KR20190104671A - The content based clean cloud systems and method - Google Patents

Abstract

The present invention relates to a content-based clean cloud management system, and to a method thereof. In a cloud storage-based service system, by checking whether content stored in a mass storage is duplicated and removing duplicated files, the content-based clean cloud management system can improve the efficiency of a storage space and provide a high quality video service, thereby providing a suitable service for discriminating copyright in accordance with distributed search rules.

Description

The content based clean cloud systems and method

The present invention relates to a content-based clean cloud management system and method, which checks whether a content stored in a mass storage is duplicated and removes duplicate files, thereby improving storage space efficiency and providing high quality video service in a cloud storage based service system. The present invention relates to a content-based clean cloud management system and a method for providing a service suitable for determining copyright existence based on distributed search rules.

Increasingly, services that transmit, store, and search not only images but also large-capacity video content through the network are increasing. Accordingly, service providers who provide services related to video content store, maintain, and manage vast amounts of content. .

In addition, as the stored content increases exponentially, service providers need a method of efficiently storing and retrieving a large amount of content in the storage.Duplicate files are detected and the same according to their quality for efficient content management. It is necessary to remove the low quality content of the content and provide only high quality content to the user.

However, various methods for removing duplicate content have been tried, but efficiency is low in terms of accuracy and time. In particular, there is an urgent need for a technology for managing duplicate content of the same content according to quality.

In addition, the development of the communication environment and digital devices, the majority of movies, music, etc. represented by MP3, Divx, etc., which creates an explosive demand for digital content is illegally shared through the online service provider (OSP), and thus copyright Cases of infringement have increased significantly, and with the advent of the Web 2.0 era, users who have simply consumed digital content are being reborn as active producers through user created content (UCC). Copyright infringement is accelerating, such as editing existing works such as movies, dramas, and music, and producing them with UCC.

In order to overcome the above problems, various technologies have been developed, and the conventional technology as described above is Korean Patent Publication No. 10-2016-0113879 (image quality based duplicate file management system and method, hereinafter 'prior art' Is called).

The prior art includes a duplicate file detection unit that improves the accuracy of detecting duplicate content by using a fingerprinting technique and detects a duplicate video file among video files stored in the file storage using a fingerprint of the video file; An image quality measuring unit measuring the image quality of the detected moving image file in accordance with an image quality standard perceived by a human; And a duplicate file management system including a duplicate file manager to move or delete some of the detected moving image files to another storage based on the measured image quality to compare image quality between duplicate files. It is a redundant file management system and method based on video quality that deletes duplicate low quality content according to the quality result, thereby efficiently using storage space and providing only high quality content to a user.

However, the prior art cannot prevent the infringement of copyrights and technical protection measures for copyright protection are required. There is a difficulty in determining whether or not the copyright for a specific content.

Republic of Korea Patent Publication No. 10-2016-0113879 (Image Quality Based Duplicate File Management System and Method) Republic of Korea Patent Publication No. 10-2017-0063077 (media content identification method)

In order to solve the above problems, the present invention improves the accuracy of detecting duplicated contents using fingerprinting technique, compares the image quality between duplicate files, and deletes the duplicated low quality contents according to the image quality result, thereby saving storage space. The purpose of the present invention is to provide a content-based clean cloud management system and method based on image quality, which can be used as a user and provide only high quality content to a user.

In addition, by distributing the database for filtering into priority work nodes and general work nodes, the content-based clean cloud management can improve the efficiency of filtering according to the increase of content for filtering by filtering according to the priority of copyright retrieval of digital content. It is an object to provide a system and a method thereof.

In order to solve the above object, the present invention provides a duplicate file detection unit for detecting a duplicate video file of the video files stored in the file storage using the fingerprint of the video file; An image quality measuring unit configured to generate an image quality measurement standard and measure image quality of the detected video file according to the generated image quality measurement standard; A duplicate file manager to move or delete some of the detected video files to another storage based on the measured image quality; And a distributed filtering unit for performing filtering according to the copyright of the digital content.

The distributed filtering unit may include a search server, a priority work node module, a general work node module, and a filtering information database, wherein the search server is a database for the priority work node having a higher priority among the plurality of priority work node databases. After selectively filtering using, control is controlled to selectively filter using a database for a lower priority task node, and the filtered result is transmitted to a search server.

The duplicate file manager stores the fingerprint of the video file having the highest measured video quality and no copyright infringement in a fingerprint database, and excluding the video file having the highest measured video quality among the detected video files. It is to delete video files that have copyright infringement.

Acquiring an ultra-high definition video or image among the moving image files stored in the file storage (S400); Generating an image converted into a brightness component value having a fixed size based on the obtained video and image (S410); Dividing the converted video and the image into N blocks and extracting the converted video and the image into blocks including sharp outlines among the divided blocks (S420); Extracting a feature vector for measuring image quality for each of the selected blocks and calculating an average and a variance of the extracted feature vectors (S430); Generating average and variance of the calculated feature vectors as a reference for measuring image quality (S440); And storing the generated image quality measurement criteria and using the same to measure the image quality of the duplicate video file (S450). After measuring the image quality, it is checked whether filtering based on the existence of copyright is requested through a search server. Step (S460); Performing a sequential search for a work node when the request for filtering according to the copyright is confirmed (S470); Sequentially determining whether filtering is performed during the sequential search (S480); Determining whether or not a copyright exists in the corresponding content (S490); And if there is a copyright in the corresponding content, first performing filtering at the work node, and then transmitting the filtering result information to the search server (S500).

Performing a search for a general task node when there is no copyright in the corresponding content (S510); Analyzing the search result in the general work node to determine whether copyright exists in the corresponding content (S520); If there is a copyright in the corresponding content as a result of the determination, first performing filtering in the work node, and then transmitting the filtering result information to the search server (S500).

According to the present invention, it is possible to improve the efficiency of the storage operation by identifying and managing the duplicate video files in the mass storage by using fingerprinting technology, high quality video content according to the quality measurement results of the duplicate video content It is possible to improve the service quality by constructing the database alone, and the copyright protection technology can reduce the storage cost of the content-based large-capacity video content cloud service.

In addition, when the content to be filtered increases by first performing a process of determining and filtering selective filtering according to the content copyright by using a database for a plurality of priority work nodes generated through distribution rules according to the filtering target information. The filtering response speed can be improved by first blocking the search volume to general work nodes.

In addition, by providing a cloud-based solution can be applied to overseas video content distribution services.

1 is a block diagram showing the structure of a content-based clean cloud management system according to an embodiment of the present invention.
2 to 3 are diagrams illustrating a method of managing a content-based clean cloud management system according to an embodiment of the present invention.

Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings so that those skilled in the art may easily implement the present invention. As those skilled in the art would realize, the described embodiments may be modified in various different ways, all without departing from the spirit or scope of the present invention.

 Therefore, a content-based clean cloud management system and method thereof according to an embodiment of the present invention will be described.

1 is a diagram showing the structure of a content-based clean cloud management system and method according to an embodiment of the present invention.

2 to 3 are diagrams illustrating a method of managing a content-based clean cloud management system according to an embodiment of the present invention.

Referring to FIG. 1, the duplicate file management system 100 includes a duplicate file detector 110, an image quality measurer 120, a duplicate file manager 130, and a distributed filtering unit 140, and a file storage 200. Denotes a storage in which a video file is stored and the fingerprint DB 300 refers to a database in which a fingerprint of a video file stored in the file storage 200 is stored.

The duplicate file detector 110 detects whether a video file identical to the corresponding file exists for the video file stored in the file storage 200.

In this case, the duplicate file refers to a video file having the same meta information such as file size and name of the video content as well as the same file even if the meta information is different.

The duplicate file detector 110 includes a fingerprint extractor 111 and a duplicate file finder 112, and may be connected to the file storage 200 to detect a duplicate file or add a new file to the file storage 200. If it is confirmed that the detection of duplicate files may be performed.

The fingerprint extractor 111 extracts a fingerprint from multiple sections of the video file requiring the management of the duplicate file among the video files stored in the file storage 200 and transmits the fingerprint to the duplicate file finder 112.

The duplicate file finder 112 compares the fingerprint received from the fingerprint extractor 111 with the fingerprint stored in the fingerprint DB 300 to detect whether a duplicate video file exists in the file storage 200.

There is no limitation on the fingerprinting algorithm used to detect duplicate files, but an embodiment of the present invention uses a method of using an audio signal included in a video file as a fingerprint, and the method uses a video signal as a fingerprint. It offers superior advantages in processing speed and accuracy over the method.

The duplicate file finder 112 transmits the information about the duplicate video file to the image quality measuring unit 120 when a duplicate file is detected.

In this case, if a duplicate file is not found, information about the corresponding video file may be transmitted to the duplicate file manager 130.

 When the image quality measuring unit 120 receives information on the duplicate video file from the duplicate file detecting unit 110, the image quality measuring unit 120 measures the image quality of the duplicate files and outputs the measured result.

In detail, the image quality measuring unit 120 generates image quality and measurement criteria based on the video file stored in the file storage 200 and measures the quality of the video file according to the generated image quality measurement criteria.

In addition, the video file transmitted to the duplicate file manager 130 is transmitted to the distributed filtering unit 140 to determine whether the copyright infringement.

The distributed filtering unit 140 may include a search server 142, a work node module 144, a general work node module 146, a filtering information database 148, and the like.

The search server 142 selectively filters the database for the priority work node having a higher priority among the plurality of priority work node databases, and then uses the database for the priority work node having a lower priority. To selectively filter and forward the filtered results to the search server.

The filtering information database 148 includes file information, hash information, meta information, and feature information.

The distributed filtering unit 140 is a device for selecting filtering according to the copyright of digital content, and the search server 142 is based on the filtering information database 148. ) And a plurality of general work node databases (DB ') are distributed, and if a request for filtering based on copyright is requested from the search server 142, the work node module 144 and the general work node module are first applied according to a search rule. 146 controls to perform a process such as searching, filtering presence determination, copyright check, filtering, and the like, and accordingly, the priority work node module 144 performs priority work according to the database for a number of priority work nodes. Performs a sequential search on the node, and if there is a copyright by sequentially determining whether filtering is performed during the sequential search, filtering the corresponding content. After that, the result is transmitted to the search server 142.

In addition, if the general work node module 146 of the distributed filtering unit 140 does not determine whether copyright exists through a sequential search for the work node, the general work node module 146 may be assigned to the general work node according to the database for the plurality of general work nodes (DB ′). Search for, analyze the search results, and if the copyright is copyrighted and then filter the corresponding content, the result information is transmitted to the search server 142, if there is no copyright, the corresponding search result information is returned to the search server 142 Will be delivered.

The presence or absence of copyright transmitted to the search server 142 is a duplicate file manager 130 so that the files correspond to a clean file having no copyright infringement on the highest definition file among the files of the same contents stored in the file storage 200. The fingerprint of the file is stored and managed in the fingerprint DB 300.

The duplicate file manager 130 stores the fingerprint of the video file having the highest measured image quality and no copyright infringement in the fingerprint DB 300, and has the highest measured image quality among the detected video files. Except for video files, video files with copyright infringement are deleted.

Meanwhile, referring to FIGS. 2 to 3, the image quality measuring unit 120 acquires an ultra high definition video or an image (eg, a 1080p or 2K or higher resolution image) from a video file stored in the file storage 200 (S400). In operation S410, an image converted into brightness component values having a fixed size is generated for the acquired image.

In the exemplary embodiment of the present invention, 200 high quality images having a fixed size (1920 * 1080) are used, but the size and the number of used images can be changed.

The image quality measuring unit 120 divides the converted image into N blocks, and selects only the blocks including the sharp outlines among the divided blocks (S420).

In the process of selecting blocks including sharp outlines, first, a block having the highest variance of pixel values in each block in the converted image is searched for, and a predetermined level or more (eg, the highest) is determined based on the variance of the retrieved block. Blocks that are greater than or equal to 75% of the variance value are selected with blocks that contain clear outlines.

The feature vector for image quality measurement is extracted for each selected block, and the average and the variance of the extracted feature vectors are calculated (S430). In this case, the feature value used uses normalized contour information in units of image pixels.

The calculated mean and variance are generated as a reference for measuring image quality (S440).

The image quality measuring unit 120 stores the generated image quality measurement standard and utilizes the image quality measurement standard to measure the image quality of the duplicate video file (S450). The image quality measurement standard is stored at a predetermined time interval or the file storage 200. The new file may be updated depending on whether the number of uploads is greater than or equal to a preset number.

After measuring the image quality, in the standby mode of the filtering service system, the search server 142 of the distributed filtering unit 140 checks whether filtering based on the presence or absence of copyright is requested (S460).

As a result of confirming the filtering request according to the existence of copyright (S460), when filtering based on the existence of copyright is requested, the search server 142 transmits filtering target information (for example, file information, hash information, meta information, and characteristic information). ) Is transmitted to the priority work node module 144, and the priority work node unit 144 performs a sequential search on the priority work node according to a search rule using a plurality of databases for the priority work nodes (S470). After determining whether filtering is performed during the sequential search (step 480), it is checked whether or not the corresponding content is copyrighted (S490).

Here, the application of the search rule is applied to the priority according to the weight (e.g., a real value of 0 < weight < 1) in the database for a plurality of priority work nodes, for example DB 1, If DB2 and DBL are weighted and have a high priority, search, filter presence, and copyright check are performed using them, and then search, filtering, and copyright check are performed sequentially using the database for the remaining work node. It can be done in a way.

If the check result of the step is copyrighted in the content, the work node module 144 first performs filtering of the content and then transfers the filtering result information to the search server 142 (S500).

If the result of the check in the above step is that the content is not copyrighted (that is, it is not possible to determine whether there is a copyright) in the general work node module 146, the general work node according to the database (DB ') for a plurality of general work nodes. Search for (S510).

The general work node module 146 analyzes the search result to determine whether the copyright exists in the corresponding content (S520).

If the check result of the step is copyrighted in the content, the work node module 144 first performs filtering of the content and then transfers the filtering result information to the search server 142 (S500).

The general work node module 146 analyzes the search result and determines whether the copyright exists in the corresponding content (S520). If the copyright does not exist in the corresponding content when determining the existence of copyright, the check result of the step is If the content is not copyrighted (that is, if the copyright is not determined), the general work node module 146 performs a search for the general work node according to a database (DB ') for a plurality of general work nodes ( Returning to the step S510 is formed to be repeated.

The content type, release time, popularity ranking, filtering search ranking, and filtering target information are classified using the database for the plurality of priority work nodes, feature information, meta information, hash information, and filtering keyword information, and the plurality of general operations. The node database is preferably generated to include at least one of feature information, meta information, hash information, and filtering keyword information.

The filtering target information is formed to include file information, hash information, meta information, and feature information.

Therefore, after generating the filtering information database distributed among the database for the plurality of preferred work nodes and the database for the common work nodes, the filtering request using the database for the plurality of preferred work nodes is selected according to the filtering request, By performing filtering search using the database for work nodes, filtering can be efficiently performed through distributed search and data processing.

In the above description has been presented by presenting the various embodiments of the present invention, but the present invention is not necessarily limited thereto, and those skilled in the art to which the present invention pertains within the scope not departing from the technical spirit of the present invention It will be readily appreciated that various substitutions, modifications and variations are possible.

100: Duplicate File Management System
110: duplicate file detection unit
120: image quality measuring unit
130: duplicate file management unit
140: distributed filtering unit

Claims (6)

A duplicate file detection unit detecting a duplicate video file among the video files stored in the file storage using a fingerprint of the video file;
An image quality measuring unit configured to generate an image quality measurement standard and measure image quality of the detected video file according to the generated image quality measurement standard;
A duplicate file manager to move or delete some of the detected video files to another storage based on the measured image quality; And
Contents-based clean cloud management system comprising a; distributed filtering unit for selecting filtering according to the copyright of digital content
The method of claim 1,
The distributed filtering unit;
Includes priority node nodes, generic job node modules, and a database of filtering information.
The search server selectively filters the database for the priority work node having a higher priority among the plurality of priority work node databases, and then selectively uses the database for the priority work node having a lower priority. Content-based clean cloud management system, characterized in that the control to filter by the filtered and forwarded filtered results to the search server.
The method of claim 1,
The duplicate file management unit
Storing the fingerprint of the video file having the highest measured image quality and without copyright infringement in a fingerprint database,
Content-based clean cloud management system, characterized in that for deleting the video file having a copyright infringement, except for the video file having the highest measured video quality of the detected video file.
Acquiring an ultra-high definition video or image among the moving image files stored in the file storage (S400);
Generating an image converted into a brightness component value having a fixed size based on the obtained video and image (S410);
Dividing the converted video and the image into N blocks and extracting the converted video and the image into blocks including sharp outlines among the divided blocks (S420);
Extracting a feature vector for measuring image quality for each of the selected blocks and calculating an average and a variance of the extracted feature vectors (S430);
Generating average and variance of the calculated feature vectors as a reference for measuring image quality (S440);
And storing the generated image quality measurement criteria and using the same to measure the image quality of the duplicate video file (S450).
Determining whether filtering based on the presence or absence of copyright is requested through a search server after measuring the image quality (S460);
Performing a sequential search for a work node when the request for filtering according to the copyright is confirmed (S470);
Sequentially determining whether filtering is performed during the sequential search (S480);
Determining whether or not a copyright exists in the corresponding content (S490);
And if there is a copyright in the corresponding content, first performing filtering in a work node and then transmitting the filtering result information to a search server (S500).
The method of claim 4, wherein
Performing a search for a general task node when there is no copyright in the corresponding content (S510);
Analyzing the search result in the general work node to determine whether copyright exists in the corresponding content (S520);
If there is a copyright of the corresponding content as a result of the determination, first filtering the work node and then transmitting the filtering result information to the search server (S500). The content-based clean cloud may further include a. How to manage.
The method of claim 5,
The content-based clean cloud management method of claim 1, wherein if it is determined whether the copyright exists in the S520, if there is no copyright in the corresponding content, the process returns to S510.

Images