JP4948505B2 - Digital watermark synchronization recovery device and digital watermark synchronization recovery program - Google Patents

Description

  The present invention relates to a digital watermark synchronization recovery apparatus and a digital watermark synchronization recovery program that detect the positions of a plurality of reference points embedded in a document and identify the embedded position of the digital watermark.

Conventionally, for example, there is a technique for improving the security of a printed material by superimposing a light pattern on a document such as a printed material and concealing information in a microscopic shape of the pattern.
For example, the following Patent Document 1 discloses a method of forming a pattern with dots and concealing binary information depending on the presence or absence of a dot at a specific position.
Patent Document 2 below discloses a technique in which blocks having a plurality of types of waveforms called information units are arranged and printed on a document, and information is embedded depending on which type of block exists at a specific position. ing. A pattern in which this information is embedded is often called a digital watermark.

In these methods, information is embedded by associating the position of a document with a specific dot (or block), so that the position information at the time of embedding must be recovered in order to detect the embedded information.
That is, at the time of embedding and detection, it is necessary to perform embedding processing and detection processing using the same coordinate axis. This is called digital watermark synchronization recovery.

Many methods have been proposed as methods for performing synchronization recovery of digital watermarks, but a typical method is a method of embedding a synchronization recovery pattern simultaneously with a digital watermark.
Patent Document 1 below discloses a method in which a rectangular information block is constituted by a set of dots, and reference patterns for synchronization recovery are arranged at four corners (or a part thereof) of the information block.
Since this information block is repeatedly arranged on the document, the reference pattern is also periodically arranged similarly to the information block.
A conventional digital watermark synchronization recovery device uses a pattern matching technique to extract a reference pattern that is periodically arranged, and determines the coordinates based on the reference pattern to perform digital watermark synchronization recovery. I am doing so.

JP-A-11-112787 (paragraph numbers [0012] to [0013], FIG. 13) JP 2003-209676 A (paragraph numbers [0010] to [0012], FIG. 1)

Since the conventional digital watermark synchronization recovery device is configured as described above, if a part of a document is cut, it is possible to identify which part is cut out even if some reference patterns are extracted. There is a problem that the digital watermark synchronization recovery cannot be performed.
In addition, it is necessary to periodically arrange the reference pattern in the same manner as the information block. However, since the human eye is sensitive to regularity, the periodic reference pattern is easily noticeable, and the confidentiality of the digital watermark deteriorates. There were some issues.

  The present invention has been made in order to solve the above-described problems. Even when a part of a document is cut out and a part of reference points is lost, the digital watermark can be synchronized and recovered. It is an object of the present invention to obtain a digital watermark synchronization recovery apparatus and digital watermark synchronization recovery program that can arrange dots regularly and perform digital watermark synchronization recovery without degrading confidentiality.

  The digital watermark synchronization recovery apparatus according to the present invention provides a distance between the length of a line segment connecting the reference points acquired by the original paper image reference point position acquisition means and the reference point calculated by the print paper image reference point position calculation means. A line segment ratio calculating unit that calculates a line segment ratio that is a ratio to the length of a connecting line segment, and a direction angle formed by a line segment that connects the reference points acquired by the original paper image reference point position acquiring unit and a printed paper surface image Direction angle difference calculating means for calculating a direction angle difference, which is a difference from a direction angle formed by line segments connecting the reference points calculated by the reference point position calculating means, and a line segment ratio calculated by the line segment ratio calculating means And a correspondence relationship for determining a correspondence relationship between a plurality of reference points embedded in the original paper image and a plurality of reference points embedded in the printed paper image from the direction angle difference calculated by the direction angle difference calculation means Decision And a geometric conversion parameter calculating means for calculating a geometric conversion parameter indicating a geometric change between the original paper surface image and the printed paper surface image based on the correspondence relationship of the plurality of reference points determined by the correspondence relationship determining means The print paper image correction means geometrically converts the print paper image using the geometric conversion parameter calculated by the geometric conversion parameter calculation means, thereby correcting the print paper image geometrically. It is what you do.

  According to this invention, the length of the line segment connecting between the reference points acquired by the original paper surface image reference point position acquisition means and the length of the line segment connecting between the reference points calculated by the print paper surface image reference point position calculation means. A line segment ratio calculating unit that calculates a line segment ratio, and a direction angle formed by a line segment connecting the reference points acquired by the original paper surface image reference point position acquiring unit and a printing paper surface image reference point position calculating unit. Direction angle difference calculating means for calculating a direction angle difference, which is a difference between the direction angles formed by the line segments connecting the calculated reference points, and the line segment ratio and direction angle difference calculating means calculated by the line segment ratio calculating means. A correspondence determining means for determining a correspondence between a plurality of reference points embedded in the original paper image and a plurality of reference points embedded in the printed paper image from the direction angle difference calculated by Decisive factor Provided with a geometric conversion parameter calculation means for calculating a geometric conversion parameter indicating a geometric change between the original paper image and the printed paper image based on the correspondence relationship between the plurality of reference points determined by Since the correcting means is configured to geometrically correct the printed paper image by geometrically converting the printed paper image using the geometric conversion parameter calculated by the geometric conversion parameter calculating means. Even if a part of the document is cut out and some reference points are lost, digital watermark synchronization can be recovered, and the reference points are regularly arranged without degrading confidentiality. There is an effect that digital watermark synchronization recovery can be performed.

Embodiment 1 FIG.
FIG. 1 is a block diagram showing a digital watermark synchronization recovery apparatus according to Embodiment 1 of the present invention.
In FIG. 1, a digital watermark embedding apparatus 1 is composed of, for example, a personal computer or a printer. The digital watermark embedding apparatus 1 embeds a digital watermark in an original paper image and uses a plurality of known arrangements for specifying the embedded position of the digital watermark. A process of embedding the reference point in the original paper image and printing the original paper image on the paper is performed.
Here, although the arrangement of the reference points will be described in the second embodiment, for example, a reference pattern as disclosed in Patent Document 2 may be arranged (see FIG. 4).

The original page image reference point position acquisition unit 2 includes an information input interface such as a communication device or a USB port, and acquires the positions of a plurality of reference points embedded in the original page image from the digital watermark embedding apparatus 1. Perform the process. The original page image reference point position acquisition unit 2 constitutes an original page image reference point position acquisition unit.
Incidentally, when the original paper image is a document, the digital watermark embedding apparatus 1 embeds a reference point as a digital watermark. In this case, position information indicating the position of the reference point embedded as a digital watermark is acquired from the digital watermark embedding apparatus 1. Alternatively, the image data (electronic data) of the original paper image is received from the digital watermark embedding apparatus 1, and the position of the reference point is calculated from the image data.
If the original paper image is a natural image obtained by photographing a digitized natural object, the reference point may not be embedded as a digital watermark. When the reference point is not embedded as a digital watermark, a feature point is extracted from an image signal such as a natural image using an appropriate filter or the like, and the feature point is used as the reference point.

  The original page image vector calculation unit 3 is composed of an arithmetic unit such as an MPU, for example, and selects any two reference points from among a plurality of reference points acquired by the original page image reference point position acquisition unit 2. A vector (line segment) connecting two reference points is determined, and a process for calculating the length of the vector is performed.

The printed paper surface image acquisition unit 4 is composed of, for example, a digital scanner and performs a process of scanning the paper surface printed by the digital watermark embedding apparatus 1 and acquiring a printed paper surface image on the paper surface.
The printing paper image reference point position calculation unit 5 is composed of an arithmetic device such as an image detector or MPU, and is embedded in the printing paper image acquired by the printing paper image acquisition unit 4 by performing, for example, a pattern matching method. A process of detecting a plurality of reference points and calculating positions of the plurality of reference points is performed.
The printing paper surface image acquisition unit 4 and the printing paper surface image reference point position calculation unit 5 constitute printing paper surface image reference point position calculation means.

The printing paper image vector calculation unit 6 is composed of an arithmetic unit such as an MPU, for example, and selects any two reference points from a plurality of reference points detected by the printing paper image reference point position calculation unit 5. A vector (line segment) connecting two reference points is determined, and a process for calculating the length of the vector is performed.
The line segment ratio calculation unit 7 is composed of an arithmetic unit such as an MPU, for example, and the vector length calculated by the original page image vector calculation unit 3 and the vector length calculated by the print page image vector calculation unit 6. A process of calculating a line segment ratio, which is the ratio of.
The original paper surface image vector calculation unit 3, the printing paper surface image vector calculation unit 6, and the line segment ratio calculation unit 7 constitute a line segment ratio calculation unit.

The direction angle difference calculation unit 8 is composed of an arithmetic unit such as an MPU, for example. The direction angle formed by the vector determined by the original page image vector calculation unit 3 and the vector determined by the print page image vector calculation unit 6 are the same. A process of calculating a direction angle difference that is a difference from the formed direction angle is performed.
The original sheet image vector calculating unit 3, the printing sheet image vector calculating unit 6, and the direction angle difference calculating unit 8 constitute a direction angle difference calculating unit.

  The reference point correspondence determining unit 9 is composed of an arithmetic unit such as an MPU, for example, and is based on the line segment ratio calculated by the line segment ratio calculating unit 7 and the direction angle difference calculated by the direction angle difference calculating unit 8. Then, a process of determining correspondence between a plurality of reference points embedded in the original paper image and a plurality of reference points embedded in the printed paper image is performed. The reference point correspondence determination unit 9 constitutes correspondence determination means.

  The geometric conversion parameter calculation unit 10 is composed of an arithmetic unit such as an MPU, for example, and based on the correspondence between a plurality of reference points determined by the reference point correspondence determination unit 9, an original paper image and a printing paper image A process of calculating a geometric transformation parameter indicating a geometric change between them (a geometric transformation generated between the original paper image and the printed paper image) is performed. The geometric conversion parameter calculation unit 10 constitutes a geometric conversion parameter calculation unit.

The printing paper image correction unit 11 is configured by an arithmetic device such as an MPU, for example, and uses a geometric conversion parameter calculated by the geometric conversion parameter calculation unit 10 to geometrically convert the printing paper image. Then, a process of geometrically correcting the printed paper image is performed.
The printing paper surface image correcting unit 11 constitutes a printing paper surface image correcting unit.

In the example of FIG. 1, an original paper surface image reference point position acquisition unit 2, an original paper surface image vector calculation unit 3, a printing paper surface image acquisition unit 4, and a printing paper surface image reference point position calculation unit, which are components of the digital watermark synchronization recovery apparatus. 5, each of the print sheet image vector calculation unit 6, the line segment ratio calculation unit 7, the direction angle difference calculation unit 8, the reference point correspondence determination unit 9, the geometric transformation parameter calculation unit 10, and the print sheet surface image correction unit 11 is dedicated. However, when the digital watermark synchronization recovery apparatus is configured by a computer, the original paper surface image reference point position acquisition unit 2, the original paper surface image vector calculation unit 3, the printing paper surface image Acquisition unit 4, printing paper surface image reference point position calculation unit 5, printing paper surface image vector calculation unit 6, line segment ratio calculation unit 7, direction angle difference calculation unit 8, reference point correspondence determination unit 9 A digital watermark synchronization recovery program describing the processing contents of the geometric transformation parameter calculation unit 10 and the printing paper image correction unit 11 is stored in the memory of the computer, and the CPU of the computer stores the electronic data stored in the memory. A watermark synchronization recovery program may be executed.
FIG. 2 is a flowchart showing the processing contents of the digital watermark synchronization recovery apparatus according to Embodiment 1 of the present invention.

Next, the operation will be described.
The outline of the digital watermark synchronization recovery apparatus according to the first embodiment will be described. The length and direction of a vector connecting two feature points on the original paper image and the two feature points on the printed paper image are connected. By performing a generalized Hough transform that compares the length and direction of the vector, and determining the correspondence between the feature points on the original paper image and the feature points on the printed paper image, the geometric transformation parameters In addition, the digital watermark synchronization recovery is performed by correcting the printed paper image by geometric conversion using the parameters of the geometric conversion.

Here, the Hough transform is a technique for detecting a specific figure from an image, and is basically a pattern matching technique, but in order to determine corresponding points by voting, compared to the brute force method, It is characterized by high processing efficiency.
The Hough transform was originally proposed as a method for detecting a straight line, but can also be used for detecting a figure formulated with a small number of parameters such as a circle.
The generalized Hough transform is an extension of the Hough transform so that a figure having an arbitrary shape can be detected by registering the coordinates of each point of the contour and the direction of the contour at each point.
In the first embodiment, this method is used as a digital watermark reference point embedded in an original paper image and a printed paper image that is electronic data obtained by scanning a printed matter on which the original paper image is printed. It is used to determine the correspondence with the reference point detected from

The processing contents of the digital watermark synchronization recovery apparatus will be specifically described below.
First, the original page image reference point position acquisition unit 2 acquires position information indicating the positions of a plurality of reference points embedded in the original page image from the digital watermark embedding device 1 (step ST1). At this time, the image data of the original paper image may be received from the digital watermark embedding apparatus 1 and the position of the reference point may be calculated from the image data.
Here, for convenience of explanation, it is assumed that position information indicating the positions of M reference points P _k1 (k1 = 0, 1, 2,..., M−1) is acquired.

When the original paper surface image reference point position acquisition unit 2 acquires the positions of the M reference points P _k1 , the original paper image vector calculation unit 3 _selects any two reference points P _pi from the M reference points P _k1. = (X _pi , y _pi ), P _pj = (x _pj , y _pj ) is selected to determine a vector P _pi P _pj connecting the two reference points P _pi , P _pj, and the following equation (1) As shown in FIG. 4, the length u _pipj of the vector P _pi P _pj is calculated (step ST2). FIG. 3 is an explanatory diagram showing the direction of a line segment connecting two reference points and the coordinates of the midpoint of the line segment.
Vector: P _pi P _pj = (x _pi −x _pj , y _pi −y _pj ) = (p _x , p _y )
The length of the vector _{P pi P pj: u pipj =} ((p x) 2 + (p y) 2) 1/2
(1)
The original page image vector calculation unit 3 calculates the vector length for all the combinations of the reference points P _k1 (k1 = 0, 1, 2,..., M−1). However, for the same combination of two reference points, only the length of one vector is calculated, and the length of the vector in the reverse direction is not calculated.

The printed paper surface image acquisition unit 4 scans the paper surface printed by the digital watermark embedding apparatus 1 and acquires electronic data indicating the printed paper surface image on the paper surface (step ST3).
When the printing paper surface image acquisition unit 4 acquires electronic data indicating the printing paper surface image, the printing paper surface image reference point position calculation unit 5 performs, for example, a pattern matching method and a plurality of embedded paper surface images. A reference point is detected, and the positions of a plurality of reference points are calculated (step ST4).
Here, for convenience of explanation, the positions of N reference points Q _k2 (k2 = 0, 1, 2,..., N−1) are calculated.
Incidentally, the coordinate system used when the printing paper surface image reference point position calculation unit 5 calculates the position of the reference point Q _k2 may be tentatively determined. However, in general, in the Hough transform, if the origin is set at the center of the document, the performance is improved. Will improve.

Here, when an original paper image is printed on paper by the digital watermark embedding device 1 and the paper surface is scanned to obtain electronic data indicating the printed paper image on the paper, the original paper image and the printed paper image A geometric change occurs between the two.
In the first embodiment, it is assumed that a geometric conversion parameter indicating a geometric change is represented by (a, b, c, d, e, f).
At this time, the positions (x _k1 , y _k1 ) of the M reference points P _k1 acquired by the original paper surface image reference point position acquisition unit 2 and the N pieces of N calculated by the printing paper surface image reference point position calculation unit 5. The position (x _k2 , y _k2 ) of the reference point Q _k2 is expressed as follows.
x _k1 = a × x _k2 + b × y _k2 + c
y _k1 = d × x _k2 + e × y _k2 + d
Note that a certain amount of error may be included in the detection of the N reference points Q _{k2 by} the printing paper surface image reference point position calculation unit 5, and such an error corresponds to the reference point in the following step ST9. It is eliminated at the stage of processing to determine the relationship.

When the printing paper surface image reference point position calculating unit 5 detects N reference points Q _k2 , the printing paper surface image vector calculation unit 6 detects any two reference points Q _qi = (out of the N reference points Q _k2. x _qi , y _qi ), Q _qj = (x _qj , y _qj ) are selected to determine a vector Q _qi Q _qj connecting the two reference points Q _qi , Q _qj , and is shown in the following equation (2) Thus, the length v _qiqj of the vector Q _qi Q _qj is calculated (step ST5).
Vector: Q _qi Q _qj = (x _qi -x _qj , y _qi -y _qj ) = (q _x , q _y )
The length of the vector Q _qi Q _qj : v _qiqj = ((q _x ) ² + (q _y ) ² ) ^1/2
(2)
Note that the printing paper surface image vector calculation unit 6 calculates the vector length for all combinations of the reference points Q _k2 (k2 = 0, 1, 2,..., N−1). However, for the same combination of two reference points, only the length of one vector is calculated, and the length of the vector in the reverse direction is not calculated.

Segment ratio calculation unit 7, the original paper image vector calculation unit 3 calculates the length u _Pipj vector P _pi P _pj, the print paper image vector calculation section 6 calculates a length v _Qiqj vector Q _qi Q _qj Then, as shown in the following equation (3), a line segment ratio S _PQ which is a ratio of the length u _pipj of the vector P _pi P _{pj and} the length v _qiqj of the vector Q _qi Q _qj is calculated (step ST6). ).
_Line segment ratio: S _PQ = v _qiqj / _upipj (3)

When the original paper image vector calculation unit 3 determines the vector P _pi P _pj and the printing paper surface image vector calculation unit 6 determines the vector Q _qi Q _qj , the direction angle difference calculation unit 8 is expressed by the following equation (4). Thus, the direction angle difference θ _PQ that is the difference between the direction angle formed by the vector P _pi P _pj and the direction angle formed by the vector Q _qi Q _qj is calculated (step ST7).
Inner product = vector P _pi P _pj × vector Q _qi Q _qj
= U _pipj × v _{qiqj cosθ PQ}
= P _x × q _x + p _y × q _y
∴ θ _PQ = cos ^-1 ((p _x × q _x + p _y × q _y ) / (u _pipj × v _qiqj )
(4)

When the line segment ratio calculation unit 7 calculates the line segment ratio S _PQ and the direction angle difference calculation unit 8 calculates the direction angle difference θ _PQ , the reference point correspondence relationship determination unit 9 calculates the line segment ratio S _PQ and the direction angle. Correspondences between a plurality of reference points embedded in the original paper image and a plurality of reference points embedded in the printed paper image are determined from the difference θ _PQ (step ST8).
Hereinafter, the reference point P _k1 (k1 = 0, 1, 2,..., M−1) on the original paper image in the reference point correspondence determining unit 9 and the reference point Q _k2 (k2 = 0) on the printed paper image. , 1, 2,..., N-1) will be specifically described.

First, the reference point correspondence determination unit 9 prepares N−1 planes of two-dimensional buffers h (s, θ) used for voting in the Hough transform.
This two-dimensional buffer h is prepared for each combination of the values (s, θ) that the geometric transformation parameters s, θ can take for each k2 of k2 = 0 to N−1. Therefore, the buffer h can be regarded as a three-dimensional buffer of (k2, s, θ). Note that s means the line segment ratio S _PQ , and θ means the direction angle difference θ _PQ .

Next, the values of the two-dimensional buffer h (s, θ) for all k2 of k2 = 0 to N−1 are all initialized to “0”.
At the reference point P _k1 (k1 = 0, 1, 2,..., M−1) on the original paper surface image, all of the other than k1 = pi on the paper image from the reference point P _pi with k1 = pi. A vector P _pi P _pj (pi = 0, 1, 2,..., M−1) to the reference point P _pj (pj = 0, 1, 2,..., M−1), pj is excluding pi, k1 = 0, 1, 2,..., M−1 (total (M−1) × (M−1)) is as follows.

Vector: P ₀ P ₁ , P ₀ P 2,..., P ₀ P pj,..., P ₀ P _M-2 , P ₀ P _M−1 ← M−1 Vectors: P ₁ P ₀ , P ₁ P ₂ ,. P ₁ Ppj, ..., P ₁ P _M-2 , P ₁ P _M-1 <-M-1 vector: P ₂ P ₀ , P ₂ P ₁ , ..., P ₂ Ppj, ..., P ₂ P _M-2 , P ₂ P _M-1 <-M-1 vector: P ₃ P ₀ , P ₃ P ₁ , P ₃ P 2,..., P ₃ Ppj,..., P ₃ P _M-2 , P ₃ P _M-1 ← M-1 vector: …………………………………… P _p iPpj ………… ← M-1 vector: P _M-1 P ₀ , P _M-1 P ₁ , P _{M - 1P2, ..., P M-} 1 Ppj, ..., P M-1 P M-2 ← M-1 pieces

Next, at the reference point Q _k2 (k2 = 0, 1, 2,..., N−1) on the printed paper image, from the point Q _qi where k2 = qi, all except k2 = qi on the same paper image. Is a vector Q _qi Q _qj (qi = 0, 1, 2,..., N−1) to a reference point Q _qj (qj = 0, 1, 2,..., N−1), and qj excludes qi. k2 = 0, 1, 2,..., N−1 (total (N−1) × (N−1)) is as follows.
Vector: Q ₀ Q ₁ , Q ₀ Q2, ..., Q ₀ Qqj, ..., Q ₀ Q _N-2 , Q ₀ Q _N-1 ← N-1 Vectors: Q ₁ Q ₀ , Q ₁ Q2, ..., Q ₁ Qqj, ..., Q ₁ Q _N-2 , Q ₁ Q _N-1 ← N-1 Vectors: Q ₂ Q ₀ , Q ₂ Q ₁ , ..., Q ₂ Qqj, ..., Q ₂ Q _N-2 , Q ₂ Q _N-1 <-N-1 vector: Q ₃ Q ₀ , Q ₃ Q ₁ , Q ₃ Q 2,..., Q ₃ Qqj,..., Q ₃ Q _N-2 , Q ₃ Q _N-1 ← N-1 vector: …………………………………… Q _q iQqj ………… ← N-1 vector: Q _N-1 Q ₀ , Q _N-1 Q ₁ , Q _{N - 1Q2, ..., Q N-} 1 Qqj, ..., QN-1QN-2 ← N-1 pieces

Here, k1 (= pi) is set to 0 to N−1 with respect to N−1 vectors Q _qi Q _{qj in} which k2 is fixed to qi and qj is changed by 0 to N−1 (excluding qi). In the case of changing to p, _Pj was obtained between (M−1) × (M−1) vectors P _pi P _pj with 0 to M−1 (excluding pi). The value of the two-dimensional buffer h corresponding to the line segment ratio S _PQ and the direction angle difference θ _PQ is incremented by “1” (see FIG. 7).

The qi fixed relative to 0~N-1 N-1 single vector Q _qi Q _qj which was varied (excluding qi) and qj, all the calculation of the line segment ratio S _PQ and direction angle difference theta _PQ (M−1) × (M−1) vectors P _pi P _pj , and the values of the two-dimensional buffers h (s, θ) of k1 = 0 to M−1 are compared. When the value of k1 having the two-dimensional buffer h including the largest value is pi, the reference point Q _k2 (k2 = qi) on the printed paper image is the reference point P _k1 (k1 = pi) on the original paper image. ).
By performing this for all the reference points Q _k2 from 0 to N−1, the reference point P _k1 on the original paper image corresponding to all the reference points Q _k2 detected on the printed paper image can be determined. it can.

In the above example, the value of k1 having the “two-dimensional buffer h including the largest value” is simply searched for and the correspondence relationship is determined. For example, a certain threshold is set for the value of the buffer h. If it is set and no buffer h greater than the threshold is found, the reference point P _k1 (k1 = pi) on the original paper image corresponding to the reference point Q _k2 (k2 = qi) on the printed paper image is not found. Also good.

In the geometric transformation parameter calculation unit 10, the reference point correspondence determination unit 9 uses the reference point P _k1 (k1 = 0, 1, 2,..., M−1) on the original paper image and the reference point on the printed paper image. When the correspondence relationship with Q _k2 (k2 = 0, 1, 2,..., N−1) is determined, the document shows a geometric change occurring in the communication path based on the correspondence relationship between the reference points. Geometric transformation parameters (a, b, c, d, e, f) are calculated (step ST9).

That is, the geometric conversion parameter calculation unit 10 has the following relationship between the feature points (x _k1 , y _k1 ) on the original paper image and the feature points (x _k2 , y _k2 ) on the print paper image: Assuming that approximation can be performed by affine transformation, geometric transformation parameters (a, b, c, d, e, f) indicating geometric changes are determined.
x _k1 = a × x _k2 + b × y _k2 + c
y _k1 = d × x _k2 + e × y _k2 + d
The error of the sum of squares of the coordinates before conversion and the coordinates after conversion is expressed as follows.
S _x = Σ _{i = 1} ⁿ (ax _k2 + by _k2 + c−x _k1 ) ²
S _y = Σ _{i = 1} ⁿ (cx _k2 + dy _k2 + e−x _k1 ) ²
Here, n is the number of feature points corresponding to the original paper image and the printed paper image.
The geometric conversion parameter calculation unit 10 performs the calculation of the least square method so that the errors S _x and S _{y of} these square sums are minimized, and the geometric conversion parameters (a, b, c, d, e, f) is determined.

When the geometric conversion parameter calculation unit 10 determines the geometric conversion parameters (a, b, c, d, e, f), the printed paper image correction unit 11 determines the geometric parameters (a, b, c, d). , E, f) as affine transformation parameters, the printed paper image is geometrically corrected by executing the affine transformation of the printed paper image (step ST10).
Further, the print paper surface image correction unit 11 determines where the corrected print paper image corresponds to the original paper image.

  In the latter-stage digital watermark detection apparatus (not shown), when detecting the digital watermark embedded in the print paper image, the digital paper synchronization is recovered by referring to the print paper image corrected by the print paper image correction unit 11. Do.

  As is apparent from the above, according to the first embodiment, the length of the line segment connecting the reference points acquired by the original paper image reference point position acquisition unit 2 and the print paper image reference point position calculation unit 5 are calculated. A line segment ratio calculation unit 7 that calculates a line segment ratio that is a ratio to the length of a line segment that connects between the reference points, and a line segment that connects the reference points acquired by the original paper surface image reference point position acquisition unit 2 A direction angle difference calculation unit 8 that calculates a direction angle difference that is a difference between a direction angle formed by a line segment that connects between the reference angle calculated by the printing paper surface image reference point position calculation unit 5 and a line segment; Based on the line segment ratio calculated by the ratio calculation unit 7 and the direction angle difference calculated by the direction angle difference calculation unit 8, a plurality of reference points embedded in the original paper image and a plurality of images embedded in the printing paper image. Criteria for determining correspondence with reference points Based on the correspondence between the correspondence determining unit 9 and the plurality of reference points determined by the reference point correspondence determining unit 9, a geometric transformation parameter indicating a geometric change between the original paper image and the printed paper image is calculated. The geometric conversion parameter calculation unit 10 is provided, and the print paper image correction unit 11 uses the geometric conversion parameter calculated by the geometric conversion parameter calculation unit 10 to geometrically convert the print paper image. Since the printed paper image is geometrically corrected, even if a part of the document is cut out and some reference points are lost, the digital watermark can be recovered and the reference points can be recovered. Are arranged regularly, and there is an effect that synchronization of digital watermark can be recovered without degrading confidentiality.

In addition, the correspondence between the reference points embedded with the digital watermark is determined by using the generalized Hough transform based on the direction of the vector connecting the reference points and the length of the vector. There is an effect that digital watermark synchronization recovery can be performed without arranging them.
Further, since the calculation of the geometric transformation parameters (a, b, c, d, e, f) is performed based on the coincidence of the main reference points, when some reference points are lost or an error occurs. Even when a certain reference point exists, there is an effect that the correct geometric transformation parameters (a, b, c, d, e, f) can be obtained stably.

In the first embodiment, the reference point correspondence determination unit 9 determines the reference point P _k1 on the original paper image corresponding to the maximum value buffer h (s, θ) and the reference point on the printed paper image. Although the case where the correspondence relationship is determined as that corresponding to Q _k2 is shown, if there is an error in the position of the reference point Q _k2 detected by the printing paper surface image reference point position calculation unit 5, the vote is true. It may be scattered around and the detection accuracy may decrease.
This problem of a decrease in detection accuracy can be reduced by smoothing the two-dimensional buffer h (s, θ).
Further, by taking the average of the parameters near ± 1 of the two-dimensional buffer h (s, θ) to obtain the final value of the two-dimensional buffer h, a decrease in detection accuracy can be reduced.

Embodiment 2. FIG.
In the first embodiment, reference is not particularly made to the arrangement of reference points with respect to the original paper image, but a plurality of reference points may be randomly arranged with respect to the original paper image.
In other words, by determining the positions of a plurality of reference points embedded in the original paper image by the digital watermark embedding device 1 using, for example, random numbers, the plurality of reference points are randomly assigned to the original paper image. It may be arranged.

FIG. 5 is an explanatory diagram showing reference points arranged at random.
In FIG. 5, “x” indicates the position of the reference point, and a plurality of reference points are randomly arranged in the plane, and the arrangement of a part of the reference points is less likely to overlap with other parts. Yes.
For this reason, when a part of the printed matter is found, it can be estimated with high probability from only the arrangement of the reference points, which part of the original paper image corresponds to the part.

For example, when only the central region surrounded by the dotted line in FIG. 5 is cut out, there is a low possibility that the arrangement of the reference points inside the region overlaps with other portions.
On the other hand, as shown in FIG. 6, when the reference points are regularly arranged, the arrangement of some of the reference points overlaps with the other parts, and the cut fragment is the original document. It cannot be determined from the reference point which part corresponds to.

  As is apparent from the above, according to the second embodiment, since a plurality of reference points are randomly arranged with respect to the original paper surface image, even if a part of the document is cut out, the digital watermark It is possible to determine the correspondence between the reference points.

Embodiment 3 FIG.
In Embodiment 1 described above, the digital watermark embedding device 1 acquires position information indicating the positions of a plurality of reference points embedded in the original paper image, but the digital watermark embedding device 1 or the digital watermark embedding is described. An original paper image as electronic data is received from the storage memory of the apparatus 1 via, for example, a communication path, and a plurality of reference points are detected from the original paper image by performing a pattern matching method or the like. Also good.
In addition, by executing the same algorithm as the reference point embedding position determination algorithm executed when the digital watermark embedding apparatus 1 embeds a plurality of reference points, the positions of the plurality of reference points embedded in the original paper image are determined. It may be determined.

1 is a configuration diagram showing a digital watermark synchronization recovery apparatus according to Embodiment 1 of the present invention; FIG. It is a flowchart which shows the processing content of the synchronous recovery apparatus of the digital watermark by Embodiment 1 of this invention. It is explanatory drawing which shows the direction of the line segment which connects between two reference points, and the coordinate of the midpoint of a line segment. It is explanatory drawing which shows arrangement | positioning of a reference pattern. It is explanatory drawing which shows the reference point arrange | positioned at random. It is explanatory drawing which shows the reference point arrange | positioned regularly. It is explanatory drawing which shows the determination method of the correspondence between the reference point on an original paper surface image, and the reference point on a printing paper surface image.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 1 Digital watermark embedding apparatus, 2 Original paper surface image reference point position acquisition part (Original paper surface image reference point position acquisition means), 3 Original paper surface image vector calculation part (Line segment ratio calculation means, Direction angle difference calculation means), 4 Printing paper surface Image acquisition unit (printing paper surface image reference point position calculating unit), 5 printing paper surface image reference point position calculating unit (printing paper surface image reference point position calculating unit), 6 printing paper surface image vector calculating unit (line segment ratio calculating unit, direction angle) Difference calculating means), 7 line segment ratio calculating section (line segment ratio calculating means), 8 direction angle difference calculating section (direction angle difference calculating means), 9 reference point correspondence determining section (corresponding relation determining means), 10 geometry Conversion parameter calculation unit (geometric conversion parameter calculation unit), 11 printing paper surface image correction unit (printing paper surface image correction unit).

Claims (3)

  For specifying the embedding position of the digital watermark, original paper image reference point position acquisition means for acquiring the positions of a plurality of reference points embedded in the original paper image, and on the paper on which the original paper image is printed A printing paper surface image reference point position calculating means for calculating the positions of a plurality of reference points embedded in the printed paper image, and a length of a line segment connecting the reference points acquired by the original paper image reference point position acquiring means; A line segment ratio calculating unit that calculates a line segment ratio that is a ratio to a length of a line segment that connects between the reference points calculated by the printed paper surface image reference point position calculating unit; and the original paper surface image reference point position acquiring unit. This is the difference between the direction angle formed by the line segment connecting the acquired reference points and the direction angle formed by the line segment connecting the reference points calculated by the printing paper surface image reference point position calculating means. A direction angle difference calculating means for calculating a direction angle difference, a line segment ratio calculated by the line segment ratio calculating means, and a direction angle difference calculated by the direction angle difference calculating means are embedded in the original paper surface image. A correspondence relationship determining means for determining a correspondence relationship between a plurality of reference points and a plurality of reference points embedded in the printed paper image, and a correspondence relationship between the plurality of reference points determined by the correspondence relationship determining means. A geometric conversion parameter calculating means for calculating a geometric conversion parameter indicating a geometric change between the original paper image and the printed paper image, and a geometric conversion parameter calculated by the geometric conversion parameter calculating means; The printing paper surface image is geometrically converted by using a printing paper surface image correcting unit that geometrically corrects the printing paper surface image. Watermark of synchronization recovery device.   2. The digital watermark synchronization recovery apparatus according to claim 1, wherein a plurality of reference points are randomly arranged with respect to the original paper image.   The original paper image reference point position acquisition processing procedure for acquiring the positions of a plurality of reference points embedded in the original paper image for specifying the embedding position of the digital watermark, and on the paper surface on which the original paper image is printed A line segment connecting between the reference points acquired by the original paper image reference point position acquisition processing procedure and the original paper image reference point position acquisition processing procedure for calculating the positions of a plurality of reference points embedded in the print paper image of A line segment ratio calculation processing procedure for calculating a line segment ratio that is a ratio of the length of the line segment connecting the reference points calculated by the print paper image image reference point position calculation process procedure, and the original paper image image reference A line connecting the direction angle formed by the line segment connecting the reference points acquired by the point position acquisition processing procedure and the reference point calculated by the printing paper surface image reference point position calculation processing procedure. Direction angle difference calculation processing procedure for calculating a direction angle difference that is a difference from the direction angle formed by the line segment ratio calculated by the line segment ratio calculation processing procedure and the direction calculated by the direction angle difference calculation processing procedure Correspondence determination processing procedure for determining correspondence between a plurality of reference points embedded in the original paper image and a plurality of reference points embedded in the print paper image from the angle difference, and the correspondence determination A geometric transformation parameter calculation processing procedure for calculating a geometric transformation parameter indicating a geometric change between the original paper image and the printed paper image based on a correspondence relationship between a plurality of reference points determined by the processing procedure; By using the geometric conversion parameter calculated by the geometric conversion parameter calculation processing procedure, the printing paper image is geometrically converted, whereby the printing paper surface Electronic watermark synchronization recovery program for executing the printing paper image correction processing procedure of geometrically correct image to the computer.

Images