KR102114136B1 - method of providing object sequence management for compressed video by use of blockchain of motion vectors - Google Patents

Abstract

The present invention relates to a technique for obtaining sequence information for objects present in general compressed videos such as H.264 AVC and H.265 HEVC videos and effectively managing the objects so that the sequence information cannot be forged or falsified. More specifically, the present invention relates to a technique which classifies and manages objects so that the same is difficult to forge or falsify as the objects become important by differently setting the difficulty of a blockchain according to the importance of the objects by adopting a method for identifying the objects in compressed videos generated by, for example, a CCTV camera or a vehicle black box, managing the same, and checking whether the sequence information for the objects is forged or falsified by managing, as the blockchain, motion vectors for the compressed videos without the need for identification experts to view a playback screen or use special identification information inserted into the compressed videos as in a conventional technique, so as to verify whether the sequence information for the objects has been forged or falsified. A method for managing an object sequence for compressed videos comprises a step of determining whether the object sequence has been forged or falsified on the basis of a comparison result between a first motion vector sequence and a second motion vector sequence.

Description

{Method of providing object sequence management for compressed video by use of blockchain of motion vectors}

The present invention generally relates to a technique for obtaining sequence information of objects existing in an image for compressed images such as H.264 AVC and H.265 HEVC and effectively managing them so that they are not tampered with.

More specifically, in the present invention, for example, in order to identify and manage objects in a compressed image generated by a CCTV camera or a vehicle black box, and verify that the sequence information of these objects has been forged, reviewing or compressing playback screens by the expert experts as in the prior art It adopts a method to check for forgery or alteration by managing the motion vector of a compressed image as a block chain without using the special identification information embedded in the image.It sets the difficulty of the blockchain differently according to the importance of the object, so that the forgery forgery becomes more important. It is about the technology to manage the classification so that it is more difficult.

Recently, it is common to establish a CCTV video control system to prevent crime or secure post-evidence. In the state where multiple CCTV cameras are installed in each region, the video generated by these CCTV cameras is displayed on a control monitor and stored in a storage device. When a control agent finds a scene in which a crime or accident occurs, it immediately responds appropriately, and searches for the video stored in the storage device to secure post-mortem evidence after the crime or accident occurs.

In addition, it is common to mount a black box on a vehicle to stand in an advantageous position in handling traffic accidents. When the vehicle is in operation, the camera continuously photographs the surrounding situation through the camera and stores it in the memory card. When a traffic accident occurs, the driver checks the image of the black box installed in his vehicle and accidents if it is advantageous to him. The image at the time of occurrence is used as evidence. In order to use these CCTV images or black box images as evidence, or conversely, to deny their evidence, it is necessary to be able to check whether the images have been forged.

In addition, a video or image sequence related to a specific person (eg, animation GIF, etc.) is separately edited and created and published on an Internet site (eg, YouTube). Among them, by deleting a part of the image or slightly changing the movement line of a specific person, it may produce a result that is completely different from the original image or a result that is similar to the original image but with significantly different nuances. In order to remove such content from the Internet, it is necessary to check whether the video is forged or not.

Using computer graphics software to delete one or two people from a video or move it a little is not that difficult. Through this, even though three people actually committed a crime, only one of them can be made a criminal by deleting one of them from the video. Alternatively, it is possible to cut a 3 second portion in the middle taken by a specific person from the image and encode the image newly. Alternately, it is also possible to fix it as if it was rubbed sideways despite being hit. This happens in the black box video as well. In the event of a traffic accident, the accident party obtains a video at the time of the accident from the black box mounted on his vehicle, cuts out the parts that are unfavorable to him, and submits only the advantageous parts to the insurance company or the police.

However, conventionally, it has been very difficult to discern whether an image has been forged. It has been common for sensational experts to repeatedly check the image to be inspected and check whether the screen is natural. In addition, by forcing special identification information such as a watermark into the image, it was determined whether the forgery has been forged.

However, this method is not suitable for images that are generated on a large scale, such as CCTV images or black box images, and which employs a lossy compression technique. Since expert discrimination is slow in discrimination, it is not only suitable for discriminating a small amount of images, but also has a problem that people with adverse results do not readily agree. The identification information method is difficult to apply in the field in which lossy compression technology is applied because the identification information embedded in the image should not be broken in the image processing process.

Therefore, there is a need for a technique for effectively identifying image forgery and alteration for compressed images such as CCTV images and black box images. In particular, it would be useful if a technique capable of discriminating whether or not image forgery occurred in individual object units when a plurality of objects are captured in the image is implemented.

An object of the present invention is to provide a technique for effectively obtaining sequence information of objects existing in an image for compressed images such as H.264 AVC and H.265 HEVC, and effectively preventing them from unauthorized forgery.

In particular, the object of the present invention is to identify and manage objects in a compressed image generated by, for example, a CCTV camera or a vehicle black box, and to verify whether the sequence information of these objects has been forged or altered, expert users look at the playback screen or view compressed images as in the prior art. It adopts a method to check for forgery or alteration by managing the motion vector of the compressed image as a block chain without using the special identification information inserted in the block, and by setting the difficulty of the blockchain differently according to the importance of the object, forgery is more important It is to provide the technology to manage the classification so that it is difficult.

In order to achieve the above object, a method for managing an object sequence of a compressed image using a motion vector block chain according to the present invention includes: a first step of receiving a first compressed image that is a target for managing forgery or alteration of an image; A second step of parsing the first compressed image and sequentially obtaining a plurality of motion vectors ('first motion vectors') over time; A third step of identifying one or more object unit areas, which are chunks of images to be managed for forgery or not, in the first compressed image, and identifying first object identification information for discriminating and identifying each of the object unit areas; In a preset block chain update cycle unit, a sequence of motion vectors ('motion vector object sequence') related to each object unit area among a plurality of first motion vectors is identified, and first object identification information for each one or more object unit areas and A fourth step of generating an object unit trajectory by combining motion vector object sequences, and generating block data by combining one or more object unit trajectories; A fifth step of updating the main blockchain by reflecting block data at each blockchain update cycle; A sixth step of receiving a second compressed image including an image sequence ('object sequence') of an object to be checked for forgery of the image; A seventh step of parsing the second compressed image and sequentially obtaining a plurality of motion vectors ('second motion vectors') over time; An eighth step of extracting second object identification information for classifying and identifying the object to be checked for forgery of the image from the second compressed image; A ninth step of obtaining a first motion vector sequence for an object sequence by identifying and linking a sequence of motion vectors related to second object identification information among a plurality of second motion vectors; Query the main blockchain to extract one or more block data having an object unit trajectory corresponding to the second object identification information, and analyze the motion vector object sequence of the object unit trajectory included in the extracted block data to analyze the object sequence. A tenth step of obtaining a second motion vector sequence for the operation; And an eleventh step of determining whether the object sequence is forged or not based on the result of comparing the first motion vector sequence and the second motion vector sequence.

At this time, the object sequence management method of the compressed image using the motion vector block chain according to the present invention is performed after the fifth step, the first object identification information of a plurality of object unit trajectories stored in the block data of the main blockchain A twelfth step of backing up and copying a plurality of backup blockchains having different difficulty levels corresponding to a group of objects to which they belong; When the second object identification information, which is performed between the ninth step and the tenth step, is backed up and copied to the backup blockchain, the backup blockchain corresponding to the object group to which the second object identification information belongs among the plurality of backup blockchains Second motion for the object sequence by searching and extracting one or more block data having the object unit trajectory corresponding to the second object identification information, and analyzing the motion vector object sequence of the object unit trajectory included in the extracted block data And a 13th step of acquiring a vector sequence and proceeding to the 11th step.

At this time, the twelfth step may include setting up a plurality of backup blockchains having different difficulty levels; Setting a plurality of object groups corresponding to a plurality of backup blockchains; In the main blockchain, each unit trajectory stored in each block data is selected as a corresponding block data by selecting one backup blockchain among a plurality of backup blockchains in correspondence with the object group to which the first object identification information belongs. Copying a backup; And configuring block data according to respective difficulty levels of a plurality of backup blockchains.

In this case, the object unit area may be set as one or more moving object images extracted from the target compressed image through decoding, downscale resizing, and image analysis, in which the data of the first compressed image or the second compressed image is each target compressed image. have.

In addition, the object unit region may be identified by a syntax-based moving object region extraction process using a first compressed image or a second compressed image as each target compressed image. At this time, the syntax-based moving object region extraction process includes: a step A of parsing a bitstream of a target compressed image to obtain a motion vector and a coding type for a coding unit; A step B of obtaining a motion vector accumulation value for a preset time for each of a plurality of image blocks constituting a target compressed image; A step C of comparing a motion vector accumulation value with a preset first threshold value for a plurality of image blocks; A step D of marking an image block having a motion vector accumulation value exceeding a first threshold as a moving object region; And an E step of setting an individual chunk of an image block marked as a moving object area as an object unit area.

At this time, the syntax-based moving object region extraction process includes: a step of identifying a plurality of adjacent image blocks ('neighbor blocks') centered on the moving object region, which is performed between steps D and E; A step b of comparing a motion vector value obtained in the first step with a second preset threshold value for a plurality of neighboring blocks; A step c of additionally marking a neighboring block having a motion vector value exceeding a second threshold as a moving object area as a result of the comparison of step b among the plurality of neighboring blocks; A d step of additionally marking a neighboring block having an intra-picture coding type as a moving object area among a plurality of neighboring blocks; And performing an interpolation on a plurality of moving object areas to additionally mark a predetermined number of non-marked image blocks surrounded by the moving object areas as moving object areas.

In addition, the syntax-based moving object region extraction process includes: a step of identifying a plurality of adjacent image blocks (“neighbor blocks”) around the moving object region; A step b of comparing a motion vector accumulation value with respect to a plurality of neighboring blocks to a second threshold that is preset to a value less than the first threshold; A step c of additionally marking a neighboring block having a motion vector accumulation value exceeding a second threshold as a moving object area as a result of the comparison of step b among the plurality of neighboring blocks; A d step of additionally marking a neighboring block having an intra-picture coding type as a moving object area among a plurality of neighboring blocks; And performing an interpolation on a plurality of moving object areas to additionally mark a predetermined number of non-marked image blocks surrounded by the moving object areas as moving object areas.

On the other hand, the computer program according to the present invention is stored in a medium in order to execute the object sequence management method of the compressed image using the motion vector block chain as described above in combination with hardware.

According to the present invention, it is possible to clearly verify whether a video is forged or altered in units of individual objects in a CCTV video or a black box video, thereby enhancing the reliability of video evidence as post-evidence such as crime investigation or traffic accident analysis. In particular, since the process is processed based on the syntax information of the compressed image, there is an advantage in that the computation speed is very fast since no complicated image processing is required.

In addition, according to the present invention, by selecting a part of a plurality of objects existing in the image and managing the difficulty of the blockchain by increasing the difficulty, forgery of the images for these can be set more difficult, and thus the reliability of the main object is relatively high. It has the advantage of being manageable. Conversely, by managing the number of objects that are not specially managed by lowering the difficulty level of the blockchain, there is an advantage of improving the operational efficiency of the blockchain while maintaining the reliability of the entire system.

1 is a diagram conceptually showing an object sequence management system of a compressed image using a motion vector blockchain according to the present invention.
2 is a flow chart showing the entire process of a method for managing an object sequence of a compressed image using a motion vector blockchain according to the present invention.
3 is a diagram showing an example of a motion vector obtained in each video frame over time in a CCTV image.
4 is a diagram showing the concept of generating a motion vector object sequence, object unit trajectory, and block data in the present invention.
5 is a diagram showing the concept of differently setting the difficulty of the backup blockchain for each object group when backing up the main blockchain in the present invention.
6 is a diagram showing the concept of differently setting the difficulty level of the blockchain.
7 is a flowchart conceptually showing a first embodiment of setting an object unit area for a compressed image in the present invention.
8 is a block diagram showing a general configuration of a video decoding apparatus.
9 is a flowchart illustrating a second embodiment of setting an object unit area for a compressed image in the present invention.
10 is a flowchart illustrating an example of an implementation of a process for detecting effective motion from a compressed image in the present invention.
11 is a diagram showing an example of a result of applying an effective motion area detection process according to the present invention to a CCTV image.
12 is a flowchart illustrating an example of implementation of a process for detecting a boundary area for a moving object area in the present invention.
13 is a view showing an example of a result of applying the boundary area detection process according to the present invention to the CCTV video image of [11].
14 is a diagram showing an example of a result of arranging a moving object area through interpolation with respect to the CCTV video image of [FIG. 13].

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

1 is a diagram showing an object sequence management system of a compressed image using a motion vector blockchain according to the present invention, and [FIG. 2] is a flowchart showing the entire process of the object image sequence management method according to the present invention.

Referring to FIG. 1, the object image sequence management system for a compressed image according to the present invention is a compressed image forgery prevention processing subsystem of the image management server 100, a blockchain subsystem, an object of the image verification server 300 And a sequence verification processing subsystem. In this case, the image management server 200 and the image verification server 300 may be the same device or separate devices depending on the implementation.

First, the image management server 200 receives captured images from a plurality of CCTV cameras 210 and stores them in the image storage 220. In addition, the image management server 200 utilizes a blockchain so that it can later check for forgery or alteration of the sequence of individual objects included in the CCTV image. If the main block chain puts information on the movement trajectories of individual objects in a CCTV image, and later, if it is a matter of authenticity about a series of image images containing a series of actions of the object for a specific object, It is to find information related to the movement trajectory and use it to confirm forgery and alteration. In this specification, such an image image is referred to as an 'object sequence', and may be in the form of a compressed file or a series of image files packaged with additional data (related motion vectors).

At this time, a method of storing the entire CCTV image in the image storage 220 and then verifying it on the display screen while reproducing the original CCTV image for a specific object sequence in which authenticity is a problem may be considered. However, it is a difficult approach to automate identification of authenticity through computer processing, as well as time-consuming because it is necessary to find the original CCTV video in the large-scale video storage 220 and look at the content while playing the CCTV video. If it is necessary to perform image analysis in order to automate through computer processing, it is difficult in reality.

In addition, it is possible to think of a method in which the entire image data is put in the blockchain or the object sequence is put in the blockchain, but in this case, it is difficult in reality because the block data is enlarged and the operation efficiency of the blockchain is reduced.

In consideration of these matters, the present invention adopts a method of managing motion vectors with a blockchain. Since the motion vector has a very small amount of data, the size of the block data can be reduced, which can increase the operational efficiency of the blockchain. In addition, since the motion vector is information obtained by simply parsing data without decoding the compressed image, the data processing speed of the image management server 200 can be increased. In addition, since the motion vector has a format of digital data, it is suitable for automation through computer processing if only critical conditions are properly set.

In the present specification, an image in which the image management server 200 performs object sequence management is displayed as a first compressed image. In the implementation example in which the image management server 200 receives a plurality of compressed images through a plurality of channels, 'image meta information' which is information for distinguishing them from each other is extracted from the first compressed image. For example, a combination of device unique ID and recording time information of the CCTV camera 210 that generated the corresponding video may be set as video meta information.

Also, the image management server 200 obtains a motion vector (motion vector) from the first compressed image. In this specification, a motion vector obtained from a first compressed image is called a first motion vector. In general, as time passes, a plurality of first motion vectors are obtained from the first compressed image as shown in FIG. 3.

Subsequently, the image management server 200 identifies a unit image chunk for handling a motion vector for the purpose of managing forgery or alteration of the first compressed image, and is referred to as an 'object unit region' in this specification. The object unit area is regarded as one object in the present invention, and the concept of obtaining an object unit area from a compressed image will be described later with reference to [FIG. 7] to [FIG. 9]. Generally, a plurality of object unit areas are identified from the first compressed image. In addition, the image management server 200 extracts 'object identification information', which is information for classifying and identifying object unit areas. For example, in the process of extracting an object unit region from a compressed image, object identification information allocated in software may be set as object identification information.

In the above, the image management server 200 obtained a plurality of first motion vectors and a plurality of object unit areas from the first compressed image. Subsequently, the image management server 200 identifies motion vectors related to each object unit region among a plurality of first motion vectors, and sorts them by object unit region and arranges them in chronological order. The sequence of motion vectors sorted for each object unit region is referred to as a 'motion vector object sequence' in this specification. Then, the object unit trajectory is obtained by combining the object identification information and the motion vector object sequence for each object unit area.

For example, it is assumed that the image management server 200 is connected to 1000 CCTV cameras 210 and the blockchain update cycle is 30 seconds. The image management server 200 receives 1000 CCTV images of 30 seconds each from the CCTV camera 210 every block chain update cycle, which corresponds to the first 1000 compressed images. Assuming, for example, that 250 first motion vectors and 7 object unit areas are obtained from one CCTV image (30 seconds), one object identification information will be assigned to each of these 7 object unit areas, and these 250 products 7 Motion vector object sequences are obtained by sorting 1 motion vector based on the corresponding object unit area. By combining object identification information and motion vector object sequences for each object unit area, 7 object unit trajectories are obtained from a corresponding CCTV image. Since 1000 CCTV images are provided from the CCTV camera 210, if an average of 7 object unit trajectories is obtained from one CCTV image, the image management server 200 acquires 7000 object unit trajectories for each blockchain update cycle.

The image management server 200 includes these object unit trajectories to generate block data and updates the main blockchain by reflecting the block data according to a preset schedule. In this embodiment, each CCTV image is modeled as a first compressed image. For example, it is also possible to model all 1000 CCTV images as a first compressed image. Through the above process, the operation performed by the image management server 200 to manage the object sequence of the first compressed image has been completed.

Block chains generally have the concept of an open digital transaction ledger in which transaction information generated on a network is replicated and stored as many nodes as shared among network participants. Blockchain technology is currently actively applied around cryptocurrencies. In the present invention, the blockchain technology is used to manage the object sequence of compressed images, and for this purpose, the image management server 200 and the image verification server 300 are used. As a peer, he participates in the digital ledger management of the blockchain.

In this specification, the term motion vector distributed ledger 100 is used to indicate a distributed ledger managed by a blockchain for object sequence management. That is, the motion vector distributed ledger 100 is a digital transaction ledger for a blockchain, and in the present invention, it contains information necessary for object sequence management of compressed images.

The motion vector distributed ledger 100 is a form in which a plurality of transaction data are configured as blocks and connected to each other. One block is peers that link to the blockchain in the corresponding blockchain update cycle (eg, 30 seconds). That is, one or more image management servers 200 correspond to newly generated block data. Referring to FIG. 1, such blocks may include image meta information and object unit trajectories. Due to the nature of the blockchain, which is based on a distributed network, data reflected in the motion vector distributed ledger 100 can be trusted because it is virtually impossible to modify it without permission.

In the present invention, the blockchain can be configured by dualizing it into a main blockchain and a backup blockchain. [Figure 1] shows the main blockchain, and [Figure 5] shows the relationship between the main blockchain and the backup blockchain. Did.

The main blockchain is a blockchain for managing information obtained by processing the first compressed image data, and the backup blockchain is a blockchain that backs up and copies the information stored in the main blockchain for long-term storage. As described later with reference to FIGS. 5 and 6, after preparing a plurality of backup blockchains, it is possible to set and store the difficulty of the backup blockchain differently for each object group. For example, objects with preset characteristics (e.g., objects with a moving speed of 3 to 5 meters per second, objects with 3 or more objects moving together) are managed by a backup blockchain with a high difficulty level, which further improves forgery and forgery. Difficult to manage. Conversely, it is possible to lower the overall burden of blockchain processing by managing the backup blockchain with low difficulty for most objects that have no special features.

Next, the image verification server 300 is requested from the administrator computer 400 to confirm whether the object sequence included in the specific image is forged or how reliable (eg, how much the reliability index is). In this specification, the image including the object sequence that the image verification server 300 needs to check for forgery or alteration is referred to as a second compressed image. The specific object to be checked may be in the form of a compressed image that appears on the playback screen, or a motion vector obtained corresponding to the corresponding image sequence from the image sequence of the specific object and the original image may be packaged. The second compressed image is an image in which it is necessary to check for forgery or alteration, for example, for verifying integrity as a crime evidence, among object sequences included in the image previously processed by the image management server 200. At this time, the length of the object sequence to be checked may be longer or shorter than the blockchain update cycle.

In order to evaluate the reliability of the object sequence, the image verification server 300 acquires motion vector sequences in two paths for the second compressed image and compares them to see how much correspondence is there. Conceptually, the degree of correspondence between two motion vector sequences corresponds to the reliability of the object sequence. At this time, if the motion vector sequence completely corresponds, it can be clearly determined that the corresponding object sequence has not been forged. In addition, if the motion vector sequence corresponds to, for example, 90%, the reliability of the corresponding object sequence can be evaluated as 90%. Preferably, when the reliability is higher than a certain threshold (for example, 97%), it may be determined that the object sequence has not been forged.

First, the first acquisition path will be described.

Conceptually, the first acquisition path is to obtain a sequence of motion vectors from the second compressed image by similarly applying the procedure performed by the image management server 200 to the first compressed image for the second compressed image.

That is, the image verification server 300 parses the second compressed image and extracts 'image meta information', which is information for distinguishing the second compressed image, and extracts a plurality of motion vectors from the second compressed image, that is, the second motion vector. To acquire. Subsequently, information for identifying the object sequence to be checked is extracted from the second compressed image. In general, the object sequence may be composed of object identification information and an object image sequence. In this step, object identification information (ie, second object identification information) of the object sequence is useful. At this time, the image meta information and the object identification information acquired for the second compressed image are preferably utilized when querying the main blockchain in the second acquisition path.

The image verification server 300 identifies a sequence of motion vectors related to the second object identification information among a plurality of second motion vectors, and connects them to obtain a first motion vector sequence for the object sequence to be identified. If the object image sequence is included in the original image in relation to the object, the image management server 200 previously put the object identification information and the first motion vector sequence of the object sequence to be checked into the main blockchain. Because it will be, the image verification server 300 will be able to obtain from the main blockchain.

Next, the second acquisition path will be described.

The image verification server 300 searches the main blockchain to extract one or more block data corresponding to the second object identification information. If there is no forgery and alteration, the second object identification information has been previously processed by the image management server 200 to process the object sequence, so the object unit trajectory for it must exist somewhere in the main blockchain. Therefore, the image verification server 300 extracts one or more block data corresponding to the second object identification information by, for example, analyzing the motion vector distribution ledger 100 managed by the user. At this time, image meta information may be referenced. For example, if the combination of the device identification information and the shooting time information of the CCTV camera 210 that generated the video, the second compressed video can be discriminated from among a plurality of video images, thereby enabling the search scope of the main blockchain. It can be narrowed to improve the operation speed.

Then, a second motion vector sequence for the object sequence is obtained from the extracted block data. In the block data, motion vector information, that is, object unit trajectories, previously organized by the image management server 200 for each object unit area is stored from the original image of the object sequence. Accordingly, the image verification server 300 may obtain the object unit trajectory corresponding to the second object identification information from the storage contents of one or more block data, and analyze the motion vector object sequence stored in the object unit trajectory to the object sequence. The second motion vector sequence for is obtained.

Then, the image verification server 300 compares the first motion vector sequence and the second motion vector sequence and determines whether the object sequence is forged or how reliable it is based on the result. Conceptually, the sequence of the motion vector extracted by the image verification server 300 from the second compressed image provided as the object of checking for forgery and alteration is previously extracted by the image management server 200 for the original image and put in the main blockchain. This is to prepare a sequence of motion vectors. At this time, alternative results such as counterfeiting or the same as the original may be derived, and the reliability index may be determined.

Next, with reference to FIG. 2, the entire process of the object sequence management method of the compressed image using the motion vector blockchain according to the present invention will be described.

Steps (S1000, S1100): First, the image management server 200 receives a first compressed image, which is a management target for forgery or alteration of an object sequence, for example, from a plurality of CCTV cameras 210. The image management server 200 acquires image meta information for the first compressed image and parses the bitstream of the first compressed image to sequentially acquire a plurality of motion vectors ('first motion vector') over time. .

3 is a diagram illustrating an example of a motion vector obtained in each video frame over time in a CCTV video. As time elapses (t = t0 ~ t4), several objects appear in the CCTV image and move and disappear. Looking at the CCTV image, the image chunks of objects are found. When parsing the CCTV image, multiple motion vectors are extracted in relation to these image chunks. The motion vector is information generated by the CCTV camera 210 and inserted into the CCTV image to reduce the data size by increasing the image compression rate in the video compression technology standards such as H.264 AVC and H.265 HEVC. In the present invention, such a motion vector is used for object sequence management.

Step (S1200): The image management server 200 identifies one or more object unit areas, which are chunks of images to be managed for forgery or not in the first compressed image. Looking at the CCTV image of [Fig. 3], objects, especially moving object images, are found, and these are object unit areas. In addition, the image management server 200 extracts object identification information ('first object identification information'), which is information for classifying and identifying an object unit region within the corresponding image. For example, object identification information allocated to an object unit region in a software processing process may be used as object identification information. An embodiment of the object unit region in the present invention will be described later with reference to [FIG. 7] to [FIG. 9].

Steps (S1300, S1400): The video management server 200 identifies a sequence of motion vectors related to each object unit area among a plurality of first motion vectors in units of a preset block chain update cycle (for example, 30 seconds), and thus a motion vector Set to object sequence. For example, 250 first motion vectors and 7 object unit regions are obtained from the first compressed image (for example, a 30-second CCTV image), and the 250 first motion vectors are sorted and sorted for 7 object unit regions. To get 7 motion vector object sequences.

Subsequently, the image management server 200 generates an object unit trajectory by combining its first object identification information and a motion vector object sequence for each object unit area. That is, the object unit trajectory is the information for distinguishing and identifying the corresponding object and the motion vectors obtained from the first compressed image in connection with the object are connected over time. In the above example, 7 object unit trajectories are obtained from a 30-second CCTV image.

Subsequently, the image management server 200 collects these object unit trajectories (that is, seven object unit trajectories) to generate block data, and updates the main blockchain by reflecting block data according to a preset schedule.

Step (S1500): The image verification server 300 receives a second compressed image including an object sequence that needs to be checked for the forgery. The second compressed image may be provided from the external, for example, the administrator computer 400, or may be provided from the internal, for example, the image storage 220.

Step (S1600 ~ S1800): The image verification server 300 extracts 'image meta information', which is information for distinguishing and identifying the second compressed image, and multiple motion vectors from the second compressed image ('second motion vector') Are acquired sequentially over time.

Subsequently, the image verification server 300 extracts object identification information ('second object identification information') as information for identifying the object to be checked from the second compressed image. The object sequence is included in the second compressed image as an object that needs to be checked for forgery or alteration, and the object sequence may preferably consist of object identification information and an object image sequence. The object identification information included in the object sequence may be used as the second object identification information in (S1700).

Subsequently, the image verification server 300 identifies a sequence of motion vectors related to the second object identification information among a plurality of second motion vectors obtained from the second compressed image, and connects them in chronological order. Through this, the first motion vector sequence for the object sequence that is the forgery check target is obtained.

Step (S1900): The image verification server 300 searches the main blockchain and extracts one or more block data corresponding to the second object identification information. The object unit trajectory is stored in the block data, and the object unit trajectory includes object identification information. Therefore, the image verification server 300 analyzes the main blockchain, for example, the motion vector distributed ledger 100 managed by itself, finds and extracts block data having an object unit trajectory corresponding to the second object identification information. .

Then, the image verification server 300 obtains a second motion vector sequence for the object sequence by analyzing the motion vector object sequence of the object unit trajectory included in the extracted block data. By identifying the object sequence identification information, that is, one or more object unit trajectories corresponding to the second object identification information, from the block data, the motion vector object sequences constituting these object unit trajectories are connected in chronological order to eliminate the object sequence. 2 Obtain a motion vector sequence.

Step (S2000): The image verification server 300 compares the first motion vector sequence and the second motion vector sequence, and based on the result, whether the object sequence included in the second compressed image is forged from the original or the same as the original Calculate how reliable you can be. Conceptually, the degree of correspondence between two motion vector sequences corresponds to the reliability of the object sequence.

The second motion vector sequence obtained from the main blockchain is theoretically identical to the motion vectors obtained from the original image in relation to the object. On the other hand, since the second compressed image is a package of an image sequence and a motion vector of a corresponding object, it may correspond to sampling of the original according to an implementation. In addition, even in the time range, the second compressed image may not correspond to an integer multiple of the original blockchain update cycle and may correspond to a portion thereof. For example, the second compressed image is an image sequence obtained from 105 seconds of the original image, whereas the block data to be contrasted with it is 150 seconds, 120 times or 5 times the blockchain update cycle (eg 30 seconds). Corresponds to the quantity.

Therefore, the first motion vector sequence obtained from the second compressed image has a subset relationship with respect to the second motion vector sequence. Accordingly, comparing the first and second motion vector sequences and calculating the corresponding degree has a wider concept than seeing whether they are completely the same. In more general terms, how many items of the first motion vector sequence are included in the second motion vector sequence, or, conversely, how many of the items of the first motion vector sequence are not in the second motion vector sequence. It can be expressed as calculating.

As a result of the comparison, if the first and second motion vector sequences completely correspond, it can be determined that the corresponding object sequence has not been forged. In addition, if the motion vector sequence corresponds to, for example, 90%, the reliability of the corresponding object sequence can be evaluated as 90%. Preferably, when the reliability is higher than a certain threshold (for example, 97%), it may be determined that the object sequence has not been forged.

4 is a diagram illustrating the concept of generating motion vector object sequences, object unit trajectories, and block data in the present invention.

In the present invention, the image management server 200 obtains a motion vector object sequence for each object unit area from the first compressed image and generates an object unit trajectory therefrom. [Fig. 4] shows an example of obtaining a motion vector object sequence and an object unit trajectory for a CCTV image having a block chain update period of 30 seconds, and for 30 seconds. Normally, in a 30-second CCTV image, multiple object unit areas can be identified, but it is assumed that two object unit areas A and B are identified for convenience of explanation.

In response to the passage of time (t = t0 to t4), a series of motion vectors may be extracted in relation to object unit regions A and B. In [Fig. 4], the object unit areas A and B have changed their positions in the CCTV image over time, and a series of motion vectors are extracted for A and B according to the movement of the position. In the example of [Fig. 3], it is observed that objects appear and disappear and move in the CCTV image over time, and motion vectors have been extracted for each object to respond.

When the motion vectors obtained from the compressed image are sorted and sorted for each object unit area for a time of 30 seconds, a motion vector object sequence is obtained for each of the object unit areas A and B as shown in FIG. 4. For example, the motion vector object sequence in the object unit area A is [mva0 mva1 mva2 mva3 mva4] by sorting a sequence of motion vectors mva0 to mva4 extracted for 30 seconds in relation to A.

Subsequently, an object unit trajectory is generated for each object unit area. Preferably, the object unit trajectory is obtained by combining the object identification information "obj_ida", "obj_idb" and the motion vector object sequence [mva0 mva1 mva2 mva3 mva4] and [mvb0 mvb1 mvb2 mvb3 mvb4]. For example, the object unit trajectory of the object unit area A can be composed of [obj_ida mva0 mva1 mva2 mva3 mva4] by combining A's object identification information "obj_ida" and A's motion vector object sequence [mva0 mva1 mva2 mva3 mva4]. have.

Subsequently, object unit trajectories obtained from the compressed image are combined to generate block data for a blockchain update period (eg, 30 seconds) for the compressed image. Since two object unit areas A and B have been identified from the compressed image, the block data is combined by combining the two object unit trajectories and the identification information "video_id76" of the compressed image [video_id76; obj_ida mva0 to mva4; obj_idb mvb0 to mvb4].

If the image management server 200 receives 1000 CCTV images from 1000 CCTV cameras 210, the block data constructed by combining 1000 block data shown in [FIG. 4] is reflected in the main blockchain and updated. do.

Meanwhile, in [FIG. 4], the motion vectors mva0 to mva4 and mvb0 to mvb4 do not each mean one motion vector, but a set of motion vectors extracted in relation to object unit regions A and B at the corresponding point in time. For example, mva3 may mean, for example, a set of 17 motion vectors extracted from a compressed image in relation to the object unit region A at time t = t3.

5 is a diagram conceptually showing a process of differently setting the difficulty of a backup blockchain for each object group while backing up the main blockchain in the present invention, and [FIG. 6] is generally set to differently the difficulty of the blockchain. This is a diagram showing an example of concept and program source code.

In the present invention, the information stored in the main blockchain is backed up and copied to the backup blockchain for long-term storage. At this time, the difficulty is set differently, and then a plurality of backup blockchains are prepared, and then copied by backup group to the corresponding backup blockchain for each object group. As a result, the difficulty of the blockchain is set differently for each object group to be configured to store the backup. It is possible to do.

Referring to FIG. 5, three backup blockchains are provided, and the backup blockchain managing the upper backup distributed ledger 510 has the highest difficulty, and the backup blockchain managing the middle backup distributed ledger 520 is provided. The difficulty is medium, and the backup blockchain that manages the sub-backup distributed ledger 530 has the lowest difficulty. The concept of setting the difficulty level for the blockchain will be described later with reference to [FIG. 6].

Then, when backing up and copying the information of the main blockchain to the backup blockchain, it designates a backup blockchain that is the destination of the path to copy the information in correspondence to the group to which the object belongs. At this time, the classification of the object group may be implemented in various forms according to the designer's intention. For example, a group of objects can be classified according to importance. The important objects are classified into the first object group, and the less important objects are classified into the second object group or the third object group according to the degree.

Also, object groups may be classified according to the properties of the objects. For example, objects having a certain characteristic (eg, objects having a moving speed of 3 to 5 meters per second, objects in which three or more objects move together) may be classified into a first object group. Other properties may be defined for the second object group, and the remaining objects that do not correspond may be classified as the third object group.

In addition, the administrator may select objects belonging to the first object group and the second object group from among a plurality of objects through mouse manipulation. At this time, the remaining unselected objects are classified into a third object group.

Referring to FIG. 5, object unit trajectories for three object identification information a, b, and c are stored in the mail blockchain. The first block data includes object unit trajectories of object identification information a, b, c, the second block data contains object unit trajectories of object identification information a, c, and the third block data includes object identification information b, c. Object trajectory of is included. At this time, object identification information a was classified into a first object group, object identification information b was classified into a second object group, and object identification information c was classified into a third object group.

Under these conditions, when the backup information of the main blockchain is copied to the backup blockchain, the object unit trajectory of object identification information a belonging to the first object group is copied to the backup blockchain of difficulty (upper level). Likewise, the object unit trajectory of object identification information b belonging to the second object group is copied to the backup blockchain of difficulty (medium), and the object of object identification information c belonging to the third object group is copied to the backup blockchain of difficulty (below). The unit trajectory is copied. At this time, the object unit trajectory does not end with being copied, but the field of block data must be filled according to each difficulty.

When the difficulty is set high in the blockchain technology, it becomes more complicated to configure the blockchain, and as a result, it becomes more difficult for a third party to forge and falsify the data of the blockchain once constructed, and eventually the data The credibility for is increased. Therefore, it is possible to manage forgery and alteration more difficult by managing an important object or an object having a special property by using a backup block chain with high difficulty. Conversely, by managing a backup block chain with low difficulty for most objects that have no special features, it is also possible to obtain an advantage of lowering the overall blockchain processing burden.

On the other hand, when the image verification server 300 tries to check the object sequence with respect to the specific second object identification information, the information corresponding to the second object identification information does not exist in the main blockchain and the backup copy to the backup blockchain If it is, the process of acquiring the second motion vector sequence should be different from (S1900) of [FIG. 2].

In this case, it is necessary to identify which object group the second object identification information belongs to, and query the backup blockchain corresponding to the object group to which the second object identification information belongs among the plurality of backup blockchains. For example, if the second object identification information belongs to the second object group, the block data must be checked by querying the backup blockchain of difficulty (medium). Through this, one or more block data having the object unit trajectory corresponding to the second object identification information is extracted from the backup blockchain, and the motion vector object sequence of the object unit trajectory included in the extracted block data is analyzed to analyze the object Obtain a second motion vector sequence for the sequence.

Hereinafter, the concept of setting the difficulty level for the blockchain will be described with reference to FIG. 6.

One block that makes up the blockchain is

-[Motion vector value detected as an object in the compressed image],

-[Time to know temporal order],

-[Hash value of the previous block],

-[A. Hash value calculated according to the difficulty],

-[B. Nonce value that satisfies the condition of hash value according to the difficulty]

The process of calculating the field values constituting this block is as shown in <step 1> to <step 4> in [Fig. 6].

At this time, the value that determines the difficulty of the blockchain is defined as 000000000 ... XXXXXX (256bit), and the more the number of 0s, that is, the less the number of Xs, the higher the difficulty of the blockchain. That is, in <step 3>, hash calculation is performed repeatedly by increasing the nonce value until a hash value smaller than the difficulty value is generated. At this time, the more the number of 0s in the difficulty value, the smaller the difficulty value and the more difficult the condition becomes. The calculation becomes complicated and the difficulty of the blockchain increases.

For example, it is assumed that the difficulty of these two blockchains A and B is 00000000XXXX and 0000XXXXXXXX. Since the number of zeros included in the difficulty value of Blockchain A is higher, Blockchain A has higher difficulty than Blockchain B. The main challenge when performing blockchain update in blockchain technology is to find a hash value that satisfies the following conditions.

Blockchain A: Hash value that satisfies a value less than 00000000XXXX

Blockchain B: Hash value that satisfies a value less than 0000XXXXXXXX

Therefore, compared to Blockchain B, it is inevitable to find hash value in Blockchain A, and the amount of computation to be performed and the time required to perform the computation are much longer to obtain the task solving value.

7 is a flow chart showing a process of identifying a moving object through video analysis in a CCTV image and setting it as an object unit area as a first embodiment of setting an object unit area for a compressed image in the present invention. ] Is a block diagram showing a general configuration of a video decoding apparatus.

The first embodiment identifies an object (eg, human A, human B, car, animal, etc.) by performing a complex and high-dimensional image analysis process on the compressed image, and each object occupies an image chunk (ie, Object image) is set as an object unit area.

8 is a block diagram showing a general configuration of a video decoding apparatus according to the H.264 AVC technology standard. 8, the video decoding apparatus according to H.264 AVC includes a parser 11, an entropy decoder 12, an inverse transformer 13, a motion vector operator 14, a predictor 15, and deblocking It comprises a filter 16. These hardware modules sequentially process the compressed image data to decompress the compressed image and restore the original image data. At this time, the parser 11 parses the motion vector and coding type for the coding unit of the compressed image. Such a coding unit is a video block such as a macroblock or subblock.

7 is a flow chart showing a process of extracting an object (moving object) from a CCTV image through image analysis and setting it as an object unit area. Referring to FIG. 7, the target compressed image is decoded according to H.264 AVC and H.265 HEVC (S10), and the frame images of the reproduced image are downscaled to a small image, for example, about 320x240 (S20). At this time, the reason for downscale resizing is to reduce the processing burden in the subsequent image analysis process. Then, after obtaining differential images for the resized frame images, a moving object is extracted through image analysis (S30). Then, the images of these moving objects are set as object unit areas for CCTV images (S40).

In the first embodiment, since the object is selected while the content of the compressed image is understood, there is an advantage in that the extraction of the object unit region is accurate and the boundary is exactly corresponding to the actual image. On the other hand, compressed video decoding, downscale resizing, and video analysis processes are very complex processes, so there is a disadvantage in that the system's image processing speed is significantly slowed down.

9 is a second embodiment of setting an object unit region for a compressed image in the present invention, extracting a moving object region based on syntax from the target compressed image and setting it as an object unit region for the target compressed image It is a flowchart showing the process.

The second embodiment is a method of identifying a portion that appears as a moving object from a compressed image based on syntax information obtained by parsing the compressed image and setting them as an object unit area. Compared to the first embodiment described above, the second embodiment does not understand the contents of the image and extract the object image, but presumes that the moving object is included based on syntax information in a state where the contents of the compressed image are completely unknown. There is a conceptual difference from the first embodiment in that it is a method of extracting an image lump. That is, since the image analysis is not performed, the moving object area is extracted without information on the image content.

In the second embodiment, the bitstream of the compressed image is parsed without the need to decode the compressed image, and a moving object region is quickly extracted through syntax information for each image block. As a video block, any one or a combination of macro blocks and sub blocks may be adopted, and motion vector and coding type are preferable as syntax information. Do. The moving object region thus obtained does not accurately reflect the boundary of the moving object existing in the image, as confirmed in the various images attached to the present specification, but has a high processing speed and high reliability.

Hereinafter, a concept of a process of quickly extracting a moving object region based on syntax from a compressed image and setting it as an object unit region for a target compressed image according to the second embodiment will be described with reference to FIG. 9.

Step (S100): First, based on the motion vector of the compressed image, effective motion that is substantially recognizable from the compressed image is detected, and the image area in which the effective motion is detected is set as a moving object area.

To this end, motion vectors and coding types of a coding unit of a compressed image are parsed according to video compression standards such as H.264 AVC and H.265 HEVC. At this time, the size of the coding unit is generally about 64x64 pixels to 4x4 pixels, and may be variously set according to the designer's selection.

For each video block, the motion vectors are accumulated for a predetermined time (eg, 500 msec), and it is checked whether the resulting motion vector accumulation value exceeds a preset first threshold (eg, 20). If such an image block is found, it is considered that an effective motion is found in the corresponding image block and is marked as a moving object area. Accordingly, even if a motion vector occurs, if the cumulative value for a certain time does not exceed the first threshold, the video change is assumed to be negligible and ignored.

Step (S200): The area around the moving object detected in the previous step (S100) is inspected based on the motion vector and the coding type to expand the boundary of these moving object areas to approximately the extent. Through this process, the moving object regions detected in the form of fragmented image blocks in (S100) are connected to each other to obtain a meaningful lump shape.

In the previous step (S100), by selecting image blocks according to strict criteria, an image block that seems to correspond to a moving object in a compressed image is detected and marked as a moving object area. In this time (S200), other image blocks located around the image block marked as the moving object area in (S100) are examined. These are referred to as one or more 'neighbor blocks' in this specification. For these neighboring blocks, it is determined whether or not they correspond to the moving object area according to the criterion that is relatively relaxed compared to the criterion applied in (S100).

In compressed images, macroblocks or subblocks are very small. Therefore, if it is a video shot of a person, a car, or an animal, such as a CCTV video, it is difficult for the moving object to appear only in one video block due to its nature, and it is expected to appear across multiple video blocks. That is, it is predicted that the probability that a moving object is stamped on an image block existing in the vicinity of the image block on which the moving object is photographed is relatively higher than that on the image block that is not. Reflecting such technical ideas, in (S200), it is determined whether or not it corresponds to the moving object area according to the relatively relaxed determination criteria for the neighboring block existing around the moving object area.

Preferably, each neighboring block is inspected, and if the motion vector value detected in the current frame is greater than or equal to a preset second threshold (eg, 0) or the coding type is an intra picture, the corresponding image block is also a moving object Mark as an area. In another embodiment, when the motion vector accumulation value previously calculated in (S100) with respect to the neighboring block is greater than or equal to the second threshold (eg, 5) or the coding type is intra picture, the corresponding video block is also marked as a moving object area. Can be. At this time, it is logically reasonable that the second threshold is set to a smaller value than the first threshold.

Conceptually, if an image block in which an effective motion is found and there is a certain amount of motion in the vicinity of the moving object area is marked as a moving object area because it is likely to be a single chunk with the preceding moving object area. In addition, in the case of an intra picture, since there is no motion vector, it is impossible to determine whether it is a moving object area based on the motion vector. Accordingly, if the intra picture is located adjacent to an image block that has already been detected as a moving object area, it is estimated that a single chunk is formed together with the extracted moving object area. This is because the loss when the moving object region is fragmented is not large, whereas the loss when the image block other than the moving object region is included in the moving object region is large.

Step (S300): The fragmentation of the moving object area is arranged by applying interpolation to the moving object area detected in the preceding (S100) and (S200). In the previous process, since it is determined whether the moving object area is in units of image blocks, there are several moving object areas because there is an image block that is not marked as a moving object area in the middle even though it is actually one moving object (eg, a person). The phenomenon of splitting into may occur. Accordingly, if there is one or a small number of non-marked image blocks surrounded by a plurality of image blocks marked as a moving object region, they are additionally marked as a moving object region. Through this, it is possible to make the moving object area divided into several pieces into one, and the effect of such interpolation is clearly seen when comparing [FIG. 13] and [FIG. 14].

Through the above steps (S100) to (S300), one or more moving object regions were obtained from the compressed image. The region of moving object is a lump in which a plurality of image blocks are aggregated as shown in [Fig. 14], and it is estimated that one or more moving objects are included therein based on syntax information. It was selected separately. For software processing, it is desirable to allocate and manage unique identification information (Unique ID) to these mobile object areas.

Step (S400): Through the above process, the moving object region is quickly extracted based on the syntax (motion vector, coding type) of the coding unit for the compressed image. In step S400, the moving object region is set as an object unit region for the target compressed image.

[Figure 10] is a flow chart showing an example of the implementation of the process of detecting an effective movement (effective movement) from a compressed image in the present invention, [Figure 11] is an example of the result of the effective motion area detection process is applied to the CCTV image It is a figure to show.

Step (S110): First, the coding unit of the compressed image is parsed to obtain a motion vector and a coding type. Referring to FIG. 8, the video decoding apparatus performs parsing (header parsing) and motion vector calculation according to video compression standards such as H.264 AVC and H.265 HEVC on a stream of compressed video. Through this process, motion vectors and coding types are parsed for the coding unit of the compressed image.

Step (S120): The motion vector accumulation value for a preset time (eg, 500 ms) is obtained for each of the plurality of image blocks constituting the compressed image.

This step was presented with the intention of detecting effective motions that are practically recognizable from the compressed image, such as a driving car, a running person, and a crowd fighting each other. Shaking leaves, ghosts that appear briefly, and shadows that change slightly due to reflection of light, etc., should not be detected as they are practically meaningless objects even though there is movement.

To this end, a motion vector is accumulated by accumulating motion vectors in one or more image block units for a predetermined time (eg, 500 msec). At this time, the video block is used as a concept including a macroblock and a subblock.

Steps (S130, S140): The motion vector accumulation value is compared with a preset first threshold value (for example, 20) for a plurality of image blocks, and an image block having a motion vector accumulation value exceeding the first threshold value is moved object area It is marked with.

If an image block having a motion vector cumulative value equal to or greater than a certain level is found, it is marked as a moving object area, as something significant motion, that is, an effective motion is found in the corresponding image block. For example, it is intended to select and detect movements that are worth the attention of the control personnel to the extent that a person jumps in the video control system. Conversely, even if a motion vector occurs, if the accumulated value for a certain time is small enough to not exceed the first threshold, the change in the image is not so large and is assumed to be insignificant and ignored in the detection step.

[Fig. 11] is an example of visually showing the result of detecting an effective motion area from a CCTV image in the present invention through the process of [Fig. 10]. In FIG. 11, an image block having a motion vector cumulative value equal to or greater than a first threshold is marked as a moving object area and displayed in red. Looking at [Fig. 11], a sidewalk block, a road, and a shadowed part are not displayed as a moving object area, while walking people or a driving car are displayed as a moving object area.

[Fig. 12] is a flowchart showing an example of the implementation of the process of detecting a boundary area for a moving object area in the present invention, [Fig. 13] is [Fig. 12] for the CCTV video image of [Fig. 11] A diagram showing an example of a result of applying a boundary region detection process according to.

Looking at [FIG. 11] above, it can be found that the moving object was not properly marked and only a part was marked. That is, if you look at a person walking or a driving car, you can find that not all objects are marked, but only some video blocks are marked. Moreover, it is also found that a plurality of moving object areas are marked for one moving object. This means that the criteria for determining the moving object area adopted in (S100) above is very useful for filtering out the general area, but it is quite strict. Therefore, a process of detecting the boundary of the moving object is necessary by looking around the area around the moving object area.

Step (S210): First, a plurality of adjacent image blocks are identified centering on the image block marked as the moving object area by the preceding (S100). These are referred to herein as 'neighboring blocks'. These neighboring blocks are parts that are not marked as moving object areas by (S100). In the process of [Fig. 12], by looking at them more, it is to check whether there are any of these neighboring blocks that may be included in the boundary of the moving object area. .

Steps (S220, S230): For a plurality of neighboring blocks, a motion vector value is compared with a preset second threshold, and a neighboring block having a motion vector value exceeding the second threshold is marked as a moving object area. If the effective motion that can give practical meaning is located adjacent to the recognized moving object area, and a certain amount of motion is also found in itself, the video block may differ from the preceding moving object area due to the nature of the captured video (eg CCTV video). It is most likely a lump. Therefore, this neighboring block is also marked as a moving object area.

As a first embodiment to implement this, each neighboring block is examined, and when the motion vector value detected in the current frame is greater than or equal to a preset second threshold (eg, 0), the corresponding video block is also marked as a moving object area.

On the other hand, as a second embodiment, if the motion vector accumulation value previously calculated in (S100) for a neighboring block is equal to or greater than a preset second threshold (eg, 5), the corresponding image block may also be marked as a moving object area. At this time, it is reasonable that the second threshold is set to a smaller value than the first threshold.

Step S240: Further, among the plurality of neighboring blocks, a coding type is an intra picture, and is marked as a moving object area. In the case of an intra picture, since a motion vector does not exist, it is impossible to determine whether a corresponding neighboring block corresponds to a moving object region based on the motion vector. If the intra picture is located adjacent to the image block that has already been detected as the moving object area, it is preferable to estimate that it is formed as a lump together with the extracted moving object area. This is because the loss when the moving object region is fragmented is not large, whereas the loss when the image block other than the moving object region is included in the moving object region is large.

13 is a diagram visually showing results applied to a boundary area detection process on a CCTV image in the present invention, and a plurality of image blocks marked as moving object areas are displayed in blue color through the above process. Looking at [Fig. 13], in the vicinity of the moving object area indicated in red in [Fig. 11], the moving object area of blue color is expanded further, and through this, it covers all moving objects when compared with the actual video captured by CCTV. You can find that it is enough to do.

14 is a diagram showing an example of a result of arranging a moving object area through interpolation for the CCTV video image of [FIG. 13].

Step (S300) is a process of arranging the division of the moving object area by applying interpolation to the moving object areas detected in the above (S100) and (S200). Referring to FIG. 13, a non-marking image block is found between regions of a moving object indicated in blue. If there are non-marking image blocks in the middle, it can be considered that they are a large number of individual mobile objects. When the moving object region is fragmented, the result of object image sequence management for the compressed image according to the present invention may be inaccurate, and the number of moving object regions increases, which complicates the process of managing the object image sequence for the compressed image according to the present invention. There is also a problem.

Accordingly, in the present invention, if there is one or a small number of non-marked image blocks surrounded by a plurality of image blocks marked as a moving object region, it is marked as a moving object region, which is called interpolation. Looking at [FIG. 14] in contrast to [FIG. 13], all non-marking image blocks existing between the moving object areas are marked as moving object areas. Through this, all the areas moving by the lump are bundled and treated as one moving object.

When comparing [FIG. 11], [FIG. 13], and [FIG. 14], it can be found that the moving object region properly reflects the situation of the real image through the boundary region detection process and the interpolation process. In [Fig. 11], if it is judged as a chunk marked with red, it will be treated as if a lot of very small objects move in the video screen, which does not correspond to the real thing. On the other hand, if it is judged as a lump marked with blue in [Fig. 14], several moving objects having a certain volume will be treated as being, and thus the actual scene is similarly reflected.

On the other hand, the present invention can be implemented in the form of computer-readable code on a computer-readable non-volatile recording medium. There are various types of storage devices such as hard disks, SSDs, CD-ROMs, NAS, magnetic tapes, web disks, cloud disks, etc., and codes are distributed and stored in multiple networked storage devices. It can also be implemented as a form of execution. In addition, the present invention may be implemented in the form of a computer program stored in a medium in order to execute a specific procedure in combination with hardware.

100: motion vector distributed ledger
200: video management server
210: CCTV camera
220: video storage
300: image verification server
400: administrator computer
510: Top Backup Distributed Ledger
520: Deputy General Ledger
530: Distributed backup ledger

Claims (9)

A first step of receiving a first compressed image that is an object of management for forgery of the image;
A second step of parsing the first compressed image and sequentially obtaining a plurality of motion vectors (hereinafter referred to as 'first motion vectors');
A third step of identifying one or more object unit areas, which are chunks of images to be managed for forgery or non-falsification in the first compressed image, and identifying first object identification information for classifying and identifying each of the object unit areas;
In a preset block chain update cycle unit, a sequence of motion vectors (hereinafter referred to as a 'motion vector object sequence') related to the object unit area among the plurality of first motion vectors is identified, and the one or more object unit areas are identified. A fourth step of generating an object unit trajectory by combining the first object identification information and the motion vector object sequence for each, and generating block data by combining the one or more object unit trajectories;
A fifth step of updating a main blockchain by reflecting the block data at each blockchain update cycle;
A sixth step of receiving a second compressed image including an image sequence (hereinafter referred to as an 'object sequence') of an object to be checked for forgery of the image;
A seventh step of parsing the second compressed image and sequentially obtaining a plurality of motion vectors (hereinafter referred to as 'second motion vectors');
An eighth step of extracting second object identification information for discriminating and identifying the object to be checked for forgery of the image from the second compressed image;
A ninth step of obtaining a first motion vector sequence for the object sequence by identifying and connecting a sequence of motion vectors related to the second object identification information among the plurality of second motion vectors;
By querying the main blockchain, one or more block data having an object unit trajectory corresponding to the second object identification information is extracted, and the motion vector object sequence of the object unit trajectory included in the extracted block data is analyzed to obtain the A tenth step of obtaining a second motion vector sequence for the object sequence;
An eleventh step of determining whether the object sequence is falsified or not based on a result of comparing the first motion vector sequence and the second motion vector sequence;
A method of managing an object sequence of a compressed image using a motion vector block chain composed of a.
The method according to claim 1,
Is performed after the fifth step,
A twelfth step of backing up and copying a plurality of object unit trajectories stored in the block data of the main blockchain to a plurality of backup blockchains of varying difficulty according to the object group to which each first object identification information belongs;
Performed between the ninth step and the tenth step,
When the second object identification information is backed up and copied to the backup blockchain, among the plurality of backup blockchains, the second object identification information is retrieved by querying the backup blockchain corresponding to the object group to which the second object identification information belongs Extracting one or more block data having an object unit trajectory corresponding to, and analyzing a motion vector object sequence of the object unit trajectory included in the extracted block data to obtain a second motion vector sequence for the object sequence and A thirteenth step proceeding to the eleventh step;
A method of managing an object sequence of a compressed image using a motion vector block chain, further comprising a.
The method according to claim 2,
The twelfth step,
Setting up a plurality of backup blockchains having different difficulty levels;
Setting a plurality of object groups corresponding to the plurality of backup blockchains;
In the main blockchain, each object unit trajectory stored in each block data corresponds to an object group to which the first object identification information belongs, and selects one backup blockchain among the plurality of backup blockchains. Backup copying to block data;
Configuring block data according to respective difficulty levels of the plurality of backup blockchains;
Object sequence management method of a compressed image using a motion vector blockchain, characterized in that comprises a.
The method according to claim 1,
The object unit area is set as one or more moving object images extracted from the target compressed image through decoding, downscale resizing, and image analysis of the first compressed image or the second compressed image as each target compressed image. A method of managing an object sequence of a compressed image using a motion vector blockchain, which is characterized by being
The method according to claim 1,
The object unit region is identified by a syntax-based moving object region extraction process using the first compressed image or the second compressed image as each target compressed image,
The syntax-based moving object region extraction process,
A step A of parsing the bitstream of the target compressed image and obtaining a motion vector and a coding type for the coding unit;
A step B of obtaining a motion vector accumulation value for a preset time for each of a plurality of image blocks constituting a target compressed image;
A C step of comparing the accumulated motion vector with respect to the plurality of image blocks with a preset first threshold;
A step D of marking an image block having a motion vector accumulation value exceeding the first threshold as a moving object area;
An E step of setting an individual chunk of the image block marked as the moving object area as an object unit area;
Object sequence management method of a compressed image using a motion vector blockchain, characterized in that comprises a.
The method according to claim 5,
The syntax-based moving object region extraction process,
Performed between the D step and the E step,
A step a of identifying a plurality of adjacent image blocks (hereinafter referred to as 'neighbor blocks') with respect to the moving object region;
A step b of comparing the motion vector values obtained in the first step with a preset second threshold for the plurality of neighboring blocks;
A step c of additionally marking a neighboring block having a motion vector value exceeding the second threshold as a moving object area as a result of the comparison of step b among the plurality of neighboring blocks;
A method of managing an object sequence of a compressed image using a motion vector block chain, further comprising a.
The method according to claim 5,
The syntax-based moving object region extraction process,
Performed between the D step and the E step,
A step a of identifying a plurality of adjacent image blocks (hereinafter referred to as 'neighbor blocks') with respect to the moving object region;
A step b of comparing the accumulated motion vector with respect to the plurality of neighboring blocks to a second threshold that is preset to a value less than the first threshold;
A step c of additionally marking a neighboring block having a motion vector accumulation value exceeding the second threshold as a moving object area as a result of the comparison of step b among the plurality of neighboring blocks;
A method of managing an object sequence of a compressed image using a motion vector block chain, further comprising a.
The method according to claim 6 or 7,
The syntax-based moving object region extraction process,
It is performed between the step c and the step E,
A d step of additionally marking a neighboring block having an intra-picture coding type as a moving object area among the plurality of neighboring blocks;
An e-step of interpolating the plurality of moving object areas to additionally mark a non-marked image block of a preset number or less surrounded by the moving object area as a moving object area;
A method of managing an object sequence of a compressed image using a motion vector block chain, further comprising a.
A computer program stored in a medium in combination with hardware to execute a method for managing an object sequence of a compressed image using a motion vector blockchain according to any one of claims 1 to 7.

Images