AU9708101A - Embedding information in the DC component of compressed image data - Google Patents

Description

S&FRef: 580454

AUSTRALIA

PATENTS ACT 1990 COMPLETE SPECIFICATION FOR A STANDARD PATENT

ORIGINAL

Name and Address of Applicant: Actual Inventor(s): Address for Service: Invention Title: Canon Kabushiki Kaisha 30-2, Shimomaruko 3-chome, Ohta-ku Tokyo 146 Japan James Philip Andrew Spruson Ferguson St Martins Tower,Level 31 Market Street Sydney NSW 2000 (CCN 3710000177) Embedding Information in the DC Component of Compressed Image Data ASSOCIATED PROVISIONAL APPLICATION DETAILS [33] Country [31] Applic. No(s) AU PR6266 AU PR2008 [32] Application Date 10 Jul 2001 11 Dec 2000 The following statement is a full description of this invention, including the best method of performing it known to me/us:- 5815c -1- EMBEDDING INFORMATION IN THE DC COMPONENT OF COMPRESSED IMAGE DATA Technical Field The present invention relates to embedding information in a digital image, and in particular in a compressed image.

Background Watermarking of digital images is used for a variety of purposes. A watermark is a signal embedded in an image (or other data), usually so that the original and watermarked image are visually almost identical. Many of the previous methods embedded the watermark in the original image, and then tested the effectiveness of the method by ooooo measuring the robustness to compression. Ensuing compression generally corrupts or changes the embedded watermark.

The publication "Watermarking of uncompressed and compressed video", Signal Processing, vol. 66, May 1998, by Hartung and Girod discloses a method for o watermarking MPEG-2 coded video. The watermark is effectively added to the decoded motion compensated frame data, and the new signal transformed, quantized and entropy o 0 0coded in the usual manner. This method suffers from the problem in that the watermark 0 is corrupted by the compression process, and any watermark extracted is generally 0.00changed by the compression process.

The publication "Embedding Image Watermarks in DC Componenets", IEEE Transactions on Circuits and Systems for Video Technology, Vol 10, No. 6, September 2000 by Jiwu Huang, Yun Q. Shi, and Yi Shi, discloses a method of embedding the watermark in the DC components of the transform coefficients of the image. The method first splits the image into non-overlapped blocks of 8x8 pixels. The method then performs discrete cosine transforms (DCT) on the respective blocks of pixels to produce corresponding blocks of transform coefficients. The method then adds the watermark 580454.doc -2image to the DC components of the blocks of transform coefficients and performs inverse discrete cosine transforms (iDCT) on the respective blocks of transform coefficients to produce the watermarked image. This method suffers from the problem that any subsequent compression process corrupts the watermarked image, and any watermark extracted is generally changed by the compression process.

The reference to the aforementioned publications is not to be taken as an admission that any information contained therein forms part of the common general knowledge.

Summary It is an object of the present invention to substantially overcome, or at least ameliorate, one or more disadvantages of existing arrangements.

According to one aspect of the invention, there is provided a method of embedding a digital watermark in a compressed image, said method comprising the steps of: transforming an image to derive a plurality of subbands, including a DC subband, and one or more AC subbands; quantizing the DC subband to form integer values; adding said 15 digital watermark to the quantized DC subband to produce a watermark embedded DC subband; and losslessly entropy coding the watermark embedded DC subband, to form o•.

part of said compressed image.

According to another aspect of the invention, there is provided a method of extracting a digital watermark from a compressed image, said method comprising the steps of: entropy decoding a quantized DC subband of the compressed image; and extracting said watermark from said quantized DC subband to reproduce the digital watermark losslessly.

According to still another aspect of the invention, there is provided a apparatus for embedding a digital watermark in a compressed image, said apparatus comprising: means for transforming an image to derive a plurality of subbands, including a DC subband, and one or more AC subbands; means for quantizing the DC subband to form integer values; 580454.doc -3means for adding said digital watermark to the quantized DC subband to produce a watermark embedded DC subband; and means for losslessly entropy coding the watermark embedded DC subband, to form part of said compressed image.

According to still another aspect of the invention, there is provided apparatus for extracting a digital watermark from a compressed image, said apparatus comprising: means for entropy decoding a quantized DC subband of the compressed image; and means for extracting said watermark from said quantized DC subband to reproduce the digital watermark losslessly.

According to still another aspect of the invention, there is provided a computer readable medium comprising a computer program for embedding a digital watermark in a •compressed image, said computer program comprising: code for transforming an image to derive a plurality of subbands, including a DC subband, and one or more AC subbands; code for quantizing the DC subband to form integer values; code for adding said digital watermark to the quantized DC subband to produce a watermark embedded DC subband; 15 and code for losslessly entropy coding the watermark embedded DC subband, to form :part of said compressed image.

According to still another aspect of the invention, there is provided a computer readable medium comprising a computer program for extracting a digital watermark from a compressed image, said computer program comprising: code for entropy decoding a quantized DC subband of the compressed image; and code for extracting said watermark from said quantized DC subband to reproduce the digital watermark losslessly.

Brief Description of the Drawings A number of embodiments of the present invention will now be described with reference to the drawings, in which: Fig. 1A shows an original image; 580454.doc -4- Fig. 1B shows subbands of a first level Discrete Wavelet Transform (DWT) of the original image shown in Fig. 1A; Fig. 2 shows subbands of a second level Discrete Wavelet Transform (DWT) of the original image shown in Fig. 1A; Fig. 3 shows subbands of a fourth level Discrete Wavelet Transform (DWT) of the original image shown in Fig. 1A; Fig. 4 shows a flow chart of a method of embedding a watermark; Fig. 5 shows a flow chart of a method of extracting an embedded watermark from the image compressed in accordance with the method of Fig. 4; and Fig. 6 is a schematic block diagram of a general-purpose computer upon which methods described can be practiced.

Detailed Description including Best Mode Where reference is made in any one or more of the accompanying drawings to steps and/or features, which have the same reference numerals, those steps and/or features have for the purposes of this description the same function(s) or operation(s), unless the contrary intention appears.

oooeo The principles of the method described herein have general applicability to the lossless embedding of a watermark in a compressed image. However, for ease of explanation, the steps of the method are described with reference to the draft JPEG 2000 image compression standard "Information Technology, JPEG 2000 Image Coding System, International Standard/ ITU-T Recommendation ISO/IEC FDIS15444-1:2000 (version 1.0 16 March 2000)" herein after referred to as "JPEG2000". However, it is not intended that the present invention be limited to the described method. For example, the invention may have application to other compression schemes utilising discrete wavelet transforms or discrete cosine transforms.

580454.doc Before proceeding with a description of the embodiments, a brief review of terminology is discussed herein. The term watermark is used in a general sense. That is a watermark is any signal that is separate from the image signal, in any domain. The present method is concerned with digital watermarks, and future reference to watermark refers to a digital watermark. The watermark is a thus a digital signal that is discrete and has integer values. Embedding a digital watermark in an image means embedding a digital signal or digital information in an image.

The method proceeds initially by means of a wavelet transform of original image data. A description of the wavelet transform process is given in many standard texts and in particular the book "Wavelets for Computer Graphics" by I. Stollinitz et. al., published .ooooi in 1996 by Morgan Kaufmann Publishers Inc. An overview of the wavelet process will now be described with reference to the accompanying drawings.

Turning initially to Figs 1A and 1B, an original image 1 is transformed utilising a Discrete Wavelet Transform (DWT) into four sub-images 3,4,5, and 6. The sub-images ooooo or subbands are normally denoted LL1, HL1, LH1 and HH1. The one suffix on the subband names indicates level 1. The LL1 subband is a low pass decimated version of the original image. Consequently it has the characteristics of a typical digital image, and in fact can be used as a lower resolution version of the original image.

The wavelet transform utilised in JPEG2000 can vary and can include either the 5/3 reversible or 9/7 irreversible filter pair. In other variations of the method, other wavelet filters could also be used Harr fitler, Gabar filter etc). The LL1 subband is in turn utilised and a second Discrete Wavelet Transform is applied as shown in Fig. 2 giving subbands LL2 HL2 LH2 HH2 This process is continued for example as illustrated in Fig. 3 wherein the LL4 subband is illustrated. Obviously, further levels of decomposition can be provided depending on the size of the input image.

580454.doc -6- The lowest frequency subband is referred to as the DC subband. In the case of Fig. 3, the DC subband is the LL4 subband. The other subbands (ie HL1, LH1, HH1, HL2 etc) are referred to as AC subbands. Note that if the original image has dimensions M x N (rows x columns) then the DC subband, for a J level DWT, has nominal dimensions M/2 J x N/2

J

Each single level DWT can, in turn, be inverted to obtain the original image 1.

Thus, a J-level DWT can be inverted as a series of J-single level inverse DWT's. Each LL or DC subband represents the original image at a particular resolution and has the characteristics of typical digital imagery.

Turning now to Fig. 4, there is shown a flow chart of a method of embedding a watermark. The method 400 embeds the watermark as part of the JPEG2000 compression standard. The method 400 is substantially the same as the JPEG2000 compression standard, except that the method 400 has two additional steps 430 and 435.

The method commences at Step 401, where preferably an RGB (24 bit per pixel) image is 15 input to the method 400. However images in other colour spaces can be used without departing from the scope of the invention. In Step 405 the image data is level shifted and transformed with the JPEG2000 irreversible component transform to produce YCbCr (like) data. If YCbCr like data, such as YCbCr4:2:0 is input to the method 400 then the component transformation is not used. After step 405, the method 400 proceeds to step 410.

In Step 410 each component (following the component transform) is transformed with a two-dimensional discrete transform (DWT). Preferably a 4 level DWT is used, and the 9/7 irreversible filters used. However a different number of levels, and different wavelet filter can be used without departing from the scope of the invention. After the step 410, the method 400 proceeds to step 420.

580454.doc -7- Each subband is then quantized in Step 420 according to the JPEG2000 standard.

In particular the DC subband of a given component is quantized according to the equation, DC sign(DC(m,n)). DC (1) where DC(m, n) is sample (m n) in the DC subband of the given component, Q is the quantization step size, and DCq n) is the resulting quantized value. The quantization takes nominally real valued coefficients and produces integer quantization values. (ie DCq n) is an integer).

As in JPEG2000 the quantization step size Q is encoded in the compressed image 10 bitstream by an exponent and a mantissa wherein Q is given by, Q=2 R e (2) where R is the (nominal) dynamic range of the DC subband. The exponent and mantissa are represented by 5 and 11 bit unsigned integers respectively. Preferably for the DC subband a value of p 0 and e 7 is used. After step 420, the method proceeds to 15 step 430.

In Step 430 a watermark is embedded into the quantized DC subbands as follows.

Preferably the watermark signal consists of the same number of components as the image (three in this case) and each watermark component has the same dimensions as the corresponding DC subband component. However, a watermark signal with one component could also be used, in which case it is embedded only in the luminance (Y) DC subband component. Let the watermark be denoted as W for a given component.

Preferably the watermark signal has range of 0,1,2,3: that is W(m,n)e Then the watermark is added to the quantized DC subband of the corresponding component giving a modified (quantized) DC subband, preferably using the equation, 580454.doc -8- DC,(m, n) 4 DCq n) W(m, n) (3) where n) is sample (m n) in the modified DC subband of the given component.

In this way 2 extra bit planes are added to the quantized DC subband. The 2 least significant (extra) bitplanes constitute the watermark, while the other bitplanes the quantized DC subband data. The watermark resides in different bitplanes to the quantized DC subband data. That is the watermark is added (substantially) independently to the quantized DC subband data. The other subbands AC subbands) are quantized according to JPEG2000. After step 430, the method proceeds to step 435.

In Step 435 the DC subband quantization step size is updated as follows. The quantization step size is preferably decreased by 2 bits by setting pu 0 and c 7+2 9.

By increasing c by 2, the two extra bit planes added are coded by the entropy encoder, 0** and a decoder dequantizes the DC subband data to the correct magnitude. After step 435, the method proceeds to step 440.

In Step 440 the quantized subbands are divided into code blocks, and the 15 method 400 proceeds to step 450.

In step 450 each code block is encoded with the three-pass bitplane encoder of JPEG2000. Each modified DC subband block is losslessly encoded. That is each compressed DC subband block can be decoded to give the original modified DC subband block. Lossless encoding is achieved by entropy coding all bit planes of the modified DC subband blocks. Preferably the blocks from other subbands (eg. AC subbands) are losslessly encoded, although the bitstreams for these code-blocks can be truncated without departing from the scope of the invention. After step 450, the method proceeds to step 460.

In Step 450 a JPEG2000 compressed image is constructed from the compressed block bit streams and relevant header information, in the resolution-layer-component 580454.doc -9progression mode. Preferably one layer is used. That is the whole bitstream for each code block is stored in a contiguous segment in the compressed image. After step 460, the method 400 terminates 470.

As images are increasingly represented in compressed form, it has become a natural format for representation of digital imagery. The described method enables the watermark to be embedded in a lossless manner, and consequently the compression process does not corrupt the information during decompression of the image.

Furthermore the described method enables the embedding of the watermark in a compressed image in an efficient but simple manner. Also, the watermark is easily extracted from the compressed image, even during the extraction of lower resolutions of 9***99 o*oo9 the compressed image.

•0 ,oat** In the above-described method, the watermark is embedded by adding two least 0@ significant bitplanes underneath the quantized DC subband data, and the DC subband data quantized by selecting and exponent 7. This exponent means that the DC subband is o 0 15 effectively represented with 7 bits of precision. The 2 extra least significant bitplanes S,(which are noise as far as the DC subband is concerned) have almost no perceptible impact. Thus a decoder, oblivious to the fact that there is a watermark embedded in the image (and thus decoding 9 bitplanes of DC subband data), would reproduce an image *999 *to**visually almost identical to a decoder that removed the watermark prior to reconstruction of the image. Typically the quantized DC subband data is not very compressible. Hence the entropy coder produces approximately the same number of output bits as the number of input bits when coding the DC subband data. Depending on the watermark the same will be true for the watermark bitplanes. Thus the 2 extra bit planes will add approximately 2 bits per DC coefficient to the compressed image. For a 4 level DWT this will add 2/(2 4 x2 4 1/128 bits per pixel to the compressed image, which is an 580454.doc insignificant amount for typical compression of still images. In this way, the addition of the watermark to the DC subband will not adversely effect the compression ratio of the image as distinct from the addition of the watermark to the AC subbands.

The separation of the watermark signal and the image data may not always be desirable, as it leaves the watermark open to simple attack. In an variation of the method, the watermark is added directly to the quantized DC subband component without the scaling of the DC data. That is DCm(m, n) DCq n) W(m, n) (4) S•In this case the mantissa and exponent representing the original quantization factor for the DC subband are used and stored in the compressed image bitstream. In this case preferably p 0, e 7 and W(m, n) e JPEG2000 proposes that the whole image be divided into one or more image tile •o••o components, each of which are then 2-D discrete wavelet transformed. The transform coefficients of each image tile component are then grouped into sub-bands, which sub- 15 bands are further partitioned into rectangular code blocks before each code block is then entropy encoded. For ease of explanation, the method 400 has been described as performing the JPEG2000 compression on the entire original image. As will be apparent to those skilled in the art, the method 400 can be extended to that situation where the original image is divided into a plurality of tiles. In these circumstances, the method 400 can add the watermark image to the lower order bitplanes of the DC subbands of one or more of these tiles.

Turning now to Fig. 5, there is shown a flow chart of a method of extracting an embedded watermark from the image compressed in accordance with the method of Fig. 4. The method 500 of extracting the watermark from the compressed image 580454.doc 11 commences at step 501, where any necessary parameters are initialised and the compressed bitstream is input. After step 501, the method 500 proceeds to step 505.

In Step 505 the method decodes the image header information of the bitstream and then proceeds to step 510. In step 510, the method extracts the compressed bitstreams for each DC subband code block using the information decoded in step 505. After step 510, the method proceeds to step 520.

In Step 520 each DC subband code block is decoded according to the JPEG2000 entropy decoding process. Preferably each DC sample is decoded in a sign-magnitude form, where the first bit of the sample is a sign bit (1 for negative and 0 for positive) and 10 subsequent bits represent the magnitude. After step 520, the method 500 proceeds to step 530.

In Step 530 the lower two bit planes of each DC subband block are extracted from each code block and used in the next Step 540 to reconstruct the watermark signal.

0 Preferably this is achieved by doing a bit-wise "and" of the lowest two bits. That is for a S 15 given component, W(m, n) DC,(m, n) 00...011 where n) is decoded DC subband component sample prior to dequantization, is the bit-wise "and" operator, and 00...011 is number 3 in sign magnitude form, and which has the same number of bits as used to represent the decoded DC subband samples.

The method 500 has the advantage that if the method 400 of embedding the watermark in the compressed image is used the watermark can be extracted without knowledge of the watermark.

In JPEG2000 (or other compression methods suitably designed) the DC subband can be decoded independently from the rest of the compressed image. That is after decoding a small amount of header information the DC subband data can be extracted 580454.doc -12from the compressed image bit stream in a simple manner, and entropy decoded without entropy decoding any AC subband data. The effort required to decode the quantized DC subband is thus substantially proportional to the size of the DC subband and independent of the size of the original image. Given that the DC subband is much smaller than the original image (for example 16 times smaller for a 4 level DWT) the watermark can be extracted with much less effort compared to that for the full image. Further, should verification of the watermark be required by some computational procedure, the computational cost is significantly reduced as compared to embedding the watermark in the full image (since the DC subband is much smaller compared to the full image).

In a variation of the method of embedding a watermark of Fig. 4, the watermark can T be embedded in an image that has already been compressed as follows. The quantized DC subband is decoded from the compressed image. The watermark is then added as described above, preferably by adding two extra bit planes. The quantization factor for the DC subband is then preferably divided by 4. The new DC subband is entropy go encoded and the resulting compressed image updated with the new compressed DC subband bitstream and quantization factor.

In a variation of the method of extracting a watermarked image of Fig. 5, the compressed lower resolution version of the image can be simply extracted from the compressed image by discarding some number of high frequency subbands, and updating the compressed image header appropriately. For example for an image compressed with a 4 level DWT, the part of the compressed bit stream representing the HL1, LH1 and HH1 subband is discarded, and the header updated, to give a compressed image representing the original image at 1/2 the size in each dimension. The watermark remains in the compressed lower resolution image, since the DC subband data remains the same.

Further, where a watermark is used for authentication purposes and is resistant, for these purposes, to scaling (and other image processing operations) it will be resistant to 580454.doc -13more radical resolution change of the image. For example, the watermark could persist following decompressing the image, scaling, and then recompressing.

The aforementioned method(s) comprise a particular control flow. There are many other variants of the method(s) which use different control flows without departing the spirit or scope of the invention. Furthermore, one or more of the steps of the method(s) may be performed in parallel rather sequential. In addition, the method(s) may be performed in a pipeline fashion.

The methods of embedding a watermark, and of extracting an embedded watermark from the compressed image are preferably practiced using a general-purpose computer

S.

system 600, such as that shown in Fig. 6 wherein the processes of Figs. 4 or 5 may be implemented as software, such as an application program executing within the computer system 600. In particular, the steps of embedding a watermark, and of extracting an embedded watermark from the compressed image are effected by software, comprising instructions in the form of code, that is carried out by the computer. The software may be divided into two separate parts; one part for carrying out the embedding and extracting methods; and another part to manage the user interface between the latter and the user.

The software may be stored in a computer readable medium, including the storage devices described below, for example. The software is loaded into the computer from the computer readable medium, and then executed by the computer. A computer readable medium having such software or computer program recorded on it is a computer program product. The use of the computer program product in the computer preferably effects an advantageous apparatus for embedding a watermark, and for extracting an embedded watermark from the compressed image.

The computer system 600 comprises a computer module 601, input devices such as a keyboard 602 and mouse 603, output devices including a printer 615 and a display device 614. A Modulator-Demodulator (Modem) transceiver device 616 is used by the 580454.doc -14computer module 601 for communicating to and from a communications network 620, for example connectable via a telephone line 621 or other functional medium. The modem 616 can be used to obtain access to the Internet, and other network systems, such as a Local Area Network (LAN) or a Wide Area Network (WAN).

The computer module 601 typically includes at least one processor unit 605, a memory unit 606, for example formed from semiconductor random access memory (RAM) and read only memory (ROM), input/output interfaces including a video interface 607, and an I/O interface 613 for the keyboard 602 and mouse 603 and i optionally a joystick (not illustrated), and an interface 608 for the modem 616. A storage device 609 is provided and typically includes a hard disk drive 610 and a floppy disk •'""drive 611. A magnetic tape drive (not illustrated) may also be used. A CD-ROM drive 612 is typically provided as a non-volatile source of data. The components 605 ooo i Sto 613 of the computer module 601, typically communicate via an interconnected bus 604 and in a manner, which results in a conventional mode of operation of the computer 15 system 600 known to those in the relevant art. Examples of computers on which the described arrangements can be practised include IBM-PC's and compatibles, Sun Sparcstations or alike computer systems evolved therefrom.

Typically, the application program is resident on the hard disk drive 610 and read and controlled in its execution by the processor 605. Intermediate storage of the program and any data fetched from the network 620 may be accomplished using the semiconductor memory 606, possibly in concert with the hard disk drive 610. In some instances, the application program may be supplied to the user encoded on a CD-ROM or floppy disk and read via the corresponding drive 612 or 611, or alternatively may be read by the user from the network 620 via the modem device 616. Still further, the software can also be loaded into the computer system 600 from other computer readable medium including magnetic tape, a ROM or integrated circuit, a magneto-optical disk, a radio or infra-red 580454.doc transmission channel between the computer module 601 and another device, a computer readable card such as a PCMCIA card, and the Internet and Intranets including email transmissions and information recorded on websites and the like. The foregoing is merely exemplary of relevant computer readable mediums. Other computer readable media may alternately be used.

The methods of embedding a watermark, and of extracting an embedded watermark from the compressed image may alternatively be implemented in dedicated hardware such as one or more integrated circuits performing the functions or sub functions of the methods. Such dedicated hardware may include graphic processors, digital signal 10 processors, or one or more microprocessors and associated memories.

Industrial Applicability It is apparent from the above that the arrangements described are applicable to the •oooo image processing industries.

*o The foregoing describes only some embodiments of the present invention, and modifications and/or changes can be made thereto without departing from the scope and spirit of the invention, the embodiment(s) being illustrative and not restrictive.

In the context of this specification, the word "comprising" means "including principally but not necessarily solely" or "having" or "including" and not "consisting only of'. Variations of the word comprising, such as "comprise" and "comprises" have corresponding meanings.

580454.doc

Claims (2)

16- The claims defining the invention are as follows: 1. A method of embedding a digital watermark in a compressed image, said method comprising the steps of: transforming an image to derive a plurality of subbands, including a DC subband, and one or more AC subbands; quantizing the DC subband to form integer values; adding said digital watermark to the quantized DC subband to produce a watermark embedded DC subband; and 10 losslessly entropy coding the watermark embedded DC subband, to form part of said compressed image. 2. A method as claimed in claim 1, wherein the digital watermark is added substantially independently of the quantised DC subband data. 3. A method as claimed in claim 1, wherein the digital watermark is added as a number of extra bitplanes less significant than bitplanes of the quantized DC subband data. 4. A method as claimed in claim 3, wherein said extra bitplanes are included in the compressed image by using a modified DC subband quantization step size. A method as claimed in claim 1, wherein said DC subband is coded into the compressed image substantially independently of the other subbands.

580454.doc 17- 6. A method as claimed in any one of the preceding claims, wherein said transforming step uses a discrete two-dimensional wavelet transform. 7. A method of extracting a digital watermark from a compressed image, said method comprising the steps of: entropy decoding a quantized DC subband of the compressed image; and extracting said watermark from said quantized DC subband to reproduce the digital watermark losslessly. 10 8. A method as claimed in claim 7, wherein said extracting step is achieved by a bit- wise AND operation. 9. Apparatus for embedding a digital watermark in a compressed image, said apparatus comprising: S 15 means for transforming an image to derive a plurality of subbands, including a DC subband, and one or more AC subbands; means for quantizing the DC subband to form integer values; means for adding said digital watermark to the quantized DC subband to produce a watermark embedded DC subband; and means for losslessly entropy coding the watermark embedded DC subband, to form part of said compressed image. Apparatus as claimed in claim 9, wherein the digital watermark is added substantially independently of the quantised DC subband data. 580454.doc -18- 11. Apparatus as claimed in claim 9, wherein the digital watermark is added as a number of extra bitplanes less significant than bitplanes of the quantized DC subband data. 12. Apparatus as claimed in claim 11, wherein said extra bitplanes are included in the compressed image by using a modified DC subband quantization step size. 13. Apparatus as claimed in claim 9, wherein said DC subband is coded into the compressed image substantially independently of the other subbands. 14. Apparatus as claimed in any one of the preceding claims 9 to 13, wherein said transforming means uses a discrete two-dimensional wavelet transform. Apparatus for extracting a digital watermark from a compressed image, said apparatus comprising: means for entropy decoding a quantized DC subband of the compressed image; and S means for extracting said watermark from said quantized DC subband to reproduce the digital watermark losslessly. 16. Apparatus as claimed in claim 15, wherein said extracting means performs a bit- wise AND operation. 17. A computer program for embedding a digital watermark in a compressed image, said computer program comprising: code for transforming an image to derive a plurality of subbands, including a DC subband, and one or more AC subbands; 580454.doc -19- code for quantizing the DC subband to form integer values; code for adding said digital watermark to the quantized DC subband to produce a watermark embedded DC subband; and code for losslessly entropy coding the watermark embedded DC subband, to form part of said compressed image. 18. A computer program as claimed in claim 17, wherein the digital watermark is added substantially independently of the quantised DC subband data. S 10 19. A computer program as claimed in claim 17, wherein the digital watermark is added as a number of extra bitplanes less significant than bitplanes of the quantized DC subband data. A computer program as claimed in claim 19, wherein said extra bitplanes are S 15 included in the compressed image by using a modified DC subband quantization step size. 00 21. A computer program as claimed in claim 17, wherein said DC subband is coded into the compressed image substantially independently of the other subbands. 22. A computer program as claimed in any one of the preceding claims 17 to 21, wherein said transforming code uses a discrete two-dimensional wavelet transform. 23. A computer program for extracting a digital watermark from a compressed image, said computer program comprising: code for entropy decoding a quantized DC subband of the compressed image; and 580454.doc code for extracting said watermark from said quantized DC subband to reproduce the digital watermark losslessly. 24. A computer program as claimed in claim 23, wherein said extracting code performs a bit-wise AND operation. A method of embedding a digital watermark in a compressed image, said method substantially as described herein with reference to Fig. 4 of the accompanying drawings. 26. A method of extracting a digital watermark from a compressed image, said method substantially as described herein with reference to Fig. 5 of the accompanying drawings. o •o SC 27. Apparatus for embedding a digital watermark in a compressed image, said apparatus substantially as described herein with reference to Figs. 4 and 6 of the accompanying drawings. 0* 28. Apparatus for extracting a digital watermark from a compressed image, said method substantially as described herein with reference to Figs. 5 and 6 of the accompanying drawings. 29. A computer program for embedding a digital watermark in a compressed image, said computer program substantially as described herein with reference to Figs. 4 and 6 of the accompanying drawings. 580454.doc -21 A computer program for extracting a digital watermark from a compressed image, said computer program substantially as described herein with reference to Figs. 5 and 6 of the accompanying drawings. DATED this FIFTH Day of DECEMBER 2001 CANON KABUSHIKI KAISHA Patent Attorneys for the Applicant SPRUSON&FERGUSON a@ 0 a 0. .9000 580454.doc