FORM 2
THE PATENT ACT 1970 (39 of 1970)
&
The Patents Rules, 2003 COMPLETE SPECIFICATION
[See Section 10, and rule 13)

TITLE OF INVENTION
THE PROCESS FOR THE SECURED DISTRIBUTION OF FIXED
IMAGES

NUMERICAL

APPLICANT(S)
a) Name :
b) Nationality :
c) Address :

MEDIALIVE
FRENCH Company
111, AVENUE VICTOR HUGO,
F-75116 PARIS
FRANCE

3. PREAMBLE TO THE DESCRIPTION

The following specification particularly describes invention and the manner in which it is to be performed

;he

ADAPTIVE AND PROGRESSIVE PROTECTION OF FIXED IMAGES ENCODED IN WAVELETS
The present invention concerns the domain of the processing of numerical images encoded in wavelets.
In the present invention, it is proposed to provide a system that allows the visual scrambling and the recomposition of a fixed image encoded in wavelets.
The general problem is to provide a procedure that is capable of transmitting, in a secured way, numerical data that correspond to high quality images in whatever numerical format, from an encoding in wavelets, live or pre-recorded to a visualization screen and/or for being recorded on a hard disk or a completely different back-up device belonging to a box that connects the remote transmission network to a display of type monitor or a television screen, all while preserving the visual quality but avoiding all fraudulent utilization such as the possibility of making pirate copies of the numerically encoded images and recordings on back-up support of the aforementioned decoder box.
The traditional techniques of cryptography consist in general of combining (according to operations of the type addition or subtraction) the data of the source with values generated in a pseudo-random way and from an initialization key (encryption key). The simple possession of this key thus allows the complete decryption of the encrypted data, those that contain in substance the totality of the original information.
In the European patent request, referenced as EP 1011269A1 and titled "System for processing an information signal," the authors detail a method for the encryption of an information signal that can be applied in the case of fixed images. The method consists of adding to the original signal, that is not compressed, a pseudo-random noise to obtain a new signal. The signal encrypted in this way is next compressed with the aid of standard, adequate algorithms, then transmitted. The key itself is transmitted, in a secured manner, to the future user of the encrypted signal. The authors indicate that this method can be applied in the case of encoded images of the standard JPEG.
In this document of older techniques, no reference is made to the case of images encoded by wavelets. Moreover, the possession of the key conditions the decryption of the transmitted signal.
In the European patent referred to as EP 0920209A1, the authors describe two processes of scrambling as well as their associated descrambling procedures in order to protect the video data encoded according to standard MPEG-2. These processes act on at least a block of coefficients DC and AC resulting from the transformation DCT and for at least an image intra I. According to the first procedure, the collection of coefficients AC of at least a block of coefficients (DC, AC) of at least an image I, undergo a permutation of order R. The associated descrambling procedure consists of transmitting the parameter R in order to
2

execute the inverse and rebuild the original MPEG-2 stream. The collection of original data is therefore always present in the protected stream. According to the second procedure, the coefficient DC of at least a block of coefficients (DC, AC) of at least an image I, is recalculated with the aid of an action command that could be random. Nevertheless, for the descrambling, all the data of the reconstitution of the aforementioned original stream (the order R, the original coefficients DC encrypted with the aid of a key), in such a way that all the data that constitute it are comprised in the aforementioned protected MPEG-2 stream and are dispatched together and through the same route, the aforementioned protected MPEG-2 stream that contains the information for the reconstruction in being the unique output data of the scrambling process. Consequently, this former method does not satisfy the criteria of high security, the objects of the present invention.
Another reference to a former method is an article written by Zing W et al., published in the ACM Multimedia Proceedings of the International Conference on October 1999 and titled "Efficient Frequency Domain Video Scrambling for Content Access Control"; in this article, the authors describe a method to protect the numerical data that encode multimedia content. The method is based on a collection of three basic pseudo-random operations (bit inversion, permutation and rotation of the block of coefficients) that can be combined and controlled by the encryption keys. The content collection resides thus in the protected stream and can be accessed by the encryption keys that will be transmitted to the authorized users. In this former method, the collection of original data are present in the protected stream and the access to the original content is completely conditioned upon the possession, or not, of the encryption keys. This former method thus does not solve the security problem presented in the presented invention.
The present invention reports in particular a device capable of transmitting in a secured manner a collection of numerical images of high visual quality to a display and/or for storing in the memory of a back-up device of a box that connects the transmission network to the display, this all while preserving the visual quality but avoiding the possibility that these images can be copied in an illicit way.
The invention concerns a procedure for the secured distribution of numerical, fixed images according to a nominal format resulting from encoding in wavelets, represented by a binary stream that consists, at least, of a packet (relative to the organization of the binary sequence) that contains, at least, a block that contains simple elements (for example, coefficients) numerically encoded according to a mode specified inside the concerned stream and utilized by all the decoders that are capable of reconstructing it or decoding it in order to be able to display it correctly. This procedure consists of:
• a preparatory stage that consists in modifying, at least, one of the aforementioned simple elements according to, at least, a substitution operation that consists of the extraction of this simple element, followed by its replacement by data called lure,
• a transmission stage
3

of a principal stream conform to a nominal format, consisting of the blocks and packets modified during the preparatory stage and
through a channel separate from that of the aforementioned principal stream of complementary numerical information that allows the reconstruction of the original stream from a calculation on the destination equipment, and as a function of the aforementioned principal stream and the aforementioned complementary information. One defines the aforementioned complementary information as a collection that constitutes data (for example, coefficients that describe the original numerical stream or extracts of the original stream) and functions (for example, the substitution or permutation function). One function is defined to contain, at least, an instruction that matches up the data and the operators. The aforementioned complementary information describes the operations to execute to recover the original stream at the start of the aforementioned modified stream.
In the present invention, one defines the notion "stream" as a structured binary sequence, constituting simple and ordered elements that represent the base form of the encoding of the data and compliant with a standard or a norm for audiovisual data.
The act of having removed a part of the original data of the stream of origin, at the time of the generation of a modified principal stream, does not allow the reconstruction of the aforementioned stream of origin, from only data of the aforementioned modified principal stream, the modified stream is thus called "secured stream" [Tr.- French: flux securise]. The "secured distribution" [Tr.- French: distribution securisee] is a distribution of secured stream.
In the present invention one understands by the term "scrambling" the modification of a numerical binary stream according to appropriate methods in a way that this stream remains in conformity with a standard with which it was encoded numerically, all for displaying it on a screen, but altered from the viewpoint of the human visual perception.
In the present invention, one understands by the term "descrambling" the reconstruction process the initial stream by appropriate methods, the binary stream restored after descrambling being identical, therefore without loss, to the initial binary stream.
Advantageously, the binary stream restituted after the descrambling has a size that has a form that is different from that of the initial binary stream, following the addition of neutral data for the decoder, but it remains in conformity with the standard or norm of the original binary stream, and produces a decoded image that is visually and numerically strictly equal to the image decoded from the original binary stream. One defines the notion "scalabilite granulaire" from the start as the expression in English "granular scalability." The scalability property characterizes an encoder that is capable of encoding or a decoder that is capable of decoding an ordered collection of binary streams in a way such as to produce or reconstruct a multi layer sequence. One defines the granularity as the amount of information that can be transmitted by a layer of a system characterized by whatever scalability; thereby the system is also qualified as "granular."
The present invention proposes the protection of the numerical image encoded in
4

wavelets based integrally on the structure of the bitstream [Tr.- French: flux de bits] (binary sequence), protection that consists in modifying the targeted parts of the bitstream relating to the modeling by wavelets and their characteristics. The real values are extracted from the bitstream and are stored as well as the complementary information, and are in their turn sent the random values or calculated or the permuted values, of the same size as the original values, and that for the totality of the numerical stream. Advantageously, the aforementioned replacement values have a size different from the original values. Thus, the scrambler adds "lures" for the decoder, which receives as input a binary stream completely in conformity with the numerical format of origin, but from which the decoded and displayed image is not acceptable from the point of view of the human visual perception. The scrambling module executes an analysis of the bitstream and selects the locations where it introduces perturbations. One defines by modification the action of introducing perturbations. A perturbation is defined as a change (for example, of the value, sign inversion, saturation, thresholding), or a substitution by a random or calculated value, or a permutation. The achieved scrambling-descrambling procedure is without quality loss of the original image.
Advantageously, the scrambling operation is also achieved with a partial decoding-encoding of the bitstream that represents the encoded image. The modifications are then directly applied to the coefficients of the decoded wavelets, the original wavelet coefficients are stored as complementary information and the coefficients of the modified wavelets are re-encoded in order to generate a principal stream that is modified conform to a standard or the norm of the initial stream.
Contrary to the majority of the encoding systems already known by the expert, the principle described below makes it possible to ensure a high level of protection while reducing the volume of information necessary for decoding.
The protection, realized in a manner conform to the invention, is based on the principle of suppression and replacement of the information that encodes the original visual signal by whatever method, that is to say: substitution, modification, permutation, or displacement of the information. This protection is equally based on the knowledge of the structure of the stream at the output of the visual encoder: the scrambling depends on the content of the aforementioned numerical stream.
The reconstitution of the original stream takes place on the destination equipment from the modified principal stream that is already present or available (for example, on a CD or DVD) or dispatched in real time on the destination equipment and the complementary information dispatched Jn real time at the moment of the visua|[zatioii ^^comjpris£jl^data and the functions executed with the aid of numerical procedures (collection of instructions).
Moreover, when the order of the packets within the modified principal stream is
5

changed according to an operation of code conversion after the preparatory stage and/or before, during or after the transmission stage, the complementary information, stored at the time of the preparatory stage and transmitted on the destination equipment, equally allows reconstruction of the original content from the modified principal stream.
In the same way, when a sub collection of packets of the modified principal stream is received on the destination equipment, only the part of the complementary information that corresponds to this sub collection of modified packets is transmitted on the destination equipment.
Advantageously, the reconstruction stage introduces, according to a fixed mode, data, neutral for the decoder, in the stream reconstructed from the modified principal stream, in order to modify the size of the reconstructed stream all while preserving its conformity vis-avis the norm or the native standard, as well as the numerical and qualitative equality of the decoded image and the image decoded from the original stream.
Knowing the way in which the modeling, the compression and the encoding in wavelets of the image by the encoder in wavelets and/or the standard or norm given, is carried out, it is always possible to extract from the bitstream the principal parameters which it describes and which are dispatched to the decoder.
Once these parameters are identified they are modified in a way such that the numerical stream, generated by the encoder and/or the standard given, is modified, but in conformity with this encoder and/or this standard. The structured binary stream represents the image encoded in wavelets. Advantageously, it could represent a succession of numerical images resulting from encoding in wavelets. The modifications render the image, resulting from the encoding of the aforementioned stream, unusable for the user because it is scrambled. However, the aforementioned stream can be understood and interpreted by the appropriate decoder, without that the latter is perturbed, and displayed on a screen.
The modification of one or various components of the aforementioned numerical stream brings about a degradation of the image from a visual point of view and transforms it into a signal that is completely incomprehensible and unpleasant from the viewpoint of human visual perception.
In order to correct the various defects of the former methods, the present invention concerns a process for the protected distribution of numerical fixed images according to a nominal format resulting from numerical encoding in wavelets, represented by an original stream consisting of at least one packet (relative to the organization of the binary sequence) that contains at least one block that regroups the simple elements (for example, coefficients) numerically encoded according to a mode specified inside the concerned stream and utilized by all the decoders that are capable of reconstructing it or decoding it in order to be able to correctly display the image that is characterized by that it consists of:
• a preparatory stage that consists in modifying at least one of the aforementioned simple elements according to at least a substitution operation that consists in the extraction of
6

this simple element, followed by its replacement by data called lure, • a transmission stage
- of a principal stream modified conform to a nominal format, consisting of the blocks and packets modified during the preparatory stage and
- by a route that is separated from the aforementioned principal stream of complementary numerical information that allows the reconstruction of the original stream from calculations, on the destination equipment, and as a function of the aforementioned principal stream and the aforementioned complementary information.
In a variation, the modified principal stream is recorded or available on the destination equipment before the transmission of the complementary information on the destination equipment.
In another variation, the modified principal stream and the complementary information are transmitted together in real time at the moment of the visualization.
Moreover, the format of the nominal flux resulting from encoding in wavelets is defined by a standard or an encoder common to a community of users.
Advantageously, this procedure contains a stage of analysis of at least a part of the original stream, the aforementioned analysis stage determining the nature of the modifications to be made to the aforementioned simple elements.
Moreover, the stage of determination of the modification of the simple elements, to be carried out at the time of the aforementioned stage of analysis, takes into account the concrete structure of at least a part of the original image.
According to one kind of realization, the procedure comprises a stage before the analog/numerical conversion under a structured format, the procedure being applied to an analog visual signal.
Advantageously, the complementary information comprises at least a numerical routine that is capable of executing a function.
According to one mode of implementation, the aforementioned preparatory stage of modification is carried out in an adaptive way on the aforementioned original stream according to at least a part of the characteristics representative of the structure, the contents and the parameters of the original image, the profile of the recipient, and external events.
Moreover, the modified principal stream conforms to a standard or a native standard of the original stream.
Advantageously, the size of the modified principal stream is strictly equal to the size of the original stream.
Advantageously, the modifications carried out at the time of the aforementioned preparatory stage are carried out directly on the packets of the initial binary stream gathering blocks of encoded simple elements.
Preferably, the modifications carried out at the time of the aforementioned preparatory stage are carried out directly on at least one packet of the simple elements obtained by decoding at least one packet of the initial binary stream.
7

In a particular mode of realization, the modifications are carried out on at least one of the layers of spatial scalability of the fixed images encoded in wavelets.
In another mode of realization, the modifications are carried out on at least one of the layers of qualitative scalability of the fixed images encoded in wavelets.
Preferably, the modifications are carried out on at least one collection of at least one simple element as a function of the parameters of the, beforehand carried out, modifications on at least one collection of at least one simple element within the same stream.
In a variation, the aforementioned fixed images constitute a succession of images in time.
Moreover, the aforementioned modification of the aforementioned simple elements is different for at least two of the aforementioned succession of images.
Advantageously, the aforementioned modification of the simple elements of an image of the aforementioned succession of images includes the modification of the simple elements of the preceding images in the temporal order of the succession, while being based on the properties of the spatial and qualitative scalability of the transformations into wavelets.
According to one mode of implementation, the aforementioned modification of the aforementioned simple elements is applied to only the spatial part of the image, determined by being based on the properties and characteristics of the transformation into wavelets.
According to another mode of implementation, the aforementioned modification of the aforementioned simple elements is applied to at least a spectral component resulting from the transformation into wavelets in order to target in this way the part of the visual stream thus to scramble.
Advantageously, this procedure is without loss of quality. According to one mode of realization, this procedure comprises a supplementary stage of the reconstruction of the aforementioned original stream that comprises:
a reconstruction stage by application of a reconstruction function from complementary information while providing a route that is separate from that of the aforementioned principal stream;
a decoding stage of the aforementioned stream reconstructed by a procedure adapted for the aforementioned nominal stream.
In one mode of implementation, the size of the stream reconstructed on the destination equipment is equal to the size of the original stream.
In another mode of implementation, the size of the stream reconstructed on the destination equipment is different from that of the original stream.
Advantageously, the complementary information allows the reconstruction of the original stream from the modified principal stream in which the order of the packets has been changed at the time of a code conversion carried out after the preparatory stage and before the reconstruction stage. The act of having removed a part of the original data of the stream. The invention concerns equally equipment to implement this procedure, comprising an encoder
8

according to the aforementioned nominal format and the means for transmitting a numerical stream, characterized by that it comprises moreover a means for the processing of the original stream that consists in modifying at least one of the simple elements of the principal stream, moreover, the equipment comprises the means to transfer the complementary information that corresponds tot the aforementioned modification.
The invention relates to another piece of equipment for the implementation of the procedure, comprising a decoder according to the aforementioned nominal format and the means of reception of a numerical stream, characterized by that it comprises, moreover, one means of reception of complementary information associated with the principal stream and a means for the reconstruction of original stream by processing of the aforementioned principal stream and of the aforementioned complementary information.
Finally, the invention concerns a system for the distribution of numerical fixed images comprising at least equipment for distributing of fixed images encoded numerically in wavelets and a plurality of equipment for restitution and means of telecommunication connecting the equipment for distributing and the equipment for restitution characterized by that the broadcasting equipment comprises a means to modify at least one of the simple elements of the principal stream, and by that these means of telecommunication are composed of a first means for the transmission of the complementary information that corresponds to the aforementioned modification, the reconstruction equipment comprises the means to reconstruct the original stream by processing the aforementioned principal stream and the aforementioned complementary information and means for decoding information that come from the first transmission means.
PRINCIPLE OF THE TRANSFORMATION INTO WAVELETS
The -transformation of an image (a signal spatial in two dimensions) into wavelets consists in applying to the original image a succession of high pass and low pass filters elaborated from the characteristics of the wavelets of the analysis. The synthesis operation, which consists in reconstructing the image from the collection or a sub collection of the wavelet coefficients, obeys a schema of inverse filtering.
The application of the stage, of the transformation into wavelets, to a numerical image (which could be composed of one or several matrices of real or integer values) is equivalent to an operation of filtering on the rows and the columns of the matrix or the matrices of values following a dyadic reduction (division by two) in the size. It thus generates in each stage 4 new matrices of wavelet coefficients, called sub-bands and whose width and height are equal to half of the transformed width and the height of the matrix (dyadic progression). Suppose an image of width L and height H and resolution r. The application of a stage of transformation into wavelets thus generates 4 matrices of wavelet coefficients of dimension (L/2, H/2): the sub-band LLr-1 result of the low pass filtering horizontally (lines) and high pass vertically (columns) of the image I, the sub-band LHr-1 result of the low pass filtering horizontally and high pass vertically, the sub-band HLr-1 result of the high pass filtering
9

horizontally and low pass vertically, and the sub-band HHr-1 result of the high pass filtering horizontally and vertically.
One considers the transformation into wavelets at level R (equivalent to R stages) of an image. A transformation into wavelets at level R is associated with R+1 levels of resolution, numbering from R to 0, with R and 0 corresponding at the level of the highest (initial image) and the lowest resolution. Each sub-band resulting from the decomposition of the image I into wavelets is thus identified by its orientation (LL or HL or LH or HZ) and its level of the corresponding resolution (0,...,R).
The original image is considered as the sub-band LLR. At each level i of the decomposition into wavelets (except the last i=0) the sub-band LLj is thus decomposed into 4 new sub-bands LLj.i, HLj.i, LHj.i and HHj.i and whose size is divided by two due to LLj. The process is iterated until the sub-band LLo has been obtained. Therefore, for a transformation into wavelets at level R, 3R+1 sub-bands of wavelet coefficients are generated: LLo, HLi, LHi, HHi, HL2, LH2, HH2,.... HLR, LHR, HHR.
The reconstruction of the image (synthesis) from the 3 R+1 sub-bands consists in applying an inverse filtering operation on the wavelet coefficients followed by a dyadic increase in size. A progressive reconstruction of the image according to different levels of resolution can also be performed. By adding, for example, to the reconstructed image of resolution r-1, the 3 sub-bands of wavelet coefficients HLr-i, LHr_i, HHr.i a new image of resolution r is obtained.
The unique sub-band of wavelet coefficients LLo is an approximation of the original image LLR and whose resolution is 2R times inferior to the original image.
The 3R sub-bands of wavelet coefficients HLr-i, LHr.i, and HHr-i, correspond in themselves to the details in the image, extracts on resolution level r-1. The larger the value of r, the more are the wavelet coefficients of these sub-bands characteristic of details more and more fine (small) in the original image.
The wavelet coefficients resulting from the transformation of an image into wavelets are thus the local characteristics of frequential information. The more r decreases the more the spatial extent, characterized by a wavelet coefficient, increases (multiplication by a factor 4 at each stage).
A transformation of an image into wavelets at level R generates an approximate "image" of inferior resolution 2R and 3R "images" of details at different resolutions (0 to R).
From this modeling, the encoding in wavelets is characterized by its spatial and qualitative scalability (or granularity), which makes it possible to obtain various resolutions for the encoded images and allows also the realization of an adaptive and progressive scrambling. An example application is for a sequence of fixed images encoded with the "Motion JPEG-2000" standard. One applies a progressive scrambling, so that the user begins to watch the succession of unscrambled images. Next, one starts to apply a weak scrambling that one increases more and more until the images are entirely scrambled. The sought after goal is to attract the interest of the user for the succession of images but to deny him the rights to watch them if the latter has not purchased them. One realization of this application is to
10

scramble—the, succession of images with one or several given algorithms that modify progressively the parameters of the scrambling for a time period determined in a way so as to increase the discomfort until the images are. completely scrambled and not visible.
The scrambling procedure of the images whose compression scheme utilizes the transformations into wavelets consists in introducing perturbations in the wavelet coefficients that belong to one or several different sub-bands. One defines a perturbation as being a change of value or a permutation. Advantageously, the perturbation can also be a substitution by a random or calculated value. The amplitude and the position of the wavelet coefficients, being characteristics of the frequencies and the spatial information present in the image, and this at several levels of resolution, frequential and spatial perturbations will be generated on an image that is reconstructed from modified wavelet coefficients. Therefore, a coefficient of large amplitude signals the presence of a local discontinuity in the intensity (detail) in the image of which the "size" is measured as a function of the resolution r of the sub-band: The finer discontinuities will be extracted for large r and the lager for small r. By modifying the amplitude and/or the position of one or several coefficients at one or several resolutions r, visual perturbations are introduced in the reconstructed image.
The human eye is essentially sensitive to the discontinuities in the image. They are the ones that carry the maximum amount of visual information. These discontinuities can be due to physical and/or geometrical properties of the objects present in the image. By perturbing the wavelet coefficients belonging to one or several of the 3x(R-l) sub-bands of details LH, HL, and HH, it is the primordial information with the general interpretation of an image that is perturbed to the extent of being destroyed. Advantageously, the wavelet coefficients of the band LLo are modified.
Another example of scrambling directed spatially, consists in modifying wavelet coefficients, which are attached to a particular entity "tile-precinct." One tile is a composite that results from the partition of the original image into several rectangular connected regions. A "precinct" {enceinte in French) is a rectangular zone that regroups three sub-bands of wavelet coefficients belong to the sub-bands LH, HL, and HH, respectively, for a spatial region limited in the image and for a fixed resolution. Another example of scrambling oriented spectrally, consists in modifying the wavelet coefficients generated by the decomposition of one of the spectral or colorimetric components of the image.
Perturbations that are oriented spectrally (choice of one component for an image color) and/or spatially (choice of a tile and/or a precinct) can thus be applied.
Another example of scrambling consists in working at different resolutions. The perturbation is applied to the wavelet coefficients belonging to one of the or the sub-bands of the same level of resolution r. Moreover, the effects of the perturbation will be visible on each tile of the image that is reconstructed from the perturbed coefficients and higher resolutions i.e. r+1, r+2, ,„ , R. The smaller r is, that is the more the perturbed wavelet coefficients belonging to sub-bands of lower and lower resolutions, the more the influence of the perturbations will be important and capable of being more visible and awkward as for the
11

comprehension and/or the utilization of the original information contained in the image.
From its nature, encoding in wavelets provides multiple possibilities of simple or progressive scrambling, this as a function of the belonging to one of the sub-bands of the coefficients that carry the disturbances: by modifying the wavelet coefficients belonging to a sub-band HL the vertical discontinuities of the reconstructed image are degraded, by modifying the wavelet coefficients belonging to a sub-band LH the horizontal discontinuities of the reconstructed image are degraded, by modifying the wavelet coefficients belonging to a sub-band HH the diagonal discontinuities of the reconstructed image are degraded.
Therefore, as a function of belonging to a sub-band of the wavelet coefficients which carry the perturbations, the geometrical aspect of the defects that will be introduced can be controlled.
Therefore, as a function of the intensity and the nature of the perturbations introduced in the coefficients of the modified, substituted, or permuted wavelets, a progressive and qualitative scrambling of the image can be carried out.
The scrambling of an image or a succession of images encoded as wavelets can be controlled according to 4 axes of action: a spectral axis (choice of one or several components), a spatial axis (choice of one or several tiles, of one or several precincts), a resolution axis (level r of resolution to scrambling and extended in the scrambling), and finally, a qualitative axis (choice of one or several sub-bands from LL, HL, LH, or HH, choice of the nature and the intensity of the perturbations).
An adaptive and progressive scrambling is carried out as a function of the parameters of the initial encoding of the image (number of components, number R of the level of decomposition). Different scenarios can therefore be envisioned: scrambling of a specific zone of an image (choice of a tile or a precinct), scrambling of a specific component of an image (luminance or chrominance, red, green, or blue), scrambling of a certain level of resolution, scrambling according to the variable directions (as a function of the perturbed sub-band). The numerous scenarios combine the preceding variations or according to the 4 axes defined previously are equally possible.
The different perturbations and modifications of the envisioned wavelet coefficients are, for example, the modification of the values of the coefficients by thresholding, the addition/substitution of noise, the permutation, the substitution by one or several random values, the change of sign.
A thresholding operation consists in forcing to a new value called and one or several wavelet coefficients Cj if their images calculated by a function F (defined according to a heuristic method) are below a threshold T fixed earlier:
If F(CJ) < T then Cj=nval.
Different functions F, different thresholds as a function of F and Another perturbation consists in adding/substituting a noise, that is in adding/substituting to the value of one or
12

several wavelet coefficients q a random noise or pseudo-random from a known statistical law. As a function of the power of the noise, the intensity of the perturbation is adjustable.
Another perturbation consists of permuting, according to a deterministic or random schema, the position of the coefficients in a sub-band or several fixed sub-bands.
In the present invention, the scrambling of the images encoded in wavelets is carried out as well in an adaptive way as function of the client's profile. One applies an adaptation of the parameters and algorithms of the cited scrambling as a function of the profile of the destination, as a function of the behavior of the destination during the connection to the server (for example, the regularity and the acceptance of payments),as a function of his habits (for example, time, hour of connection), as a function of his characteristics (for example, age, sex, religion, community), or as a function of the data communicated by a third party (for example, belonging to associations or a walk-on role based on the data of the consumers). One applies also an adaptation of the parameters and algorithms of the scrambling as a function of external events, for example, hour of the broadcast, the size of the audience, social-political events, and perturbations during the connection.
The invention will be better understood with the aid of the description made hereafter on a purely explanatory basis, of a mode of realization of the invention, in reference to the attached figure:
• figure 1 illustrates a mode of realization particular to the client-server system
in conformity with the invention.
The image encoded (1) according to the JPEG-2000 standard, that one wishes to secure is sent to a system of analysis (121) and of scrambling (122) intended for the generation of the modified principal stream (124) and complementary information (123).
The original stream (1) can be directly in a numerical form (10) or in an analog form (11). In this latter case, the analog stream (11) is converted by an encoder that is not represented in a numerical format (10). In the following of the text, we will note (1) the binary input stream.
A first stream (124) in JPEG-2000 format, of a format identical to the numerical input stream (1) apart from certain coefficients, values and/or vectors that have been modified, is placed in an output memory buffer (125). The complementary information (123), of whatever format, contains the references to the parts of the elements that have been modified and is placed in the buffer (126). As a function of the characteristics of the input stream (1), the systems of analysis (121) and of scrambling (122) decide what scrambling applies and which parameters of the stream are to be modified as a function of the type of encoder with which it has been encoded, for example JPEG-2000, JJ2000, JASPER, Kakadu, Motion JPEG-2000, Still Texture in MPEG-4, etc.).
The modified JPEG-2000 stream is next transmitted from the buffer (125), via a high bandwidth network (4) of the type wireless, cable, satellite, etc., to the client (8), and, more precisely, to its memory (81) of the type RA, ROM, or hard disk. When the client desires to see an image present in the memory (61),the destination equipment (8) sends a request to the
13

server (12) which verifies the rights of this user. Two eventualities are therefore possible:
- or the destination (8) does not posses the necessary rights to watch the image. In this case, the stream (124) generated by the scrambling system (122) present in its memory (81) is dispatched to a synthesis system (82) that does not modify it and that transmits it identical to a reader (83) and the contents of the image, strongly degraded visually are displayed on a visualization screen (9).
or the destination (8) does have the rights to view the image. As a function of the rights of the user, the server (12) transmits the appropriate complementary information (123) that is stored, in totality or partially, in the buffer (126) by the link (6). In this case, the synthesis system issues a request for visualization to the server (12) that contains the complementary information (123) necessary for the reconstitution of the original image (1). The server (12) then sends by the connection (6) via telecommunications networks (6) of the type standard analog or numerical telephone line, DSL (Digital Subscriber Line), BLR [Tr.-French: Boucle Locale Radio] (Local Radio Loop), DAB (Digital Audio Broadcasting) or of mobile numerical telecommunication (GSM, GPRS, UMTS), the complementary information (123) that allows the reconstruction of the initial image in a manner that the client (8) can watch the image. The synthesis system (82) proceeds therefore with the descrambling of the image by the reconstruction of the original stream by combining the modified principal stream (124) and the complementary information (123). The image obtained in this way at the output of the synthesis system (82) is next transmitted to a traditional reader (83) that displays it on a screen (9).
The present invention concerns more particularly the analysis (121) and the scrambling (122) modules, given the complexity and the property of multi-layers and scalability of the streams encoded in wavelets.
An example of scrambling carried out by the scrambling module is the saturation of the wavelet coefficients relative to the details. This method consists in forcing the value of one or several wavelet coefficients c belonging to one or several of the sub-bands HLr, LHr, HHr (r remains fixed) to a value nval if their amplitudes (modules) are higher than threshold:
If |c| > threshold, c = nval
The value threshold influences the number of modified wavelet coefficients in a sub-band: the smaller threshold is, the more important will be the number of modified coefficients and the larger will be the degradations.
The scrambling is adaptive as a function of the modified sub-band. The distribution of the values of the wavelet coefficients in a sub-band can be modeled by a Laplacian distribution. The larger r is, the more spread out is the distribution, that is, one finds more and more wavelet coefficients of large amplitudes. Therefore, for a fixed threshold, the number of modified coefficients in every sub-band is not constant.
The value nval rules the intensity of the degradation introduced in the reconstructed
14

image. The larger the absolute value \nval\, the more important will be the visual intensity of the introduced artifacts.
If the modified wavelet coefficients belong to the sub-band HL the artifacts appear in a vertical direction in the reconstructed image.
If the modified wavelet coefficients belong to the sub-band LH the artifacts in a horizontal direction in the reconstructed image.
If the modified wavelet coefficients belong to the sub-band HH, the artifacts appear in a checkerboard pattern in the reconstructed image.
The value r rules as to the level of the resolution where the defects are introduced. The smaller r is, the more will the degradation be spatially extended in the reconstructed image.
The complementary information (123) is then composed from the wavelet coefficients c replaced by rival and their position in the transformed image.
Another example of scrambling realized by the module (122) is the substitution of the original coefficients by the wavelet coefficients to which one adds noise. The idea is to add/substitute to/in the coefficients one or several bands a random or pseudo-random value generated according to given probability law. For example, in the case of white additive noise, ci = ci + a ei,
with ei a random variable generated according to a law N(0,1) (normal law with average zero and variance 1.0) and a the power of the noise.
The complementary information (123) is then composed from the characteristics of the added noise and from the subtraction function of this noise for the modified coefficients.
Another realization of the scrambling is the permutation of the wavelet coefficients. The idea is basically to permute, in a deterministic or random way, the coefficients belonging to one or several sub-bands. The permutations can be carried out in a deterministic (and characterized by a permutation matrix P) or in a random manner on the totality of the rows and columns of coefficients, or still between the sub-collections of coefficients. For example, permutations that are deterministic, of the lines in the sub-bands HL, HH, and which are random, of the lines in the sub-bands LH (horizontal details).
The obtained visual effect is a strong degradation of the image in the permutated zones, the effects of blurring and the degradations of the contours.
In this type of scrambling, the complementary information (123) constitutes of inverse permutation functions of the permutations realized by the scrambling device (123).
Another mode of realization of the scrambling is to substitute in the wavelet coefficients belonging to a resolution level r, random wavelet coefficients chosen according to a uniform probability law. The obtained visual effect is a destruction of the details of the image at resolutions r, r+1,..., R, therefore bringing about a strong visual degradation. According to this type of scrambling, the complementary information (123) constitutes
15

original wavelet coefficients and their placements in the initial stream.
The decomposition into wavelets also permits to carry out a localized scrambling of an image or a succession of images. The idea is to degrade, according to one of the three preceding methods (threshold, permutation, noise), a part of the image, in targeted way, that is uniquely the zones of interest that make the image not acceptable from the viewpoint of human visual perception. In this way one obtains a disagreeable effect, all while limiting the amount of complementary information. An example of a realization that illustrates the principle is described below.
Let (L,H) be the size of the original image and (l,h) the size of the zone that one wants to scramble on the original image. To simplify, one will suppose that the zone to scramble is situated in the center of the image, that is the zone where in general the information that is more pertinent is located in the image.
Let r be the resolution level of the sub-band of the wavelet coefficients that one wants to modify Therefore, the following formulas indicate the intervals [is, js] and [ie, je] the indices of the wavelet coefficients to modify in the sub-band of the considered resolution r:
is = nlr / 2 - ( 1 / 2r+1) , je = ncr / 2 - ( h / 2r+1),
ie = nlr/2 + (l/2r+1),je = ncr/2 + (h/2r+1),
where (nlr, ncr) are the numbers of rows and columns, respectively, of the matrix of the wavelet coefficients in the considered sub-band.
The cited examples are not limiting and illustrate the principle of modifications of a
numerical stream resulting from the encoding of an image in wavelets with the goal to protect
them, this principle is also applicable to a whole succession of numerical images encoded in
wavelets. ~ . ———___

16

WE CLAIM
1. A process for the secured distribution of fixed numerical images according to a nominal
format resulting from the numerical encoding in wavelets, represented by a original stream
comprising a packet relating to organization of a binary sequence that contains at least a
block that regroups numerically encoded simple elements according to a mode specified
inside the stream and utilized by decoders that are capable of reconstructing or decoding it
to be able to correctly display the image comprising:
• modifying at least one of the simple elements according to at least a substitution operation comprising extracting the simple element, followed by its replacement by lure data, and
• modifying a principal stream to conform to a nominal format comprising modified blocks and packets, and by a path that is separate from the principal stream of complementary numerical information and allowing reconstruction of the original stream from calculations, on destination equipment, as a function of the principal stream and the complementary information.

2. The process according to claim 1, wherein the modified principal stream is recorded or available on the destination equipment before the transmission of the complementary information on the destination equipment.
3. The process according to claim 1, wherein the modified principal stream and the complementary information are transmitted together in real time at a moment of visualization.
4. The process according to claim 1, wherein the format of the nominal stream resulting from encoding in wavelets is defined by a standard or an encoder common to a group of users.
5. The process according to claim 1, further comprising analyzing at least a part of the original stream and determining the nature of modifications to be made to the simple elements.
6. The process according to claim 5, wherein determining modification of the simple elements is carried out during the analysis and accounts for at least a part of the original image.
7. The process according to claim 1, applied to an analog visual signal before analog/numerical conversion.
17
fresh of original

8. The process according to claim 1, wherein the complementary information of the modification comprises at least one numerical routine capable of executing a function.
9. The process according to claim 1, wherein modification is adaptive on the original stream as a function of a part of at least representative characteristics of structure, content and parameters of the original image, profile of the destination, and external events.
10. The process according to claim 1, wherein the modified principal stream conforms to a norm or a native standard of the original stream.
11. The process according to claim 1, wherein the size of the modified principal stream is equal to the size of the original stream.
12. The process according to claim 1, wherein modifications are made directly on the packets of an initial binary stream regrouping blocks of encoded simple elements.
13. The process according to claim 1, wherein modifications are made on at least one simple elements obtained by decoding at least one packet of an initial binary stream.
14. The process according to claim 1, wherein the modifications are carried out on at least one spatial scalability levels of fixed images encoded in the wavelets.
15. The process according to claim 1, wherein modifications are carried out on at least qualitative scalability of fixed images encoded in the wavelets.
16. The process according to claim 1, wherein modifications are carried out on at least one collection of at least one simple element as a function of parameters of modifications made before on at least one collection of at least one simple element inside the same stream.
17. The process according to claim 1, wherein the fixed images constitute a succession of images in time.
18. The process according to claim 17, wherein modification of the simple elements is different for at least two successions of images.
19. The process according to claims 17, wherein modification of the simple elements of an image of the succession includes modification of the simple elements of preceding images in temporal order of succession founded on spatial and qualitative scalability properties of
18

transformations into wavelets.
20. The process according to claim 1, wherein modification of the simple elements is applied to a spatial part of an image only, determined while being based on properties and characteristics of a transformation into the wavelets.
21. The process according to claim 1, wherein the modification of the simple elements is applied to at least one spectral component resulting from transformation into wavelets to target a part of the stream to scramble.
22. The process according to claim 1, wherein modification occurs without loss of quality.
23. The process according to claim 1, further comprising: applying a reconstruction function from complementary information while providing a route that is separate from a vector of the principal stream; and decoding the stream reconstructed by a procedure adapted for the nominal format.
24. The process according to claim 23, wherein the size of the stream reconstructed on the destination equipment is equal to the size of the original stream.
25. The process according to the claim 23, wherein the size of the stream reconstructed on the destination equipment is different from the size of the original stream.
26. The process according to claims 23, wherein the complementary information allows reconstruction of the original stream from the modified principal stream in which the order of the packets has been changed during an operation of code conversion.
27. Apparatus for the secured distribution of fixed numerical images according to a nominal stream format used in a process according to claim 1, comprising an encoder that sets into motion a numerical encoding into wavelets according to the nominal format, means of transmission of a numerical stream, means for processing an original stream that comprises modifying at least one of the simple elements of the principal stream, and means to transfer the complementary information corresponding to the modification.
28. Apparatus for the restitution of fixed numerical images according to a nominal format of images used in a process according to claim 1, comprising a decoder according to the nominal format, means of reception of a numerical stream, means of reception of complementary information associated with the principal stream, and means for
19

reconstruction of the original stream by processing the principal stream and the complementary information.
29. A system for the protected distribution of fixed numerical images comprising at least one distribution device of fixed images numerically encoded into wavelets and a plurality of devices of restitution and means of telecommunication linking broadcast equipment and the restitution equipment, wherein the broadcast equipment comprises means to modify at least one of the simple elements of the principal stream, wherein the means of telecommunication are composed of the first means for transmission of principal information encoded into wavelets and second means to transfer complementary information that corresponds to the modification, the restitution equipment comprises means for the reconstruction of the original stream by processing the principal stream and the complementary information and means to decode the information coming from the first means of transmission.

20