Detecting information hidden out-of-phase in color channels

Abstract

The present invention relates to detecting steganographically hiding images and information. In a preferred embodiment, an image is hidden within a media signal. The media preferably includes a cyan (C) color plane, a magenta (M) color plane, a yellow (Y) color plane and a black (K) color plane. In an alternative embodiment, the media includes a spot color. The image is converted into a black color channel image and is then applied to the media's K channel. The black channel image is inverted and the inverted image is applied to the media's CMY (or spot) color planes. For detection, the embedded media is illuminated with infrared light, which emphasizes the black color plane. An input device captures a digital image of the illuminated media. The image is detected or displayed based on the emphasized black plane.

Description

FIELD OF THE INVENTION

The present invention relates to hiding data, and is particularly illustrated with reference to hiding information out-of-phase in color channels.

BACKGROUND AND SUMMARY OF THE INVENTION

The above mentioned parent applications disclose various techniques for embedding and detecting out-of-phase digital watermarks.

Digital watermarking technology, a form of steganography, encompasses a great variety of techniques by which plural bits of digital data are hidden in some other object, preferably without leaving human-apparent evidence of alteration.

Digital watermarking may be used to modify media content to embed a machine-readable code into the media content. The media may be modified such that the embedded code is imperceptible or nearly imperceptible to the user, yet may be detected through an automated detection process.

Digital watermarking systems typically have two primary components: an embedding component that embeds the watermark in the media content, and a reading component that detects and reads the embedded watermark. The embedding component embeds a watermark pattern by altering data samples of the media content. The reading component analyzes content to detect whether a watermark pattern is present. In applications where the watermark encodes information, the reading component extracts this information from the detected watermark. Assignee's U.S. patent application Ser. No. 09/503,881, filed Feb. 14, 2000, discloses various encoding and decoding techniques. U.S. Pat. Nos. 5,862,260 and 6,122,403 disclose still others. Each of these U.S. patent documents is herein incorporated by reference.

Now consider our inventive out-of-phase digital watermarking techniques with reference to FIGS. 1 a and 1 b . In FIG. 1 a , the dash/dot C, M, Y and K lines represent, respectively, cyan, magenta, yellow and black color channels for a line (or other area) of a media signal (e.g., a picture, image, media signal, document, etc.). The FIG. 1 a lines represent a base level or a particular color (or gray-scale) level (or intensity). Of course, it is expected that the color (or gray-scale) level will vary over the media signal. FIG. 1 b illustrates the media of FIG. 1 a , which has been embedded with an out-of-phase digital watermark signal. The watermark signal is preferably applied to each of the color component dimensions C, M and Y.

In FIGS. 1 a and 1 b , the M and Y channels are represented by one signal, since these color components can be approximately equal, but separate signals. Of course, it is not necessary for these components to be equal, and in many cases the yellow and magenta components are not equal. The illustrated “bumps” (or “tweaks”) in FIG. 1 b represent the digital watermark signal, e.g., upward and downward signal adjustments in relation to a respective color channel at given points over the media signal. The tweaks are preferably applied at the same level (or signal strength). Alternatively, the bumps are applied with a different signal strength (or tweak level) when compared to one another. Of course, these tweaks can be embedded over a color channel in a predetermined pattern, a pseudo random fashion, a random fashion, etc., to facilitate embedding of a digital watermark signal.

For the K dimension (or channel), the digital watermark signal is preferably embedded to be out-of-phase with respect to the CMY channels. Most preferably, the K channel is approximately 180 degrees out-of-phase (e.g., inverted) with the watermark signals in the CMY color channels, as shown in FIG. 1 b . For example, if a digital watermark signal modifies each of the color channels at a media' first location with a tweak level of say 7, then a tweak level of −7 correspondingly modifies the K channel at the media's first location. This inventive digital watermark technique is referred to as our out-of-phase (or “K-phase”) digital watermark. (We note that if a watermark signal is determined in terms of luminance, we can assign or weight corresponding tweak levels to the respective color plane pixel values to achieve the luminance value tweak. Indeed, a tweak can be spread over the CMY channels to achieve a collective luminance at a given media location. The luminance attributable to the CMY tweak is preferably cancelled or offset by the luminance effect attributable to a corresponding inverted K channel tweak at the give media location.).

Our inventive watermarking scheme greatly reduces watermark perceptibility. Since the watermark signal for the K channel is applied approximately 180 degrees out-of-phase, when compared to the respective tweaks applied to the C, M and/or Y channels, the watermark visibility is greatly reduced. The visibility reduction is achieved by the effective cancellation of perceived luminance changes when the CMYK image is viewed or printed. Indeed, combining an inverted watermark signal “tweak” or “bump” in a K channel with a corresponding non-inverted watermark signal tweak or bump in the CMY channels effectively cancels an overall perceived luminance change for a given area (e.g., a pixel or block of pixels)—greatly reducing visibility of the digital watermark.

The present invention discloses a new data hiding technique based on our out-of-phase technology. According to one implementation of the present invention, an image is hidden in or carried by a media signal. The hiding is accomplished with our out-of-phase embedding techniques. The image can be a photograph, a graphic, a barcode (1-D or 2-D), etc., etc. Another aspect of our inventive techniques is used to improve the visibility characteristics of our out-of-phase embedding techniques.

The foregoing and other aspects, features and advantages of the present invention will be even more apparent from the following detailed description, which proceeds with reference to the accompanying drawings.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 a is a diagram illustrating CMYK channels.

FIG. 1 b illustrates the color CMYK channels of FIG. 1 a , embedded with information.

FIG. 2 illustrates hiding an image in media.

FIG. 3 is a flow diagram illustrating an embedding method according to one implementation of the present invention.

FIGS. 4 and 5 are graphs showing hidden signal strength in terms of luminance.

FIGS. 6 and 7 are graphs showing hidden signal strength in terms of color saturation.

FIG. 8 illustrates limiting a signal tweak in low CMY areas to reduce hidden signal visibility.

FIG. 9 illustrates the segmentation of media into blocks.

FIG. 10 illustrates a feedback loop in an embedding process.

FIG. 11 illustrates feedback for the FIG. 10 feedback loop.

FIGS. 12 a and 12 b illustrate detection apparatus.

FIG. 13 illustrates orientation fiducials hidden in a media signal with our out-of-phase embedding techniques.

FIG. 14 illustrates out-of-phase embedding of a spot color.

FIG. 15 illustrates a printer calibration process.

DETAILED DESCRIPTION

Image Embedding

With reference to FIG. 2 , an image 10 is steganographically hidden within media 12 . Of course, media 12 may represent digital media such as an image, photograph, video frame, graphic, picture, logo, product tag, product documentation, visa, business card, art work, brochure, document, product packaging, trading card, banknote, deed, poster, ID card (including a driver's license, member card, identification card, security badge, passport, etc.), postage stamp, etc., etc. And image 10 can correspond to a digital representation of a photograph, picture, graphic, text, orientation fiducial, object, barcode, message, digital watermark, outline, symbol, etc., etc. In the FIG. 2 example, image 10 includes a close-up photograph, and the media includes a driver's license or passport photograph. The hiding (or embedding) is accomplished using our inventive out-of-phase techniques.

With reference to FIG. 3 , our K-phase hiding is preferably initiated by converting image 10 to a black channel image 10 ′ (step 30 —FIG. 3 ). Most digital imaging software tools such as Adobe's Photoshop facilitate such a black channel conversion. The black channel image 10 ′ includes a set of black pixel values (e.g., grayscale values) 10 ′. A location in the media 12 is selected to place the black channel image (step 32 ). The dashed circle 13 in FIG. 2 represents this selected location. The media 12 location can be represented by sets of media 12 pixels. (For example, a first set of pixels corresponds to the selected location's black channel values, a second set corresponds to the selected location's cyan channel values, a third set corresponds to the selected location's magenta channel values, and a fourth set corresponds to the selected location's yellow channel values.). The set of black channel image 10 ′ values is applied to the black channel pixels in the selected location of media 12 —effectively modifying media 12 (step 34 ). For example, if an image 10 ′ pixel includes a gray-scale value of 3, this gray-scale value is applied to a corresponding pixel in the selected media 12 location to raise that corresponding pixel value by 3. In an alternative implementation, instead of adjusting the corresponding pixel in the selected media 12 location by the gray-scale value, we replace that corresponding pixel value with the black image 10 ′ gray-scale value. In another implementation, the corresponding media 12 pixel is modified to achieve the gray-scale value of the image 10 ′ pixel. Of course we can scale and/or weight the gray-scale value as needed prior to modifying pixels in the selected location of media 12 .

The black channel image 10 ′ is inverted to produce a set of signal tweaks (step 36 ). For example, if a black channel pixel is tweaked by a grayscale value of say 24, then a corresponding, inverted CMY tweak value is −24. (As an alternative implementation, image 10 is converted into corresponding C, M and Y images and such images are applied to their respective channels. These signal tweaks are then used to modify or change the color values in their respective CMY color channels (step 38 ). Most preferably, in the above example, the −24-tweak value is applied to each of the CMY color channels. The overall luminance cancellation can be effected as such. In another implementation we unevenly spread the tweak value over the CMY channels to achieve an overall luminance change in a given media location to cancel the +24 tweak in the black channel. For example, if using a luminance equation of: L=0.3*C+0.6*M+0.1*Y, we can achieve an overall luminance tweak of −24 by tweaking C=−15, M=−30 and Y=−15. Of course there is a vast range of other color combinations to achieve the same collective luminance change. Care should be taken, however, to minimize a color shift when using this tweak-spreading alternative. The CMY pixels and the K pixels are thus out-of-phase with respect to one another—resulting in a local cancellation of the perceived luminance change. Accordingly, image 10 is successfully hidden or carried by media 12 .

The selected location 13 can be determined manually, e.g., via editing software tools (like Adobe's Photoshop). Or the selection process can be automated.

Image Hiding Enhancements

We have developed improvements to enhance our out-of-phase hiding techniques. These improvements apply to hiding both images and digital watermark signals (in this section both will be referred to as a hidden signal). While these techniques are not necessary to carry out our out-of-phase hiding techniques, they generally reduce the visibility of a hidden signal. Consider our following inventive improvements.

High Luminance Areas

Media 12 may include areas of low CMY and/or K ink (or signal intensity). In a first case, an area includes little or no C, M and/or Y ink. This results in an inability to counteract (or cancel) an inverted signal in a corresponding channel(s). Accordingly, we can sample the luminance of a media 12 area (or pixel) and, based on the luminance level, determine whether to scale back the hidden signal strength. For example, we begin to scale back the signal strength once the luminance reaches a predetermined threshold (e.g., in a range of 70-95% luminance). We can scale back the signal strength for a given area according to a linear reduction, as shown in FIG. 4 , or we can scale the signal strength in a non-linear manner, e.g., as shown in FIG. 5 . The illustrated scaling signal strength applies to both the K channel and CMY channels. In a related implementation, we determine the luminance of the yellow channel. We base our scaling decisions on the yellow luminance percentage.

Saturated Color

Hiding signals in a saturated color area can also result in increased hidden signal visibility concerns. For this document the term “saturation” refers to how pure a color is, or refers to a measure of color intensity. For example, saturation can represent the degree of color intensity associated with a color's perceptual difference from a white, black or gray of equal lightness. We determine the color saturation level in a color plane (e.g., the yellow color plane), and then scale back a hidden signal strength as the color saturation level exceeds a predetermined level (e.g., 80% yellow color saturation). As with the FIGS. 4 and 5 implementations, we can scale the signal strength in a linear manner ( FIG. 6 ) or in a non-linear manner (FIG. 7 ).

Low or High Luminance Areas

We have found that we can even further improve the visibility characteristics of our hidden signals by considering the amount of luminance at a given pixel or other media 12 area. A low luminance may indicate that there is insufficient CMY to compensate for a K channel tweak. For example, a 10% luminance in CMY for a given pixel implies that the pixel can accommodate only about a 10% signal tweak (e.g., remember the simplified luminance relationship mentioned above: L=0.3*C+0.6*M+0.1*Y). With reference to FIG. 8 , we can cap (or limit) the positive K tweak signal level in such low CMY areas to ensure that the CMY levels can be sufficiently decreased to counteract or cancel the positive K channel signal.

Similarly, in an area of high CMY luminance, a negative K channel tweak can be capped (or limited) to ensure a sufficient range to increase the CMY values.

Equalizing Detectability

Now consider an implementation where media 12 is segmented into a plurality of blocks (FIG. 9 ). Here a block size can range from a pixel to a group of pixels. We redundantly embed an image or watermark signal in each of (or a subset of) the blocks. As shown in FIG. 10 , we preferably use signal feedback (k) to regulate the embedding process. A signal feedback (k) method is shown in FIG. 11. A black (K) channel image or watermark signal (in this section hereafter both referred to as a “watermark”) is embedded in block i of media 12 (step 110 ), where “i” is an integer ranging from 1-n and where n is the total number of blocks. The watermark signal is inverted (step 112 ) and embedded in the CMY channels of block i (step 114 ). At this point, we preferably perform a detection process of the signal embedded within the i th block (step 116 ). The detection process determines whether the signal is sufficiently detectable (step 118 ). The term “sufficient” in this context can include a plurality of levels. In one, “sufficient” implies that the signal is detectable. In another, the detectability of the signal is ranked (e.g., according to error correction needed, ease of detectability, or a detection-reliability metric, etc.). The term sufficient in a ranking context also implies that the detection ranking is above a predetermined threshold. The process moves to embed a new block i+1 if the embedding is sufficient ( 120 ). Otherwise the signal strength is increased or otherwise altered (step 122 ) and the embedding of block i is repeated.

Such a signal feedback process helps to ensure consistent embedding throughout media 12 .

Infrared Image Detection

An infrared detection method is illustrated with reference to FIG. 12 a . In particular, the illustrated detection method employs infrared illumination to facilitate image (or watermark) detection. Media 12 is illuminated with an infrared illumination source 14 . The media 12 is embedded as discussed above, for example, to include various components in a multicolor dimension space (e.g., CMYK). A first component (or image) is preferably embedded in the CMY channels. A second component (or image) is embedded in the K channel. The second component is preferably inverted (or is out-of-phase) with respect to the CMY channels.

Infrared illumination source 14 preferably includes a light emitting diode, e.g., emitting approximately in a range of 800 nm-1400 nm, or a plurality of light emitting diodes (“LED”). Of course, there are many commercially available infrared diodes, and such may be suitable used with our present detection techniques. It will be appreciated that many commercially available incandescent light sources emit light both in the visible and infrared (“IR”) spectrums. Such incandescent light sources may alternatively be used as infrared illumination source 14 . Indeed, infrared watermark detection may be possible in otherwise normal (“daylight”) lighting conditions, particularly when using an IR-pass filter.

A conventional power source powers the infrared illumination source. (We note that a variable trim resistor and a small wall transformer can be optionally employed to control illumination source 14 .). Power alternately can be supplied from a battery pack, voltage or current source, or by directly tapping a power source of a camera, e.g., internally drawn from a parallel, USB, or corded power lines. For a consumer device, a battery pack or a single power cord that is stepped down inside a digital watermark reader housing can also be used.

Returning to the composition of an out-of-phase hidden image (or watermark), a first image (or watermark) component is embedded in a K (or black) channel. A second image component, e.g., which is out-of-phase with respect to the K channel, is embedded in the CMY channels. These characteristics have significance for infrared detection. In particular, C, M and Y inks will typically have high transmission characteristics in the infrared spectrum when printed, which render them nearly imperceptible under infrared illumination. Yet conventional black inks absorb a relatively high amount of infrared light, rendering the black channel perceptible with infrared illumination. We note that standard processing inks, such as those conforming to the standard web offset press (SWOP) inks, include black ink with IR detection properties. Of course, there are many other inks that may be suitably interchanged with the present invention.

As discussed above our out-of-phase embedding provides an effective cancellation of perceived luminance changes when the CMYK image is viewed in the visible spectrum. Indeed, combining an inverted watermark signal “tweak” or “bump” in a K channel with a corresponding non-inverted watermark signal tweak or bump in the CMY channels effectively cancels an overall perceived luminance change. However, under infrared illumination, the hidden image (or watermark) component in the black (K) channel becomes perceptible without interference from the C, M and Y channels. An infrared image primarily portrays (e.g., emphasizes) the black channel, while the C, M and Y channels are effectively imperceptible under infrared illumination.

In one implementation, camera 16 captures an image of media 12 . Preferably, camera 16 includes an IR-Pass filter that passes IR while filtering visible light. For example, the Hoya RM90 filter available from M&K Optics L.L.C. is one of many IR-Pass/Visible Opaque filters suitable for daylight detection. Another suitable filter is the RG850 filter, part number NT54-664, available from Edmund Scientific. These filters are offered as examples only, and certainly do not define the entire range of suitable IR-pass filters. Of course there are many other IR-Pass filters that are suitably interchangeable with the present invention.

In yet another implementation, a conventional digital camera (or web cam) is modified so as to capture infrared light. In particular, most digital cameras and web cams include an IR filter, which filters out IR light. Removing the IR filter allows the camera to capture light in the IR spectrum. Consider a visibly dark environment (e.g., an enclosed case, shielded area, dark room, etc.). Media 12 is illuminated by infrared illumination source 14 in the visibly dark environment. Camera 16 (without an IR filter) effectively captures an infrared image (i.e., the K channel image) corresponding to the illuminated media 12 .

The captured image is communicated to computer 18 . Preferably, computer 18 includes executable software instructions stored in memory for execution by a CPU or other processing unit. If media 12 includes a digital watermark, the software instructions preferably include instructions to detect and decode the embedded digital watermark. Otherwise, the instructions preferably include instructions to display the K-phase image. The software instructions can be stored in memory or electronic memory circuits. Of course, computer 18 can be a handheld computer, a laptop, a general-purpose computer, a workstation, etc. Alternatively, computer 18 includes a hard-wired implementation, which precludes the need for software instructions.

With reference to FIG. 12 b , a detection housing 20 can be provided to house an infrared illumination source 14 and digital camera (both not shown in FIG. 12 b , since they are within the opaque housing 20 ). The housing 20 is preferably opaque to shield (or otherwise constructed to filter) the camera and media 12 from visible light. The housing 20 has an opening 20 a to receive the media 12 . In a first case, opening 20 a is adapted to engulf media 12 . This allows media 12 to be placed on a surface (e.g., table, imaging station, or counter) and the housing opening 20 a to be placed over media 12 , effectively shielding media 12 from visible light. In a second case, the opening 20 a receives media 12 into (e.g., slides media through opening 20 a ) and positions media 12 within the opaque housing 20 . In either implementation, the infrared illumination source 14 illuminates media 12 , and the digital camera 12 captures an image of the illuminated media (e.g., captures as image of the K-channel image). The digital camera 12 communicates with computing device 14 , which detects and decodes a digital watermark embedded with media 12 , if present, or otherwise displays the image.

In another illustrative embodiment, the above described infrared detection technique is carried out in a visibly dark environment, such as a dark room, shielded area, etc. An out-of-phase image (or digital watermark) is embedded in media. The media is illuminated with an infrared illumination source, and a digital camera captures an image of the illuminated media.

In still another illustrative embodiment, the above described infrared detection technique is carried out in a visibly lighted environment. An out-of-phase image (or watermark) is embedded in media. The media is illuminated with an infrared illumination source, and a digital camera captures an image of the media. Preferably, the camera includes an IR-pass filter. The digital camera communicates with a computing device, which detects and decodes an out-of-phase image (or digital watermark) embedded in the media.

Infrared detection is an elegant solution to detect out-of-phase images or digital watermarks, since high transmission colors in the IR spectrum are effectively washed out, allowing detection of a low transmission color channel. Specialized inks are not required to embed the out-of-phase digital watermark. Indeed most multicolor printer ink packs, offset ink, process inks, dye diffusion thermal transfer inks, such as inks conforming to the SWOP standard include black inks that allow infrared detection. Some of these inks include a carbon-based black ink, furthering the absorption of IR. While infrared detection is ideal for out-of-phase images or digital watermarks, this method is also applicable to detection of conventional digital watermarks. For instance, a watermark signal can be embedded only in a black channel of media. Infrared illumination helps to reveal the embedded watermark in this black channel. Alternatively, a digital watermark is embedded across many color planes, while detection is carried out in only those color planes that are perceptible with IR illumination. Additionally, while we have discussed infrared detection techniques, we note that ultraviolet (UV) detection is also possible. In this case, one of the color channels (including the K channel) preferably includes UV pigments or properties. A UV detection process is carried out in a manner analogous to that discussed above. (We also note that a CMY color can include IR/UV pigments or properties to facilitate detection of that color with respective IR or UV detection methods.).

Applications

Now consider a few applications of our inventive out-of-phase hiding techniques.

Identification Documents (e.g., Passports, Driver's Licenses, etc.)

An out-of-phase image is hidden in an identification document to provide enhanced security. For example, a hidden image is a gray-scale version of the identification document's photograph. An airport screener, or law enforcement officer, illuminates the out-of-phase image with infrared (or ultraviolet) light for comparison of the hidden image to the printed photograph. Or, instead of a photograph, the hidden image may include text, which can be compared with the visibly printed text on the identification document.

In assignee's U.S. application Ser. No. 10/094,593, titled “Identification Document Including Embedded Data,” filed Mar. 6, 2002 (Attorney Docket No. P0587), we disclosed various security and authentication improvements. One disclosed improvement ties machine-readable code such as barcode information to a digital watermark. Our inventive out-of-phase hiding techniques can be used with the techniques disclosed in the above-mentioned application. For example, instead of hiding an out-of-phase image in the identification document, we instead embedded an out-of-phase digital watermark. The digital watermark includes a payload, which has information corresponding to the printed information or to information included in a barcode. In one implementation, the information includes a hash of the barcode information. In another implementation, we hid a barcode in the identification document as discussed below.

Hiding Bar Codes in Out-of-Phase Channels

Over the years, a number of standards organizations and private entities have formed symbology standards for bar codes. Some examples of standards bodies include the Uniform Code Council (UCC), European Article Numbering (EAN, also referred to as International Article Numbering Association), Japanese Article Numbering (JAN), Health Industry Bar Coding Counsel (HIBC), Automotive Industry Action Group (AIAG), Logistics Application of Automated Marking and Reading Symbols (LOGMARS), Automatic Identification Manufacturers (AIM), American National Standards Institute (ANSI), and International Standards Organization (ISO).

The UCC is responsible for the ubiquitous bar code standard called the Universal Product Code (UPC). AIM manages standards for industrial applications and publishes standards called Uniform Symbology Standards (USS). Some well know bar code schemes include UPC and UCC/EAN-128, Codabar developed by Pitney Bowes Corporation, 12 of 5 and Code 128 developed by Computer Identics, Code 39 (or 3 of 9) developed by Intermec Corporation, and code 93.

Some bar codes, such as UPC, are fixed length, while others are variable length. Some support only numbers, while others support alphanumeric strings (e.g., Code 39 supports full ASCII character set). Some incorporate error checking functionality.

While the bar codes listed above are generally one-dimensional in that they consist of a linear string of bars, bar codes may also be two-dimensional. Two dimensional bar codes may be in a stacked form (e.g., a vertical stacking of one-dimensional codes), a matrix form, a circular form, or some other two-dimensional pattern. Some examples of 2D barcodes include code 49, code 16k, Data Matrix developed by RVSI, QR code, micro PDF-417 and PDF-417.

For more information on bar codes, see D. J. Collins, N. N. Whipple, Using Bar Code-Why It's Taking Over, (2d ed.) Data Capture Institute; R. C. Palmer, The Bar Code Book, (3 rd ed.) Helmers Publishing, Inc., and P. L. Grieco, M. W. Gozzo, C. J. Long, Behind Bars, Bar Coding Principles and Applications, PT Publications Inc., which are herein incorporated by reference.

A hidden, out-of-phase image can include a barcode. Consider the vast possibilities. A barcode is often disdained for aesthetic reasons, but a hidden, out-of-phase barcode can carry relatively large amounts of information while remaining virtually imperceptible. In one implementation, a barcode is redundantly hidden or titled throughout media using our out-of-phase embedding techniques. This allows for robust barcode detection even if only a portion of the media is recoverable. In another implementation one or more barcodes are placed in predetermined areas throughout the image. In still another implementation, a barcode reader, such as those provided by Symbol (e.g., the VS4000 and P300IMG models) or Welch Allyn (e.g., the Dolphin model), is augmented with an infrared illumination source and/or IR-filters. Once illuminated, the barcode reader detects and decodes a barcode hidden in a K channel.

Fiducials and Orientation Signal

In some digital watermarking techniques, the components of the digital watermark structure may perform the same or different functions. For example, one component may carry a message, while another component may serve to identify the location or orientation of the watermark in a signal. This orientation component is helpful in resolving signal distortion issues such as rotation, scale and translation. (Further reference to orientation signals can be made, e.g., to previously mentioned application Ser. No. 09/503,881.). In some cases, channel capacity is congested by an orientation signal.

One improvement is to embed an orientation signal using our out-of-phase hiding techniques. The message component of a digital watermark can then be embedded using out-of-phase or non-out-of-phase embedding techniques. This improvement will increase message capacity, while improving visibility considerations. Scale, orientation, and image translation can be resolved based on the orientation of the fiducial.

A related improvement embeds a plurality of fiducials or orientation markers 54 in an out-of-phase channel of media 12 (FIG. 13 ). A watermark detection module detects the fiducials to identify distortion.

Spot Colors

We have found that our inventive techniques are not limited to process colors. Indeed, our out-of-phase techniques can be extended to spot colors. (See Assignee's U.S. patent application Ser. No. 10/074,677, filed Feb. 11, 2002, for a further discussion of spot colors and digitally watermarking spot colors. The Ser. No. 10/074,677 application is herein incorporated by reference.). With reference to FIG. 14 , and preferably (but not limited to) relatively darker spot colors, e.g., violets, blues, etc., we counteract a watermark signal (or image) embedded in the spot color channel with an inverted signal in a K channel. Preferably, the K channel base intensity is subtle (e.g., 0% as represented by the K channel base level dashed line in FIG. 14 ) in comparison to the base level spot color intensity (e.g., 100% intensity as represented by the spot color maximum level dashed line in FIG. 14 ). The watermark signal (or image) signal is embedded through a combination of negative spot color tweaks and positive, offsetting, K channel tweaks. Infrared illumination facilitates detection of the K-channel watermark tweaks. (Embedding a spot color need not be limited to negative tweaks. Indeed, if the spot color is not at 100% intensity, positive spot color tweaks and corresponding negative K channel tweaks can facilitate embedding.).

Paper Information and Printing Processes

Another improvement is to carry printing process information and/or paper characteristics with a digital watermark. For example, a digital watermark may include signal gain or embedding characteristics that are specific to a printing press, printing process, process ink type or paper characteristics. The digital watermark can be embedded in a digital file, which is analyzed prior to a print run. The embedding process is adjusted according to the watermark data. Or the watermark signal can be analyzed after printing one or more test copies. The signal strength or payload metric can be analyzed to determine whether the process should be adjusted.

Our out-of-phase digital watermark can be used to detect a misalignment in a printing process. With reference to FIG. 15 a printer 150 outputs a CMYK (or spot color, etc.) printed sheet 152 . The printed sheet includes an out-of-phase digital watermark or image hidden therein. An input device 154 captures an image of sheet 152 . Preferably, input device 154 captures a visible spectrum image of sheet 152 . The input device provides the captured image (e.g., digital scan data) to a watermark detector 156 . The watermark detector 156 analyzes the captured image in search of the embedded out-of-phase digital watermark. The watermark detector 156 should not be able to detect the embedded watermark if the printing of the CMY and K are aligned, due the localized cancellation of the signal tweaks (or luminance changes). The term aligned in this context implies that the CMY and K are sufficiently inverted to allow localized cancellation. A misalignment is identified if the watermark detector 156 reads the digital watermark. Such a misalignment is optionally communicated from the watermark detector 156 to the printer 150 or otherwise provided to announce the printing misalignment. Of course other alignment and color balance information can be identified from the detection of the digital watermark.

Color Channel Keys

A related inventive technique embeds a key in one color channel for decoding a watermark in a second color channel. Consider an implementation where a first digital watermark is embedded in a first color channel. The first digital watermark includes a payload including a key. The key is used to decode a digital watermark embedded in a second color plane. The term decode in this context includes providing a reference point to locate the second watermark, providing a key to unlock, decrypt, decode or unscramble the second digital watermark payload, etc. Of course this inventive technique is not limited to our out-of-phase digital watermarks.

Fragile Security

Our out-of-phase hiding techniques are fragile since a signal processing operation that combines the K channel with the CMY channels effectively cancels the hidden signal. A fragile watermark is one that is lost or degrades predictably with signal processing. Conversion to other color spaces similarly degrades the watermark signal. Take a typical scan/print process for example. Digital scanners typically have RGB image sensors to measure the image color. Scanning an out-of-phase embedded CMYK image degrades the embedded watermark due to the combination of K with CMY in a local area, effectively canceling the watermark. When the RGB image representation is converted to CMYK and printed, the watermark signal is effectively lost. Similarly, other conversions, such as to a L*a*b color space, degrade the out-of-phase watermark due to the combination of K with CMY throughout local areas. Nevertheless, the watermark signal is detectable from an original CMYK media, since the K channel can be detected separately by viewing, e.g., in the near infrared.

A fragile watermark has utility in many applications. Take counterfeiting, for example. The inventive fragile watermark is embedded in original CMYK media. If the media is copied, the embedded fragile watermark is either lost or degrades predictably. The copy is recognized as a copy (or counterfeit) by the absence or degradation of the fragile watermark. Fragile watermarks can also be used in conjunction with other watermarks, such as robust watermarks. The fragile watermark announces a copy or counterfeit by its absence or degradation, while the other robust watermark identifies author, source, links and/or conveys metadata or other information, etc. In other embodiments, a fragile watermark is an enabler. For example, some fragile watermarks may include plural-bit data that is used to enable a machine, allow access to a secure computer area, verify authenticity, and/or link to information. This plural-bit data is lost or sufficiently degrades in a copy, preventing the enabling functions.

Another inventive feature is to embed a hash or other representation of a product (e.g., product code or serial number) in a digital watermark payload or message. The digital watermark is then tied or linked directly to the product. If the product includes a barcode having the product code, such can be compared with the digital watermark.

Conclusion

Preferably, an out-of phase watermark signal is embedded 180 degrees out-of-phase with corresponding channels. However, some cancellation will still be achieved if the signal is approximately 180 degrees, for example, in a range of ±0-50% from the 180-degree mark. The term “inverted” includes values within this range. We note that while the present invention has been described with respect to CMYK process inks, the present invention is not so limited. Indeed, our inventive techniques can be applied to printing processes using more than four inks with the K channel canceling the three or more color channels. Similarly, as shown above under the spot color discussion, our inventive techniques are also applicable to printing processes using less than four inks. Of course our techniques can be used with a variety of printing techniques, including offset printing, dye diffusion thermal transfer (D2T2), other thermal transfers, process ink printing, etc., etc., etc.

The section headings in this application are provided merely for the reader's convenience, and provide no substantive limitations. Of course, the disclosure under one section heading may be readily combined with the disclosure under another section heading.

To provide a comprehensive disclosure without unduly lengthening this specification, the above-mentioned patents and patent applications are hereby incorporated by reference, along with U.S. patent application Ser. No. 09/694,465, filed Oct. 22, 2000. The particular combinations of elements and features in the above-detailed embodiments are exemplary only; the interchanging and substitution of these teachings with other teachings in this application and the incorporated-by-reference patents/applications are also contemplated.

The above-described methods and functionality can be facilitated with computer executable software stored on computer readable media, such as electronic memory circuits, RAM, ROM, magnetic media, optical media, memory sticks, hard disks, removable media, etc., etc. Such software may be stored and executed on a general purpose computer, or on a server for distributed use. Data structures representing the various luminance values, out-of-phase embedded signals, embedded color planes, color signals, data signals, luminance signals, etc., may also be stored on such computer readable media. Also, instead of software, a hardware implementation, or a software-hardware implementation can be used.

In view of the wide variety of embodiments to which the principles and features discussed above can be applied, it should be apparent that the detailed embodiments are illustrative only and should not be taken as limiting the scope of the invention. Rather, we claim as our invention all such modifications as may come within the scope and spirit of the following claims and equivalents thereof.

Claims

1. A method to detect a first image hidden in a second image, the second image comprising a plurality of color channels, wherein the hiding of the first image in the second image comprises at least hiding a first image component in a first of the plurality of color channels, and in the same second image location, hiding the component in a second of the plurality of channels, the second channel component being inverted with respect to the first channel component, said method comprising:illuminating the embedded second image with infrared light; capturing a digital image of the illuminated second image; and analyzing the captured illuminated digital image to detect the first image.

2. The method according to claim 1, wherein the first image comprises at least one of a photograph, graphic, logo, barcode, digital watermark, image, text, message and orientation fiducial.

3. The method according to claim 2, wherein the first color channel comprises black, and the second color channel comprises cyan, magenta and yellow channels, wherein the second channel component is embedded in each of the cyan, magenta and yellow components, the second channel component being inverted in each of the cyan, magenta and yellow channels with respect to the first channel component.

4. The method according to claim 2, wherein the first color channel comprises black and the second color channel comprises cyan, magenta and yellow channels.

5. The method according to claim 4, wherein at least said black color channel conforms to standard web offset press (SWOP) standards.

6. The method according to claim 2, wherein the first color channel comprises black, and the second color channel comprises a spot color.

7. The method of claim 2, wherein the inverted nature of the second channel component with respect to the first channel component cancels localized luminance attributable to the hiding of the first channel component.

8. The method according to claim 1, further comprising shielding the media from visible light.

9. The method according to claim 1, further comprising the step of filtering visible light before said step of capturing a digital image.

10. A method to detect a digital watermark embedded in media, wherein the media includes a plurality of color planes and the digital watermark includes at least a first component and a second component, wherein the second component is embedded in a second color plane at a first media location so as to be inverted in the same media location with respect to the first component embedded in a first color plane so as to cancel luminance attributable to the first component, said method comprising illuminating the media with infrared light or ultraviolet light, and detecting the digital watermark from the illuminated media.

11. The method according to claim 10, wherein the plurality of color planes includes at least a black color plane and the second component is embedded in the black color plane.

12. A method of detecting an orientation fiducial steganographically hidden in a second image to help resolve image distortion, the second image comprising a plurality of color channels, wherein the hiding of the orientation fiducial in the second image comprises hiding the fiducial in a first of the plurality of color channels, and hiding the fiducial in a second of the plurality of channels at the same location as the hidden first channel fiducial, the second channel fiducial being inverted with respect to the first channel fiducial so as to cancel localized luminance attributable to the first channel fiducial, said method comprising the steps of illuminating the second image with infrared light, capturing a digital image of the illuminated second image, and analyzing the captured illuminated digital image to detect the fiducial.

13. The method of claim 12, wherein the fiducial comprises an indication of an original scale and orientation of at least one of the orientation fiducial and second image.

14. The method of claim 13, further comprising comparing the orientation and scale of the detected fiducial to an expected orientation and scale to rescale and reorient the second image.

15. The method of claim 15, wherein said image further comprises a third channel with the fiducial hidden therein at the same spatial location as the hidden first channel fiducial, the image further comprising a fourth channel with the fiducial hidden therein at the same spatial location as the hidden first channel fiducial, wherein the second channel fiducial is inverted with respect to the third channel fiducial and the fourth channel fiducial so as to cancel at least some localized liminance attibutable to the third channel fiducial and fourth channel fiducial.

16. A method to detect a digital watermark embedded in printed media, wherein the printed media includes a plurality of color planes and the digital watermark includes at least a first component and a second component, wherein the second component is embedded so as to visually effect a first spatial area, wherein the embedded second embodiment component reduces at least some visual artifacts at the first spatial area that are attributable to the embedded first component, said method comprising illuminating the media at least one of infrared light and ultraviolet light, and detecting the digital watermark from an image captured from the illuminated media.

17. The method according to claim 16, wherein the plurality of color planes includes at least a black color plane and the second component is embedded in the black color plane.

18. The method of claim 17, wherein the plurality of color planes includes a cyan color plane, a magenta color plane and a yellow color plane, and the first component is embedded in each of the cyan color plane, magenta color plane and yellow color plane.

19. The method of claim 18, wherein the first component is embedded at different strengths in at least one of the cyan color plane, magenta color plane and yellow color plane.

20. The method of claim 16, wherein the visual artifacts comprise a change in luminance that is attributable to the embedded first component.

21. A method to detect a first image hidden in a second image, the second image comprising a plurality of color channels, wherein the hiding of the first image in the second image comprises at least hiding a first image component in a first of the plurality of color channels, and hiding the component in a second of the plurality of channels, the second channel component being hidden so as to be spatially registered with the first image component to reduce at least some visual artifacts that are attributable to the first image component, said method comprising;illuminating the second image with at least of infrared light and ultraviolet light; optically capturing an image of the illuminated second image; and analyzing the captured illuminated image to detect the first image component.

22. The method according to claim 21, wherein the second color channel comprises black, and the first color channel comprises cyan, magenta and yellow channels, wherein the first image component is embedded in each of the cyan, magenta and yellow channels, the second image component being spatially registered with the first image components to reduce at least some visual artifacts that are attributable to the first image component in each of the cyan, magenta and yellow channels.

23. The method of claim 21, wherein the second channel component is hidden so as to be spatially registered with the first image component to cancel at least some localized luminance that is attributable to the hiding of the first image component.

24. The method according to claim 21, wherein the first image comprises at least one of a photograph, graphic, logo, barcode, digital watermark, image, text, message and orientation fiducial.