WO2006126153A2

WO2006126153A2 - Signal combination method

Info

Publication number: WO2006126153A2
Application number: PCT/IB2006/051600
Authority: WO
Inventors: Alphons A. M. L. Bruekers
Original assignee: Koninklijke Philips Electronics N.V.
Priority date: 2005-05-25
Filing date: 2006-05-19
Publication date: 2006-11-30
Also published as: WO2006126153A3

Abstract

A method of generating output values from first and second input values. The method comprises receiving the first input value which is a sample of a first signal, and receiving the second input value which is a sample of the second signal. A portion of the first input value and a portion of the second input value are merged such that the output value is an approximation of the first input value when read in a first predetermined manner, and an approximation of the second input value when read in a second predetermined manner.

Description

Signal combination method

The present invention relates to a method of generating an output value from first and second input values. More particularly, but not exclusively, the invention relates to the embedding of a first signal within a second signal such that both the first and second signals can be read from the combined signal. It is known that data can be embedded within an information signal. For example, watermark data is often embedded in an information signal for purposes of controlling distribution of the information signal, and preventing unauthorised copying. Such embedding is particularly applicable in the case of digital media files such as music files or video files. In such cases watermark data can effectively mitigate the problems of piracy which cause loss of revenue to legitimate owners of copyright in the media files.

In order to extract embedded watermark data, complex processing of the information signal is typically required. Such processing typically comprises a number of operations carried out in a predetermined order including various filtering of the information signals. Such processing can typically be computationally inefficient.

It is an object of the present invention to provide an alternative method for combining data signals.

According to the present invention, there is provided, a method of generating an output value from first and second input values, the method comprising receiving said first input value being a sample of a first signal receiving said second input value being a sample of a second signal and generating an output value comprising a plurality of output digits wherein said output value is an approximation of said sample of said first signal when said output digits are read in a first predetermined manner and said output value is an approximation of said sample of said second signal when said output digits are read in a second predetermined manner.

Therefore, the invention provides a method for generating an output value comprising a plurality of output digits, the output value being such that the digits can be read in different ways so as to obtain approximations of the first input value and the second input value. That is, direct reading of the digits of the output value without further processing allows to different values to be obtained.

The output value may be generated by concatenating a portion of the first input value with a portion of the second input value. The method may further comprise reversing the portion of the second input value prior to said concatenating. The portion of the first input value used in said concatenation may be the most significant portion. The first input value may comprise a plurality of first digits and the portion used for concatenation may be a subset of said first digits. The subset of first digits is preferably a predetermined number of most significant digits of the first input value. Generating the output value by concatenation is preferred in some embodiments of the present invention given that it has a relatively low computational complexity.

In alternative embodiments of the invention, the method may comprise determining a pair of values comprising first and second determined values. The pair of values is selected so as to satisfy a predetermined relationship and is such that the first and second determined values are respectively approximations of the first input value and the second input value. The output value may then be the first determined value. The first and second determined values may be determined by "minimum distance" method so as to obtain determined values which are good approximations of the first and second input value. The predetermined relationship may be that the second determined value is a reversal of the first predetermined value. For example, the first and second determined values may each be a sequence of digits, and the predetermined relationship may be that the sequence of digits of the first predetermined value is a reversal of the sequence of digits of the second determined value. Proximity of the first determined value to the first input value may be preferred to proximity of said second determined value to said second input value. Such a method is desirable when one of the first and second input value is more susceptible to error than the other of the first and second input values.

In some embodiments of the present invention, a plurality of first input values may be processed, each being a sample of a first signal. A plurality of second input values may be processed, each being a sample of the second signal.

It will be appreciated that the method described above can be applied to a variety of different types of signals. In particular, in some embodiments of the invention, at least one of the first and second signals is an image signal or an audio signal. The first and second input values may be unsigned binary values. Alternatively the first and second input values may be signed binary values, such as for example two's complement binary values.

The invention further provides a carrier medium carrying computer readable program code configured to cause a computer to carry out a method as set out above. There is also provided a computer apparatus for generating an output value from first and second input values. The computer apparatus comprises a program memory containing processor readable instructions, and a processor configured to read and execute instructions stored in the program memory. The processor readable instructions comprise instructions configured to cause the computer to carry out a method as set out above.

According to a further aspect of the present invention, there is provided a display device having first and second display surfaces, each defining a plurality of light emitting pixels, the display device comprising, a plurality of light sources interposed between said first and second display surfaces, the light sources being arranged in sets, each set comprising a plurality of light sources and being associated with a pixel of said first display surface and a pixel of said second display surface, an attenuator positioned between at least one pair of light sources of each set of light sources, said attenuator applying a predetermined attenuation factor, wherein light emitted from a pixel of said first display surface comprises light emitted from a first light source of the respective set of light sources and light emitted from a second light source of the respective set of light sources after attenuation, and light emitted from a pixel of said second display surface comprises light emitted from said second light source and light emitted from said first light source after attenuation.

Each set of light sources may be arranged so as to interpose said first and second display surfaces and an attenuator may be positioned between each light source of each set of light sources. The plurality of light sources may be arranged in a three dimensional grid. Indeed, one of or each of said sets of light sources may be arranged in a row interposing the first and second display surfaces. In such a case, the row of light sources defines a first dimension of said three-dimensional grid and the first and second display surfaces define a further two dimensions of the three dimensional grid. Each light source may emit light at a predetermined plurality of different intensities, and the attenuator may apply an attenuation factor which is equal to the reciprocal of said predetermined plurality. Each light source may emit light at two predetermined intensities (i.e. on or off) and the or each attenuator may apply an attenuation factor of a half. The display device may further comprise means for receiving data associated with a light emitting pixel of said first display surface and for applying said data to a respective set of light sources. Each light source may emit light at two predetermined intensities (i.e. on or off) and the or each attenuator may apply an attenuation factor of a half. In such a case, the data may comprise a binary value. A further aspect of the present invention provides a data carrier carrying an information signal comprising a plurality of samples each sample comprising a plurality of digits, said digits of each sample being readable in first and second predetermined manners, such that when digits of each sample of said information signal are read in a first predetermined manner a first output information signal is obtained, and when digits of each sample of said information signal is read in a second predetermined manner, a second output information signal is obtained.

Embodiments of the present invention will now be described, by way of example, with reference to the accompanying drawings, in which:

Figure 1 is a schematic illustration of a process for combining a pair of samples to generate an output sample in accordance with the present invention;

Figure 2 is an illustration of Voronoi cells defined by a pair of unsigned 4-bit integers; Figures 3 A and 4A are images suitable for processing using an exemplary embodiment of the present invention;

Figures 3B and 4B are images generated by bit-reversing each pixel value of the images of Figures 3 A and 4A respectively;

Figures 5A and 5B are images read from data comprising the images of Figures 3 A and 4A combined together using a first embodiment of the present invention;

Figures 6A and 6B are images read from data comprising the images of Figures 3 A and 4A combined together using a second embodiment of the present invention;

Figure 7 is a schematic illustration of a method for weighting sample values prior to combination in accordance with the present invention; Figure 8 is an illustration of Voronoi cells defined by a pair of unsigned 4-bit integers and modified in accordance with weighting applied using the method of Figure 7;

Figure 9 is a schematic illustration of pre-processing which is carried out in some embodiments of the invention; and Figures 10 and 11 are partial cross-sections through display devices configured in accordance with the present invention.

An embodiment of the present invention is now described with reference to 4- bit integers, it will however be appreciated that the invention is not restricted to such representations.

Table 1 shows, in its first column, the sixteen binary sequences x which can be created using four bits. A second column shows values of V(x) where V is a function which provides the decimal value of the binary sequence x, when x is interpreted as an unsigned integer. A third column shows values V(x) obtained when the sequence of bits x is read in reverse order, when x is interpreted as an unsigned integer. It can therefore be seen that x is used to indicate the bitwise reversal of x. That is, when x = 0101 as is the case in the sixth row of table 1, x = 1010. In this case V(x) = 5, and V(x ) = 10.

X V(x) V(x) V₂(X) V₂(X)

0000 0 0 0 0

0001 1 8 1 -8

0010 2 4 2 4

0011 3 12 3 -4

0100 4 2 4 2

0101 5 10 5 -6

0110 6 6 6 6

0111 7 14 7 -2

1000 8 1 -8 1

1001 9 9 -7 -7

1010 10 5 -6 5

1011 11 13 -5 -3

1100 12 3 -4 3

1101 13 11 -3 -5

1110 14 7 -2 7 mi 15 15 -1 -1

TABLE 1 The fourth and fifth columns of table 1 are concerned with interpretation of the bit sequences x as signed integers, using a conventional two's complement representation. The fourth column contains values of a function V₂ (x) which returns the decimal value of the binary sequence of the first column, when this sequence is read from left to right and interpreted as a two's complement value. The fifth column contains values of the function V₂ (x) which provides the decimal value of the binary sequence x of the first column, when this sequence is read in reverse order (i.e. from right to left).

The described embodiment of the invention represents two four bit samples A, B as a single four bit sample X, such that

X = f(A,B) (1)

where:

A = A = X = V(x) (2) B ~ B = X = V(x) (3)

That is, reading the value X will produce an approximation of the sample A and reading the value X will produce an approximation of the sample B.

It should be noted that the value of A is generally noise, and the value of B will similarly generally be noise. Accordingly, reading X as indicated above will generally produce good approximations of A and B.

An example of a first method for implementing the function/to generate a value X from the samples A and B is now described using unsigned 4-bit integers.

Let:

(A,B) = (S,6) (4)

such that

(a,b) = (1000,0110) (5)

To generate a binary value x (such that V(x)=X), the two most significant bits of a (10) are concatenated with the two most significant bits of b (01). However the two most significant bits of b are reversed before concatenation, such that 10 is concatenated with 10 to generate a value of x such that:

x = 1010 (6)

According to equations (2) and (3), V(x) is an approximation of A, while V{ x ) is an approximation of B. In this case, V{x) = 10, and V{ x ) = 5. It can therefore be seen that in this simple case:

V(x) = 5 - 6 = 5 (8)

The example set out above can be formally defined for any pair of Nbit numbers (a, b), where N is even as follows:

Let: a = (a_N__γ,a_N_₂ ...,a_γ,a₀) (9)

then x is defined by: x = (a_N__λ,...,a_NI2,b_NI2,...,b_N__λ) (11)

For cases when N is odd, a similar formulation can be generated, although in such cases the number of bits taken from a and b to generate x will differ.

The processing described above is schematically illustrated in Figure 1. The samples A, B are input to a first process 1 and a second process 2 respectively. The processes 1, 2 respectively select parts of the samples A and B as described above. The selected parts of the samples are then combined by a combination unit 3 to generate an output sample X. An alternative method for generating the sample value X as defined by equation (1) using a "minimum distance method" and is now described. Here f(X) is defined by equation (12): f(A,B) = X :: min{(Λ - X)² + (fl - Xf } (12) where: min {g(X)} is the minimal value of g (X) over all X.

X

That is, X is selected such that:

(X,X) = (A,B) ~ (A,B) (13)

and such that [X, X) is at a minimum distance from (A,B).

Figure 2 shows a grid defined by broken horizontal and vertical lines. Points representing possible values of [X, X), are shown as emboldened points on the grid. X is shown on the x axis and X is shown on the y axis. It can be seen that there are sixteen distinct emboldened points corresponding the sixteen distinct values shown in Table 1. Each point is within a polygonal cell (or Voronoi cell) defined by continuous lines. The polygonal cell for any particular emboldened point contains all grid points which are such that the enclosed emboldened point is the closest emboldened point to those grid points. It should be noted that grid points situated on lines defining the polygons (e.g. (4,4)) are equidistant from two emboldened points.

It should be noted that experiments (described below) have found that the method described with reference to Figure 2 provides more accurate results from a perceptual point of view than the concatenation method described above with reference to Figure. Indeed, in some cases the minimum distance method described with reference to Figure 2 provides considerably smaller errors.

For example, if (A,B) = (3,3) (a point marked in Figure 2 by a small square) then the method described with reference to Figure 2 will output the point (4,2) or (2,4), both of which are equidistant from (A,B). The concatenation method would however output a value of (0,0) which is less good. It should however be noted that the concatenation method described above has a lower computational complexity than minimum distance method described with reference to Figure 2.

It should be noted that the samples A, B will usually be taken from respective signals, all samples of these signals are processed in turn.

Examples of the processing described above are now presented with reference to images which are shown in Figures 3 A and 4A. Each of these images is defined by a grid of 256 x 256 pixels, each pixel having an associated 8-bit unsigned pixel value. Figures 3B and 4B show the bitwise reversal of each pixel value of the images of Figures 3 A and 4A respectively, and it can be seen that the result of these reversals is effectively the generation of noise patterns.

Referring now to Figures 5 A and 5B, this pair of images was generated using one set of 256 x 256 8-bit unsigned pixel values. This set of pixel values was generated using the concatenation method described above to combine the data of Figure 3 A with that of Figure 4A. In order to obtain the image of Figure 5 A each pixel value of the combined data was read in a conventional manner, while in order to obtain the data for the image of Figure 5B, each pixel value of the combined data was read in reversed order.

Referring now to Figures 6A and 6B, this pair of images was again generated using one set of 256 x 256 8-bit unsigned pixel values. However, in this case, the values were generated using the minimum distance method described above with reference to Figure 2. In order to obtain the image of Figure 6A each pixel value of the combined data was read in a conventional manner, while in order to obtain the image of Figure 6B, each pixel value was read in reversed order.

Reviewing Figures 5A and 5B, and Figures 6A and 6B it can be seen that the second combination method produces images with less perceptual degradation. It should be noted that the image of Figure 3 A does show some degradation in both combination methods. However it should also be noted that the image of Figure 3 A was specifically selected given that this is the case, and many images (e.g. that of Figure 4A) show far less degradation.

Experiments have also been carried out using PCM audio signals, where each sample value is represented as a signed integer. The distortion which can be heard when audio signals are combined is in some cases audible, although generally at a low level.

In some cases, where two signals are to be combined in the manner described above, one signal will be more susceptible to errors caused by the combination process than the other signal. In such cases the signals can be combined in an unequal manner. For example, using the concatenation method described above, less bits may be selected from one signal than from the other signal. Such weighting can also be applied when using the minimum distance method described above with reference to Figure 2. In this case, the function of equation (12) is modified as follows:

f(A,B, $) = X : : min V(A -Xf ₊-(B -X) (14) β

where β is a balance control signal.

Where the sample A is less sensitive to degradation than the sample B, the value of β should be selected such that β < 1. In cases where the sample A is more sensitive to degradation than the sample B, the value of β should be selected such that β > 1. It should be noted that the value of β may vary in time for different samples of temporal signals such as audio signals. Similarly B may vary on the basis of position within spatial signals such as image signals. Such variance can be controlled by a perception model. Such perception models produce output signals which are combined to obtain the value of β for a particular sample. This is illustrated in Figure 7.

Referring to Figure 7, samples A and B are input to respective perception models 4, 5. The perception models generate output signals which are input to a balance controller 6. The balance controller 6 is configured to output a value β which should be used in combination of the samples A, B in accordance with equation (14). The values of the samples A and B, and the value of β output from the balance controller 6 are input into a combination unit 7 configured to apply the function defined by equation (14).

Referring back to Figure 2, the illustrated polygonal cells are drawn for the case where no weighting is added, that is the case where β is set to be equal to one. Referring now to Figure 8, a further set of cells is illustrated. The polygonal cells defined by broken lines in Figure 8 are those for the case where β is to be set equal to one, that is those shown in Figure 2. The cells defined by solid lines are those for the case where β is set to be equal to 0.5. It should be noted that the variation of the value of β can be applied both in the minimum distance method described above and in the concatenation method. It should also be noted that during playback of the combined signal the value of/? is not required. Referring now to Figure 9, pre-processing operations which can be carried out in embodiments of the present invention are described. Samples A and B are input to respective pre-processors 8, 9. The values output from these pre-processors are then input into an embedder 10. The pre-processing carried out will typically compand the received sample so as to 'move' information from the least significant bits (given that this information will be lost in combination) towards more significant bits, where the information will be retained. Experiments have shown that such companding provides significant improvements when applied to audio signals which are combined in accordance with the invention. The companding can be carried out using known techniques such as the A- law or μ-law. Such techniques are described in Saywood, K: "Introduction to data compression", 2^nd Edition, Morgan Kaufinann Publishers, pages 243 to 249.

The embodiments of the invention described above have been based upon bit reversal operations. For example, in the concatenation method of combination described above a generated value would typically be of the form:

x = (a₁,a₆,a₅,a₄,b₄,b₅,b₆,b₁) (15)

Where the order of the most significant bits of b has been reversed. In such embodiments of the invention the value of x read in a conventional manner will be approximately equal to the value of A, while the value of x read in reverse bit order will give an approximation of B.

However, in alternative embodiments of the invention the signals may be embedded using bit rotation operations. In such circumstances a combined signal x may have the form shown in equation (16):

x = (a₁,a₆,a₅,a_A,b₁,b₆,b₅,b_A) (16)

To read the value of B, various types of processing can be carried out. For example rotation operations can be carried out to yield x = {b₁,b₆,b₅,b_A,a₁,a₆,a₅,a_A) . B can then be read as described above. Such rotation operations can alternatively yield x = (b₁,b₆,b₅,b₄,a_l,a_J ,a_k,a_l) .It should also be noted that more than two signals may be combined. For example, where three signals combined using the concatenation method described above values of x as shown in equation (16) and (17) are generated. x = (a₁,a₆,a₅,b₁,b₆,b₅,c₁,c₆,c₅) (17)

x = (a₇,a₆,a_s,b_s,c_s,b₆,c₆,b₇c₇) (18)

A can be read directly from eq 17 and 18. B and C can be read, in a similar manner to that described above with reference to equation (16).

The minimum distance method described above with reference to Figure 2 is also applicable to the combination of more than two signals, as is shown in equation (19).

f(A,) = X :: miμ ∑{A_ι -T_ι{x))² (19) i

Where A₁ is a signal to be combined, and T₁ is the transformation operation used in the combination process.

In general it is possible to apply different error measures. The power of two in equation (19) can be replaced by any power p. The p-th root may be included.

The embodiments of the invention described above have all been based around the binary number system. More particularly, embodiments described have been based upon binary representations of unsigned integers and binary representations of two's complement integers. The invention is however also applicable to other number basis such as ternary or decimal. Further examples of such alternative number bases are described below.

When combined samples have been generated in the manner described above various signal processing operations can be carried out such as filtering, noise addition. However, in general terms, the application of such techniques to a sample when processed so as to obtain the first signal will degrade the value of the obtained second signal. However, some particular types of processing can improve one signal while not degrading the other. These processing operations include rotation and mirroring of images, time reversing of audio, selection of only part of the combined signal to be read and lossless encoding and encryption. It should be noted that linear interpolation will in general introduce additional distortion, but it is be possible to preserve the property of mutually embedded signals. The invention also provides a novel display device. This is now described with reference to figure 10.

Referring to Figure 10 there is illustrated a cross section through a display device 11. The display device has a first display surface 12 and a second display surface 13. The portion of the display device 11 shown in cross section in Figure 12 comprises five pixels po, pi, P2, Ps, P4 on the first display surface 12, and five pixels qo, qi, q2, qs, q4 on the second display surface 13. Each of these pixels is defined by four light sources arranged in a row, one in each of four columns xo, xi, X2 and X₃. In the described embodiment, each of these light sources can either be turned off or on (i.e. each light source has one of two intensities). Each pixel of the display device is controlled using a four bit value x of the form:

{x_n,x_λ,x₂,x^) (20)

Here, the value of X to be displayed on the first surface 12 is defined by equation (21):

X = X, 3 + — ₂ X 2_? + — ₄ X, 1 + - g X_n 0 = Xr 3, + + + - ₂ X_n 0 (21)

Similarly, the value of X to be displayed on the second surface 13 is defined by equation (22):

+ - 1 Xr, (22)

Each light source illustrated in Figure 10 is separated from other light sources of the same pixel by means of an optical attenuator. Three such attenuators 14, 15, 16 are shown in Figure 10. Each of the attenuators 14, 15, 16 applies a attenuation factor of a half. Thus, assuming that a value x of the type defined by equation (20) is applied to one of the pixels of Figure 10, a viewer viewing the first surface 12 of the display device 11 will see a value for that pixel defined by equation (21), given the action of the optical attenuators 14, 15, 16. Similarly, a viewer viewing the second surface 13 of the display device 11 will see the bit reversed value of equation (22).

The display device described above can be used so as to display a lull resolution image (i.e. 4 bits per pixel) on a first side of the display device while displaying a noisy pattern on the second side of the display device. However, using combination methods such as those described above both sides of the display device can display different images, albeit with reduced resolution.

It should be noted that by introducing some spatial distortion into the attenuators 14, 15,16, such as by slightly changing the attenuation factor over the plane, a constant watermark maybe embedded within the signal. Such a watermark may depend upon viewing direction.

It was described above that although some embodiments of the present invention have been described in terms of binary representation, the invention is not limited to such representations. For example, the display device 10 could be modified such that each light source receives information indicating one of four intensities (e.g. 0, 1, 2 or 3). Such modified display device is shown in Figure 11 and is indicated by reference numeral 17. This display device again has a first display surface 18 and a second display surface 19. Five pixels Po, Pi, P 2, P 3, P 4 of the first display surface 18 are shown similarly five pixels Qo, Qi, Q2, Qs, Q₄ of the second display surface 19 are shown. In this case, each pixel is defined by two light sources in columns yoyi. Here, the light sources are separated by an attenuator 20 which applies an attenuation factor of a quarter. In such a case, viewing the display device from the first side 18 a viewer sees a pixel value defined by (22):

Whereas, a viewer viewing the display device from the second side 19 views pixels defined by equation (23):

That is, two low resolution displays can be combined to obtain a single double resolution display or a double sided display.

Although preferred embodiments of the present invention have been described above, it will be appreciated that various modifications can be made to the described embodiments without departing from the spirit or scope of the present invention.

Claims

CLAIMS:

1. A method of generating an output value from first and second input values, the method comprising: receiving said first input value being a sample of a first signal; receiving said second input value being a sample of a second signal; and - generating an output value comprising a plurality of output digits; wherein said output value is an approximation of said sample of said first signal when said output digits are read in a first predetermined manner and said output value is an approximation of said sample of said second signal when said output digits are read in a second predetermined manner.

2. A method according to claim 1, wherein reading said output digits in the first predetermined manner comprises reading said output digits in a first predetermined order, and reading said output digits in the second predetermined manner comprises reading said output digits in a second predetermined order.

3. A method according to claim 1 or 2, wherein reading said output digits in the first predetermined manner comprises reading said digits in a first direction, and reading said output digits in the second predetermined manner comprises reading said digits in a second direction, said second direction being a reversal of said first direction.

4. A method according to claim 1, 2 or 3, wherein said output value is a binary number, and each of said output digits is a bit.

5. A method according to any preceding claim, wherein generating said output value comprises: concatenating a portion of said first input value and a portion of said second input value.

6. A method according to claim 5, further comprising reversing said portion of said second input value prior to said concatenating.

7. A method according to claim 5 or 6, wherein said portion of at least one of said first and second input values is the most significant portion.

8. A method according to any one of claims 1 to 4, comprising: determining a pair of values comprising a first determined value and a second determined value, the pair of values satisfying a predetermined relationship, and being such that said first and second determined values are respectively approximations of said first input value and said second input value; wherein said output value is said first determined value.

9. A method according to claim 8, wherein said predetermined relationship is such that said second determined value is a reversal of the first predetermined value.

10. A method according to claim 8 or 9, wherein proximity of said first determined value to said first input value is preferred to proximity of said second determined value to said second input value.

11. A method according to any preceding claim, wherein at least one of said first and second signals is an image signal or an audio signal.

12. A method according to any preceding claim, wherein at least one of said first and second input values is an unsigned binary value.

13. A method according to any one of claims 1 to 11 , wherein at least one of said first and second input values is a signed binary value.

14. A carrier medium carrying computer readable program code configured to cause a computer to carry out a method according to any preceding claim.

15. A computer apparatus for generating an output value from first and second input values, the computer apparatus comprising: a program memory containing processor readable instructions; and a processor configured to read and execute instructions stored in said program memory; wherein said processor readable instructions comprise instructions configured to cause the computer to carry out a method according to any one of claims 1 to 13.

16. A display device having first and second display surfaces, each defining a plurality of light emitting pixels, the display device comprising: a plurality of light sources interposed between said first and second display surfaces, the light sources being arranged in sets, each set comprising a plurality of light sources and being associated with a pixel of said first display surface and a pixel of said second display surface. an attenuator positioned between at least one pair of light sources of each set of light sources, said attenuator applying a predetermined attenuation factor; wherein light emitted from a pixel of said first display surface comprises light emitted from a first light source of the respective set of light sources and light emitted from a second light source of the respective set of light sources after attenuation, and light emitted from a pixel of said second display surface comprises light emitted from said second light source and light emitted from said first light source after attenuation.

17. A display device according to claim 16, wherein each set of light sources is arranged so as to interpose said first and second display surfaces, and wherein an attenuator is positioned between each light source of each set of light sources.

18. A display device according to claim 17, wherein said plurality of light sources are arranged in a three dimensional grid.

19. A display device according to claim 18, wherein lights of one of said sets of light sources are arranged in a row interposing said first and second display surfaces, and said display surfaces are defined by two dimensions of said grid.

20. A display device according to any one of claims 16 to 19, wherein each light source emits light at a predetermined plurality of different intensities, and the or each attenuator applies an attenuation factor which is substantially equal to the reciprocal of said predetermined plurality.

21. A display device according claim 20, wherein each light source emits light at two predetermined intensities, and the or each attenuator applies an attenuation factor of substantially a half.

22. A display device according to any one of claims 16 to 21, comprising means for receiving data associated with a light emitting pixel of said first display surface, and for applying said data to a respective set of light sources.

23. A display device according to claim 22, wherein each light source emits light at two predetermined intensities, the or each attenuator applies an attenuation factor of substantially a half, and said data comprises a binary value.

24. A display device according to any one of claims 17 to 23, wherein said attenuator applies different attenuation factors at different pixels of said first and second display surfaces.

25. A display device according to claim 24, wherein said differing attenuation factors are configured to apply a predetermined watermark to images displayed on said first and/or second display surface.

26. A data carrier carrying an information signal comprising a plurality of samples each sample comprising a plurality of digits, said digits of each sample being readable in first and second predetermined manners, such that when digits of each sample of said information signal are read in a first predetermined manner a first information output signal is obtained, and when digits of each sample of said information signal is read in a second predetermined manner, a second information output signal is obtained.

27. A data carrier according to claim 26, wherein reading digits of each of said samples in the first predetermined manner comprises reading said digits of said samples, and reading digits of each of said samples of said information signal in the second predetermined manner comprises reading a reversal of each of said digits.

28. A data carrier according to claim 27, reading said digits of said samples in the first predetermined manner comprises reading said digits in a first direction, and reading said digits of said samples in the second predetermined manner comprises reading said digits in a second direction, said second direction being a reversal of said first direction.

29. A data carrier according to any one of claims 26 to 28, wherein each of said samples comprises a first portion and a second portion, said first portion being derived from samples of said first signal, and said second portion being derived from samples of said second signal.

30. A data carrier according to any one of claims 28 to 29, wherein at least one of said first and second signals is an image signal or an audio signal.