Method and apparatus for preventing forgery of data, method and apparatus for detecting forgery of data

Patent metadata

Field	Value
Publication number	US-2022014800-A1
Application number	US-202117366570-A
Country	US
Kind code	A1
Filing date	Jul 2, 2021
Priority date	Jul 7, 2020
Publication date	Jan 13, 2022
Grant date	—

Abstract

A method for preventing forgery of data according to an embodiment includes determining a noise level based on metadata of original data, generating noise by applying the determined noise level to a preset noise pattern, generating transformed data of the original data by adding the generated noise to the original data, and transmitting the transformed data and the metadata to a server.

First claim

What is claimed is:

1 . A method for preventing forgery of data, comprising:
determining a noise level based on metadata of original data; generating noise by applying the determined noise level to a preset noise pattern; generating transformed data of the original data by adding the generated noise to the original data; and transmitting the transformed data and the metadata to a server.

2 . The method of claim 1 , wherein the metadata comprises data related to at least one of an attribute of the original data, a generation time of the original data, a location where the original data is generated, a device that generates the original data, and a user who generates the original data.

3 . The method of claim 1 , wherein the determining comprises determining the noise level from the metadata according to a method agreed upon with the server in advance.

4 . The method of claim 1 , wherein the preset noise pattern is a noise pattern selected from among a plurality of noise patterns agreed upon with the server in advance.

5 . The method of claim 4 , wherein the generating of the noise comprises:
selecting one of the plurality of noise patterns based on the metadata; and generating the noise by applying the determined noise level to the selected noise pattern.

6 . The method of claim 5 , wherein the selecting comprises selecting one of the plurality of noise patterns according to a selection method agreed upon with the server in advance, based on the metadata.

7 . The method of claim 5 , wherein the transmitting comprises transmitting, to the server, identification information on the selected noise pattern together with the transformed data and the metadata.

8 . A method for detecting forgery of data, the method comprising:
receiving transformed data of original data and metadata of the original data from a client device; determining a noise level based on the metadata; generating estimated data for the original data from the transformed data based on a preset noise pattern; estimating noise added to the original data to generate the transformed data, based on the transformed data and the estimated data; estimating a noise level corresponding to the estimated noise based on the noise pattern and the estimated noise; and determining whether the transformed data is forged based on the determined noise level and the estimated noise level.

9 . The method of claim 8 , wherein the metadata comprises data related to at least one of an attribute of the original data, a generation time of the original data, a location where the original data is generated, a device that generates the original data, and a user who generates the original data.

10 . The method of claim 8 , wherein the generating of the estimated data comprises generating estimated data for the original data from the transformed data by using a denoise model corresponding to the preset noise pattern.

11 . The method of claim 10 , wherein the preset noise pattern is a noise pattern selected from among a plurality of noise patterns agreed upon with the client device in advance.

12 . The method of claim 11 , wherein the generating of the estimated data comprises:
selecting one of the plurality of noise patterns based on the metadata; and generating estimated data for the original data from the transformed data by using a denoise model corresponding to the selected noise pattern.

13 . The method of claim 11 , wherein the receiving comprises receiving, from the client device, identification information on a noise pattern corresponding to the noise added to the original data together with the transformed data and the metadata; and
the generating of the estimated data comprises:
selecting a noise pattern corresponding to the identification information from among the plurality of noise patterns; and
generating the estimated data for the original data from the transformed data by using a denoise model corresponding to the selected noise pattern.

14 . The method of claim 8 , wherein the estimating of the noise level comprises estimating the noise level corresponding to the estimated noise from the noise pattern and the estimated noise by using a pre-trained noise level estimation model.

15 . The method of claim 14 , wherein the estimating of the noise level comprises:
concatenating an image corresponding to the preset noise pattern to an image corresponding to the estimated noise; dividing the concatenated images into preset image patch units; and estimating a noise level for each region included in a sliding window of a preset size by using the noise level estimation model while sequentially moving the sliding window in the concatenated images, and the determining comprises determining whether the transformed data is forged based on the estimated noise level and the determined noise level for each region included in the sliding window.

16 . An apparatus for preventing forgery of data, comprising:
a noise level determiner configured to determine a noise level based on metadata of original data; a noise applier configured to generate noise by applying the determined noise level to a preset noise pattern, and generate transformed data of the original data by adding the generated noise to the original data; and transmitter configured to transmit the transformed data and the metadata to a server.

17 . The apparatus of claim 16 , wherein the metadata comprises data related to at least one of an attribute of the original data, a generation time of the original data, a location where the original data is generated, a device that generates the original data, and a user who generates the original data.

18 . The apparatus of claim 16 , wherein the noise level determiner is further configured to determine the noise level from the metadata according to a method agreed upon with the server in advance.

19 . The apparatus of claim 16 , wherein the preset noise pattern is a noise pattern selected from among a plurality of noise patterns agreed upon with the server in advance.

20 . The apparatus of claim 19 , wherein the noise applier is further configured to select one from among a plurality of noise patterns based on the metadata and generate the noise by applying the determined noise level to the selected noise pattern.

21 . The apparatus of claim 20 , wherein the noise applier is further configured to select one from among the plurality of noise patterns according to a selection method agreed upon with the server in advance, based on the metadata.

22 . The apparatus of claim 20 , wherein the transmitter is further configured to transmit, to the server, identification information on the selected noise pattern together with the transformed data and the metadata.

23 . An apparatus for detecting forgery of data, comprising:
a receiver configured to receive transformed data of original data and metadata of the original data from a client device; a noise level determiner configured to determine a noise level based on the metadata; a noise estimator configured to generate estimated data for the original data from the transformed data based on a preset noise pattern and estimate noise added to the original data to generate the transformed data based on the transformed data and the estimated data; a noise level estimator configured to estimate a noise level corresponding to the estimated noise based on the noise pattern and the estimated noise; and a determine

Assignees

• Samsung Sds Co Ltd

Inventors

• Jeong Yong Hyun
• Kim Do Yeon
• Hong Min Ki
• Park Chang Hyun
• Park Se Hyeon
• Yoon Kwan Sik
• Hwang Sol Bi
• Lee Jae Hyeon
• Park Hyun Chul

Classifications

• G06F18/214

  Generating training patterns; Bootstrap methods, e.g. bagging or boosting · CPC title

• G06V20/95

  Pattern authentication; Markers therefor; Forgery detection · CPC title

• G06T2201/0051

  Embedding of the watermark in the spatial domain · CPC title

• G06T1/0028 Primary

  Adaptive watermarking, e.g. Human Visual System [HVS]-based watermarking · CPC title

• H04N21/440254 Primary

  by altering signal-to-noise parameters, e.g. requantization · CPC title