KR-20260062250-A - METHOD AND APPARATUS FOR IDENTIFYING SOFTWARE OF VIDEO SOURCE BASED ON METADATA EXTRACTION FROM MULTIMEDIA FILE

KR20260062250AKR 20260062250 AKR20260062250 AKR 20260062250AKR-20260062250-A

Abstract

A video source software identification device according to one embodiment can perform the operation of acquiring a data set including a video file transmitted by a predetermined software; the operation of extracting a value of metadata including a container file format, internal data of the container file, and encoding parameters of each video file; and the operation of creating a classification model based on a deep learning algorithm that learns a parameter reflecting the correlation between the value of metadata of each video and the software to which each video file was transmitted.

Inventors

박정흠
김준호
양효민

Assignees

고려대학교 산학협력단

Dates

Publication Date: 20260507
Application Date: 20241028

Claims (10)

In a method performed by a video source software identification device operated by a processor, The operation of acquiring a data set including a video file transmitted by a predetermined software; An operation to extract the value of metadata including the container file format, internal data of the container file, and encoding parameters of each video file; and A method comprising the operation of generating a classification model based on a deep learning algorithm that learns parameters reflecting the correlation between the metadata values of each video and the software to which each video file was transmitted. method.
In paragraph 1, The operation of extracting the above metadata is An operation to extract the value of metadata specifying the type and order of the top-level container file format as the container file format of each video file; An operation to extract the values of General metadata for the container file, Video metadata for the video stream, and Audio metadata for the audio stream as internal data of the container file for each video file; An operation to extract the value of metadata including SPS (Sequence parameter set) and PPS (Picture parameter set) as encoding parameters for each video file; and Includes the operation of labeling the values of metadata extracted by metadata and labeling a class that specifies the software to which the video was transmitted as the correct answer class. method.
In paragraph 2, The value of the above General metadata is Including internal data of ftyp, moov, and udta boxes among container file formats, method.
In paragraph 2, The value of the above Video metadata is Among container file formats, including trak with a component subtype value of 'vide' at the path moov/trak/mdia/hdlr, method.
In paragraph 2, The value of the above Audio metadata is Among container file formats, including trak with a component subtype value of 'soun' at the path moov/trak/mdia/hdlr, method.
In paragraph 2, The value of the metadata including the above SPS and PPS is, Among container file formats, including bit sequences within the avcC box of the moov/trak/mdia/hdlr path, method.
In paragraph 2, The above labeling operation is If the extracted value for the metadata contains a string, An operation to define a predetermined integer that specifies each word included in the above string; The operation of encoding each word into an integer index based on the integers defined above; An operation to reduce the dimensionality of the integer index of the above metadata based on the Principal Component Analysis (PCA) technique and convert it into a low-dimensional numeric sequence; and A process including the operation of labeling the lower-dimensional numeric sequence in the above metadata, method.
In paragraph 2, The above labeling operation is If the extracted value for the metadata does not contain a string, The operation of labeling the number sequence in the above metadata, method.
In paragraph 1, The above method is, The method further includes inputting a target video file into the above classification model to determine the source of the software from which the target video file was transmitted. method.
Memory containing instructions; and It includes a processor that performs a predetermined operation based on the above instructions, and The operation of the above processor is, The operation of acquiring a data set including a video file transmitted by a predetermined software; An operation to extract the value of metadata including the container file format, internal data of the container file, and encoding parameters of each video file; and A method comprising the operation of generating a classification model based on a deep learning algorithm that learns parameters reflecting the correlation between the metadata values of each video and the software to which each video file was transmitted. Video source software identification device.

Description

Method and apparatus for identifying video source software based on metadata extraction from multimedia file The present invention relates to a technology for extracting metadata within a multimedia file and identifying the software that transmitted the video file based thereon, and can be utilized in various application fields such as digital forensics, cybercrime investigation, copyright protection, and prevention of the distribution of illegally filmed material. Video files can be generated on various digital devices and transmitted and shared through various software. Accordingly, technology that identifies the software through which a video file was transmitted plays a crucial role in cybercrime investigations, resolving copyright infringement issues, and preventing the distribution of illegally filmed content, as it enables the identification of the file's source. With the recent popularization of smartphones and compact cameras, cases of illegal video filming and distribution have surged. Furthermore, as the development of internet technology has created an environment where high-capacity, high-definition videos can be easily shared, criminal acts resulting from the illegal distribution of videos are becoming increasingly frequent. As video generation media, generation methods, and transmission software become increasingly diverse, there is a need for technology to accurately identify the various software through which video files are transmitted to evolve alongside them. Previously, the primary method used to identify the software from which a video file was transmitted involved checking specific metadata. Metadata provides various technical information contained within the video file, allowing one to determine details such as the device that generated the file, the codec used, and the resolution. Furthermore, some software modifies metadata during the process of transmitting or processing video files, making it possible to estimate the source software to some extent based on these changes. However, existing video metadata analysis technologies have several limitations. First is the issue of metadata variability. Various software programs may re-encode metadata, delete certain information, or alter it during the video file transmission process, which can cause confusion in identifying the source of the video file. Additionally, video file metadata may use different formats depending on the software, and these differences can make it difficult to clearly identify the source of a specific software. However, existing metadata analysis methods rely on verifying the presence or absence of specific metadata to estimate the source, often failing to reflect the software's unique processing processes. Therefore, to address these issues, there is a need for a new approach that analyzes video file metadata in greater detail and reflects the differences in re-encoding processes or metadata processing methods across various software. FIG. 1 is a configuration diagram of a video source software identification device according to one embodiment. FIG. 2 is a flowchart of the operation performed by a video source software identification device according to one embodiment. FIG. 3 is an example of an ISOBMFF-based multimedia container according to one embodiment. FIG. 4 is a flowchart showing the operation of step S1020 according to one embodiment in detail. FIG. 5 is an exemplary diagram illustrating the operation of extracting metadata including a container file format according to one embodiment. FIG. 6 is an exemplary diagram illustrating the operation of extracting metadata containing internal data of a container file according to one embodiment. FIGS. 7 and FIGS. 8 are exemplary diagrams for explaining the operation of extracting metadata including encoding parameters according to one embodiment. FIG. 9 is an example diagram showing the types of metadata extracted according to the embodiments of FIG. 5 to 8. FIG. 10 is a flowchart showing the operation of step S1024 according to one embodiment in detail. FIG. 11 is an example diagram illustrating the concept of data conversion according to step S1024 in one embodiment. Detailed information regarding the purpose, technical configuration, and resulting effects of the present invention will be more clearly understood through the following detailed description based on the drawings attached to the specification of the present invention. An embodiment according to the present invention will be described in detail with reference to the attached drawings. The embodiments disclosed herein should not be interpreted or used to limit the scope of the invention. It is obvious to those skilled in the art that the description including the embodiments herein has various applications. Accordingly, any embodiments described in the detailed description of the invention are illustrative for better explaining the invention and are not intended to limit the scope of the invention t