The present invention relates to encoding data processing apparatus and methods for generating marked copies of material items by introducing a payload data word into copies of the material items.
Correspondingly, the present invention also relates to detecting data processing apparatus and methods operable to detect code words, which may be present in marked material items.
The present invention also relates to media terminals for generating marked copies of material items for reproduction. In some applications, the material items may be video material items, and the media terminal may form part of a digital cinema projector.
Generally, a technique for embedding data in material to the effect that the embedded data is perceptible or imperceptible is referred to as watermarking. Code words are applied to copies of material items for the purpose of identifying the copy of the material item or for conveying data represented by the code words. In some applications, watermarking can provide, therefore, a facility for identifying a recipient of a particular copy of the material.
A process in which information is embedded in material for the purpose of identifying a specific copy of the material is referred to as finger printing. A code word, which identifies the material, is combined with the material in such a way that, as far as possible, the code word is imperceptible in the material. As such, if the material is copied or used in a way, which is inconsistent with the wishes of the owner, distributor or other rights holder of the material, the copy of the material can be identified from the code word, so that the right holder can take appropriate action. The code word is therefore used to convey a payload data word, which can be used to identify uniquely the copy of the material. In the following description and claims, the term “water mark” and “finger printing” may be used interchangeably, particularly in respect of the encoding and detection techniques disclosed.
In co-pending UK patent application number 0327854.6 an encoding data processor is disclosed for application to for example digital cinema in which payload data having two or more fields is represented by watermarking an image with one or more code words. Each value of a first field of a payload data word is associated with a first code word, and each value of a second or subsequent data field is represented by a second code word, which is selected in dependence upon the value of the first data field. As such a detected code word can be used to identify one of a plurality of sub-sets into which the data words are divided, each data word in the sub-set having the same value in the first field. A second or subsequent field of the data word from the sub-set can be then identified by detecting a second code word from the material item. The second code word identifies a value in the second data field as well as the value of the first data field.
The first and second fields can be assigned to different associated parameters such as for example address parameters. The first field could therefore identify the country of distribution of the material, whereas the second field could identify a particular distribution outlet within the country. A watermarking or finger printing system can be formed for identifying a point of distribution of copies of video material.
UK patent application number 0424225.1 discloses an apparatus and method for embedding a payload data word into video images for application to digital cinema. The payload data word includes a data field providing a representation of a date and time of reproduction of the video images, and a data field providing a representation of an identifying codeword of the projector.
Generally, it is desirable to reduce a likelihood of an embedded code word being detectable. If a code word is detected, the payload data word may be determined. As a result the watermarking system could be compromised by attackers wishing to remove a code word or include a false code word to disrupt detection of the payload data.
According to the present invention there is provided an encoding data processing apparatus for generating a marked copy of a material item by representing a payload data word as a code word and embedding the code word into a copy of the material item. The apparatus comprises a code word generator including a pre-processor and an encryption processor. The pre-processor is operable to divide the payload word into a plurality of parts, to form a plurality of initial values from the plurality of parts. Each of the initial values is formed by adding parts of the payload data word to each initial value successively starting with the first part added to the first initial value, and adding the first and second to the second initial value so that each subsequent initial value includes all the parts of the payload data word of those of the previous initial values in addition to a new part of the payload data word. The encryption processor is operable to receive each of the initial values and to generate an encrypted data stream from each of the initial values using a payload encryption key to provide a plurality of encrypted data streams. The combiner is operable to select parts of the encrypted data streams, and to combine the selected parts of the encrypted data streams with the frames of the material item.
Embodiments of the present invention provide an arrangement in which a payload data word can be embedded within a material item to the effect of allowing the payload data word to be recovered from the material item without an exhausting search which could otherwise be computationally prohibitive. Furthermore, the likelihood of an attacker being able to discover, alter or destroy the payload data word in the material item is reduced. To this end, a plurality of initial values are used to generate a plurality of encrypted data streams, which are substantially orthogonal or are at least arranged to provide improved orthogonality between each other when being detected by correlation. As such, the presence of one encrypted part of the payload data word is less likely to interfere with the detection of another part of the payload data word. The payload data word is divided into a plurality of parts and each of the parts is successively added to each of the plurality of initial values. Thus starting with the first initial value, the first initial value is formed from the first part of the payload data word. The second initial value is formed from the first and second parts of the payload data word, the third initial value is formed from the first, second and third parts of the payload data word and so on. Thereafter, the n-th initial value IVn is formed from the parts of the initial values as expressed as follows:
IVn=Σl=0l=n-1part(n−l)
Thus each subsequent initial value is formed from a further code word part and all the previous code word parts.
In some examples, each of the initial values includes a field index providing an identification of that initial value.
The number of data symbols included in the initial values maybe set to the effect that each of the initial values provide at least some degree of orthogonality with respect to each other, when the parts of the payload data word are being detected. Each of the encrypted data streams is made substantially orthogonal with respect to one another because encryption is used to generate each of the data streams and as a result of the length of each of the data streams being sufficiently long to appear substantially “random-like”. Encrypting the data streams as part of their generation has an effect that a change of one bit in the initial value input to the encryption circuit has an effect of producing a very different encrypted data stream with respect to the encrypted stream, which would have been generated for the unaltered initial value. Furthermore, by arranging for the encrypted data streams to be sufficiently long, an encrypted data stream will correlate with itself producing a high correlation output value, with the unwanted encryption data streams producing a lower correlation value for longer encrypted streams. Thus, by setting the initial values to a predetermined length which is long enough to cause a correlation with unwanted encrypted data streams to be a substantially low value, a likelihood of successfully detecting the wanted encrypted data stream is increased.
Since the parts of the payload are divided into each of the initial values the detection of one of the parts can be used to ensure the detection of another of the parts. Thus, by detecting the first part of the of the payload data word from one or more of the initial values, subsequent parts of the payload data word can be successively detected.
Video images are one example of material, which can be protected by embedding a digital code word. Other examples of material, which can be protected by embedding a code word, include still image, audio data, software programs, digital documents (optionally reproduced on paper or other media) and any other information-bearing signal.
According to an aspect of the present invention there is provided a detecting data processing apparatus operable to detect a payload data word embedded within a marked copy of a material item by the encoding data processing apparatus. The detecting data processing apparatus comprises an initial value regenerator operable to generate each possible value of each of a plurality of parts of the payload data word, starting with the first part, to form, for each possible value of the part of the payload data word an initial value. An encryption processor is operable to receive each of the initial values and to regenerate an encrypted data stream from each of the initial values using a payload encryption key. A correlator is operable under control of the control processor to detect the part of the payload data word by correlating each of the regenerated encrypted data streams with a part of the code word, corresponding to the encrypted data stream for the regenerated initial value, recovered from the marked version of the material item, and detecting the part of the payload from a result of the correlation. The control processor is arranged to store the detected part of the payload in a data store and to use the detected part of the payload to form subsequent initial values using the detected part of the code word in combination with all possible values of a subsequent part of the payload data word and the corresponding encrypted data streams, for detecting the subsequent part of the payload.
Various further aspects and features of the present invention are defined in the appended claims.
Embodiments of the present invention will now be described by way of example only with reference to the accompanying drawings, where like parts are provided with corresponding reference numerals, and in which:
An encoding data processing apparatus according to an example embodiment of the present technique is shown in
In the present example shown in
In
Returning to the shuffle data store 18, according to the present technique, the shuffle processor 19 is arranged to read out the water mark code word formed from the encrypted data streams substantially randomly for embedding in the frames of the video material. To this end, an address of the shuffle data store 18 for reading out the encrypted data stream is generated by a random address generator 24. The random address generator 24 generates for each frame of the video input signal a random address, which selects randomly parts of the 512 encrypted initial values. In one example, the random address generator 24 is arranged to generate the address using an encryption algorithm, so that not only is the address generated randomly, but the address is generated in a way which does not allow a pattern of the addresses generated to be easily identified.
For the example shown in
The shuffle data store 18 is arranged to store the encrypted data streams and the shuffle processor 19 is arranged to shuffle the streams for embedding into the video image frames. In one example, a fraction of the content of the data store is randomly selected and embedded into each frame of the video signal. For example, for each of 512×128-bit initial values, a 512×8-bit encrypted stream is produced. Thus the shuffle data store is arranged to store 512×512×8-bit bytes, which is 262144 bytes. The shuffle processor 19 is operable to read out ⅛-th of the shuffle data store for each frame, which is 32678 bytes, so that the combiner 22 can embed the 32678 bytes in each frame.
In accordance with the present technique because the encryption data streams are read out randomly from the shuffle data store 18 and embedded in a corresponding frame of the video material, a likelihood of an attacker determining and identifying a particular input data stream from the watermarked video signal is substantially reduced. Furthermore, passing the encrypted data streams through a Gaussian look-up table, has an effect of making the encrypted data streams appear more noise like and therefore less likely to be detected.
As will be explained shortly, advantageously the encrypted data streams may by substantially orthogonal with the effect that the detection of one of the data streams with respect to the other data streams by correlation can be effected with reduced interference from the other data streams. A formation of the encrypted data streams will now be described in more details with reference to
As shown in
IVn=Σl=0l=n-1byte(n−l)
As shown in
Thus, as shown in
As shown in
In accordance with the present technique forming the encrypted data streams from the initial values shown in
According to the present technique, by forming the initial values in the way set out above, the detection of byte 1 can be used to detect byte 2. The detection of byte 1 will require only 28 correlations. When detecting byte 2, a correlation can be performed for all possible values of byte 2 only, since byte 1 which is also present will then be known. The detection of byte 2 can therefore be made also with only 28 correlations, so that the detection of bytes 1 and 2 will require only 2×28 correlations as opposed to 216 correlations, which can therefore be made more easily. As will be explained shortly therefore, by recovering each byte of the payload starting with the first and successively increasing the parts of the payload recovered by increasing the initial value number which is searched from 1 to 16, an amount of processing which is required in order to detect successfully all 14 bytes from the first 16 initial values is reduced to a practical level. Correspondingly the next 14 bytes from the subsequent 16 initial values can be recovered until the entire payload is recovered. For each byte to be detected by the detecting data processor only 28 possible correlations are required as opposed to 24096 to detect the 4096-bits of a payload data word by correlation, which could otherwise be prohibitively computationally difficult.
Since the payload data word includes a digital signature 6, if the recovered digital signature is authenticated then the payload can be considered to have been successfully recovered. It is for this reason that error correction encoding can be applied to the digital signature, whereas error correction encoding is not applied to the projector identifier (ID) or the time stamp. Adding error correction to the projector ID and the time stamp may compromise the integrity of a false positive detection probability or false negative detection probability which is established to prove the presence of the water mark payload for a finger printing system. However, since the digital signature is self authenticating, in that if the digital signature is authenticated with a corresponding private key of the digital signature private key/public key pair, then it must be the digital signature concerned, without a need to rely on the integrity of the watermarking arrangement which conveyed that code word. Therefore, error correction can be applied to the digital signature to improve a likelihood of recovering the digital signature.
In
In order to avoid streams 0 to 31 producing the same encrypted data stream which is added to the video material items, the stream number is provided in the field index FI, which is changed to identify each of the initial values. As a result since each initial value will have a different stream number, a result of passing the initial value through an encryption circuit will produce a different encrypted data stream. Thus, each of the resulting encrypted streams produced by the initial values 0 to 31 will produce different encrypted data streams, which can thereby be more easily detected by correlation, even though each initial value otherwise conveys the same payload.
By arranging for the initial values to each be of a predetermined length and setting the predetermined length to as long as possible, an improvement is provided of a likelihood of detecting each of the encrypted data streams by correlation. This is because, whilst an encrypted data stream will correlate with itself producing a high correlation output value, the other encrypted data streams should preferably produce a low correlation output value so that as far as possible these other encrypted data streams appear as noise with respect to a wanted encrypted data stream during correlation. The unwanted encryption data streams will produce a lower correlation value for longer encrypted streams. Thus, by setting the initial values to a predetermined length which is long enough to cause a correlation with unwanted encrypted data streams to be a substantially low value, a likelihood of successfully detecting the wanted encrypted data stream is increased.
As shown in
In
The hierarchical encoding arrangement provided for the initial values 32 to 191 for the media terminal ID and the time stamp is provided in order to remove an ambiguity in a situation where more than one payload data word is present in a watermarked copy of a video material item. For an example in which a collusion attack has taken place in which an attacker combines two versions of the same watermarked image, each being produced from a different digital projector, detecting each byte from for example the first set of streams 31 to 63 may produce two detected bytes. If the next set of streams were to identify the second bytes of for example media terminal ID without the presence of the first byte there would be some ambiguity in the order of those two bytes within the material. In order to avoid this ambiguity the hierarchical encoding arrangement shown in
The ambiguity in the order of the detected bytes is not so relevant to detecting the digital signature. This is because since a digital signature can be uniquely identified with a corresponding key of a private key/public key pair, although there may be some ambiguity in the order of the bits, if one of these orders of the bits correctly identifies a digital signature then that digital signature will, with a very high likelihood, correctly identify the authenticity of that digital signature. As such, for the example of
An example encoding processor in accordance with the present technique is shown in
A further example encoding data processing apparatus in accordance with the present technique is shown in
The payload generator 8.2 generates a write address on a channel 65 and writes data on a channel 66 in order to store the encrypted data streams into a “payload RAM” forming a shuffle data store 18.2. A finger print shuffle processor 67 generates a random address using a random address generation processor such as that shown in
An example of payload generator 8.2, which forms part of the example encoder shown in
As shown in
An example of the finger print shuffle processor 67, which appears in
Co-pending UK patent application 0424225.1 discloses a technique for weighting coefficients of a water mark or digital finger print code word in accordance with a relative ability of a corresponding part or pixel of an image frame to which the code word is being added, to carry that coefficient without introducing a viewer perceivable effect on the image. A perceptual weighting factor generator for generating the weighting factors for weighting the 8-bit values from the encrypted data streams is not shown in the Figures. An example of such a perceptual weighting generator is provided in the abovementioned UK patent application 0424225.1.
An example of the finger print output generator 61, shown in
The term “samples” is used to refer to discrete samples from which an image is comprised. The samples may be luminance samples of the image, which are otherwise produced from the image pixels. Therefore, where appropriate the term samples and pixels are inter-changeable.
A detecting data processing apparatus, which is arranged to detect code words, which have been embedded in video material items by the encoding data processing apparatus of
The detecting data processing apparatus shown in
As mentioned above, in order to recover the payload data word, the detector shown in
The presence of zeros or indeed any other default value which is used by the encoder and known to the decoder for parts of the initial values which do not contain parts of the payload data word can have an effect of improving the orthogonality of the encrypted data streams with respect to one another.
As indicated above, the detection of one part of the payload data word can be used to detect other parts of the payload data word. For example, starting with the first initial value identified by the index field having a value of 1, the first byte 1 can be detected using 28 re-generated encrypted data streams and corresponding correlations. The detection of byte 1 can be used to detect byte 2, because all possible values of byte 2 can be combined with the value from byte 1 read out from the recovered payload data store 102. As a result, a number of correlations which is required to detect the payload data word can be substantially reduced with respect to what would be required if a correlation was being made for all possible values of the payload alone. This is because each byte of the payload is recovered starting with the first and successively increasing the parts of the payload recovered by increasing the initial value number which is searched from 1 to 16, in order to recover the 14 bytes from the first 16 initial values. Correspondingly the next 14 bytes from the subsequent 16 initial values can be recovered using the same process until the entire payload is recovered.
As mentioned above, in one example, the initial values used by the encoding data processing apparatus to form the encrypted data stream include a field index identifying each of the encryption data streams with respect to the others. For this example, the initial value regenerator is operable to re-generate the initial values for the part of the payload data word being detected using a corresponding field index value in order to produce exactly the encrypted stream for which correlation is being performed. Furthermore, the field index may be used to detect the subsequent part of the payload data word.
As mentioned above, the payload data word includes a digital signature 6. The other parts of the payload data word can be considered to have been successfully recovered with a false positive and false negative detection probability established for the water marking/finger printing technique described above. However, if the recovered digital signature is authenticated then the payload can be considered to have been successfully recovered, because authentication as a digital signature requires a corresponding key from a public key/private key pair. Thus combining the signature with a corresponding key from the pair will authenticate the recovery digital certificate. Furthermore, for the example shown in
Various further aspects and features of the present invention are defined in the appended claims. Various modifications can be made to the embodiments herein before described without departing from the scope of the present invention.
As mentioned above, embodiments of the present invention find application with embedding data in any kind of information material. For example, although the above embodiments have been described with respect to representing a payload data word as a code word and embedding the code ward in the form of encrypted data streams in the frames of the video material, the present technique is also applicable to embedding data in audio material. For the example of audio material, in digital form, audio material typically comprises a plurality of data frames, in which the encrypted data streams will be embedded.
Number | Date | Country | Kind |
---|---|---|---|
0611127.2 | Jun 2006 | GB | national |
Filing Document | Filing Date | Country | Kind | 371c Date |
---|---|---|---|---|
PCT/GB07/01605 | 5/3/2007 | WO | 00 | 3/24/2009 |