Наука

Поиск

Video content verification using blockchain technology

Скачать в формате pdf

Viacheslav Voronin

Moscow State University of Technology

«STANKIN»

Moscow, Russian Federation

voronin_sl@mail.ru

Evgenii Semenishchev

Moscow State University of Technology

«STANKIN»

Moscow, Russian Federation

[email protected]

Aleksandr Zelensky

Pro-rector for Research Work and

R&D Politics

Moscow State University of Technology

«STANKIN»

Moscow, Russian Federation

science@stankin.ru

Iliya Svirin

CJSC Nordavind, Moscow

Moscow, Russian Federation

i.svirin@nordavind.ru

Andrey Alepko

Dept. of Radio-Electronics Systems

Don State Technical University

Rostov-on-Don, Russian Federation

alepko@sssu.ru

In distributed high-performance systems that allow

processing large data sets, the important task is the analysis

and verification of the video data. The problem of confirming

data is relevant for many areas. The need for this arises when

fixing offences, banking, remote management, confirmation of

actions, etc. In this paper, we describe an approach to verifying

video received by a mobile device like mobile phone, tablet or

PC, equipped with a camera and controlled by the operating

system (Windows, Android or iOS). The proposed algorithm

uses the procedure of entering the Swype code using the

movement mobile camera. To improving the accuracy, we use

additional information from different mobile sensors like

accelerometer, gyro, barometer, and GPS. In the process of

data verification, video transmission to the server is not

perform what ensures the privacy of the captured video data.

The server and mobile device stores data about the file size, the

date of its recording, the time, the data device and its position.

Keywords— verification data, Swype code, mobile device,

PROVER, blockchain

I. I

NTRODUCTION

The revolution of smartphones, which started ten years

ago with the advent of the iPhone and tablet computers, had

a significant impact on the ways users communicate in the

network. Everyone suddenly became the creator and

distributor of content, causing, as a consequence, his

exponential growth. Such data has a large dimension. They

can be multi-layered and contain data on the gradient of

temperature, 3D, stickers and explanations, and so on. Big

data requires large calculations, which are most often

impossible on a mobile device. To analyze such data it is

necessary to use Smart Cloud.

This has significantly changed and continues to change

many sectors of the economy, launching the processes of

digitizing the world around us. Photo and video content are

actively used not only for entertainment and educational

purposes but also for other needs, including economic and

legal nature - financial, insurance, judicial, medical and other

services. In this regard, there is a serious need for an

independent, decentralized service that objectively

guarantees the authenticity of the created video content and

protects it from possible forgery and unfair editing.

The authenticity of digital video recordings of events and

facts of commercial and legal value is often questionable

because video files can be edited for forgery purposes, by

using a virtual camera (emulator). Attributes such as the date

of the video could also be artificially altered.

In this connection, the actual task is to operate on

verification data. Verification will guarantee the originality

of video data, confirm the time, location and duration of the

video.

II. SOLUTIONS

FOR

VIDEO

VERIFICATION

The authenticity of the content and the rights to it

(copyright) today are mostly being approved at the level of

platforms for storing and demonstrating video materials. An

example of such a solution is Google content ID, which

allows only to confirm the time of downloading a video on

YouTube, this is the basis for the assumption of its

originality, based on the principle of presumption of

authorship (a person is considered as an author until the real

author disputes this fact).

The drawback of this and similar solutions is that they do

not allow to restore the time of real video recording, its

originality, and integrity. Also, these solutions work within

their platforms "only on YouTube," are provided manually

by service administrators upon application and always

depend on the opinion of the service administrators. That is,

there is always a place for the human factor - subjectivity or

simply error [1, 2].

The authors try to protect the content by watermarks and

logos on the video screen, but this can only help them in the

subsequent contestation of its authorship, but not to prove the

fact of forgery. That means that in legal and financial matters

this approach is completely inapplicable.

At the present, the blockchain community already has

various online electronic notary services, which make

possible to certify the existence of "Proof of existence" and

the authorship of "Proof of ownership" of all kinds of files

and documents and digital content:

• Block Notary - a service that helps to create "Proof

of existence" of any content (photos, files, any

media) using the TestNet3 or Bitcoin network. The

frontend system is a mobile application for iOS that

registers a document hash in a blockchain.

• Emercoin DPO Anti-fake - technology based on the

Emer platform allows to create the item (product) a

unique digital passport stored in a decentralized

database - blockchain and provides services for

208

2018 IEEE International Conference on Smart Cloud

DOI 10.1109/SmartCloud.2018.00042

managing this passport. It is focused mainly on the

offline segment - it helps to register individual

details (VIN, IMEI numbers) in the system to the

shield of real goods from fraud.

• Stampery - blockchain technology that can verify e-

mail or any files. This simplifies the process of

verifying letters by simply sending them by e-mail to

a specially created mailing address for each

customer. Law firms use Stampery technology for a

very cost-effective way of document verification.

• https://www.ascribe.io/ - service for registration of

content. It is positioned for digital works of art. It

offers to register a work, put it on sale in a secure

marketplace, and then monitor its use

(demonstration).

• https://letsnotar.me/ - an easy service that

automatically stores to blockchain a hash of files that

were uploaded to it. It could launch on a smartphone,

allows to get access the camera, take photos and

videos and to hash them. However, it cannot

guarantee that the video is recorded from a real, not a

virtual camera - so it does not protect against

forgery.

All these services unite one thing: they can assure a hash

of a file that has already been written before or supposedly

written down from the camera of the device, but they cannot

guarantee its originality, integrity, and authenticity of the

video. They do not protect against forgery and unfair editing

because they do not have the appropriate technology to

verify the video content being created [3].

Everyone can record video at any time. It is fast, simple,

available and convenient. Many companies use video content

in their business processes – it gives them a competitive

advantage. But technologies allow others to edit and fake

videos, which seriously undermines the credibility of that

format.

In this work we use PROVER technology allows

guaranteeing that this content was created at a specific time

and place, from the camera of a device, to ensure that there

are no signs of forgery and editing. The PROVER can

become a functional addition, extending the capabilities of

the services discussed above and the level of trust to them.

Having the opportunity to 100% guarantee the authenticity of

the created video file, these services will, for example,

provide services for notarization of the authenticity of video

statements.

PROVER technology use cases:

• Fintech, when it comes to authentication of loaner’s

identity. Banks and financial services providers can

verify clients during the customer onboarding

procedure with lower risks of identity theft. Clients

can perform remote actions within the system.

• Auto insurance, dealing with fraud of security

addition in contractual terms. Clients can record

video evidence in case of insurance loss for

insurance on demand and car sharing services.

Insurance companies can receive video evidence of

performed services covered by insurance (medical,

repairs).

• Simple video proof of ownership, working for

bloggers as example. Individuals and organizations

can store a timestamped video hash on Blockchain as

a digital proof of ownership for original and

authentic video content.

• Public statements, to keep secure what is said before

editing. Public speakers, celebrities and businessmen

preventing reputational damage from montage, CGI

and rapidly growing sophisticated machine learning

algorithms and tools able to edit or generate fake

video statements.

• Crowdsourced media platforms, to keep ownership

of actual content makers. Public and crowdsourced

news platforms can validate the authenticity,

exclusivity and timing of video news submitted by

individual contributors.

• Video platforms with user-generated content, of

exclusive contributions. Both users and platforms

can prove the authenticity and exclusivity of user

generated video content and share monetization

proceeds.

• Online dating, keeping users from fake content.

Users can be sure that they are chatting with a real

person on video dating websites and services.

• Outsourced reporting, carrying out remote inspection

of actual performance. Employers and contractors

can exchange authentic and time stamped work

reports.

• Accident reports, to proof someone’s position in

court using evidence. Both parties involved in a

traffic accident can rely on a video recording to

prove authenticity of time, date and record of the

accident.

• Notary actions, allowing people verify video content

without visiting special parties. Parties can maintain

a Blockchain video database of trusted “handshake”

agreements.

• Home education and exams control, allowing remote

authentication of online courses. Video recordings

can be used to confirm authenticity of a person

taking an online exam.

III. THE

TECHNOLOGY

PROVER

The developed service contains the following

components (www.prover.io):

• A mobile app that installs on a smartphone and

launches with the camera turned on or initiates the

launch of the camera itself.

• A set of algorithms and utilities for integrating

PROVER technology into third-party solutions and

services.

• Smart Contract PROOF (only for implementation

based on the Ethereum platform).

The presented technology allows performing the

following actions:

• Video footage is produced by a real video camera

integrated into the mobile device, and not emulated

by a virtual video camera;

• The video material is complete, not edited, without

gluing and insertions;

• The record was made in a certain period.

To verify the data, two approaches are used, based on

changes in sensor-fixed mobile devices and analysis of video

data received by the camera. The first approach uses data

obtained with the help of an accelerometer, gyro, barometer,

and GPS. The second approach is based on preliminary

identification with visual Swype code.

209

The algorithm of verification with Swype code

the classical Swype code, which is being entered by

moving user's finger on the touchscreen, forming a

continuous line connecting the points, shown on the screen,

in PROVER technology, the Swype code is being entered by

moving the smartphone with the camera in a record mode.

On Figure 1, we show an example to enter Swype code

using video stream from the mobile camera.

Figure 1. Graphical representation to enter Swype code.

On Figure 2 shows the algorithm for search verification

using Swype code.

The following steps realize the algorithm presented in

Figure 2:

Step 1: Receiving video data. The video data is

transferred in parallel to the screen of the mobile device and

the input of the data search and verification program. Video

analysis is carried out according to the location of objects in

the frame. The algorithm of search is always looking for a

condition for launching verification. This condition is the

circular movement of the mobile device.

Step 2: Search for frame offsets. This section of the

algorithm is based on the application of the phase correlation

method. This method is described in the paper [4], end

realized by the following steps:

2.1) For two frames

()

111

and

()

222

, we find

the Fourier transformation

()

111

and

()

222

[4-7].

2.2) The cross-phase spectrum of the two spectral

functions

()

111

and

()

222

stands for the ratio:

()( )

111 2 22

Fuv F uv

uv F u v

(1)

The resulting expression is a spectral function with a unit

modulus whose phase is equal to the phase difference of the

functions.

Figure 2. The verification algorithm with Swype code.

2.3) We perform the inverse Fourier transform, which is

the phase correlation. Since on adjacent frames, there are the

same elements up to bias:

() ()

222 111

size size

ua vb

uv e Fuv

§·

−+

¨¸

©¹

(2)

In this expression,

b,a

are the peaks of the delta

function.

2.4) In the case of the similarity of the frames, peaks will

be present. The height of the peak determines the degree of

similarity, and the peak position corresponds to the

displacement of frames relative to each other.

Step 3: Search for circular motion (Fig. 3).

210

Figure 3. Determination of the circular motion.

This stage of the algorithm is performed by analyzing the

displacement vectors that are in step 2. An ideal description

of the change in the motion of the vectors as a function of the

starting point in time for each of the axes is shown

in Figure 4.

The displacement graphs relative to the 0X and 0Y axes

are two harmonic functions that have a bias on

Realizing that in real conditions, a person cannot describe the

ideal circle. We set the confidence intervals equal to ±10% of

the signal amplitude. Within the range of this interval, the

result will be accepted as correct.

Step 4: Connection to the server. At this stage, there is a

connection to the server. The server forwards the Swype

code sequence of offsets. The sequence of movements of the

mobile device relative to these elements is operation decode.

An example of such a code is shown in Figure 2. A

displacement of one unit within the three by three field is

considered correct if the movement of the vector in the

correct direction (step 2) at a time is fixed more than five

steps. Each step takes place every half second. If an error

occurs in the Swype code entry, the check operation is

terminated. To begin verification of the data, it is necessary

to repeat the action from step 3.

Step 5: Data is transferred between the server and the

mobile device. The mobile device forms a data packet. This

package includes information on the time of the start of the

data verification, the frame size, the device description, and

the file name.

Step 6: Verification. At the end of the video recording,

the resulting file gets a label. Data about the end time of the

survey, its size, and the hash function describing the file are

transmitted to the server. Any change in the file makes a

change in the information that was sent to the server.

Using this automatic algorithm for recognizing the

Swype code will allow the user at the stage of video

recording to be sure that afterward, later, if that the video

will need to be checked for authenticity, this Swype code

will be recognized by the service.

The verification algorithm using phone sensors

A stream of metadata (data from all sensors available in

mobile device, with the maximum frequency - an

accelerometer, a gyroscope, a magnetometer, GPS

coordinates, etc.) will be recorded in parallel with the video

file to prevent forgery.

To determine the position of the object in space, we

introduce the global three-dimensional Cartesian coordinate

system

XYZ0

so that the axis

coincided in direction with

the direction of gravity field power lines

and the axis

coincided with the declination of the vector of the magnetic

field of the planet

To describe the position of the sensor in the global

coordinate system (GCS) we introduce a local coordinate

system (LCS), whose axes will coincide with the

corresponding axes of the acceleration sensors and the

magnetic field. Then the position of the LCS (sensors) in the

GSC can be described by four vectors:

LCS

- the displacement vector of the origin of the LCS

relative to the GCS origin;

LCS

- directing vectors of the orthonormal

basis of LCS expressed regarding the directing vectors of the

orthonormal GCS basis.

Such description gives complete information about the

orientation of the LCS in the GCS in coordinate form. The

problem of determining the angular orientation reduces to

finding the coordinates of the vectors

LCS

in GCS.

Because the gravitational field is more stable than the

magnetic field, let us take as a basis the acceleration sensor.

The acceleration sensor readings are the coordinates of the

acceleration vector of the free fall, decomposed along the

axes of the LCS:

{

}

zyxLCS

aaaa ,,=

(3)

The normalized vector

LCS

is nothing else but a vector

GCS

, i.e., the defining vector of the OZ GSK axis, expressed

regarding the directing vectors of the LCS:

{}

zyx

LCS

GCS

kkk

k ,,==

(4)

The readings of the magnetic field sensor are the

coordinates of the magnetic field vector, decomposed along

the LCS axes:

{

}

zyxLCS

mmmm ,,=

(5)

Vector

LCS

in the general case, it may not be parallel to

the vector

GCS

: Therefore, it needed to get its normal

component:

()

LCS LCS LCS LCS LCS

nm mkk=− ⋅

GG G

(6)

The normalized vector

LCS

is nothing else but a vector

GCS

, i.e., The defining axis vector

GCS, expressed

through the directing vectors of the LCS:

{}

zyx

LCS

GCS

jjj

j ,,==

(7)

211

The defining axis vector

GCS, expressed through the

directing vectors of the LCS, is found using the vector

product:

{}

zyxGCSGCSGCS

iiikjj ,,=⋅=

(8)

We compose the matrix of the row represented by the

vectors

LCS

then transpose it and expand it into

vectors (also in rows):

zzzyyyxxx

zyxzyxzyx

kjikjikjikkkjjjiii =

(9)

Thus, the vectors:

{}

xxxLCS

kjii

{

}

yyyLCS

kjij =

(10)

{}

zzzLCS

kjik =

determining the axis of the LCS in the GCS, in other

words, determine the orientation of the LCS in the GCS.

Let's express the vector

LCS

in GCS:

{

}

x LCS y LCS z LCS x y z

aai aj ak aaa

′′′

=⋅ +⋅ +⋅ =

(11)

Vector readings are given in the quanta of the ADC of

the acceleration sensor, for further calculations we translate

them into the International System of Quantities (ISQ) and

write the instantaneous acceleration vector:

(12)

where the "range" is the range of the analog-to-digital

sensor converter, and N is the division price.

To take into account the gravitational field, it is necessary

to reduce the vertical component by the value of the

acceleration of gravity:

{

}

0, 0,

g phys

aa g=−−

(13)

Let's move from acceleration to speed:

00gg g

vadtv aTv=+≈Δ+

GGG G G

(14)

Let's move from speed to coordinates:

00gg g

rvdtr rTr=+≈Δ+

GG G G G

(15)

We use model makes it possible to determine the

orientation of the sensor relative to the global coordinate

system, associated with the physical features of the planet

and sensor readings. This method is to determine of linear

the integration movement of the sensor in the global

coordinate system and reconstruct the trajectory of the

movement of the user's mobile device.

ONCLUSION

We describe an approach to verifying video data received

by a mobile device. The proposed algorithm uses the

procedure of entering the Swype code using the movement

mobile camera. To improving the accuracy we use addition

information from different mobile sensors like

accelerometer, gyro, barometer, and GPS.

CKNOWLEDGMENT

This work was supported by PROVER project

(https://prover.io/).

EFERENCES

[1] Keller, Y., Averbuch, A., Israeli, M. “Pseudo polar-based estimation

of large translations, rotations and scalings in images,” IEEE trans. on

Image processing, 12-22 (2005).

[2] Sarvaiya, J. N., Patnaik, S., Kothari, K. “Image Registration Using

Log Polar Transform and Phase Correlation to Recover Higher Scale”

Journal of Pattern Recognition Research, vol. 7(1) (2012).

[3] Zitova, B., Flusser, J. “Image registration methods: a survey”,

Elsevier Image and Vision Computing, 977-1000 (2003).

[4] Foroosh, H., Zerubia, J. B., Berthod, M. “Extension of phase

correlation to subpixel registration.” IEEE trans. on image processing,

vol. 11(3), 188-200 (2002).

[5] Reddy, B. S., Chatterji, B. N. “An FFT-based technique for

translation, rotation, and scale-invariant image registration”, IEEE

Transactions on Image Processing, vol. 5(8), 1266–1271 (1996).

[6] Xu, H., Hua, G., Zhuang, J., & Wang, S. A. “A Frequency Domain

Approach to Fast and Accurate Image Registration”, in IEEE

International Conf. on Information and Automation, 340-345 (2009).

[7] Samritjiarapon, O., Chitsobhuk, O. “An FFT-Based Technique and

Best-first Search for Image Registration”, Communications and

Information Technologies, ISCIT, 364-367 (2008).

212