KR100768838B1 - Scalable digital fingerprinting for high-definition video streaming service - Google Patents

Abstract

The present invention relates to a digital fingerprinting method for copyright protection of digital content or to prevent illegal distribution or use of digital content. The present invention relates to a method for generating and generating a random sequence pattern by changing a key value for generating a pattern according to an inserted character. The present invention relates to scalable digital fingerprinting technology for high-definition video streaming service that is resistant to resizing attack and compression attack by applying the applied pattern to an image, and does not show image degradation.

In the scalable digital fingerprinting generation method for a high-definition video streaming service according to the present invention, a first step of generating a random sequence pattern by changing a key value for generating a pattern according to a character to be inserted, and the step 1 And inserting two steps of dithering the pattern, and inserting three steps of generating a fingerprinting pattern from the dithered pattern and the original image and inserting the pattern into the original image.

Fingerprinting, dithering, patterns, random sequences, keys.

Description

Scalable digital fingerprinting for high-definition video streaming service

1 is a state diagram of an averaging attack for attacking an image to which a conventional fingerprinting technique is applied.

2 is a state diagram of a fingerprinting insertion method according to the present invention.

3 is a state diagram for determining a key value for generating a pattern according to a message.

4 is a detailed state diagram of the expansion step (S119) of FIG.

FIG. 5 is a signal diagram of an extended pattern at the end of insertion 1 step S110 of FIG.

FIG. 6 is a signal diagram of a dithered pattern at the end of insertion 2 step S120 of FIG.

Figure 7 is a state diagram of the fingerprinting extraction method according to the present invention.

The present invention relates to a digital fingerprinting method for copyright protection of digital content or to prevent illegal distribution or use of digital content. The present invention relates to a method for generating and generating a random sequence pattern by changing a key value for generating a pattern according to an inserted character. The present invention relates to a scalable digital fingerprinting technology for high-definition video streaming service that is resistant to resizing attacks and compression attacks by applying the applied pattern to an image and exhibits no image degradation.

Digital fingerprinting is a technology that prevents illegal distribution or illegal use by inserting fingerprints, which are buyer-specific information, into the digital content such as text, images, video, and audio. . It is a technology that inserts a buyer's own fingerprint in the content sold to the buyer and extracts the fingerprint in the content in case of illegal distribution or illegal use of the content to find illegal distributors or illegal users and take legal responses or actions.

Digital fingerprinting has a disadvantage in that it is vulnerable to various attacks to remove the embedded fingerprint because the fingerprint information modulated in the form of noise is inserted into the original content.

In the conventional digital fingerprinting, since different user's fingerprints are put on the same original, each user has slightly different contents. Therefore, when two or more users compare content with each other, the difference can be easily identified, and the inserted fingerprint can be easily removed by using the difference between the two users. The averaging attack is one of the best ways to weaken the fingerprint inserted in the noise form in the content among the methods of removing the fingerprint through the content comparison.

An averaging attack is an averaging attack that can be most easily applied to an image or a video to which a fingerprinting technique is applied. As shown in FIG. 1, an averaging attack averages frames of two or more images or videos in which different finger prints are inserted. To create new images or frames. The averaging attack has the effect of reducing the fingerprint in proportion to the number of contents used in the attack. In general, when the averaging attack decreases the strength of the fingerprint signal inserted therein, there is a problem that the buyer's information is not extracted or incorrect information is extracted when the fingerprint is extracted.

The fingerprinting system, which is robust against averaging attacks, should be able to identify and inform the buyers who participated in the attack even for the contents in which the averaging attack occurred. To do this, the inserted fingerprint information must be robust to the averaging attack.

In designing the fingerprinting system, the factor to be considered along with the robustness against the averaging attack is that it must be robust against various attacks that can be applied to the image. Among them, size conversion attacks and compression are the most common attacks. An example of a size change attack is to convert a 1920x1080-sized high-definition video into a 640x480 VGA or 352x288 CIF video for viewing on various devices. Re-encoding operations, that is, compression, occur in reducing the size of an image. As the size of the image is reduced and compressed, there is a disadvantage in that the detection rate is usually lowered.

As a conventional method for preventing digital fingerprinting release due to a change in image size and compression, there is a method of inserting a fingerprint into DC regions of a DCT. In the method of inserting a fingerprint into DC regions of DCT, the fingerprint is inserted into a region that is robust to image size change and compression, but has a strong characteristic of image size change and compression, but has a problem of severe image quality deterioration.

The present invention has been made to solve the inconvenience of the prior art as described above,

An object of the present invention is to resist the change in image size by minimizing image degradation by minimizing image size by using dithering of pattern and Noise Visibility Function (NVF) filter at the same time. It is to provide a digital fingerprinting technology.

In order to achieve the above object, a scalable digital fingerprinting generation method for a high-definition video streaming service according to the present invention includes a first insertion step of generating a random sequence pattern by changing a key value for generating a pattern according to an inserted character, and And an inserting step 2 for dithering the pattern generated in the inserting step 1, and an inserting step 3 for generating a fingerprinting pattern from the dithered pattern and the original image in the inserting step 2 and inserting it into the original image. It features.

In the inserting step 1 of the method for generating scalable digital fingerprinting for a high-definition video streaming service according to the present invention, a message is inserted in a cyclically shifting pattern and a key value for generating a pattern is changed according to the inserted character. And a generating step of generating a random size sequence pattern having a size of 32 and an expanding step of expanding the generated pattern to the size of the image.

In the generating step of the scalable digital fingerprinting generation method for a high-definition video streaming service according to the present invention, a pattern is generated using one key, and an ascii difference value between the immediately preceding character and 'A' is generated in a key value for generating the pattern. Next, the next pattern is generated by using the value added as a key value of the pattern generation.

The pattern generated in the generation step of the scalable digital fingerprinting generation method for the high-definition video streaming service according to the present invention is characterized in that it is a 32x32 random number sequence having a Gaussian distribution having a mean of 0 and a variance of 1.

The expansion step of the scalable digital fingerprinting generation method for the high-definition video streaming service according to the present invention transverses the value of one pixel constituting the 32x32 size patterns generated by the number of messages to be inserted in the generation step. Expanded fingerprint pattern by creating 160x160 extended patterns by expanding 5 times horizontally and 5 times vertically by substituting 5 pixels vertically and 5 pixels vertically, and repeating the message in order to insert the generated extended patterns. It characterized in that to generate.

In the inserting step 3 of the method for generating scalable digital fingerprinting for a high-definition video streaming service according to the present invention, the inserting step 3-1 includes passing an NVF filter indicating a degree of visibility of the original image to the human visual structure; Inserting the fingerprint pattern completed in the inserting step 3-2 and the fingerprint pattern completed in the inserting step 3-2 by multiplying the resultant image generated in the inserting step 3-1 and the dithered pattern created in the inserting step 2 In addition to the original image is characterized in that it comprises a 3-3 insertion step of inserting a fingerprint pattern.

A scalable digital fingerprinting extraction method for a high quality video streaming service according to the present invention is a fingerprint extraction method of a scalable digital fingerprinting technology for a high quality video streaming service, wherein a fingerprint is inserted or an averaging attack is applied. Extracting step 1 for resizing and filtering to extract noise estimated as a fingerprint pattern, and extracting step 2 for obtaining a fingerprint pattern using correlations with the extracted pattern using the same pattern as the inserted pattern And extracting three steps of tracking legitimate users or conspirators participating in the averaging attack from the extracted fingerprint code.

The extracting first step of the scalable digital fingerprinting extraction method for a high-definition video streaming service according to the present invention comprises steps 1-1 of resizing content to 384x216 size and noise from the image reduced in step 1-1. A part that is supposed to be a fingerprint is extracted through a Wiener filter, which is a filter to be removed, and the extracted part is divided into a basic pattern size of 32x32, and the pattern of the number of fingerprint patterns inserted in the extracted content image is extracted. Steps 1 to 2 separated into one fingerprint pattern, and steps 1 to 3 to extract the first pattern of the fingerprint pattern separated in step 1-2 is characterized in that it comprises a.

In the extracting step 2 of the scalable digital fingerprinting extraction method for a high-definition video streaming service according to the present invention, the basic patterns created when the fingerprint code is inserted using a secret key are generated, and the patterns generated from the keys and the extraction are performed. The correlation between the estimated fingerprint patterns extracted in the first step may be calculated to determine that a pattern having a correlation value exceeding a predetermined threshold value among the correlation values is inserted in the image.

In the extracting step 3 of the scalable digital fingerprinting extraction method for the high-definition video streaming service according to the present invention, the message of the pattern extracted in the extracting step 2 and the key of the pattern extracted in the extracting step 2 Calculate the correlation between the patterns generated by the sum of ascii difference values and the pattern extracted in the next block. Extract the pattern, and calculate the correlation between the pattern extracted in the next block using the kit value and the message used in the extracted pattern. It is characterized by extracting the fingerprint pattern of the images used in the operation by repeating the process.

Hereinafter, a scalable digital fingerprinting technique for a high quality video streaming service according to an embodiment of the present invention will be described in detail.

Scalable digital fingerprinting technology for high-definition video streaming service extracts fingerprint patterns by filtering fingerprints and creating fingerprints on images and filtering images with fingerprints or averaging attacks. Fingerprinting extraction method for tracking legitimate users or conspirators who participated in the averaging attack.

2 is a state diagram of a fingerprinting insertion method according to the present invention.

In the fingerprinting insertion method, as shown in FIG. 2, an insertion step 1 (S110) for generating a random number sequence pattern of noise type, and an insertion step 2 (S120) for dithering the pattern generated in the insertion step 1 (S110). ), The insertion step 3 (S130) of generating a fingerprinting pattern from the dithered pattern and the original image in the insertion step 2 (S120) and inserting it into the original image.

Insertion step 1 (S110) is a step of inserting a message in a circular movement method of the pattern, and generating a random sequence pattern by varying the key value for generating the pattern according to the character to insert (S113) and extending the generated pattern It is configured to include an expansion step (S119).

In the generating step (S113), the process of inserting a message by the circular movement method of the pattern is a process of inserting a message according to how much the pattern generated by the key value is cyclically moved since the pattern has a different shape according to the key value. In an embodiment, the result pattern of moving the pattern by 10 horizontally and 5 vertically is defined as 'A', and the result pattern of moving the pattern by 20 horizontally and 10 vertically by inserting the message 'AB'.

The process of generating a random sequence pattern by changing a key value for generating a pattern according to the character to be inserted may be performed using a different key value according to the character inserted immediately when the pattern is generated according to the message to be inserted in the process of inserting the message. As will be described in detail through the embodiment of inserting the fingerprint 'CDEF' as shown in FIG.

)을 생성하고 다음 'D'에 해당하는 패턴은 앞문자가 'C'이므로 'C'에 해당하는 키값을 사용하여 'D'의 패턴 (

)을 생성하게 된다. 앞문자를 생성한 키값에 바로 앞 문자와 'A'와의 ascii 차이값을 더하여 그 값을 패턴 생성의 키값으로 삼는다. 앞문자가 'A'라면 100 + 0을, 'C'라면 100 + 2, 'Z'라면 100 + 25의 키로 'D'에 해당하는 패턴을 생성하게 된다. 즉, 예를 든 경우에서는 102를 키로 패턴을 생성한다. 그 다음 'E'의 패턴(

)을 만들기 위해서는 앞에서 키로 사용되었던 102에 앞문자가 'D'인 관계로 3을 더해서 만든 105를 키로 패턴을 생성하게 된다. 그리고, 그 다음 'F'의 패턴(

)을 만들기 위해서는 앞에서 키로 사용되었던 105에 앞문자가 'E'인 관계로 4를 더해서 만든 109를 키로 패턴을 생성하게 된다. 상기 과정을 통하여 생성되는 패턴(

,

,

,

)은 평균이 0이고 분산이 1인 가우시안 분포를 따르는 32x32 크기의 난수열이다. For example, suppose that the fingerprint 'CDEF' is inserted as shown in FIG. 3.

), And the pattern corresponding to the next 'D' is a letter of 'D' using the key value corresponding to 'C' because the preceding character is 'C'.

Will be generated. Add the ascii difference between the previous character and 'A' to the key value that generated the preceding character, and use that value as the key value for pattern generation. If the first letter is 'A', 100 + 0, 'C' is 100 + 2, and 'Z' is 100 + 25 to create a pattern corresponding to 'D'. That is, in the case of the example, a pattern is generated using 102 as a key. Then the pattern of "E" (

In order to create), a pattern is created using 105, the key created by adding 3 since 102 is the letter 'D'. Then, the pattern of "F"

In order to create), the pattern is created by using 109, which is the previous letter 'E', plus 4 since 105 is used as the key. Pattern generated through the above process (

,

,

,

) Is a 32x32 random sequence with a Gaussian distribution with mean 0 and variance 1.

,

,

,

)을 구성하는 픽셀1개의 값을 가로로 5픽셀, 세로로 5픽셀에 대입함으로써 가로로 5배, 세로로 5배 확장하여 160x160크기의 확장 패턴으로 생성한다(S116). Expansion step (S119) is to expand the pattern generated in the generation step (S111) to be robust to the image size change in the embodiment of the present invention, as shown in Figure 4 generated in the generation step (S111) 32x32 size pattern (

,

,

,

By substituting the value of one pixel constituting 5) into 5 pixels horizontally and 5 pixels vertically, it expands 5 times horizontally and 5 times vertically to generate an extension pattern of 160x160 size (S116).

Insertion step 1 (S110) generates an extended 1280x960 size fingerprint pattern by repeatedly generating the extended patterns generated as many kinds of messages to be inserted (S119).

), 2에 들어가는 패턴은 D에 해당하는 확장 패턴(

), 3에 들어가는 패턴은 E에 해당하는 확장 패턴(

), 4에 들어가는 패턴은 F에 해당하는 확장패턴(

)이 된다. In the embodiment of the present invention, the pinker print is generated by repetition of four types of expansion patterns of 1,2,3,4 as shown in FIG. 4, and the pattern entering the position of 1 is an expansion pattern corresponding to C (

), 2 goes into the expansion pattern corresponding to D (

), 3 goes into the expansion pattern corresponding to E (

), 4 goes into the expansion pattern corresponding to F (

)

Insertion step 2 (S120) is a step for preventing the blocking of the image appearing when inserting the expansion pattern made in the insertion step 1 (S110) to the original image shown in Figure 5 generated in the insertion step 1 (S110) A step in which a zigzag-shaped filter is applied to an extension pattern representing a signal as shown in FIG. 6 is converted into a dithered pattern as shown in FIG. 6.

Insertion step 3 (S130) is a step of inserting the completed pattern in the insertion step 2 (S120) into the spatial region of the original image, passing through the NVF filter showing the degree to which a 1920x1080 size original image is easily seen in the human visual structure. Insertion 3-2 to generate a fingerprint pattern by multiplying the resultant image generated in step 3-1 (S133) and the insertion image 3-1 (S133) and the dithered pattern created in step 2 (S120). In step S136 and the insertion of the fingerprint pattern completed in the insertion step 3-2 (S136) to the original image is inserted into the insertion step 3-3 (S139).

In the fingerprinting extraction method according to the present invention, as illustrated in FIG. 7, the fingerprint-inserted or averaged attack image is reduced to a basic pattern size and filtered to extract noise estimated as a fingerprint pattern. Extraction step 1 (S210), extraction step 2 (S220) of obtaining a fingerprint pattern using the extracted pattern and correlation using the same pattern as the inserted pattern, legitimate user or averaging attack from the extracted fingerprint code It comprises a three-step extraction (S230) for tracking the participating conspirators.

The first step of extracting (S210) is a step of extracting a portion of the content, which is assumed to be a fingerprint, to reduce the content so that the fingerprint pattern included in the content is 32x32, which is a basic pattern size. Steps 1-1 (S216) and steps 1-2 (S216) of extracting a portion that is assumed to be a fingerprint through a Wiener filter which is a filter for removing noise from the reduced image in step 1-1 (S213). In step S216, the fingerprint extracted is divided into basic pattern units, and a first step (S219) of extracting the divided first block is configured.

In step 1-1 (S213), if the size of the fingerprint-inserted content image does not change in size from 1920x1080, the image is resized to 1/5 in width and 1/5 in length (384x216), and the content image is inserted into the fingerprint. If the image is shrunk or expanded to any size other than 1920x1080 size, it is the step of resizing any size image to 384x216 size.

Step 1-1 (S213) is to reduce the 1280x960 size fingerprint pattern consisting of 160x160 expansion patterns 5 times horizontally, 5 times vertically extending the 32x32 size of the basic pattern 5 times horizontally, 5 times vertically To make a 256x192 size fingerprint pattern consisting of a 32x32 size base pattern, a 1920x1080 size image with a 1280x960 size fingerprint pattern inserted is reduced 5 times horizontally and 5 times vertically to resize to 384x216 size.

Step 1-2 (S216) extracts a portion that is assumed to be a fingerprint through a Wiener filter, which is a filter for removing noise from the image reduced in step 1-1 (S213), and extracts the extracted portion. A step of dividing the pattern by the number of fingerprint patterns inserted into the extracted content image into 32x32 basic pattern sizes into one fingerprint pattern.

In the embodiment of the present invention, since the number of the inserted fingerprint patterns is four, four patterns having a size of 32x32 are estimated.

Step 1-3 (S219) is a step of extracting the first pattern of the fingerprint pattern separated in step 1-2 (S216) for the correlation operation of the extraction step 2 (S220).

Since the fingerprint of the present invention exists in the form of noise in the image, it can be inferred that the noise filtered through the Wiener filter is the fingerprint inserted in the insertion process.

In addition, since the energy of the image itself is very large when extracting the fingerprint using the correlation between the image itself and the code of the extracting step 2 (S220), the extraction of the first step S210 may be performed to obtain a correlation due to the redundancy of the inserted fingerprint. Since it is not possible, the noise type fingerprint is extracted in advance through the Wiener filter.

)간의 상관관계를 이용하여 핑거프린트 패턴을 구하는 단 계로 상관관계는 다음의 수학식1과 같다.In the second step (S220), the basic patterns created when the fingerprint code is inserted using the secret key are generated, and the patterns generated from the keys (pattern 1st generated with a known key,..., (N- th)) and the first pattern of the estimated fingerprint patterns extracted in the first extraction step S210 (

). The fingerprint pattern is obtained by using the correlation between the correlations.

Zcc (X, X ') is a correlation coefficient value, and X' and X are the estimated fingerprints extracted from the image and folded into the pattern size, and the fingerprint codes generated from the key values, respectively. Are the average of X and X ', respectively.

In the second extraction step (S220), as shown in FIG. 7, a correlation between the estimated pattern and each of the patterns generated by the known key value is obtained, and a predetermined threshold is calculated from the correlation coefficients obtained from the patterns. It is a step of determining that a pattern having a correlation value exceeding a threshold is inserted in an image.

Extraction step 3 (S230) is a step of finding a legitimate user or conspirator who participated in the averaging attack of the images used in the operation by referring to the key offset from the pattern extracted in the extraction step 2 (S220).

Extraction step 3 (S230) will be described in detail through an embodiment in which two users (ABCD, CDEF) averaging attack.

)에서 추출한 패턴과 도3의 Key Offset을 참조하여 A에 대해서 100 + 0의 키 값으로 만든 패턴과 상관관계를 구해보면 B라는 메시지를 찾을 수 있다. 그리고 계속 반복을 하면 "ABCD"라는 핑거프린트 코드를 찾게 된다. 다음, C에 대하여 도3의 Key Offset을 참조하여 C에 대한 키 값 100 + 2로 만든 패턴과 상관관계를 구해보면 D라는 2번째 메시지를 찾을 수 있고 같은 방법으로 "CDEF"를 찾을 수 있게 된다. 이와 같은 메시지에 따라 다른 패턴 생성 키를 쓰는 방식을 사용하게 되면 여러 명이 평균화 공격(Collusion Attack)을 하였을 때도 여러 사용자를 다 찾아낼 수 있게 된다. If the correlation value is obtained in the second extraction step (S220) of the embodiment, the messages exceeding the threshold value in the first message are A, C values. First, the next pattern of the pattern extracted in step 1-3 (S219) for A (

Referring to the pattern extracted in Fig. 3) and the pattern made with a key value of 100 + 0 for A with reference to the Key Offset of Fig. 3, the message B can be found. And if you keep repeating, you will find a fingerprint code called "ABCD". Next, referring to the Key Offset of FIG. 3 for C and correlating the pattern made with the key value 100 + 2 for C, the second message D can be found and the "CDEF" can be found in the same way. . By using a different pattern generation key in response to such a message, multiple users can be found even when multiple attacks are performed.

Although the embodiments of the present invention have been described with reference to the accompanying drawings, those skilled in the art to which the present invention pertains may implement the present invention in other specific forms without changing the technical spirit or essential features thereof. You will understand that. Accordingly, the embodiments described above are exemplary in all respects and not restrictive.

In the digital fingerprinting method according to the present invention as described above, a random sequence pattern is generated by changing a key value for generating a pattern according to an inserted character, and the correlation between the patterns generated with the known key and the extracted pattern is detected. By calculating the relationship, even if a plurality of averaging attacks, the fingerprint can be detected.

In addition, the digital fingerprinting method according to the present invention has a robust effect on image size and compression by extending the generated pattern to the image.

In addition, the digital fingerprinting method according to the present invention has an effect that no image degradation occurs.

Claims (10)

A fingerprinting generation method of scalable digital fingerprinting technology for a high quality video streaming service, Insertion step 1 for generating a random sequence pattern by changing a key value for generating a pattern according to the character to insert; An insertion step 2 for dithering the pattern generated in the insertion step 1; And an inserting 3 step of generating a fingerprinting pattern from the dithered pattern and the original image in the inserting 2 step and inserting the pattern into the original image. The method of claim 1, Insertion step 1, A step of generating a random sequence pattern having a size of 32 * 32 by inserting a message in a circular movement method of a pattern and changing a key value for generating a pattern according to the inserted character; And a scaling step of expanding the generated pattern to the size of an image. The method of claim 2, In the generating step, the pattern is generated using one key, and the next pattern is generated by using a value obtained by adding the ascii difference value between the letter immediately preceding the letter and 'A' as the key value of the pattern generation. A scalable digital fingerprinting generation method for a high quality video streaming service. The method of claim 2, The pattern generated in the generating step is a scalable digital fingerprinting generation method for a high-definition video streaming service, characterized in that the random sequence of size 32x32 along the Gaussian distribution having a mean of 0 and a variance of 1. The method of claim 2, The expansion step is 5 times horizontally and 5 times vertically by substituting 5 pixels horizontally and 5 pixels vertically for the value of 1 pixel constituting 32x32 size patterns generated by the number of messages to be input in the generating step. Scalable digital fingerprinting for high-definition video streaming service, characterized by generating a extended pattern of 160x160 size by expanding 5 times, and generating an extended fingerprint pattern by repeatedly generating the extended pattern in order to insert the generated extended pattern. How to create. The method of claim 1, The insertion 3 step, Inserting 3-1 to pass the original image through the NVF filter showing the degree of visibility of the human visual structure; An insertion step 3-2 of generating a fingerprint pattern by multiplying the resultant image generated in the insertion step 3-1 and the dithered pattern created in the insertion step 2; And inserting a fingerprint pattern by adding the fingerprint pattern completed in step 3-2 to the original image. 3. A fingerprinting extraction method of scalable digital fingerprinting technology for a high quality video streaming service, An extraction step 1 of extracting noise estimated as a fingerprint pattern by resizing and filtering an image to which a fingerprint is inserted or subjected to an averaging attack, and An extraction step 2 of obtaining a fingerprint pattern by using a correlation with the extracted pattern using the same pattern as the inserted pattern; And extracting three steps of tracking a legitimate user or a conspirator who participated in an averaging attack from the extracted fingerprint code. 2. A scalable digital fingerprinting extraction method for a high-quality video streaming service. The method of claim 7, wherein The first step of extraction, Resizing the content to 384x216, steps 1-1, The image determined to be a fingerprint is extracted through a Wiener filter, which is a filter for removing noise from the reduced image in step 1-1, and the content image is divided and extracted into a basic pattern size of 32x32. Steps 1-2 for separating the pattern of the number of fingerprint patterns to be inserted into one fingerprint pattern, And extracting the first pattern among the fingerprint patterns separated in the step 1-2, the scalable digital fingerprinting extraction method for the high-definition video streaming service. The method of claim 7, wherein The extracting step 2 generates basic patterns created by inserting a fingerprint code using a secret key, calculates correlations between the patterns generated from the keys and the estimated fingerprint patterns extracted in the extracting step 1 A scalable digital fingerprinting extraction method for a high quality video streaming service, characterized in that it is determined that a pattern having a correlation value exceeding a predetermined threshold value is inserted into an image. The method of claim 7, wherein In the extracting step 3, the pattern generated as the sum of the key value used for pattern extraction in the extracting step 2 and the ascii difference value between the message of the pattern extracted in the extracting step 2 and the ascii difference value of 'A' and the pattern extracted in the next block It extracts the fingerprint pattern of the images used in the operation by repeating the process of calculating the correlation and extracting the pattern, and calculating the correlation between the pattern extracted in the next block using the kit value and the message used in the extracted pattern. A scalable digital fingerprinting extraction method for a high definition video streaming service.

Images