WO2003013137A1

WO2003013137A1 - Digital video copy protection

Info

Publication number: WO2003013137A1
Application number: PCT/GB2001/005202
Authority: WO
Inventors: Martyn Gilbert; Keith Wang
Original assignee: Amino Holdings Limited
Priority date: 2001-07-27
Filing date: 2001-11-26
Publication date: 2003-02-13
Also published as: GB0118353D0

Abstract

A method of protecting a digital video image from copying, comprising the following steps: selecting a part of the image to be masked, and applying a reversible operation to at least some of the pixels within the masked region. The reversible operation used or the pixels the operation is applied to, or both, being selected based on a key.

Description

Digital Video Copy Protection

This invention relates to a method of copy protection of digital video images, particularly compressed digital video images.

The advantage of producing video image signals in digital form to allow a better quality

image to be displayed and to reduce the bandwidth required to transmit the video signal

are well known. However, the ease with which copies of digital video signals can be

produced and distributed causes commercial problems for digital service providers.

The making and distribution of illicit copies of analogue video signals is intrinsically

limited because the copying of analogue video signals result in degradation. As a result,

the image produced from the copy of an analogue video signal is of lower quality than

the image produced from the original analogue video signal. Also, because analogue

video signals must be mechanically recorded onto a video tape the number of copies

which can be made is limited unless a substantial investment in hardware is made.

None of these constraints restrict illicit copying and distribution of digital video signals.

Repeated generations of copies of the digital video signals can be made without any

image degradation and the copies can be easily recorded in large numbers, for example

onto rewritable CDs, using inexpensive equipment. As a result, not only are the video

images produced by the illicit copies of just as good a quality as those produced by the

original digital video signal but the number of copies which can be produced is greatly

increased and the cost of the necessary copying equipment^' reduced.

As a result, it has been attempted to devise technical means to prevent illicit copying of

digital video signals. Most such approaches have relied upon encrypting the stored or transmitted digital video signal data so that unauthorised users cannot decrypt the digital video signal and view the video image.

In principle, a digital video signal can be protected by conventional cryptographic

techniques. However, in practice, there are a number of problems in applying

conventional cryptographic techniques to digital video signals which are encoded using

data compressing techniques because of the increase in the amount of data which needs

to be transmitted in order to send and subsequently decrypt the encrypted digital video

signal compared to the amount of data which must be transmitted to send the unencrypted

digital video signal.

There are two reasons why this increase in the amount of data which must be transmitted

occurs. The first is that encrypting the digital video signal or parts of it makes the

encrypted parts of the signal appear random. As a result, data compression techniques

are largely ineffective in compressing the encrypted digital video signal for transmission.

This increase in the amount of data which must be transmitted is very large in practice.

For example, a conventional digital video encoding technique such as MPEG achieves up

to 50 times compression, mainly by removing data redundancies within the video.

Encryption of the digital video signal appears to randomise the image data and conceals

these redundancies so that compression is less effective or ineffective.

The second reason for the increase in the amount of data which must be transmitted is the

requirement to transmit the necessary codes or keys to carry out the decryption process,

although this is a less significant problem for most encryption processes.

Accordingly, in the past copy protection of digital video signals has been carried out by

encoding the original digital video data using a conventional video coding and compression technique such as MPEG and then applying encryption to the coded and compressed video data.

Accordingly, there is a problem that digital video signals cannot be encrypted before

encoding and compression without unacceptably increasing the amount of compressed

encoded data required to represent the digital video signal.

This invention was intended to allow this problem to be overcome, at least in part.

In a first aspect, this invention provides a method of protecting a digital video image

from copying, comprising the steps of:

selecting a part of the image to be masked; and applying a reversible operation

to at least some of the pixels within the masked region, the reversible operation used or

the pixels the operation is applied to, or both, being selected based on a key.

This invention provides copy protection for digital video signals by applying a

degradation mask to a part of the digital video image. Within the mask area reversible

operations are carried out on some or all of the pixels with the operation carried out

and/or the pixels operated on being determined by a pseudo random sequence controlled

or seeded by a secret key.

The protected or masked digital video signal can be conventionally encoded, transmitted

and decoded without any significant increase in the amount of data in the encoded digital

signal.

An authorised receiver knowing the key can reverse the masking process on the decoded

video signal data to produce the original digital video image for viewing. However,

unauthorised eavesdroppers or copiers cannot restore the original digital video signal. A colour digital signal comprises chrominance and luminance information, while a

monochrome digital signal comprises luminance information only.

A first embodiment of the invention applies masking to the chrominance data only and

this is applicable only to colour video signals.

The use of chrominance only masking is preferred for colour video signals which are

intended to be coded using MPEG encoding because this achieves the desired masking

effect of making the protected video image unwatchable with minimum impact on the

possible following MPEG encoding.

MPEG coding is performed on video data having luminance and chrominance values Y

Cb and Cr. The chrominance resolution is half the resolution of the luminance

horizontally in 4:2:4 video and half the resolution both horizontally and vertically in

4:2:0 video. Accordingly, chrominance usually carries rather less information than

luminance. Human eyes are very sensitive to colour so interference with the

chrominance data is very noticeable. Further, motion compensation, which is a crucial

part of MPEG coding is based entirely on luminance and so will not be effected by a

chrominance only mask.

All these reasons make a chrominance only mask particularly effective in protecting

colour- video signals to be coded and compressed for transmission or storage using

MPEG coding. However, it should be understood that these are only reasons why a

chrominance only mask is particularly advantageous in connection with MPEG coding,

the use of a chrominance mask and its effectiveness is not dependent on a MPEG coding being used. There are a range of possible reversible operations which can be applied to pixel

chrominance data in a digital video signal. Suitable operations include swapping the Cb

and Cr values or flipping one or both of these colour values by zero. These operations

can be easily reversed without losing any information. Further, these operations survive

MPEG or other encoding and decoding well.

In a chrominance masking procedure using the above reversible operations, the

chrominance mask procedure defines a pixel block to be masked in each frame of the

digital video image and carries out one or none of the possible operations on each one of

the pixels for each frame under the control of a masking algorithm.

The reversible operations identified above allow five possible operations which can be

applied to the chrominance data of each pixel. These are, swapping the Cb and Cr

values, flipping the Cb value only, flipping the Cr value only, flipping both the Cb and

Cr values and taking no action. The chrominance masking algorithm applies one of these

possible five operations, one of the possible operations being no action, to each of the

pixels within the mask area in turn with the operation applied to each pixel being selected

in a pseudo random sequence set, seeded or otherwise controlled by the secret masking key.

For the five operations identified, if the mask is applied as a 32 x 32 pixel block having a

total of 1024 pixels, 5¹⁰²⁴ tests must be carried out in order to determine the correct colour

values.

This is a sufficiently large number of tests to deter attempts to restore the video by illicit

third parties not having the secret key to allow reversal of the masking process. The same process is repeated for the next frame and the masked video signal is MPEG encoded and transmitted.

The encoded video signal is then received and MPEG decoded. An authorised user

having the secret key which is required to remove the chrominance mask from the

received and MPEG decoded digital video signal can simply reverse the masking

algorithm to reproduce the original digital video signal.

Preferably the position of the mask area in the digital video image is regularly changed.

This may be done by changing the position of the mask for each frame or periodically or

in a pseudo random sequence when the mask may or may not move from frame to frame.

Use of a moving mask provides two advantages. Firstly, a moving mask area in which

the video signal is unwatchable is far more annoying and harder to ignore than a similarly

sized fixed mask so that the moving mask is more effective in rendering the protected

video signal valueless.

This second advantage is that colour swapping or flipping have no visible effect, or no

significant visible effect, on grey or white areas of the video image or areas of very weak

colour. It is possible that a stationary mask could lie over such an area of grey, white or

weak colour for a considerable length of time rendering the mask ineffective as a copy

protection tool.

A moving mask has the advantage that the mask is sure to be on areas of significant

colour so as to provide effective copy protection a lot, if not all, of the time.

It should be noted that the locations of the moving mask in each frame of the video does

not have to be secret because that is not what the copy protection depends upon. Thus,

where a moving mask is used there is no particular requirement that the movement of the mask be unpredictable since the location of the mask is self-evident from the protected digital video image.

Accordingly, it is not necessary to use any secret key or algorithm to control movement

of the mask. It is only necessary that the location of the mask be determinable by an

authorised user in order to allow reversal of the masking process. This will reduce the

cryptographic overhead of the process by eliminating any need to transmit secret keys or

algorithms to predict the mask movements.

One preferred approach for controlling the mask movement is to control the mask

movement using an algorithm driven by the video signal frame number of the video

frame to which the mask is applied.

For example, if the horizontal size of the picture is h size and the vertical size of the

picture is v size the location of the mask can be defined as follows:

horizontal position = frame number x (a prime number) modulo h size

vertical position = frame number x (a prime number) modulo v size

This will allow the location of the mask to be predicted and determined by the authorised

user and will ensure that the moving mask moves across all areas of the video image for a

video sequence of any significant length.

Of course, it is also possible for illicit copiers to similarly predict the location of the

mask. However, finding the mask does not assist an illicit copier in the reversing the

masking process.

In a second embodiment of the invention a luminance mask is applied to the luminance

data part of a digital video signal. The use of a luminance mask is essential for monochrome digital video signals or monochrome sequences in colour digital video signals, but can also be used for colour digital video signals.

The use of a luminance mask has a larger impact on subsequent MPEG encoding than the

chrominance mask of the first embodiment and as a result a higher bit rate is required in

order to avoid the production of artefacts in the final decoded video signal due to the

masking operation.

Similarly to the chrominance mask process of the first embodiment, a copyright

protection process employing a luminance mask defines an area of the digital video

image in each frame which is to be masked.

A second masking algorithm is then applied in which a secret key is used to generate a

pseudo random sequence of pixel numbers which are then swapped. That is, the

luminance data relating to the identical numbered pixels is exchanged.

One way of doing this is to use the key to generate a pseudo random sequence of (-1, 1)

with a mean of zero with each number corresponding to one pixel in the masking region

or block. The number of (-1) and (1) should be largely equal due to the zero mean so that

the number of pixels corresponding to (-1) is almost equal to the number of pixels

corresponding to (1). The (1) pixels can then have their luminance data swapped with the

(-1) pixels to carry out the luminance masking process. If the numbers of (-1) pixels and

(1) pixels are not exactly equal, the relatively small number of unpaired pixels can be left

unchanged.

It should be appreciated that there are many ways in which such a process of identifying

pairs of pixels and swapping their luminance data can be carried out. Where the mask block is a 32 x 32 pixel block , i.e. 1024 pixels, a huge number of

combinations must be tried to find the original image, that is, 1023 x 1021 x 1019 ... x 3 x l.

Similarly to the chrominance mask it is advantageous for the luminance mask to move

and similar movement procedures can be followed as discussed with reference to the

chrominance mask of the first embodiment.

One possible problem which might be expected to occur in identifying the mask regions

and reversing the degradation mask by authorised users is that during processing and

transmission of the digital video signal mis-alignment may occur so that the position of

the mask in the received and decoded video image signal may be different to the position

at which the mask process was applied to the original video signal. However, the mask

block can be easily located by simple and well-known image processing techniques.

Conveniently, edge detection techniques can be used to identify the discontinuities at the

mask block edges. This process of locating the mask block is made simpler by the fact

that the size of the mask is known and that in practice the degree of shift will usually be

no more than a few lines or pixels.

As explained above, where a moving degradation mask is used for copy protection it is

not necessary for the movements of the mask to be unpredictable by third parties

including potential illicit copiers.

Ideally in a copy protection regime the video image data should be protected using the

degradation mask by the original content provider so that all subsequent legitimate

copying and distribution of the video image incorporates the mask. The necessary hardware and software, algorithms and keys to allow the mask to be removed can then be distributed to legitimate end users as required.

Where this ideal situation cannot be achieved and some distribution and copying of the

digital video image is carried out in unmasked form, care should be taken that each time

the degradation mask is applied to the digital video image the moving mask is in the

same position in each frame. Otherwise, it would be possible for illicit copiers to simply

obtain several copies of the digital video signal protected by a moving degradation mask

on different occasions and splice together the unmasked parts of the different copies to

produce a single complete unmasked copy of the digital video signal.

The embodiments of the invention described above refer to MPEG encoding and

decoding of the digital video signal for convenience because this is currently the most

common form of digital video signal coding. However, the invention is not limited to

applications where MPEG coding is to be or may be used.

Preferably the size of the area of each frame of the digital video image which is subjected

to the masking process is the same as an MPEG macro block in order to minimise the

impact of the mask process on subsequent MPEG encoding and decoding. Where an

alternative digital video encoding and decoding protocol is to be used or expected to be

used it will normally be advantageous to have the mask area equal in size to the coding

element corresponding to an MPEG macro block, if there is one.

The degradation mask process according to the present invention allows the digital video

signal to be protected against copying and it can be applied to the original non-encoded

digital video image signal without any significant impact on a subsequent encoding operation and in particular without significantly reducing the degree of data compression

which can be achieved by the encoding operation.

Knowing the location of the mask area or the masking algorithm will not allow

unauthorised third parties to reverse the masking operation unless the masking and

unmasking secret key or keys are known.

As with any other key-based encryption system the keys may be changed as often as

desired to maintain security and to complicate the task of illicit copiers. Suitable

procedures, systems and protocols for carrying out secure supply of the necessary keys

and carrying out key changes at the correct times are well known.

The degradation mask process according to the invention does not cause the loss of any

of the digital video data and an authorised user can recover the original video in perfect

form by reversing the masking process using the appropriate key. The dynamic range of

the video data is not changed so that the video data can go through MPEG or other

encoding and decoding and other video processing equipment and procedures without

effect.

The degradation mask can be removed and the masking process reversed after encoding

and subsequent decoding of the digital video data, for instance by MPEG, without any

visible residual effect on the displayed video image.

The degradation mask process according to the invention has little impact on the

statistical nature of the video data so that the degree of compression which can be

achieved is largely unaffected and the size of the encoded or compressed digital video

signal, for example encoded using MPEG, is largely unaffected. In principle, the degradation mask process could be applied to the entire video image.

However, in practice using a relatively small mask area is efficient to destroy the

economic value of illicit copies, particularly if the mask area moves. The presence of a

mask area plainly identifies a copy as an illicit copy preventing it being passed off as a

legitimate copy. Further, even the customers happy to knowingly buy an illicit copy of

the video are usually unwilling to pay for a video containing a masked area.

Accordingly, it is preferred to mask only a relatively small part of the digital video image

in order to minimise the processing requirements to add or remove the mask and to

minimise the amount of data added to the video signal by the masking process.

The size of the mask area may be varied as desired. The use of 32 x 32 pixel blocks as

described in the examples is particularly advantageous when used in conjunction with

MPEG coding.

The mask processes described in the examples are examples only and the person skilled

in the art will be able to identify many other reversible operations which can be carried

out on pixel data as part of a masking process.

If desired, the type of mask, chrominance mask or luminance mask, the mask size, the

operations carried out in selected pixels, the algorithms used to determine what procedure

is to be applied to which pixels and the secret key used as a basis for the pseudo random

selection by the algorithm can all be changed for different video signals or during

masking of a single video signal as judged necessary for effective video copy protection.

In the described embodiment, the chrominance mask process carries out one of a number

of possible reversible operations on individual pixel values while the luminance mask exchanges the data values of different pixels. It should be understood that a pixel value exchanging masking process could be used in a chrominance mask and that the application of different reversible data value changes could be used in a luminance mask,

although of course the specific reversible operations described for the chrominance mask

are specific to chrominance pixel values.

The above embodiments are described by ways of examples only and the skilled person

will understand that many changes could be made to the process within the scope of the

present invention.

Claims

1. A method of protecting a digital video image from copying, comprising the steps of:

selecting a part of the image to be masked; and

applying a reversible operation to at least some of the pixels within the masked

region, the reversible operation used or the pixels the operation is applied to, or

both, being selected based on a key.

2. A method according to claim 1 , in which a number of possible reversible

operations may be applied to the pixels and the operation applied to each pixel is

selected based on the key.

3. A method according to claim 1 or claim 2, in which the selected part of the

image to be masked is a 32 x 32 pixel block.

4. A method according to any preceding claim, in which the selected part of the

image to be masked corresponds in size to a JPEG macroblock.

5. A method according to any preceding claim, in which the reversible operation

does not significantly increase the size of the video image data or the size of the

video image data after compression.

6. A method according to any preceding claim, in which the reversible operation is applied to the luminance data of the pixels.

7. A method according to any preceding claim, in which the reversible operation is

applied to the chrominance data of the pixels.

8. A method according to claim 7, in which the reversible operation swaps

different colour values of the pixel.

9. A method according to any preceding claim, in which the reversible operation

flips at least one colour value of the pixel with zero.

10 A method according to any one of claims 6 to 9, in which the reversible

operation swaps values between different pixels.

11. A method according to any preceding claim, in which a different part of the

image is selected to be masked in different frames.

12. A method according to claim 11, in which part of the image selected is

determined by the video frame number.

13. A method according to any preceding claim, in which the reversible operation is

subsequently removed using a key.

4. Apparatus adapted to carry out the method of any preceding claim.