KR100383408B1 - Apparatus for verifying the contents based watermarking of multimedia data and method thereof - Google Patents

Abstract

The present invention relates to a multimedia watermarking authentication method based on the contents of multimedia data. In the content-based watermarking authentication method of the present invention, MPEG-7 metadata is extracted from the multimedia data concealing MPEG-7 metadata according to the MPEG-7 standard using an MPEG-7 metadata extraction algorithm, and the multimedia data is extracted from the multimedia data. After detecting watermark data using a watermarking extraction algorithm, the similarity is measured by comparing the extracted MPEG-7 metadata with the detected watermark, and the watermark authentication is judged based on the measured similarity. do.

According to another embodiment of the present invention, after extracting the watermark data and the PN code from the watermarking multimedia data in which the PN code converted from the MPEG-7 metadata is inserted into the multimedia data as the watermark, the extracted watermark and the PN code are extracted. A method of authenticating watermarking by measuring similarity between livers is disclosed.

Therefore, the present invention provides a watermark based on the content of data by using meta data based on MPEG-7 which is an international standard for content feature descriptors of multimedia data as a content-based watermark or by using a PN code converted therefrom. It is effective in determining whether the authentication of the watermarking is effective and suitable for recognition.

Description

Apparatus and method for watermarking authentication based on contents of multimedia data {APPARATUS FOR VERIFYING THE CONTENTS BASED WATERMARKING OF MULTIMEDIA DATA AND METHOD THEREOF}

The present invention relates to the concealment and authentication of watermarks in multimedia data. More particularly, the present invention relates to a multimedia watermark concealment based on the contents of multimedia data, and an authentication apparatus and method thereof.

With the development of electronic publishing, computer, communication, and multimedia technologies, various media such as documents, voice, photos, and video data have been digitized by electromagnetic devices, so that they can be efficiently stored, accessed, and used. Allowed to save and send. Therefore, the form of data is gradually changing from the analog form to the digital form, and further expanded with the appearance of the World Wide Web.

The digital method has difficulty in protecting intellectual property in virtual space because it is convenient to store or transform information. As the digital market has attracted attention, such as digital digitization, the development of multimedia, and the e-commerce due to the spread of the Internet, the replication of digital data has spread, and the ownership problems of various multimedia data have been effectively protected. It is required. However, the current solution for copyright protection and authentication of digital multimedia data has not yet been found to be acceptable.

One method that can be applied for the protection of digital multimedia data is digital watermarking as a method to prevent illegal manipulation of digital information and to guarantee its ownership. Digital watermarks provide a solution for providing copyright protection, data authentication, and ownership identification when legal issues arise by incorporating copyright information, distributor information, and user information into the data. can do.

The above-described digital watermarking technology inserts a watermark that is not recognized by the human eye regardless of its contents since the mid-1990s for registration of ownership of digital multimedia data. There are Cox and wavelet methods. In the Cox method, watermark data is inserted into a DCT region, whereas in the wavelet method, watermarking data is inserted into a predetermined band of a wavelet region. will be.

Meanwhile, watermarking based on the contents of data such as objects and colors of data has been actively studied. This method, called the next generation watermarking method, is a method of obtaining the most important feature of human perception in multimedia data and concealing the watermark around it. This content-based watermarking is suitable for human's cognitive characteristics by being strong in external attack as watermark survives as long as important part is recognized by concealing watermark in important part of data. Characteristics suitable for the content-based can be represented by the international standard descriptor and technical designer called MPEG-7. This patent relates to the use of the MPEG-7 standard descriptor for content-based watermarking.

In early 1998, MPEG-7, the international standard for efficient storage and retrieval of vast multimedia data, was formed and standardized. MPEG-7, unlike MPEG-1 / 2/4, which aims to compress audio visual data, deals with how to express the content of data rather than the data itself. Standardization in MPEG-7 relates to content representation for content-based retrieval of multimedia data such as audio, voice, video, and video. In MPEG-7, standardization is a descriptor that expresses characteristics or properties of data. And a description scheme consisting of a combination of these descriptors. This feature of MPEG-7 metadata is well suited as a perceptual recognition feature by expressing the contents of the original multimedia in small amounts of data in detail.

An object of the present invention is to provide a method for generating watermarked multimedia data using content-based features of multimedia data as data for authentication, and determining whether the watermarked multimedia data is authenticated.

According to an aspect of the present invention, there is provided an apparatus and method for watermarking authentication based on contents of multimedia data, comprising: means for extracting MPEG-7 metadata from contents-based watermarked multimedia data; Means for detecting watermark data from the multimedia data using a watermarking extraction algorithm; An authentication means for comparing the extracted MPEG-7 metadata with the detected watermark to measure similarity, and determining whether watermark is authenticated based on the measured similarity; Extracting MPEG-7 metadata from content-based watermarked multimedia data; Detecting watermark data from the multimedia data using a watermarking extraction algorithm; Comparing the extracted MPEG-7 metadata with the detected watermark data to measure similarity; And determining watermark authentication based on the measured similarity.

1 is a block diagram showing the configuration of a first embodiment for content-based watermarking authentication according to the present invention;

2 is a flowchart illustrating a content-based watermarking authentication method according to the present invention;

3 is a flowchart for explaining a process of generating watermarking multimedia data used in the first embodiment of the present invention;

4 is a block diagram showing the configuration of a second embodiment for content-based watermarking authentication according to the present invention;

5 is a flowchart illustrating a content-based watermarking authentication method according to the present invention;

6 is a flowchart for explaining a process of generating watermarking multimedia data used in the second embodiment of the present invention.

<Explanation of symbols for the main parts of the drawings>

100, 1000: data source 200, 2000: authentication module

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

1 is a block diagram for multimedia data authentication using MPEG-7 metadata according to a first embodiment of the present invention, an Internet environment or an intranet environment interworking through a data source 100, for example, a database or a network. An authentication module 200 for authenticating the watermarked multimedia data provided from the source 100 and the data source 100 including an is illustrated.

In the first embodiment of the present invention illustrated in FIG. 1, watermarked multimedia data to be authenticated is data generated through a process as described in FIG. 3.

First, if multimedia data is provided from a data source (not shown) (step 110), MPEG-7 metadata is extracted from the multimedia data (step 120). MPEG-7 metadata, based on the MPEG-7 standard, is an MPEG-7 standard descriptor that describes spatial relationships, colors, textures, shapes, object movements, sounds, widths, and spectra, and their descriptor schemes. It is treated as an important part of the perceptual characteristics of multimedia data.

Next, in step 130, the metadata of the descriptor and descriptor configuration of the extracted MPEG-7 standard is concealed in the original multimedia data by using a watermarking method, thereby generating watermarked multimedia data in which the watermark is hidden. As described above, a method for generating watermarked multimedia data by inserting MPEG-7 metadata as watermark data into multimedia data is described by the same applicant as of March 27, 2000 in the MPEG-7 standard in the multimedia search system. Patent Application No. 2000-15566, entitled "Method of Hidden and Detection of Metadata and Retrieval of Multimedia Data Using the Same".

As shown in FIG. 3, the multimedia data of which the descriptor and descriptor configuration of the MPEG-7 standard is watermarked is authenticated in the authentication module 200 through the data source 100 illustrated in FIG. 1.

Referring back to FIG. 1, the authentication module 200 includes a metadata extractor 202, a watermark extractor 204, and a similarity measurer 206. The metadata extractor 202 extracts MPEG-7 metadata from the watermark hidden multimedia data provided by the data source 100, and the watermark extractor 204 uses a watermark extraction technique from the watermarked multimedia data. To detect the watermark data. The similarity measurer 206 compares the similarity between the MPEG-7 metadata and the watermark data extracted by the metadata extractor 202 and the watermark extractor 204, and extracts the similarity if the similarity is less than a predetermined amount. If the watermark is authenticated and the similarity exceeds the range, it is judged as disapproval.

The operation of the authentication module 200 described above is described in more detail with reference to the flow chart of FIG. 2.

First, when multimedia data concealing MPEG-7 metadata is input from the data source 100 to the authentication module 200 (step 210), the metadata extraction unit 202 of the authentication module 200 receives the received multimedia data. Extract metadata of standard descriptor and descriptor configuration based on MPEG-7 standard from. The MPEG-7 metadata extracted by the authentication module 200 is provided to the similarity measurer 206 and may be expressed in the form of a vector having N components, as shown in Equation 1 below.

In Equation 1, Is a vector form of MPEG-7 metadata, whose composition is to be.

Next, the watermark extractor 204 of the authentication module 200 detects the hidden watermark from the multimedia data using the digital watermarking method (step 230). At this time, the watermark extractor 204 detects the watermark using a key, wherein the key used is data indicating ownership of the multimedia data. Uniquely provided. The watermark detected by the watermark extractor 204 may be provided to the similarity measurer 206 and may be expressed in a vector form having N components as shown in Equation 2 below.

In Equation 1, Is a vector form of a watermark whose composition is to be.

In step 240, the similarity measurer 206 of the authentication module 200 compares the similarity between the MPEG-7 metadata extracted in step 220 and the watermark data detected in step 230. . The similarity comparison by the similarity measurer 206 may be calculated as an inverse of the Eucladian distance between the extracted feature vector of the MPEG-7 metadata and the detected watermark vector as shown in Equation 3 below.

In Equation 3, s quantitatively represents the similarity in the inverse of the Euclidean distance. Equation 3 above is just an example of similarity measurement for explaining the present invention, and includes a method of measuring similarity between two general vectors, such as not only being limited to the Eucladian distance, but also measuring correlation coefficient.

Thereafter, the similarity measurer 206 disqualifies the extracted watermark if the similarity s obtained in the step 240 is smaller than the preset threshold T as shown in Equation 4 below (step 250). If it is larger than the set threshold T, it is determined as an acknowledgment (step 260). In other words, the high degree of similarity means that the data of the owner of the multimedia data is deliberately or unintentionally used by unauthorized persons, for example, the watermark inserted in the multimedia data by lossy compression such as JPEG or MPEG is damaged. Even if the multimedia data is his owner, and if the similarity is less than the threshold, then the multimedia data is not his owner.

In Equation 4 described above, T represents a similarity measurement threshold.

In the present invention, the similarity measurement method using the threshold as described above has described an example of the authentication decision, and may include a general clustering method without being limited to the threshold method.

FIG. 4 is a configuration for implementing an authentication method according to a second embodiment of the present invention, and utilizes multimedia features of MPEG-7 metadata without using MPEG-7 metadata directly for watermarking. Therefore, the configuration and operation of the first embodiment illustrated in FIG. 1 are substantially the same except that they are different from the watermarking multimedia data generated using the scheme illustrated in FIG. Is omitted.

First, a process of generating watermarking multimedia data in a second embodiment of the present invention will be described with reference to FIG. 6.

If multimedia data is provided from a data source (not shown) (step 1010), MPEG-7 standard metadata is extracted therefrom. As described with reference to FIG. 3, MPEG-7 metadata extracted from multimedia data is made up of descriptors and descriptor configurations, and is involved in watermark generation used in the second embodiment of the present invention.

That is, as in step 1020, the descriptor and descriptor configuration data of the extracted MPEG-7 metadata are converted into a unique key for proving ownership of the watermark, which is expressed by Equation 5 below.

In Equation 5, hash () is a mapping function as a hash function for generating a unique key for ownership of the watermark, Denotes an MPEG-7 and MPEG-7 descriptor configuration as shown in Equation 1, and an owner number represents information of a multimedia data owner.

Thereafter, in step 1030, a PN code having a pseudo random number is generated by the converted key using Equation 6 below.

In Equation 6, Random () is a pseudorandom generator for generating a unique random number for a unique key value for watermark ownership, and also includes a random noise generator of the spread spectrum method.

The PN code generated by using Equation 6 is generated as hidden watermarked multimedia data inserted into the original multimedia data by a general watermark embedding method as shown in Equation 7 below. Or it can be used in conjunction with the intranet environment.

In Equation 7, Represents the weight of the watermark.

Referring back to FIG. 4, the authentication module 2000 includes a metadata extractor 2010, a watermark generator 2020, a watermark extractor 2030, and a similarity measurer 2040. The metadata extracting unit 2010 extracts MPEG-7 metadata from watermarked multimedia data provided from the data source 1000. The watermark generation unit 2020 converts the owner number into a key for watermarking extraction along with descriptors and descriptors of the MPEG-7 metadata extracted by the metadata extraction unit 2010, and uses the converted key to generate water. A watermark, i.e., a PN code, of the marking multimedia data is generated. The PN code generated by the watermark generator 2020 is input to the similarity measurer 2040.

Meanwhile, the watermark extractor 2030 extracts a watermark from the watermarking multimedia data provided from the data source 1000 using a digital watermarking method. Substantially, the watermark extracted by the watermark extractor 2030 includes a PN code. The watermark extractor 2030 provides the extracted watermark to the similarity measurer 2040 for authentication.

The similarity measurer 2040 measures the correlation coefficient between the PN code and the watermark provided by the watermark generator 2020 and the watermark extractor 2030 as shown in Equation 8 below to determine whether the similarity is authenticated. Judge.

The operation of the authentication module 2000 described above is described in more detail with reference to the flowchart of FIG. 5.

First, the authentication process according to the second embodiment of the present invention starts from step 2100 in which watermarking multimedia data is input from the data source 1000 to the authentication module 2000.

The metadata extracting unit 2010 of the authentication module 2000 extracts MPEG-7 metadata from the watermarking multimedia data input from the data source 1000 (step 2200).

Next, the watermark generator 2020 converts the metadata extracted by the metadata extractor 2010 into a key (step 2300) and generates a PN code using the converted key (step 2400).

Meanwhile, since the watermark extractor 2030 knows the watermarking insertion position in advance, the watermark extractor 2030 may extract the hidden watermark data from the watermarking multimedia data input from the data source 1000 (step 2500).

Next, the similarity measurer 2040 compares the PN code and the watermark provided from the watermark generator 2020 and the watermark extractor 2030 with each other in the same manner as in Equation 8 to determine whether to authenticate. (Step 2600).

As a result of the comparison, the extracted watermark is authenticated when the correlation coefficient is larger than the preset threshold (step 2700). If the correlation coefficient is smaller than the preset threshold, it is determined as disapproval (step 2800).

Therefore, the present invention generates watermarked multimedia data using watermarks formed by PN codes or using meta data based on MPEG-7 which is an international standard of content feature descriptors of multimedia data. In addition, it is effective to determine whether the watermark is authenticated based on the contents of the watermarking multimedia data and to authenticate the watermarking suitable for recognition.

Claims (14)

In the watermarking authentication method based on multimedia data content, Extracting MPEG-7 metadata from content-based watermarked multimedia data; Detecting watermark data from the multimedia data using a watermarking extraction algorithm; Comparing the extracted MPEG-7 metadata with the detected watermark data to measure similarity; Determining whether the watermark is authenticated based on the measured similarity. The method of claim 1, The content-based watermarking multimedia data is a content-based watermarking authentication method, characterized in that the multimedia data is inserted as watermark data MPEG-7 metadata indicating the characteristics of the multimedia. The method of claim 1, The similarity is measured by the inverse of the Eucladian distance, such as [Equation] MPEG-7 metadata vector, : Watermark data vector, And if the measured similarity is greater than a predetermined reference threshold, authenticating the extracted watermarking, and if the measured similarity is smaller than the reference threshold, watermarking is not recognized. In the method for generating content-based watermarking multimedia data, Extracting MPEG-7 metadata representing characteristics of the multimedia from the content-based watermarking multimedia data; Converting the extracted MPEG-7 metadata into a key for proving ownership of a watermark; Generating a PN code which is a pseudo random number using the converted key; Generating the content-based watermarking multimedia data by inserting the generated PN code into the multimedia data as the watermark. In the watermarking authentication method based on the content of the multimedia data, Extracting MPEG-7 metadata from content-based watermarking multimedia data; Converting the extracted MPEG-7 metadata into a key for proof of ownership; Generating the PN code using the converted ownership proof key; Extracting watermark data from the watermarking multimedia data using a watermarking extraction algorithm; Comparing the extracted watermark with the PN code to measure similarity; Determining whether the watermark is authenticated based on the measured similarity. The method of claim 5, The content-based watermarking multimedia data converts the MPEG-7 metadata extracted from the multimedia data into a key for proving ownership of the watermark, and generates a pseudo-random number PN code using the converted key. Content-based watermarking authentication method characterized in that the data inserted into the PN code as a watermark. The method of claim 5, The similarity is measured by calculating a correlation between the PN code and the watermark as shown in the following equation, [Equation] And if the correlation coefficient is greater than a preset threshold, authenticate the extracted watermarking, and if the correlation coefficient is less than a reference threshold, watermarking is not recognized. The method of claim 5, In the key conversion step, the key is content-based watermarking authentication method, characterized in that generated using the following equation. [Equation] hash (): a hash function that generates a unique key for ownership of the watermark, : N components ( MPEG-7 metadata vector with Owner number: Information of the owner of the multimedia data. The method of claim 5, In the PN code generation step, the PN code, content-based watermarking authentication method characterized in that it is generated using the following equation. [Equation] Random (): Pseudorandom generation function that generates a unique random number for key values for watermark ownership, or pseudo random noise generation function using spread spectrum method. In the content-based watermarking authentication device, Means for extracting MPEG-7 metadata from content-based watermarked multimedia data; Means for detecting watermark data from the multimedia data using a watermarking extraction algorithm; And authentication means for comparing the extracted MPEG-7 metadata with the detected watermark to measure similarity, and determining whether watermark is authenticated based on the measured similarity. Authentication device. The method of claim 10, The authentication means measures the similarity by calculating the inverse of the Eucladian distance as in the following equation, [Equation] MPEG-7 metadata vector, : Watermark data vector, And if the similarity is greater than a predetermined reference threshold, authenticate the extracted watermarking, and if the similarity is less than a preset reference threshold, watermarking is not recognized. An apparatus for watermarking authentication of content-based watermarking multimedia data, Means for extracting MPEG-7 metadata from the watermarked multimedia data; Means for converting the extracted MPEG-7 metadata into a key for proof of ownership; Means for generating the PN code using the converted ownership proof key; Means for extracting watermark data from the watermarking multimedia data using a watermarking extraction algorithm; And an authentication means for comparing the extracted watermark with the PN code to measure similarity, and determining whether watermark is authenticated based on the measured similarity. The method of claim 12, The content-based watermarking multimedia data converts the MPEG-7 metadata extracted from the multimedia data into a key for proving ownership of the watermark, and generates a pseudo-random number PN code using the converted key. Content-based watermarking authentication device, characterized in that the data inserted into the PN code as a watermark. The method of claim 12, The authentication means measures the similarity by calculating a correlation between the PN code and the watermark as shown in the following equation, [Equation] And if the correlation coefficient is greater than a preset reference threshold, authenticate the extracted watermarking, and if the similarity is less than the preset reference threshold, watermarking is not recognized.