KR100945726B1 - Apparatus for watermarking based on local auto-correlation function and method therefor - Google Patents

Abstract

The present invention relates to a technique for watermarking based on a local autocorrelation function. The present invention relates to a local autocorrelation for correctly extracting a watermark inserted from video content in order to extract a watermark even for illegally photographed movies by a camcorder. The present invention relates to a function-based watermarking apparatus and a method thereof.

According to the present invention, a watermark pattern generation unit for generating a low-pass filter by generating a watermark pattern, and determines the intensity of the generated watermark pattern, and adjusts the intensity of the watermark pattern according to the change of the image to the spatial domain of the video frame. An insertion unit to insert, an image input unit to receive a watermark-embedded video frame, and a watermark embedded in the input video frame, and estimate the watermark from the geometric distortion using a local autocorrelation function on the estimated watermark. And a control unit for detecting a message from the restored watermark, and a control unit for controlling the watermark pattern generation unit, the inserting unit, the image input unit, and the detection unit.

The present invention has the effect of having robustness not only in affine transformation attack but also in projection transformation attack.

Watermark pattern, local autocorrelation function, geometric distortion

Description

Apparatus for watermarking based on local auto-correlation function and method therefor}

The present invention relates to a technique for watermarking based on a local autocorrelation function, and more particularly, to extract watermarks for illegally photographed movies by a camcorder. The present invention relates to a watermarking device based on a local autocorrelation function and to a method for extracting a PB correctly.

Watermarking technology has been proposed to satisfy the requirements set out in the Digital Cinema Initiative (DCI) standard for the security of digital cinema, but the solution for piracy or distribution by camcorder attack is not perfect yet. Overseas, companies are conducting research on countermeasures against illegal filming in theaters, while domestic studies on watermarking technology that have robustness against illegal filming in theaters are still incomplete.

Here, the image captured by the camcorder undergoes a geometric transformation, which includes projection transformation as well as affine transformation.

Conventionally, a technique for watermarking is described in a paper published in 1998, M. Kutter, "Watermarking resisting to translation, rotation, and scaling," in Proc. SPIE, 3528, pp. As disclosed in 423-431, 1998, the watermarking technique using an auto-correlation function has proposed a watermarking technique having robustness against affine transformation attacks. However, this technique has a robustness against affine transformation attacks by using an autocorrelation function, but has no strength against projection transformation attacks.

The watermarking technique using the autocorrelation function can restore the image from the affine transformation or small local distortion, but cannot identify the projection transformation representing the horizontal and vertical projections from the distorted image, and the correlation of the watermarks. There is a problem in that relation maximums cannot be extracted accurately.

SUMMARY OF THE INVENTION An object of the present invention is to provide a watermarking apparatus and method based on a local autocorrelation function having robustness not only for affine transformation attacks but also for projection transformation attacks.

Another object of the present invention is to capture a watermark-embedded image with a camcorder, and extract a watermark based on a local autocorrelation function to find a synchronization of the watermark from affine transformed and projection-converted video content. A marking apparatus and method are provided.

According to an aspect of the present invention, there is provided a watermarking apparatus based on a local autocorrelation function, comprising: a watermark pattern generator for generating a watermark pattern and performing low pass filtering; An insertion unit which determines the intensity of the low pass filtered watermark pattern by the watermark pattern generator, and inserts the watermark pattern in the spatial domain of the video frame by adjusting the intensity of the watermark pattern according to the change of the image; An image input unit configured to receive a video frame including a watermark from an image acquisition device that causes geometric distortion; A detector for estimating a watermark inserted in the input video frame, estimating a geometric distortion using a local autocorrelation function on the estimated watermark, and reconstructing the watermark from the estimated geometric distortion; And a controller configured to control the watermark pattern generator, the inserter, the image inputter, and the detector. It includes.

On the other hand, the present invention comprises the steps of (a) estimating a watermark pattern signal from a video frame including the geometric distortion through a Wiener filter; (b) estimating geometric distortion from the watermark pattern signal using a local autocorrelation function; And (c) said Reconstructing the watermark by calculating a perspective projection matrix H for geometric distortion; It includes.

The present invention as described above has the effect of having robustness not only in affine transformation attack, but also in projection transformation attack.

In addition, the present invention has the effect of extracting the inserted watermark by finding the synchronization of the watermark from the affine conversion and projection-converted video content occurring in the camcorder shooting.

The features and advantages of the present invention will become more apparent from the following detailed description based on the accompanying drawings. Prior to this, the terms or words used in the present specification and claims are defined in the technical spirit of the present invention on the basis of the principle that the inventor can appropriately define the concept of the term in order to explain his invention in the best way. It should be interpreted to mean meanings and concepts. In addition, when it is determined that the detailed description of the known function and its configuration related to the present invention may unnecessarily obscure the subject matter of the present invention, it should be noted that the detailed description is omitted.

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

A watermarking apparatus based on a local autocorrelation function according to an embodiment of the present invention will be described with reference to FIGS. 1 and 8.

1 is a schematic block diagram of a watermarking apparatus based on a local autocorrelation function according to an embodiment of the present invention.

As shown in FIG. 1, a watermarking apparatus based on a local autocorrelation function according to an embodiment of the present invention, includes a watermark pattern generator 100, an inserter 200, an image inputter 300, and a detector 400. ) And the control unit 500.

First, the watermark pattern generator 100 includes a pattern generation module 110 for generating a watermark pattern using a secret key, a lowpass filter module 120 for lowpass filtering the generated watermark pattern, and a watermark pattern. And a modulation module 130 for performing modulation to have the content of the message to be inserted. According to the present embodiment, the modulation module inserts a watermark of 2 bits per video frame into an angle of 0, 90, 180, and 270 degrees according to each message 00, 01, 10, 11 to be inserted. Perform rotational modulation.

Here, since the low frequency component of the watermark pattern generated by the pattern generation module 110 is robust to a general image processing process or DA / AD conversion, the watermark pattern generator 100 generates a low pass filter module 120. Perform lowpass filtering through.

In addition, the watermark pattern generation unit 100 according to the present embodiment allows the same watermark pattern to be repeated in the horizontal and vertical directions. This is to estimate what kind of geometric transformation has occurred for watermark synchronization, and we use autocorrelation or local autocorrelation for this estimation. It must be repeated a certain number of times in the longitudinal direction.

In addition, the watermark pattern generation unit 100 according to the present embodiment is configured to express a 2-bit message through modulation for rotating the watermark pattern at any one of 0, 90, 180, and 270 degrees.

In addition, the inserter 200 determines the intensity of the watermark pattern generated by the watermark pattern generator 100, and adjusts the intensity of the watermark pattern according to the change of the image to adjust the intensity of the pattern in the spatial domain of the video frame. This function inserts a watermark pattern and a message whose intensity is adjusted. The insertion unit 200 that performs this function includes a pattern intensity determination module 210, a perceptual scaling module 220, and a message insertion module 230.

Here, the spatial domain is a pixel domain in which no conversion is made to the original video frame, and the x-axis and the y-axis are the horizontal and vertical axes of the original video frame.

The pattern intensity determining module 210 determines the intensity of the watermark pattern for each region of the image by using a noise visibility function (hereinafter referred to as 'NVF') which is a function for calculating a local characteristic of the image. At this time, the NVF value has a value between 0 and 1, has a value of 1 in an area where there is no change of the image, and a value close to 0 in an area where there is a change of the image.

In addition, when the watermark pattern is inserted into the spatial domain, the perceptual scaling module 220 performs perceptual scaling so that the watermark pattern is not visible to the human eye. Here, the perceptual scaling determines the change in the image, and inserts the watermark pattern with the intensity of the watermark pattern lowered in the region where there is no change of the image, and inserts the watermark pattern with the strength of the watermark pattern in the vicinity of the contour or the region having the change of the image.

When inserting the watermark pattern, the message insertion module 230 expresses one message to be inserted as a string of bits of 0 and 1 and inserts the message into the original video frame.

In the watermark inserting unit 200 according to the present embodiment, the watermark embedding at the i-th position of the original video frame occurs as shown in Equation 1 below.

y _i = x _i + sλ _i w _i

Here, x is a brightness component of the original frame, w is a watermark to be inserted, s is a total intensity coefficient of the watermark, and λ is a value calculated from the NVF.

In addition, the image input unit 300 performs a function of receiving a video frame in which a watermark is inserted, as shown in FIG. 2, from an image acquisition device that causes geometric distortion. In the watermarking apparatus based on a local autocorrelation function according to an embodiment of the present invention, an image acquisition device causing geometric distortion is set as a camcorder.

In addition, the detector 400 estimates a watermark inserted in a video frame input through the image input unit 300, extracts correlation maximums using a local autocorrelation function on the estimated watermark, and extracts the extracted correlation. Estimate geometric distortion from maximal values, Restores the watermark from the geometric distortion, and detects the payload.

The detection unit 400 performing this function includes a watermark estimation module 410, a geometric distortion estimation module 420, a watermark restoration module 430, and a detection module 440.

Here, the watermark estimation module 410 receives a video frame with a watermark inserted through the image input unit 300 and performs a function of estimating the inserted watermark using a Wiener filter. In this case, the Wiener filter is generally used to remove noise from the image. Since the noise includes a large amount of pattern data, the Wiener filter may be regarded as a watermark pattern estimated from the image. Such a Wiener filter is expressed as in Equation 2 below.

를 위너 필터를 통하여 추출된 워터마크 패턴 신호로 간주한다.T _L is the data to be tested, T _w is the watermark to be estimated, V is the local variance at position (i, j) corresponding to i rows j columns on the video frame, and M is the average at (i, j). Here, T _w is a blind detector, in which case we use the average of the T _L local variances. therefore,

Is regarded as a watermark pattern signal extracted through the Wiener filter.

In addition, the geometric distortion estimation module 420 extracts correlation maximums using a local auto-correlation function (LACF) on the watermark estimated by the watermark estimation module 410, and extracts the correlation. It performs the function of estimating geometric distortion by using the spacing and angle of maximal values.

The geometric distortion estimation module 420 according to the present embodiment will be described below.

Video frames by the camcorder are perspective projection distortion from a mathematical point of view. Here, the perspective projection matrix transformation is expressed as Equation 3 below.

이고, (x,y)^T는 최초 영상의 좌표를 표시하며, (x',y')^T는 투시적 투영 왜곡된 영상의 좌표를 표시한다. 비록 주어진 행렬 H가 9개의 상수로 이루어지지만, 행렬의 자유도(derree of freedom, DOF)는 8이며, 행렬 H에서 i=1이라고 가정할 수 있다. 호스트 영상이 수평으로 투영된다고 추정하면 b=h=0이며, 또한 수직으로 투영된다고 추정하면 d=g=0이다. 수평 투영 왜곡의 경우, 왜곡된 좌표의 수학식을 다음의 수학식4와 같이 나타낸다.here,

(X, y) ^T denotes the coordinates of the original image, and (x ', y') ^T denotes the coordinates of the perspective projection distorted image. Although the given matrix H is composed of nine constants, it can be assumed that the matrix of freedom (DOF) is 8 and i = 1 in the matrix H. Assuming that the host image is projected horizontally, b = h = 0, and assuming that the host image is projected vertically, d = g = 0. In the case of horizontal projection distortion, the equation of the distorted coordinates is expressed as Equation 4 below.

Therefore, only the coordinates of four points need to be known, and any three sets of them are not collinear. Here, the interval and the angle of the correlation maximums extracted from the embedded watermark pattern are used to obtain the coordinates. This is calculated by the Local Auto-Correlation Function (LACF) of two local regions parallel to each other of the image instead of calculating the autocorrelation of the entire image. The local autocorrelation function is represented by Equation 5 below.

I (j, i) represents the brightness value of the pixels in row j column i on the video frame.

The correlation maximum extracted from the video frame shown in FIG. 2 by the auto-correlation function (ACF) according to the prior art is as shown in FIG. 3. Meanwhile, the video frame illustrated in FIG. 4 is distorted and watermarked by a horizontal projection of 10 degrees, and the correlation maximum extracted by the autocorrelation function (ACF) according to the prior art is shown in FIG. 5. It appears distorted.

That is, the geometric distortion estimation module 420 according to an embodiment of the present invention executes a local autocorrelation function (LACF) of two local regions parallel to each other, as shown in FIG. The geometric maximum is estimated by extracting all correlation maximums of the estimated watermarks from the distorted watermarking image as shown in FIG. 7.

In addition, the watermark reconstruction module 430 calculates a perspective projection matrix H, which is a mathematical model that extracts four distorted points by using angles and intervals of correlation maximum values extracted by the geometric distortion estimation module 420. And restores the watermark from the estimated geometric distortion. Here, the watermark reconstruction module 430 uses the geometry shown in FIG.

The watermark reconstruction module 430 according to the present embodiment will be described on the assumption that the watermark pattern is embedded in the center of the host image as shown in FIG. 8.

과

상에 발견되었다고 가정하면, E는

의 중심점이고, M은

와

의 연장선들에 대한 교차점이다. 여기서, K를

과

의 교차점으로서 표시하고, G를

와

의 교차점으로서 표시한다. 그 후, 두 개의 평행선

과

로부터

의 길이를 거리로서 획득한다.First, four corner points A, B, C and D are designated as the initial points. Then, the coordinates of the projection distorted points A, B ', C' and D are obtained. First, the correlation maximums are line segments from the results of the local autocorrelation function (LACF).

and

Suppose that is found in

Is the center point of, where M is

Wow

Is the intersection of the extension lines of. Where K

and

As the intersection of, and G

Wow

It is expressed as the intersection of. After that, two parallel lines

and

from

Obtain the length of as a distance.

와

는 다음의 수학식6과 같이 나타낸다.here,

Wow

Is expressed as in Equation 6 below.

,

,

는 선분

상의 상관관계 극대치들 사이의 간격을 나타내며, r은 기본 워터마크 패턴의 반복 횟수이고, J를

과

연장선의 교차점으로서 나타내며, F를

과

연장선의 교차점으로서 나타낸다.At this time,

Line segment

Represents the interval between correlation maxima of the phases, r is the number of repetitions of the basic watermark pattern, and

and

Represented as the intersection of the extension lines, F

and

It is shown as an intersection of extension lines.

의 길이를 알아야 한다.In this geometry, since the positions of A and D do not change, only the coordinates of B 'and C' need to be calculated. Therefore, from the similarity of the triangle

You should know the length of.

의 길이를 획득하기 위해서 삼각형 △HKM과 △IGM의 상사성을 이용한다. 또한 워터마크 h^w의 높이에 대한 영상 hⁱ의 높이의 비율과 동일한

에 대한

의 비율을 이용하여

의 길이를 획득한다.first,

We use the similarity of triangles ΔHKM and ΔIGM to obtain the length of. Also equal to the ratio of the height of the image h ^{i to} the height of the watermark h ^w .

For

Using the ratio of

Obtain the length of.

과

는 다음의 수학식7과 같이 나타낸다.to the next,

and

Is expressed as in Equation 7 below.

의 길이는

,

및 삼각형 △JKM과 △AEM의 상사성을 이용하여 획득되며 다음의 수학식8과 같이 나타낸다.to the next,

The length of

,

And obtained by using the similarity of triangles? JKM and? AEM.

=hⁱ/2이다. 수학식7 및 수학식8을 이용하여, 다음의 수학식9와 같이 정의된다.here,

= h ⁱ / 2. Using Equations 7 and 8, it is defined as Equation 9 below.

Here, w ⁱ represents the width of the image.

Finally, the coefficient matrix of the projection transformation can be obtained using the estimated pairs of four points and FIG. 8. In FIG. 8, four pairs of points are converted as in Equation 10 below.

Nine constants of the projection matrix H in equation (3) are obtained as shown in equation (11) by substituting the coordinates in equation (4).

The detection module 440 detects the payload embedded in the watermark restored by the watermark restoration module 430. Here, the payload is copyright information for the owner or a message for the buyer. If the message is represented as a string of letters or numbers, one letter or number can be converted to an ASCII code and eventually represented as 8 bits.

In the watermarking apparatus based on the local autocorrelation function according to an embodiment of the present invention, since the pattern into one frame represents two bits, the message we want to put in the video frame can be divided into two bits. In this case, the watermark pattern expressing two bits is to rotate the watermark pattern at any one of 0, 90, 180, and 270 degrees for 00, 01, 10, and 11, respectively, in four cases in which two bits may be present. By expressing the payload.

The detection module 440 will be described below.

The detection module 440 is represented as shown in Equation 12 below.

Where y _i represents the data to be tested.

At this time, it is determined whether the watermark is detected according to the detection determination value d (Detection measurement). Here, the threshold value is a threshold value for determining whether a watermark is detected in comparison with the detection determination value d. In this embodiment, the threshold value is a value derived by using an average and a variance of the signal. False positive error rate) is set to 10 ^-6 .

If the experimental data is an image without a pattern inserted, the detection determination value d approaches zero according to Equation 11 because the pattern has little relation with the pattern.

On the other hand, if the image is a pattern is inserted it can be represented as shown in Equation 13.

Where x _i represents the original pattern generated by the key.

At this time, assuming that there is no relation between the original image and the pattern, the correlation between the original image and the pattern approaches 0, but the correlation between the same patterns has a maximum value.

The controller 500 controls the watermark pattern generator 100, the inserter 200, the image inputter 300, and the detector 400.

Experimental results of the watermarking device based on the local autocorrelation function according to an embodiment of the present invention are as follows.

In the experiment on the local autocorrelation function based watermarking apparatus according to the present embodiment, a 40 bit watermark payload was embedded in three 5 minute HD size video clips. The average peak signal to noise rate (PSNR) of the three video clips is 44.28 dB. The experiment according to the present embodiment compares the performance of the method using the autocorrelation function (hereinafter referred to as ACF) according to the prior art and the performance of the method using the local autocorrelation function (hereinafter referred to as LACF) according to the present embodiment. As an experiment to test the robustness against distortion, a result of actually performing the DA / AD conversion to obtain a result from the geometric distortion caused by camcording is shown in Table 1 below.

In Table 1, each designated number of attacks indicates how far this watermarking scheme can survive. The extraction ratio refers to the ratio of the total number of frames to the number of watermark extracted frames. As shown in Table 1, the scheme using the ACF cannot extract the watermark when the projection distortion occurs, whereas the scheme using the LACF can extract the watermark when the projection distortion occurs. In all cases the LACF approach, the correlation was higher than the threshold of 0.11, leading to a false positive error rate of less than 10 ⁻⁶ .

As shown in Table 1, according to an embodiment of the present invention, the robustness against projection distortion as well as frame rate variation, format conversion, and D-A / A-D conversion for video applications is shown.

Meanwhile, a method of using a watermarking apparatus based on a local autocorrelation function (hereinafter, referred to as a 'watermarking method') according to an embodiment of the present invention will be described with reference to FIG. 9.

9 is a general flowchart of a watermarking method according to an embodiment of the present invention.

As shown in FIG. 9, the controller 500 generates a watermark pattern through the watermark pattern generator 100 (S2).

Next, the control unit 500 inserts the watermark pattern and the message by the watermark pattern generation unit 100 into the spatial domain of the video frame (S4). Here, the spatial domain is a pixel domain which has not been transformed into the original video frame. The spatial domain is a two-dimensional spatial domain in which the horizontal and horizontal axes of the original video frame are regarded as the y-axis.

Next, the control unit 500 receives the video frame inserted with the watermark pattern through the image input unit 300 from the image acquisition device causing the geometric distortion (S6).

The controller 500 restores the watermark inserted in the spatial domain through the detector 400 in the video frame input through the image input unit 300 and detects the payload (S8).

Also, a watermark pattern generation process according to an embodiment of the present invention will be described in detail with reference to FIG. 10.

10 is a detailed flowchart illustrating a watermark pattern generation process according to an embodiment of the present invention.

As shown in FIG. 10, the control unit 500 generates a watermark pattern using a secret key through the pattern generation module 110 (S22).

Next, the control unit 500 low-pass filters the watermark pattern generated by the pattern generation module 110 through the low pass filter module 120 (S24).

Next, the control unit 500 rotates the watermark pattern at any one of 0, 90, 180, and 270 degrees through the modulation module 130 to have the content of the message to be inserted into the watermark pattern. Modulation is performed to express the message (S26).

The control unit 500 causes the same watermark pattern to be repeated in the horizontal and vertical directions through the watermark pattern generator 100 (S28).

In addition, the watermark pattern and the message insertion process according to an embodiment of the present invention will be described in detail with reference to FIG. 11 as follows.

11 is a detailed flowchart illustrating a watermark pattern and a message insertion process according to an embodiment of the present invention.

As shown in FIG. 11, the control unit 500 uses the noise intensity function (NVF), which is a function of calculating a local characteristic of an image through the pattern intensity determining module 210, to generate a watermark pattern for each region of the image. Determine the strength of (S32).

In operation S32, the value of the noise visibility function (NVF), which is a function of calculating a local characteristic of an image, has a value between 0 and 1, and has a value of 1 in an area where there is no change of the image, In the region it is close to zero.

Next, the control unit 500 determines the change of the image through the perceptual scaling module 220 (S34), and in the case where there is no change of the image, lowers the intensity of the watermark pattern by the pattern intensity determining module 210. The watermark pattern is inserted into the spatial domain (S36).

Next, when the area of the image changes in step S34, the control unit 500 determines the intensity of the watermark pattern by the pattern intensity determining module 210 through the perceptual scaling module 220 (S36). The watermark pattern is inserted into the spatial domain by making it stronger than the mark pattern intensity (S38).

The controller 500 expresses the message to be inserted through the message insertion module 230 as a string of 0,1 bits and inserts the message into the spatial domain of the original video frame (S40).

In addition, the watermark detection process according to an embodiment of the present invention will be described in detail with reference to FIG. 12 as follows.

12 is a detailed flowchart illustrating a watermark detection process according to an embodiment of the present invention.

As shown in FIG. 12, the control unit 500 receives a video frame having a watermark inserted through the image input unit 300, and a watermark inserted from the video frame through the Wiener filter of the watermark estimation module 410. Estimate (S42).

Next, the control unit 500 extracts correlation maximums using the local autocorrelation function LACF on the watermark estimated by the watermark estimation module 410 through the geometric distortion estimation module 420 (S44).

Next, the control unit 500 reconstructs the watermark by calculating the perspective projection matrix H for the geometric distortion from the geometric distortion estimated by the geometric distortion estimation module 420 through the watermark reconstruction module 430 ( S46).

The control unit 500 detects the payload embedded in the watermark restored in step S46 through the detection module 440 (S48).

1 is a block diagram of a watermarking apparatus based on a local autocorrelation function according to an embodiment of the present invention.

2 is an exemplary view of a video frame embedded with a watermark according to an embodiment of the present invention.

3 is a diagram showing correlation maxima extracted by an autocorrelation function (ACF) according to the prior art.

4 illustrates a video frame distorted and watermarked by 10 degrees of horizontal projection according to an embodiment of the present invention.

FIG. 5 illustrates a correlation maximum extracted and distorted by an autocorrelation function (ACF) of a video frame distorted and watermarked by 10 degrees of horizontal projection.

FIG. 6 is an illustration showing the execution of the autocorrelation function (LACF) of local regions of a video frame distorted and watermarked by 10 degrees of horizontal projection.

7 is an exemplary diagram showing a correlation maximum of a watermark estimated by executing a local autocorrelation function according to an embodiment of the present invention.

8 is a view showing the geometry of a mathematical model according to an embodiment of the present invention.

9 is an overall flowchart of a watermarking method according to an embodiment of the present invention.

10 is a detailed flowchart illustrating a watermark pattern generation process according to an embodiment of the present invention.

11 is a detailed flowchart of a watermark pattern and message insertion process according to an embodiment of the present invention.

12 is a detailed flowchart of a watermark detection process according to an embodiment of the present invention.

<Detailed Description of Drawing>

100: watermark pattern generation unit 110: pattern generation module

120: low pass filter module 200: insertion portion

210: pattern strength determination module 220: perceptual scaling module

230: message insertion module 300: video input unit

400: detection unit 410: watermark estimation module

420: geometric distortion estimation module 430: watermark reconstruction module

440 detection module 500 control unit

Claims (14)

In the watermarking device based on local autocorrelation function, A watermark pattern generator for generating a watermark pattern and performing low pass filtering; An insertion unit which determines the intensity of the low pass filtered watermark pattern by the watermark pattern generator, and inserts the watermark pattern in the spatial domain of the video frame by adjusting the intensity of the watermark pattern according to the change of the image; An image input unit configured to receive a video frame including a watermark from an image acquisition device that causes geometric distortion; A detector for estimating a watermark inserted in the input video frame, estimating a geometric distortion using a local autocorrelation function on the estimated watermark, and reconstructing the watermark from the estimated geometric distortion; And A controller for controlling the watermark pattern generator, an inserter, an image inputter, and a detector; Local autocorrelation function-based watermarking apparatus comprising a. The method of claim 1, The watermark pattern generator, A pattern generation module for generating a watermark pattern using a secret key; And A low pass filter module for low pass filtering the watermark pattern generated from the pattern generation module; Local autocorrelation function-based watermarking apparatus comprising a. The method of claim 1, The watermark pattern is, Local autocorrelation function-based watermarking apparatus, characterized in that the same repeated in the horizontal and vertical direction. The method of claim 1, The watermark pattern generator, A modulation module configured to rotate the watermark pattern on which the low pass filtering is performed at any one angle of 0, 90, 180, and 270 degrees; Local autocorrelation function-based watermarking apparatus further comprising a. The method of claim 1, The insertion unit, A pattern intensity determining module configured to determine an intensity of a watermark pattern generated by the watermark pattern generator by using a noise visibility function, which is a function of calculating a local characteristic of an image; And When the watermark pattern is inserted into the spatial domain, the change of the image is judged, and the intensity of the watermark pattern is lowered and inserted into the region where there is no change of the image, and the intensity of the watermark pattern is changed to the region of the change of the image. Perceptual scaling module for inserting stronger than compared to the region; Local autocorrelation function-based watermarking apparatus comprising a. The method of claim 5, wherein The insertion unit, A message insertion module for inserting a message to be inserted into the video frame by expressing one message to be inserted into bit strings of 0 and 1 when a watermark pattern is inserted; Local autocorrelation function-based watermarking apparatus further comprising a. The method of claim 5, wherein The noise visibility function (NVF), A watermarking device based on a local autocorrelation function, having a value between 0 and 1, having a value of 1 in an area where there is no change in the image, and having a value close to 0 in an area where there is a change in the image. The method of claim 1, The detection unit, A watermark estimation module that receives a video frame having a watermark embedded therein through the image input unit and estimates the inserted watermark using a Wiener filter; A geometric distortion estimation module extracting correlation maxima using a local autocorrelation function on the watermark estimated by the watermark estimation module, and estimating geometric distortion from the extracted correlation maxima; And A watermark reconstruction module for reconstructing a watermark by calculating a perspective projection matrix (H) for geometric distortion of correlation maximum values extracted by the geometric distortion estimation module; Local autocorrelation function-based watermarking apparatus comprising a. The method of claim 1, The detection unit, A detection module detecting a payload embedded in the restored watermark; Local autocorrelation function-based watermarking apparatus further comprising a. (a) estimating a watermark pattern signal from a video frame including geometric distortion through a Wiener filter; (b) estimating geometric distortion from the watermark pattern signal using a local autocorrelation function; And (c) above Reconstructing the watermark by calculating a perspective projection matrix H for geometric distortion; Watermarking method comprising a. The method of claim 10, After step (c), (d) detecting a payload embedded in the watermark reconstructed in step (c); Watermarking method characterized in that it further comprises. The method of claim 10, The watermark embedded in the video frame of step (a) is (a-1) generating a watermark pattern based on a secret key; (a-2) low pass filtering the generated watermark pattern; (a-3) performing modulation to express a 2-bit message by rotating the watermark pattern on which the low pass filtering is performed at any one of 0, 90, 180, and 270 degrees; And (a-4) forming the watermark pattern to repeat the same in the horizontal and vertical directions; Watermarking method, characterized in that generated by. The method of claim 10, The watermark embedded in the video frame of step (a) is (a-5) determining the intensity of the watermark pattern for each region of the image using a noise visibility function (NVF); (a-6) determining a change in the image through the perceptual scaling module; (a-7) in the case where the image has no change as a result of the step (a-6), lowering the intensity of the watermark pattern and inserting the watermark pattern into the spatial domain; And (a-8) As a result of the determination in the step (a-6), in the case where the image is changed, the watermark is made stronger by the intensity of the watermark pattern than that of the watermark pattern of (a-7). Inserting the pattern into the spatial domain; Watermarking method, characterized in that inserted by. The method of claim 13, After the step (a-8), (a-9) expressing a message to be inserted through a message insertion module into 0 and 1 bit strings and inserting the message into a spatial domain of a video frame; Watermarking method characterized in that it further comprises.

Images