated capacity (C).  12. Method according to claim 9 or 10, characterized in that the steps of estimating capacity are repeated from a maximum capacity to an estimated capacity, and in that the maximum capacity is modified (E502) up convergence (E501) between maximum capacity and estimated capacity.  13. Device for estimating the capacity of a digital signal for inserting a message, as a function of a predetermined probability of error on the detection of any message of length equal to the capacity, characterized in that that it includes: <Desc / Clms Page number 23>  - means of calculation (2) of a first pseudo-random sequence (Ww) as a function of a criterion on a measure of correlation of this first sequence with data representative of the signal, - means of calculation (3) of a second pseudo-random sequence (W) as a function of the first sequence and the probability of error, - means for inserting (4) the second sequence (W) into the digital signal, - estimation means (6) the capacity of the digital signal into which the second sequence was inserted.  14. Device for estimating the capacity of a digital signal for inserting a message, as a function of a predetermined probability of error on the detection of any message of length equal to the capacity, characterized in that that it comprises: - means for calculating (2) a first pseudorandom sequence (Ww) as a function of a criterion on a measure of correlation of this first sequence with data representative of the signal, - means for calculation (3) of a second pseudo-random sequence (W) according to the first sequence and the probability of error, - means for inserting (4) the second sequence into the digital signal, - means for application (5) of a distortion on the digital signal in which the second sequence has been inserted, - means for estimating (6) the capacity of the digital signal in which the second sequence has been inserted.  15. Device according to claim 13 or 14, characterized in that it is adapted to implement a criterion on a correlation measurement which is that the value of the correlation of the first sequence (Ww) with the data representative of the signal is negative and of absolute value as large as possible. <Desc / Clms Page number 24>  16. Device according to claim 13 or 14, characterized in that it is adapted to implement a criterion on a correlation measurement which is that the value of the correlation of the first sequence with the data representative of the signal is positive and as large an absolute value as possible.  17. Device according to any one of claims 13 to 16, characterized in that the means for calculating the first sequence comprise: - means for comparing, for each coefficient of the first sequence, the value of a representative coefficient of the digital signal with a comparison value, - means for determining the binary value of the coefficient of the first sequence as a function of the result of the comparison.  18. Device according to claim 17, characterized in that it is adapted to implement, for each coefficient representative of the digital signal, a comparison value which is an average calculated over a region containing said coefficient.  19. Device according to any one of claims 13 to 18, characterized in that it comprises means for filtering at least part of the signal by high-pass filtering, prior to the calculation of the first sequence.  20. Device according to any one of claims 13 to 19, characterized in that the means for calculating the second sequence comprise: - means for determining a number (Xb) of coefficients to be inverted in the first sequence, in as a function of the probability of error, <Desc / Clms Page number 25>  - Means for inverting the previously determined number of coefficients of the first sequence (Ww) to form the second sequence (W).  21. Device according to any one of claims 13 to 20, characterized in that the capacity estimation means (C) comprise: - means for calculating a correlation measure between the second sequence (W) and the digital signal in which the second sequence has been inserted according to a predetermined insertion law, - means for determining the maximum length (C) of a message corresponding to a correct detection of this message, inserted into the digital signal by modulation of the second sequence and according to the predetermined insertion law, according to a criterion on the correlation.  22. Device according to claim 21, characterized in that it is adapted to implement a criterion on the correlation which is that it corresponds to a correct detection of a predetermined proportion of the coefficients of the inserted message.  23. Device according to claim 21 or 22, characterized in that the operation of the capacity estimation means is repeated from a maximum capacity (Cmax) to an estimated capacity (C).  24. Device according to claim 21 or 22, characterized in that the operation of the capacity estimation means is repeated from a maximum capacity to an estimated capacity, and in that it is adapted to modify the maximum capacity up to '' at convergence between the maximum capacity and the estimated capacity.  25. Insertion device according to any one of claims 8 to 10, characterized in that the calculation, insertion and estimation means are incorporated in: <Desc / Clms Page number 26>  - a microprocessor (100), - a read only memory (102) comprising a program for processing the data, and - a random access memory (103) comprising registers adapted to store variables modified during the execution of said program.  26. Apparatus for processing (10) a digital image, characterized in that it includes means suitable for implementing the method according to any one of claims 1 to 12.  27. Apparatus for processing (10) a digital image, characterized in that it comprises the device according to any one of claims 13 to 25.