KR102435876B1 - Method for embedding and extraction of watermarking data - Google Patents

Abstract

A technical problem to be solved through an embodiment of the present invention is to provide a method and an apparatus to which the method is applied, in which the recognition rate is improved during extraction and security is improved by embedding data using a random number sequence. In order to solve the above technical problem, an embedding method according to an embodiment of the present invention comprises the steps of constructing a noise base image in blocks having M x M pixels, for each block, the pixels of all pixels belonging to the block converting the noise base image by unifying values and modifying pixel values of some pixels of each block using the watermark data while traversing each block, The converting may include sequentially adopting a random number from a pre-stored random number sequence while traversing for each block using the watermark data and correcting the pixel values of some of the pixels with a rule corresponding to the adopted random number. can

Description

Method for embedding and extracting watermark data

The present invention relates to a method for embedding and extracting watermark data. In more detail, it relates to a method of embedding watermark data into an image in a manner that is not recognized by human eyes, but can be extracted through image processing, and extracts the embedded watermark data from the image.

By using watermarking technology, watermark data or watermarking textures that are invisible in the original image, but are revealed during unauthorized copying, can be inserted into the original image. This watermarking technology is widely used for purposes such as authenticity authentication or authenticity authentication. However, invisible to the human eye and extraction is possible through image processing through a computing device, but in the case of unauthorized copying, a watermarking technology in a form in which extraction is impossible is not provided.

In the conventional watermark technology, by embedding data according to a rule of one method, a synchronization signal is damaged, and a problem occurs in that the recognition rate is lowered during extraction. Furthermore, since a certain pattern is formed by embedding data according to a rule of one method, there is a security concern due to leakage of a noise-based image in which data is embedded.

Korean Patent Registration No. 10-1877372

Another technical problem to be solved through an embodiment of the present invention is to provide a method and an apparatus to which the method is applied, in which the recognition rate is improved during extraction and security is improved by embedding data using a random number sequence.

Another technical problem to be solved through an embodiment of the present invention is to provide a method for providing a watermark in a form suitable for an original image by changing the shape of a noise base image, and an apparatus to which the method is applied.

The technical problem to be solved by the present invention is to provide an extraction method that determines the existence of watermark data from an image and reduces the time and computing power required for the process of extracting the watermark data.

The technical problems of the present invention are not limited to the technical problems mentioned above, and other technical problems not mentioned will be clearly understood by those skilled in the art from the following description.

In order to solve the above technical problem, an embedding method according to an embodiment of the present invention comprises the steps of constructing a noise base image in blocks having M x M pixels, for each block, the pixels of all pixels belonging to the block converting the noise base image by unifying values and modifying pixel values of some pixels of each block using the watermark data while traversing each block, The converting may include sequentially adopting a random number from a pre-stored random number sequence while traversing for each block using the watermark data and correcting the pixel values of some of the pixels with a rule corresponding to the adopted random number. can

In one embodiment, the M x M pixels are 2 x 2 pixels, and the stored random number sequence is

It is an array of random numbers corresponding to any one of six cases, and the step of modifying the pixel value of the M x M pixel is performed by traversing for each 2 x 2 pixel and using a rule corresponding to the adopted random number. The method may include modifying a pixel value of the 2x2 pixel.

The method may further include obtaining an aspect ratio of the original image, and when the aspect ratio falls within a reference range, the noise base image may be a rectangular image having a shape corresponding to the aspect ratio.

In an embodiment, the rectangle having a shape corresponding to the aspect ratio may correspond to a rectangle having a fixed aspect ratio.

In an embodiment, the rectangle having a shape corresponding to the aspect ratio may correspond to a rectangle having the same aspect ratio as the aspect ratio of the original image.

In an embodiment, the method may further include obtaining a numerical pattern of horizontal and vertical tendency of the original image, and when the numerical value is less than or equal to a reference value, the noise base image may be a rectangular image having a shape corresponding to the numerical value.

In an embodiment, the step of numerically obtaining the tendency may include obtaining a minimum distance from an axis to which a peak coordinate is closest in a magnitude spectrum of a Fourier transform frequency domain of the original image.

In an embodiment, the rectangle may be a rectangle in which a vertical length of the rectangle is longer than a horizontal length when the axis closest to the peak coordinate corresponds to the x-axis.

In an embodiment, the rectangle may be a rectangle in which a horizontal length of the rectangle is longer than a vertical length when the axis adjacent to the peak coordinate corresponds to the y-axis.

In an embodiment, when the random number of the random number sequence is continuously adopted more than a reference number of times, the method may further include replacing the random number with a number other than the adopted random number.

In order to solve the above technical problem, a data extraction method according to an embodiment of the present invention includes the steps of blocking an image in blocks composed of M x M pixels and a noise base blocked in blocks composed of M x M pixels. and extracting data while traversing each block of the image using and extracting the watermark data for each of the blocks using the determined rule.

In an embodiment, the method further includes obtaining a random number sequence identifier from the image, wherein extracting the watermark data while traversing each block of the image includes: and determining the watermark data extraction rule for each block by referring to a random number sequence.

In one embodiment, the apparatus for extracting watermark data includes an image capturing unit, a similarity determining unit and a data extracting unit, wherein the data extracting unit traverses each block in the captured image obtained from the image capturing unit, and stores the A data extraction rule may be determined for each block using a sequence, the watermark data may be extracted for each block using the determined rule, and the extracted data may be output.

1 is a conceptual diagram illustrating a situation in which a method for embedding watermark data according to an embodiment of the present invention and a method for extracting watermark data according to another embodiment of the present invention are utilized.
2 is a diagram for comparing an image in which watermark data is embedded with an original image according to a method of embedding watermark data according to some embodiments of the present invention.
3 is a flowchart of a method for embedding watermark data according to an embodiment of the present invention.
FIG. 4 is a flowchart for explaining in more detail some operations of a method for embedding watermark data, which may be understood with reference to FIG. 3 .
FIG. 5 is a diagram for explaining a result in which a noise base is blocked as a result of some operations that may be understood with reference to FIG. 4 .
FIG. 6 is a flowchart for explaining in more detail some operations of a method for embedding watermark data, which may be understood with reference to FIG. 4 .
FIG. 7 is a flowchart for explaining in more detail some operations of a method for embedding watermark data, which may be understood with reference to FIG. 6 .
8 is an exemplary diagram illustrating a rule corresponding to a random number that may be referred to in FIG. 7 .
9 is a diagram for explaining block traversal for embedding watermark data that may be referred to in some embodiments of the present invention.
10 to 13 are exemplary diagrams for illustrating in detail a method of embedding watermark data that can be understood with reference to FIG. 7 .
14 is a conceptual diagram illustrating a method for embedding watermark data according to an embodiment of the present invention.
15 and 16 are diagrams for explaining synchronization signals that may be referred to in some embodiments of the present invention.
17 is a diagram for explaining in more detail a method of embedding watermark data according to an embodiment of the present invention.
18 to 20 are exemplary diagrams specifically illustrating an embedding method of watermark data that can be understood with reference to FIG. 17 .
21 is a flowchart of a watermark data extraction method according to another embodiment of the present invention.
22 is a diagram for explaining why pre-processing of a photographed image is necessary in the method for extracting watermark data according to an embodiment of the present invention.
23A to 23C are diagrams for explaining an operation of identifying a plurality of image blocks from a captured image in the method of extracting watermark data, which may be understood with reference to FIG. 21 .
24 is a detailed flowchart for explaining in detail an operation of determining a first degree of similarity for each image block and obtaining a rotation angle and a scale in the watermark data extraction method that can be understood with reference to FIG. 21 .
25 is a diagram for explaining a preprocessing result of a photographed image.
FIG. 26 is a detailed flowchart for explaining in detail an operation of determining a second degree of similarity with respect to a selected image block and obtaining a reference point in the method of extracting watermark data, which can be understood with reference to FIG. 21 .
FIG. 27 is a detailed flowchart for describing in detail an operation related to data extraction of the method for extracting watermark data, which may be understood with reference to FIG. 21 .
28 is a block diagram of an apparatus for embedding watermark data according to another embodiment of the present invention.
29 is a block diagram of an apparatus for extracting watermark data according to another embodiment of the present invention.

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings. Advantages and features of the present invention and methods of achieving them will become apparent with reference to the embodiments described below in detail in conjunction with the accompanying drawings. However, the technical idea of the present invention is not limited to the following embodiments, but may be implemented in various different forms, and only the following embodiments complete the technical idea of the present invention, and in the technical field to which the present invention belongs It is provided to fully inform those of ordinary skill in the art of the scope of the present invention, and the technical spirit of the present invention is only defined by the scope of the claims.

In adding reference numerals to the components of each drawing, it should be noted that the same components are given the same reference numerals as much as possible even though they are indicated on different drawings. In addition, in describing the present invention, when it is determined that a detailed description of a related known configuration and/or function may obscure the gist of the present invention, the detailed description thereof will be omitted.

Unless otherwise defined, all terms (including technical and scientific terms) used herein may be used with the meaning commonly understood by those of ordinary skill in the art to which the present invention belongs. In addition, terms defined in a commonly used dictionary are not to be interpreted ideally or excessively unless clearly defined in particular. The terminology used herein is for the purpose of describing the embodiments and is not intended to limit the present invention. In this specification, the singular also includes the plural, unless specifically stated otherwise in the phrase.

In addition, in describing the components of the present invention, terms such as first, second, A, B, (a), (b), etc. may be used. These terms are only for distinguishing the elements from other elements, and the essence, order, or order of the elements are not limited by the terms. When it is described that a component is “connected” or “coupled” to another component, the component may be directly connected to the other component, but another component is “connected” or “coupled” between each component. “It should be understood that it could be.

As used herein, “comprises” and/or “comprising” refers to a referenced component, step, operation and/or element of one or more other components, steps, operations and/or elements. The presence or addition is not excluded.

Hereinafter, some embodiments of the present invention will be described in detail with reference to the accompanying drawings.

summary

1 is a conceptual diagram for explaining a situation in which a method for embedding watermark data 10 according to an embodiment of the present invention and a method for extracting watermark data 10 according to another embodiment of the present invention are utilized. First, the result image 30 created as a result of embedding the watermark data 10 in the original image 20 may be distributed through various channels. The resulting image 30 may be, for example, a business card or a label attached to the product surface. In addition, the result image 30 may be printed on the outer surface of the vehicle, such as a box.

As shown in FIG. 1 , it is difficult to visually recognize whether the watermark data 10 is embedded by looking at the result image 30 . In the watermark data embedding method according to the present embodiment, it is difficult to visually recognize whether the watermark data 10 is embedded, unlike conventional technologies such as barcodes and QR codes. There is an advantage that can contain the mark data (10). In addition, the size of the watermark data 10 according to an embodiment is 144 bits, among which body data is actually usable is 54 bits. As a result, the number of 2 ⁵⁴ cases can be expressed using the watermark data. Therefore, if the present technology is applied to a label attached to the surface of a product, even if it looks like the same label to the naked eye, it is also possible to assign a unique ID to every product and embed the ID in the label as watermark data. Furthermore, more variable data can be created by increasing usable data compared to the existing ²⁴⁸ .

Although it is difficult to visually recognize whether the watermark data is embedded by looking at the result image 30, if the user terminal 900 executing the method for extracting watermark data according to some embodiments of the present invention is used, the result image 30 ), the watermark data 10 can be extracted. For example, the watermark data 10 may be extracted by photographing the result image 30 using a camera provided in the user terminal 900 and executing the watermark data extraction logic on the photographed result. .

The watermark data 10 that may be embedded in some embodiments of the present invention may have the format of Table 1, for example.

4 bit 3 bit 8 bit 54 bit 7 bit 68 bit control parameters body data CRC8 Body data error correction bit version code CRC4 Control parameter error correction bit

The watermark data 10 may include a control parameter of 15 bits, body data of 54 bits, a verification code of 7 bits (CRC 8), and an error correction code of 68 bits. can A control parameter in Table 1 may indicate the type of body data. For example, the user terminal 900 may extract the watermark data 10, recognize 54-bit body data among them, and transmit a content request containing the body data and control parameters to the content server. The content server may store a content DB that manages each content to which a unique ID is assigned by type. The content server may provide the content of the type corresponding to the control parameter among the content of the ID matching the body data to the user terminal 900 .

In an embodiment, the control parameter may include a 4-bit version code, a 3-bit verification code for the corresponding region (CRC4), and an 8-bit error correction code. By separating and managing the watermark data 10 into ²⁴ versions according to the version code, the version can be updated and managed or multiple versions can be operated at the same time.

The 68-bit Error Correction Bit may include an error correction code to compensate for the possibility that the watermark data 10 may be erroneously extracted. For example, the error correction code may be a Bose-Chaudhuri-Hocquenghem (BCH) code.

2 is a diagram for comparing an image in which watermark data is embedded and an original image according to a method of embedding watermark data according to some embodiments of the present invention. As shown in FIG. 2 , the resulting image 30 is not visually different from the original image 20 . However, compared to the monochrome version 20a of the original image, noise stands out in the resultant image 30a converted to monochrome. This is because the watermark data according to an embodiment of the present invention is embedded in the original image in the form of noise. In this regard, it will be described in more detail with reference to some drawings.

Embedding watermark data

Hereinafter, a method for embedding watermark data according to an embodiment of the present invention will be described in its entirety with reference to FIG. 3 . The embedding method of watermark data according to the present embodiment may be executed, for example, by a computing device.

The computing device loads the watermark data (S100) and loads the noise base image (S120). In addition, the original image to contain the watermark data is also loaded (S160). The noise base may be understood as a kind of data container to contain watermark data. The noise base may be loaded from an image file or may be loaded in the form of data indicating a two-dimensional matrix in a frequency domain. Also, in an embodiment, it may be loaded in the form of a magnitude spectrum and a phase spectrum in the frequency domain.

The noise base image is converted according to a pattern determined using the stored watermark data and the stored random number sequence (S140), and the original image is adjusted using the converted noise base image, thereby generating an image embedded with watermark data. and output (S180, S200). For example, the image in which the watermark data is embedded may be output in the form of an image file or may be printed on a printing target such as paper by a printing device. According to an embodiment, the output may be made in the form of being painted on the outer surface of the printing result by the 3D printer.

As described with reference to FIG. 3 , in this embodiment, the noise base plays a key role. The noise base image generated through the frequency domain maintains the characteristics of the synchronization signal even if the image is partially damaged according to the characteristics of the frequency, so that the extraction success rate of watermark data can be increased. In addition, after the original noise-based image is blocked, conversion is performed by the watermark data, so that the original noise-based image is not easily leaked through analysis of the resulting image.

In the conventional watermark technology, by embedding data according to a rule of one method, the synchronization signal is damaged and a problem of a reduction in recognition rate during extraction occurred. Furthermore, as a certain pattern is formed by embedding data according to a rule of one method, there is a security concern due to leakage of a noise-based image in which data is embedded. Hereinafter, a method of embedding watermark data for solving the above problem will be described in more detail with reference to FIGS. 4 to 16 .

FIG. 4 is a more detailed flowchart of the noise-based image conversion step S140 shown in FIG. 3 .

The noise base image is partitioned in blocks of M x M pixels (S141). One block is preferably formed in a rectangular shape composed of an even number of pixels in the horizontal and vertical directions. For example, one block may be 2 x 2 pixels in size. A block of 2 x 2 pixels is a block of the smallest size that can be configured. By configuring one block in a size of 2 x 2, an effect capable of responding to images of various sizes is obtained. However, the present invention is not limited thereto, and one block may be formed in a square shape including an even number of pixels in the horizontal and vertical directions.

In some embodiments, the size of a block may be adjusted according to an embedding setting. When the size of the block increases, there is an effect of reducing the time required for embedding. When the size of the block decreases, there is an effect that more data can be embedded in small pixels.

Next, for each block, the average of the pixel values belonging to the block is calculated, and if the average value is greater than or equal to a reference value (for example, the interim value between a pixel value indicating white and a pixel value indicating black), all of the pixel values in the block are averaged. The color is unified with the first color (eg, black) ( S143 ). When the average value is less than the reference value, all colors of the corresponding block are unified as the second color (eg, white) (S143).

FIG. 5 shows a blocked noise base image 41 as a result of segmentation and unification of the noise base image 40 described with reference to FIG. 4 . Referring to the pixel-by-pixel enlarged figure of the blocked noise base image 41, every square unit block having a size of 2 x 2 pixels is unified with the same pixel value. The current block 41a is assigned a pixel value of 0 and corresponds to white color.

Blocking described with reference to FIGS. 4 and 5 prevents the formation of a repeating pattern that occurs by inserting only watermark data into the noise base image, and the recognition rate at the time of extraction that occurs only by inserting watermark data in units of one pixel deterioration can be prevented. Furthermore, by performing the blocking, it is possible to maintain a synchronization signal corresponding to the frequency characteristic of the original noise base image.

Referring back to FIG. 4 , the blocked noise base image 41 is converted ( S145 ) by modifying some pixel values of the block according to a pattern determined by the stored random number sequence by reflecting the watermark data. Transformation of the blocked noise base image 41 is embodied by some embodiments to be described later.

6 is a flowchart for specifically explaining the transformation ( S145 ) of the noise base image 41 described in FIG. 4 . The embedding method of watermark data according to the present embodiment may be executed, for example, by a computing device.

The computing device loads a pre-stored random number sequence to transform the noise base image ( S1451 ). The pre-stored random number sequence may be one in which repeatedly generated random numbers are stored in the order of generation. However, the present invention is not limited thereto. In one embodiment, the number of random numbers in the random number sequence required for embedding 144-bit watermark data is 144. For example, the random number may correspond to any one selected from among natural numbers 1 to 6.

In the example described above, since one block is divided into 2 x 2 pixels, a total of 6 rules for selecting 2 pixels among 4 pixels must be allocated to represent 1 bit. However, the present invention is not limited thereto, and the number of random numbers may be changed according to the scale of the block.

Meanwhile, any one of two or more sets of random number sequences may be selectively used in the watermark data embedding and extraction process. A plurality of sets of random number sequences are shared in advance between an apparatus for embedding watermark data and an apparatus for extracting watermark data.

In another embodiment, the generated random numbers may be stored as a random number sequence according to the order in which they occur. It is stored as a random number sequence during embedding, and the stored random number sequence is used during extraction to perform extraction according to a corresponding rule. In another embodiment, the random number sequence may be stored in an encrypted form. As it is stored in an encrypted form, it may reduce security concerns due to leakage.

In an embodiment, when the random number of the stored random number sequence is continuously adopted more than a reference number of times, a step of replacing the random number with a number other than the adopted number may be included. A plurality of rules, not one rule, are applied so that the synchronization signal can be maintained by embedding watermark data, and security can be strengthened. It may include a step of substituting a number other than the number continuously adopted.

Next, by using the loaded watermark data and the adopted random number, the pixel value of the block is corrected by a rule corresponding to the random number (S1453). Referring to FIG. 8 , in the case of the 2×2 pixel illustrated above, a total of six rules correspond to preset random numbers.

Hereinafter, with reference to FIG. 7 , the step ( S1453 ) of correcting the pixel value of the block according to the rule corresponding to the random number described in FIG. 8 will be described in more detail. The embedding method of watermark data according to the present embodiment may be executed, for example, by a computing device.

When the current block is white (S1453-1), a rule corresponding to the nth random number is loaded (S1453-2). Referring to FIG. 8 , in the case of 2×2 pixels, a rule corresponding to the number of 6 cases may be confirmed. Next, some pixels of the current block are converted to black to match the corresponding rule (S1453-3).

When the current block is black (S1453-1), a rule corresponding to the nth random number is loaded (S1453-5). Referring to FIG. 8 , in the case of 2×2 pixels, a rule corresponding to the number of 6 cases may be confirmed. However, since the corresponding rule is illustrated based on the case in which the current block is white, the black and white inverted rule corresponds to the rule in the case where the current block is black. Next, some pixels of the current block are converted to white to match the corresponding rule (S1453-6).

Next, if the processing of all watermark data is not completed (S1453-7), the same process is repeated using the watermark data and the random number while traversing each block (S1453-4).

9 illustrates a process of embedding watermark data while traversing each block described above (S1453-4). In order to embed 144-bit watermark data, 144 blocks are configured, and a minimum of 576 pixels is required. In some embodiments to be described later, the noise base image may be formed in a rectangle rather than a square.

Hereinafter, a specific example of the embedding method described in FIG. 7 will be described with reference to FIGS. 10 to 14 . For convenience of description below, it is assumed that a block corresponds to 2 x 2 pixels, and an adopted random number corresponds to any one of natural numbers corresponding to 1 to 6.

Referring to FIG. 10 , the first watermark data is 0, and the adopted random number corresponds to 1. Since the transformation object block 43a is blocked in white, the pixel value of the transformation object block 43a is corrected according to 0 of case 1 corresponding to the random number with reference to FIG. 8 .

Referring to FIG. 11 , the second watermark data is 1, and the adopted random number corresponds to 3. Since the transformation object block 43b is black, the pixel value of the transformation object block 43b is corrected according to the inverted form of 1 in Case 3 corresponding to the random number with reference to FIG. 8 .

Referring to FIG. 12 , the third watermark data is 1, and the adopted random number corresponds to 4. Since the transformation object block 43c is blocked in white, the pixel value of the transformation object block 43c is corrected according to Case 4 1 corresponding to the random number with reference to FIG. 8 .

Referring to FIG. 13 , the fourth watermark data is 0, and the adopted random number corresponds to 5. Since the transformation object block 43d is black, the pixel value of the transformation object block 43d is corrected according to the inverted form of 0 in case 5 corresponding to the random number with reference to FIG. 8 .

7 to 13, a specific method for embedding watermark data on a noise base image 41 that is blocked using random numbers and rules corresponding to random numbers has been described. Unlike the conventional watermark technique of embedding with a single rule, an embodiment according to the present invention does not damage the synchronization signal, so that the recognition rate during extraction can be improved. Furthermore, since a certain pattern is formed by embedding data according to a rule of one method, security concerns due to leakage of a noise-based image in which data is embedded can also be eliminated.

Hereinafter, a method for embedding watermark data according to an embodiment of the present invention described in detail with reference to FIGS. 6 to 13 will be described as a conceptual diagram shown in FIG. 14 . In one embodiment, when embedding the watermark data 10 in the blocked noise base image 41, a pre-stored random number sequence is used, and some pixel values of the block are used as a rule corresponding to the adopted random number while traversing each block. to obtain a transformed noise base image 50 .

The converted noise base image 50 that can be referenced in some embodiments maintains synchronization signal characteristics better than the conventional noise base image converted by one rule, and security is enhanced because a constant pattern is not formed.

Hereinafter, a synchronization signal that may be referenced in some embodiments will be described with reference to FIGS. 15 and 16 . In the noise base image generated through the frequency domain, the synchronization signal characteristics are maintained even if the image is partially damaged according to the characteristics of the frequency. Therefore, geometric distortion can be corrected using the similarity between the synchronization signal and the captured image during the extraction process.

As shown in FIG. 15 , when the noise base image 40 is Fourier transformed, a magnitude spectrum 45 and a phase spectrum 47 in the frequency domain can be obtained. The illustrated noise base image 45 corresponds to an image generated by the synchronization signals 45a, 45b, 45c, 45d. As shown in FIG. 16 , synchronization signals 51aa , 51ab , 51ac and 51ad are maintained in the magnitude spectrum 51a in the frequency domain according to the Fourier transform of the transformed noise base image 50 . As described above, the synchronization signal is maintained even after the steps of blocking and embedding the watermark data, so that the noise base image corresponding to the container containing the data can be recognized in the extraction step.

Referring back to FIG. 3 , an image in which watermark data is embedded is generated by adjusting the original image using the converted noise base image ( S180 ). The adjusting includes adjusting the first pixel only when the R channel value, the G channel value, and the B channel value of the first pixel of the original image all exceed or all fall below the limit value. In this case, adjusting the current pixel of the original image includes first adjusting the first pixel when the second pixel of the converted noise base image is white, and adjusting the first pixel when the second pixel is black and a second reconciliation with a different rule than the first reconciliation. For example, the first adjustment is to decrease the R-channel value of the first pixel, increase the G-channel value, and decrease the B-channel value, and the second adjustment is to increase the R-channel value of the first pixel. increase, decrease the G-channel value, and increase the B-channel value.

In the above adjustment, the values of three constants α, β, and γ are utilized. The values of the three constants are ratios to be used for monochrome in order to maximize the noise of the image in the watermark data extraction process.

Assuming (R, G, B = each pixel value of the original R channel, G channel, and B channel before being updated, R, G, B' = each pixel value of the R channel, G channel, and B channel after update), The following reconciliation rules may be considered.

Case 1: When the pixel value of noise is 0 and the pixel value of the original image is greater than 250 for all three channels R, G, B, R' = R(1 - α), G'= G(1 + β), B' = B(1 - γ)

Case 2: When the pixel value of noise is 0 and the pixel value of the original image is less than 250 for all three channels R, G, B, R' = R(1 - α), G'= G(1 + β), B' = B(1 - γ)

Case 3: When the pixel value of noise is 255 and the pixel value of the original image is greater than 250 for all three channels R, G, B, R' = R(1 + α), G'= G(1 - β), B' = B(1 + γ)

Case 4: When the pixel value of noise is 255 and the pixel value of the original image is less than 250 for all three channels R, G, B, R' = R(1 + α), G'= G(1 - β), B' = B(1 + γ)

The three constants α, β, and γ used to insert noise into the original image depend on the original image. This technology has a characteristic that it is difficult to identify it with the human eye. In order to satisfy this, the α, β, and γ values should be varied according to the pixel values of the original image into which the noise is to be inserted.

Hereinafter, specific examples in which the noise base image corresponds to a rectangle rather than a square in the watermark data embedding method according to an embodiment of the present invention will be described with reference to FIGS. 17 to 20 .

Hereinafter, a method of embedding watermark data based on the aspect ratio of the original image will be described with reference to FIGS. 17 and 18 .

In one embodiment, the step of obtaining the aspect ratio of the original image may be further included, and when the aspect ratio of the original image falls within the reference range, watermark data using a noise base image of a rectangular image instead of a square corresponding to the aspect ratio of the original image can be embedded. In response to the design characteristics of the original image, the watermark can be flexibly modified and embedded. Since the design characteristics of the original image are reflected, recognition performance can be improved when extracting watermark data. For example, when the aspect ratio (=horizontal/length) is 4 or more or 0.25 or less, watermark data can be embedded using a noise base image of a rectangular image.

In an embodiment, when the aspect ratio of the original image falls within the reference range, watermark data may be embedded using a rectangular noise-based image having a fixed aspect ratio.

In an embodiment, when the aspect ratio of the original image falls within the reference range, watermark data may be embedded using a rectangular noise-based image having the same aspect ratio as the aspect ratio of the original image. For example, referring to FIG. 9 , if the conventional square watermark consists of 12×12 blocks and is configured in units of 24×24 pixels, the unit watermark corresponding to the aspect ratio of 4 may consist of 48×12 pixels. . The unit watermark corresponding to the aspect ratio of 0.25 may consist of 12 x 48 pixels.

17 shows the aforementioned rectangular watermark. A square watermark 70 is generated by the square unit watermark 60 of 24 x 24 pixels. A rectangular watermark 70a whose width is longer than the length is generated by the rectangular unit watermark 60a having a width of 48 x 12 pixels longer than the length. A rectangular watermark 70b whose length is longer than the width is generated by the rectangular unit watermark 60b having a length of 12 x 48 pixels longer than the width. As described above, since the watermark is generated by reflecting the design characteristics of the original image, it is possible to improve recognition performance when extracting watermark data.

18 corresponds to an exemplary view showing a specific example in which the above-described rectangular watermark is utilized. Since the watermark is not recognized by the human eye, it can be used together by locating the company's CI or QR code in the center 60c. In particular, when used together with a QR code, it can be used for the purpose of strengthening the security of commerce using the QR code in the future. The rectangular unit watermark 60a having a length longer than the width may be located at the upper and lower ends, and the rectangular unit watermark 60b having a length longer than the width may be located at the left and right ends to constitute a watermark that is not recognized by the human eye. have.

Hereinafter, a method of embedding watermark data based on the pattern tendency of the original image will be described with reference to FIGS. 19 and 20 .

In one embodiment, the step of obtaining the pattern horizontal and vertical tendency of the original image as a numerical value may be further included, and when the numerical value is equal to or less than a reference value, the watermark data is embedded using a rectangular noise base image having a shape corresponding to the numerical value. can be

In an embodiment, the step of numerically obtaining the fringe horizontal and vertical tendency may include obtaining a minimum distance from an axis having the closest peak coordinate in a magnitude spectrum of a Fourier transform frequency domain of the original image. Referring to FIG. 19 , as a result of the Fourier transform of the original image 53 in which vertical fringes are continuous, a magnitude spectrum 53a in which a peak point is shown on the x-axis may be obtained. As a result of the Fourier transform of the original image 55 in which horizontal fringes are continuous, a magnitude spectrum 55a in which a peak point is shown on the y-axis may be obtained. In this example, since it corresponds to the peak point located on the axis, the minimum distance to the adjacent axis corresponds to 0.

In one embodiment, when the minimum distance between the peak coordinate and the adjacent axis is equal to or less than the reference value, when the adjacent axis corresponds to the x-axis, the watermark data is generated using a noise base image in which the length of the rectangle is longer than the width. can be embedded. In one embodiment, when the minimum distance between the peak coordinate and the adjacent axis is equal to or less than the reference value, when the adjacent axis corresponds to the y-axis, the watermark data is generated using a noise base image in which the width of the rectangle is longer than the length. can be embedded. Referring to FIG. 19 , as a result of 53a , a watermark longer in length than in width may be embedded in the original image 53 . As a result of 55a, a watermark longer in width than length may be embedded in the original image 55 . By adjusting the size of the unit watermark in response to the pattern tendency of the original image, the synchronization signal is maintained strong, thereby improving the recognition rate during extraction.

20 shows a specific example of disposing a watermark corresponding to the pattern of the original image described above. A unit watermark 60a whose width is longer than that of length is disposed on the horizontal pattern original image 20b. A unit watermark 60b whose length is longer than that of the width is disposed on the vertical pattern original image 20c. As described above, by adjusting the size of the unit watermark in response to the pattern tendency of the original image, the synchronization signal is strongly maintained to improve the recognition rate during extraction and reduce the recognition speed.

Extraction of watermark data

Hereinafter, a method of extracting watermark data according to another embodiment of the present invention will be described with reference to FIGS. 21 to 27 . The watermark data extraction method according to the present embodiment is applicable only to watermark data embedded by the watermark data embedding method according to the embodiment of the present invention described above. First, it will be described with reference to FIG. 21 .

The watermark data extraction method according to the present embodiment starts by receiving a photographed image and identifying a plurality of image blocks therefrom ( S200 ).

If, like the existing digital watermarking technology, the watermark is extracted on the PC with the digital image file embedded with the watermark as it is, the watermark can be extracted by performing the embedding process exactly in reverse. However, since the technology of the present invention supports the recognition of a watermark using a camera of a mobile terminal such as a smartphone camera, many image preprocessing steps are required before extracting the watermark. Referring to FIG. 22 , the user of the smart phone 900 may photograph the medium 30 ′ on which the image in which the watermarking data is embedded is printed without paying special attention to extracting the watermarking data. The watermark data extraction method according to the present embodiment was developed to enable extraction of watermark data even in this case.

The pre-processing process makes it possible to maintain a high recognition rate even in the presence of many variables such as the position, size, rotation, brightness, and damage of the object that will determine the existence of a watermark in the photo taken by the camera. In this process, using the synchronization signal inserted in the embedding process, the existence of a watermark is checked and the image is transformed into the optimal condition for data extraction.

Before performing the pre-processing in earnest, pre-calibration may be performed. For example, when an object for determining the existence of a watermark is illuminated using a smartphone camera, a preview coming on the smartphone screen is captured in 1920x1080 size. After that, all pixel values of the captured image are changed to RGB format. This is because embodiments of the present invention perform operations based on RGB.

Next, a captured 1920x1080 image (hereinafter, referred to as a “shot image”) is divided into a plurality of image blocks. Here, each of the plurality of image blocks is an image unit to be processed for the first similarity determination in step S210 to be described later, and whether or not watermark data exists in any one image block selected as a result of the first similarity determination is determined and watermark data extraction is attempted.

23A is a diagram illustrating an example in which a photographed image 90 is divided into a total of nine image blocks 91 to 99 having the same size. If the captured image is an image 90 having a size of 1920x1080, the image blocks 91 to 99 may be, for example, a part of the captured image, each having a size of 256x256 pixels or 512x512 pixels. The size of each of the image blocks may be set to any size so that watermark presence determination and watermark data extraction can be smoothly performed in steps S210 to S250 to be described later, but at least watermark embedding It should be set to be larger than the size of the noise base image used for In this case, the size of the noise base image is appropriately selected in consideration of the size (number of bits) of watermark data to be embedded.

In this case, the plurality of image blocks 91 to 99 may partially overlap each other. In the example shown in FIG. 23A , the image blocks 1, 2, and 3 have regions (areas indicated by dark shades) overlapping with the image blocks 4, 5 and 6, respectively, and the image blocks 7 and 8 , 9) have regions (areas indicated by dark shades) overlapping the image blocks 4, 5, and 6, respectively.

A first similarity determination step S210, which will be described later, is a process of finding an image block in which watermark data is most likely to exist in a state where it is unknown in which area within the captured image 90 the watermark data is present. Accordingly, as many image blocks are set to overlap each other in the captured image 90 , the possibility that the watermark present in the captured image 90 cannot be recognized decreases. However, as the number of image blocks increases, computing power and time consumed for the first similarity determination performed on each image block increases. Conversely, as fewer image blocks are set in the captured image 90, the computing power and time consumed in step S210 to be described later are reduced, but the possibility of not recognizing the watermark present in the captured image 90 increases. do. If the watermark is not recognized despite the presence of the watermark in the photographed image 90, eventually, the watermark recognition process must be performed again for a new photographed image, so the finally consumed computing power and delay time may further increase. have.

Meanwhile, although it is preferable that all areas of the captured image 90 are covered by at least one of the plurality of image blocks 91 to 99 , this is not necessarily the case. When a user illuminates an object with the camera, since there is a very high possibility that the object is centered on the screen, some of the outer regions of the captured image 90 are not covered by the plurality of image blocks 91 to 99 described above. Even if it does, the effect on the watermark recognition result may be relatively low.

On the other hand, in some embodiments, the image block to be used in steps (S210) to (S250) to be described later by performing the Deblur process on each of the plurality of image blocks (91 to 99) identified from the previously captured image (90) can make them clearer.

Although an example of identifying a total of nine image blocks 91 to 99 from the captured image 90 has been described in FIG. 23A , the present invention is not limited to such an embodiment. The purpose of the present invention, the performance of the device in which the watermark data extraction function will be implemented, the time required for watermark recognition and watermark data extraction, and computing power, etc., are taken into consideration in consideration of various factors, such as more or fewer You can choose to use image blocks.

23B and 23C show other examples in which a plurality of image blocks are identified from the captured image 90 . As described above, when a user illuminates an object with the camera, there is a high probability that the object will be projected so that the object is in the center of the screen. can do. 23B is a view showing an example in which a plurality of image blocks are arranged around the center of the captured image 90, and FIG. 23C is a plurality of image blocks arranged to overlap each other a lot near the center of the captured image 90 It is a drawing showing an example.

On the other hand, in some embodiments in which any one of two or more random number sequence sets is selectively used in the watermark data embedding and extraction process, a watermark from the captured image 90 in step S250 to be described later An identifier for determining a random number sequence to be referred to when extracting data may be identified from the captured image 90 . Note that the identifier is not a random number sequence itself, but an identifier that distinguishes a plurality of random number sequences from each other. A plurality of sets of random number sequences are shared in advance between an apparatus for embedding watermark data and an apparatus for extracting watermark data. Meanwhile, the identifier may be identified from the captured image 90 using various conventional methods for identifying data from the image.

As described above, a step (S200) of acquiring a captured image and identifying a plurality of image blocks therefrom is performed.

Next, a step S210 of determining a first similarity with a noise base image for each of the plurality of image blocks 91 to 99 is performed. In step S210, the first similarity, rotation angle, and scale of each image block 91 to 99 are determined by comparing each image block 91 to 99 with a noise base image through processing on the frequency domain. . This operation will be described in more detail with reference to FIG. 24 .

First, one image block among the plurality of image blocks 91 to 99 is transformed into the frequency domain (S211), and an image MS2 of a 2D magnitude spectrum of the image block is generated (S212). Also, an image MS1 of a noise-based 2D magnitude spectrum to be compared with the image block is prepared (S213). The noise-based 2D magnitude spectrum image MS1 may be generated from a pre-stored noise-based image or may be simply retrieved from pre-stored data.

Next, the noise-based 2D magnitude spectrum image MS1 and the 2D magnitude spectrum image MS2 of the image block are LogPolar-transformed (S215). Since the LogPolar transform has properties that are invariant to the two-dimensional rotation and size transformation, by using the LogPolar transform, it is possible to obtain an effect of healing the flaws in the photographed image that are reliable in terms of rotation and size.

Next, each image created as a result of LogPolar transformation is transformed into the frequency domain (S215), multiplied by the result of the transformation into the frequency domain (S216), and then as a magnitude matrix of the result of transformation into the frequency domain (f_LP2) of the captured image divided to generate a first result matrix for the image block (S217). The first result matrix may be understood as a matrix indicating how many noise bases are included in the image block and the ratio thereof. Also, the first result matrix may be understood as a matrix indicating a degree of similarity between the image block and the noise-based image.

Next, a first reference image for the image block is generated by transforming the first result matrix for the image block into the spatial domain ( S218 ). A first degree of similarity of the image block is calculated using the first reference image (S219). The first similarity is a maximum value among pixel values of the first reference image with respect to the image block. The rotation angle is the X-coordinate of the pixel having the maximum pixel value, and the scale is the Y-coordinate of the pixel (S219).

Steps S211 to S219 are performed for each of the plurality of image blocks 91 to 99 in the captured image 90 , and as a result, a first degree of similarity for each image block is calculated.

It will be described again with reference to FIG. 21 .

In step S220, a first image block is selected from among the plurality of image blocks 91 to 99 based on the first similarity calculation result in step S210, and the rotation angle and scale are reflected to the first image block. pre-process

In step S220 , first similarities calculated for each of the plurality of image blocks 91 to 99 in step S210 are compared with each other. Then, an image block having the largest first similarity is selected as the first image block. An image block having the highest first similarity in the captured image 90 may be understood as an area in which a watermark is most likely to exist in the captured image 90 . In step S220, the watermark precision recognition and watermark data in steps S230 to S250, which will be described later, by selecting the first area where the watermark is most likely to exist among the plurality of image blocks 91 to 99. The target of the extraction-related processing is limited to a partial region within the captured image 90 rather than the entire captured image 90 . Accordingly, computing power and processing time consumed in the process of performing steps S230 to S250 to be described later may be reduced.

In some embodiments, when the first similarities calculated for each of the plurality of image blocks 91 to 99 are all less than a threshold, it is determined that the watermark data does not exist in any of the plurality of image blocks 91 to 99. may be implying that Therefore, if the first similarity of the first image block is less than the threshold, the watermark data extraction operation for the photographed image 90 is stopped, and the process returns to step S200 to automatically obtain a new photographed image. have. For example, an image displayed as a preview on the screen of a smartphone camera that is photographing an object for which the existence of a watermark is to be determined may be automatically acquired as a new photographed image. Subsequently, a plurality of image blocks in the new captured image are identified, and operations after step S210 may be repeated.

In step S220, pre-processing is performed on the selected first image block.

As described above with respect to step S210, in the process of calculating the first similarity for each of the plurality of image blocks 91 to 99, the rotation angle of each image block (the maximum value among pixels of the first reference image is The pixel's X-coordinate) and scale (the Y-coordinate of the pixel) are obtained together. In step S220, preprocessing is performed on the first image block using the rotation angle and scale obtained with respect to the first image block.

The result of the pretreatment may be understood with reference to FIG. 25 . As a result of the pre-processing, the orientation of the photographed image 90 is matched, and the size is also corrected to match the comparison with the noise base.

It will be described again with reference to FIG. 21 .

In step S230, operations for calculating a second degree of similarity with respect to a first image block whose rotation angle and scale are corrected (hereinafter, referred to as a “corrected image”) and determining a watermark extraction reference point are performed. In step S230, by comparing the corrected image and the noise base image through processing on the frequency domain, a second degree of similarity of the corrected image is calculated, and a reference point that is a reference point for extracting a watermark from the corrected image is determined. This operation will be described in more detail with reference to FIG. 26 .

The second similarity calculation operation for the corrected image in step S230 is compared with the first similarity calculation operation for each of the plurality of image blocks 91 to 99 in step S210, It is similar except that it does not perform a LogPolar transformation on .

Specifically, converting the corrected image into a frequency domain to obtain a frequency domain matrix f2 (S231), querying or generating a noise-based frequency domain matrix f1 (S232), and combining the f1 and f2 Generating R', which is a multiplication matrix (S233), dividing R' by a magnitude matrix of f2 to create a second result matrix (S234), transforming the second result matrix into a spatial domain to obtain a second reference image A generating step (S235) is performed. In this case, the second result matrix may be understood as a matrix indicating how much the noise base is included in the corrected image and the ratio thereof.

In step S236, by analyzing the second reference image generated in step S235, a second degree of similarity between the corrected image and the noise base image and information on the position of the noise base in the corrected image are obtained. Specifically, the coordinates of the reference pixel having the largest pixel value among the pixel values of the second reference image are determined as the reference point, which is the watermark data extraction position. Also, a pixel value of the reference pixel is determined as a second similarity degree of the corrected image.

It will be described again with reference to FIG. 21 .

In step S240, the second similarity calculated for the corrected image in step S236 and a threshold are compared.

If the second degree of similarity is less than the threshold, this may imply that no watermark is present in the corrected image, or that watermark data cannot be extracted from the corrected image. In this case, the watermark data extraction operation for the corrected image (an image obtained by correcting the first image block, which is a partial region selected from the photographed image 90 in step S200), is stopped.

In some embodiments, when the second similarity is less than the threshold, the process returns to step S200 and a new captured image may be automatically acquired. For example, an image displayed as a preview on the screen of a smartphone camera that is photographing an object for which the existence of a watermark is to be determined may be automatically acquired as a new photographed image. Subsequently, a plurality of image blocks in the new captured image are identified, and operations after step S210 may be repeated.

In some other embodiments, when the second similarity is less than the threshold, operations after step S220 may be repeated for an image block other than the first image block selected in step S210 . As described above, in the previous step S210, the image block having the highest first similarity is selected as the first image block based on the first similarity calculation result between the plurality of image blocks 91 to 99 and the noise base image. is selected, and subsequent steps ( S220 to S230 ) are performed based on the first image block. If the second similarity calculated based on the first image block is less than the threshold, the image block having the second highest similarity calculated in step S210 is identified as the second image block, and the second image block Subsequent steps ( S220 to S230 ) may be performed for .

As in the above-described embodiment, the embodiment in which some operations (steps S220 to S250) of the watermark data extraction process are performed on the second image block in the existing photographed image 90 by acquiring a new photographed image Compared with the embodiment in which the entire process (steps S200 to S250) of watermark data extraction is performed, computing power and processing time may be reduced. This is because the process of dividing a new captured image into a plurality of image blocks and calculating the first similarity for each block (steps S200 to S210) can be omitted.

It will be described again with reference to FIG. 21 .

In step S240, if the second similarity is equal to or greater than the threshold, the process proceeds to step S250.

In step S250, watermark data is extracted on the basis of the watermark data extraction reference point obtained in step 240. It will be described with reference to FIG. 27 .

First, the corrected image is monochrome (S251). In this case, the three constants α, β, and γ used to insert noise into the original image are the same. As a result, the noise representation in the corrected image will be maximized.

Then, the monochrome corrected image is partitioned into blocks having a size of 2 x 2 pixels (S252).

In subsequent steps S253 to S256, watermark data is extracted while traversing each block (2 x 2 pixels) of the monochrome corrected image.

First, in step S253, the data extraction rule of the current block of the monochrome corrected image is determined. The data extraction rule corresponds to the data embedding patterns of the six cases described for example with reference to FIG. 8 . In other words, the rule used when embedding the watermark data in each block is the data extraction rule of each block.

The data extraction rule of each block may be determined with reference to a pre-generated random number sequence. For example, the data extraction rule indicated by the first number of the random number sequence (for example, any one of the six patterns illustrated in FIG. 8) is applied to the first block of the monochrome corrected image, and the second block indicated by the second number of the random number sequence is applied to the second block. Data extraction rules apply. It will be understood that the random number sequence exactly matches the random number sequence referenced to determine a data embedding rule in the process of embedding watermark data in an image. In some embodiments, the random number sequence may be shared in advance between a device for embedding watermark data in an image and a device for extracting watermark data from an image.

On the other hand, in some embodiments in which any one of two or more random number sequence sets is selectively used in the watermark data embedding and extraction process, among the plurality of random number sequence sets, the A random number sequence indicated by the identified random number sequence identifier from the photographed image 90 may be identified and used to determine a rule for extracting watermark data from the monochrome corrected image.

Subsequently, in step S254 , the watermark data of the current block is extracted according to the data extraction rule determined in the previous step S253 with reference to the noise base image that is blocked in units of 2×2 pixels. The operation of extracting the watermark data from the current block with reference to the blocked noise base image may be understood as a reverse order of the embedding process described with reference to FIGS. 7 to 13 .

In step S255, it is determined whether watermark data extraction is complete for all blocks of the monochrome corrected image, and if data extraction is complete, data extracted from each block is concatenated, and the entire watermark data is created The entire generated watermark data can be displayed, transmitted, or output.

In step S255, if extraction of all blocks of the monochrome corrected image has not been completed, it moves to the next block (S256). The order of traversing the monochrome corrected image block is the same as the block traversal order of the embedding process. Subsequently, in step S253 again, a data extraction rule for the next block of the monochrome corrected image is determined with reference to the random number sequence, and the process proceeds to step S254 to extract the watermark data embedded in the next block. Repeat.

In the watermark data extraction method according to some embodiments of the present invention described with reference to FIGS. 21 to 27 so far, the entire process of determining whether a watermark exists and attempting to extract watermark data for the captured image 90 (S200 to Instead of performing S250), after calculating the first similarity of each of the plurality of image blocks that are part of the photographed image 90 (S210), a subsequent operation is performed on only one image block having the highest first similarity The steps S220 to S250 are performed. Accordingly, computing power consumed in the operations S220 to S250 is reduced, and a processing delay time caused by the execution of the operations S220 to S250 is reduced. In addition, it is possible to satisfy the needs of users who want to complete the recognition while facing the watermark recognition device in the image in which the watermark is embedded.

In the watermark data extraction method according to some embodiments of the present invention, each bit of watermark data is embedded in a noise-based image or each bit of watermark data is extracted from an image in which a watermark is embedded, Instead of using one and the same embedding/extraction rule for bits, any one embedding/extraction rule among a plurality of embedding/extraction rules indicated by each random number in the pre-generated and shared random number sequence is used. Accordingly, even if the noise base image original 40 is leaked, a third party can use the data embedded in the watermark image without permission, unless the random number sequence used for embedding each bit of the watermark data in the noise base image is leaked. cannot be extracted. That is, the security of the watermark data embedding/extraction algorithm is greatly improved. In addition, the problem that a pattern such as a wave pattern is generated in the image as a result of embedding the watermark is solved.

The watermark data extraction method according to some embodiments of the present invention uses a different random number sequence set for each different watermark, and identifies an appropriate random number sequence set when extracting watermark data so that it can be used for data extraction, Even if one set of random number sequences is leaked, a third party cannot extract data embedded in a watermark image generated using another random number sequence without permission. That is, the security of the watermark data embedding/extraction algorithm can be further improved.

The methods according to the embodiments of the present invention described so far may be performed by executing a computer program implemented as computer readable code. The computer program may be transmitted from the first computing device to the second computing device through a network such as the Internet and installed in the second computing device, thereby being used in the second computing device. The first computing device and the second computing device include all of a server device, a physical server belonging to a server pool for cloud services, and a stationary computing device such as a desktop PC.

The computer program may be stored in a recording medium such as a DVD-ROM or a flash memory device.

The method for embedding and extracting watermark data according to an embodiment of the present invention may be modified and implemented in a form in which different noises are recognized depending on the viewing direction as shown in FIG. 25 .

Hereinafter, a configuration and operation of a data embedding apparatus according to another embodiment of the present invention will be described with reference to FIG. 28 . 28, the data embedding apparatus 300 according to the present embodiment includes a memory (RAM) 330, an image processor 320 that executes a data embedding software operation loaded in the memory 330, and a storage ( 340). In an embodiment, the embedding apparatus 300 may further include an image sensor 310 , a network interface 370 , a system bus 350 , and a printer interface 360 .

The storage 340 stores an executable file (binary file) 342 of data embedding software that implements the data embedding method described with reference to FIGS. 1 to 20 in the form of software. The executable file 342 of the data embedding software may be loaded into the memory 330 via the bus 350 . 28 shows an operation 332 of data embedding software loaded into memory 330 .

The storage 340 may further store the noise base data 341 . In an embodiment, the storage 340 stores the noise base data 341 and stores it in an encrypted form to prevent leakage of the original noise base data to the outside. In an embodiment, the storage 340 may include the noise base data in the form of a resource in the data embedding software executable file to prevent the noise base source from being leaked to the outside.

The storage 340 may further store one or more sets of random number sequences indicating rules for embedding watermark data. In an embodiment, the storage 440 may store the random number sequence in an encrypted form, and may store it in the form of a resource in a data embedding software executable file.

The data embedding device 300 receives watermarking data from an external device through the network interface 370 , or directly from a user through an input unit (not shown) such as a keyboard or mouse, or watermarking data stored in the storage 340 . (not shown) may be read. The data embedding device 300 obtains the original image, which is the target for embedding the watermarking data, from the image sensor 310 photographing the target object, or receives it from an external device through the network interface 370, or the storage 340 It is possible to read the watermarking data (not shown) stored in the .

The data embedding device 300 transmits the resultant image calculated by inputting the watermarking data and the original image into data embedding software to an external device through the network interface 370, or displays it on a display device (not shown), Printing may be performed through a printing device connected through the printer interface 360 .

The data embedding device 300 according to the present embodiment may be, for example, a computing device or a printing device. When the data embedding apparatus 300 is a printing apparatus, the data embedding apparatus 300 may include components for performing a printing function instead of the printer interface 360 .

Hereinafter, the configuration and operation of a data decoding apparatus according to another embodiment of the present invention will be described with reference to FIG. 29 . In this specification, decoding of watermarking data and extraction of watermarking data refer to substantially the same operation. As shown in FIG. 29, the data decoding apparatus 400 according to this embodiment includes a memory (RAM) 430, an image processor 420 that executes a data decoding software operation loaded in the memory 430, and a storage ( 440). In an embodiment, the decoding apparatus 400 may further include an image sensor 410 , a network interface 470 , a system bus 350 , and a watermarking data interpretation processor 460 .

The storage 440 stores an executable file (binary file) 442 of data decoding software that implements the data extraction method described with reference to FIGS. 21 to 27 in the form of software. An executable file 442 of data decoding software may be loaded into memory 430 via bus 450 . 29 shows an operation 432 of data decoding software loaded into memory 430 .

The storage 440 may further store the noise base data 441 . In one embodiment, the storage 440 stores the noise base data 441, but stores the noise base data 441 in an encrypted form, or includes the noise base data in the form of a resource in the data decoding software executable file to provide noise to the outside. It is possible to prevent the base original from being leaked.

The storage 440 may further store one or more sets of random numbers indicating rules for extracting watermark data. In an embodiment, the storage 440 may store the random number sequence in an encrypted form, and may store it in the form of a resource in a data decoding software executable file.

The data decoding apparatus 400 obtains the data of the photographed image from the image sensor 410 photographing the target object on which the image in which the watermarking data is embedded is printed, or receives it from an external device through the network interface 470, or storage Data of the captured image stored in 440 may be obtained. The data of the captured image is loaded into the memory 430 through the system bus 450 and referenced when the data decoding software operation 432 is executed.

The data decoding device 400 displays the watermark data extracted as a result of the execution of the data decoding software operation 432 on a display (not shown), transmits it to an external device through the network interface 470, or the storage 440 You can save it to , or print it in some other way.

In one embodiment, the data decoding apparatus 400 may include a watermarking data interpretation processor 460 that performs a process using the extracted watermark data. For example, the content server (not shown) extracts the body data of the watermark data (see Table 1), and sends a content request containing the body data and control parameters (see Table 1) through the network interface 470 . can be sent to A display device (not shown) of the data decoding apparatus 400 may display the content received from the content server as a response to the content request.

The data decoding apparatus 400 according to the present embodiment may be, for example, a mobile terminal equipped with a camera.

Although embodiments of the present invention have been described above with reference to the accompanying drawings, those of ordinary skill in the art to which the present invention pertains can realize that the present invention can be embodied in other specific forms without changing the technical spirit or essential features. can understand Therefore, it should be understood that the embodiments described above are illustrative in all respects and not restrictive.

Claims (13)

A method of embedding watermark data in an original image, the method comprising:
composing a noise base image in units of blocks having M x M pixels;
unifying the pixel values of all pixels belonging to the block for each block; and
Transforming the noise base image by traversing each block and correcting pixel values of some pixels of each block using the watermark data,
Further comprising the step of obtaining an aspect ratio of the original image,
The noise base image is
A first rectangular image of a shape corresponding to the aspect ratio of the original image,
How to embed data. The method of claim 1,
Transforming the noise-based image includes:
A step of sequentially adopting a random number of a pre-stored random number sequence while traversing for each block using the watermark data and correcting the pixel value of some of the pixels according to a rule corresponding to the adopted random number,
How to embed data. 3. The method of claim 2,
The M x M pixels are
2 x 2 pixels,
The stored random number sequence is
It is an array of random numbers corresponding to any one of six cases,
The step of modifying the pixel value of the M x M pixel comprises:
Iterating for each 2 x 2 pixel and modifying the pixel value of the 2 x 2 pixel with a rule corresponding to the adopted random number,
How to embed data. The method of claim 1,
A first rectangular image of a shape corresponding to the aspect ratio,
A rectangular image with a fixed aspect ratio,
How to embed data. The method of claim 1,
A first rectangular image of a shape corresponding to the aspect ratio,
A rectangular image having the same aspect ratio as the aspect ratio of the original image,
How to embed data. The method of claim 1,
Further comprising the step of obtaining numerically the horizontal and vertical tendency of the pattern of the original image,
When the numerical value is less than or equal to the reference value, the noise base image is a second rectangular image having a shape corresponding to the numerical value,
How to embed data. 7. The method of claim 6,
The step of obtaining the tendency numerically is,
Including the step of obtaining the minimum distance from the axis to which the peak coordinate is closest in the magnitude spectrum of the Fourier transform frequency domain of the original image,
How to embed data. 8. The method of claim 7,
The second rectangular image,
When the axis closest to the peak coordinate corresponds to the x-axis, a rectangular image in which the vertical length of the rectangle is longer than the horizontal length,
How to embed data. 8. The method of claim 7,
The second rectangular image,
When the axis adjacent to the peak coordinate corresponds to the y-axis, a rectangular image in which the horizontal length of the rectangle is longer than the vertical length,
How to embed data. 3. The method of claim 2,
When the random number of the random number sequence is continuously adopted more than a reference number of times, the method further comprising the step of substituting a number other than the adopted random number,
How to embed data. A data extraction method for extracting watermark data embedded in an image, the data extraction method comprising:
blocking the image in blocks of M x M pixels; and
Extracting data while traversing each block of the image using a noise base image that is blocked in blocks of M x M pixels,
The noise base image is
A first rectangular image of a shape corresponding to the aspect ratio of the image or a second rectangular image of a shape corresponding to the pattern tendency of the image,
How to extract data. 12. The method of claim 11,
Extracting data while traversing each block of the image includes:
determining a data extraction rule for each block using the stored random number sequence; and
extracting the watermark data for each of the blocks using the determined rule;
How to extract data. 13. The method of claim 12,
Further comprising the step of obtaining a random number sequence identifier from the image,
Extracting the watermark data while traversing each block of the image comprises:
determining an extraction rule of the watermark data for each block by referring to a random number sequence indicated by the random number sequence identifier among the stored random number sequences;
How to extract data.

Images