Embedding and extracting watermark in video data

Patent metadata

Field	Value
Publication number	US-12579598-B2
Application number	US-202217951898-A
Country	US
Kind code	B2
Filing date	Sep 23, 2022
Priority date	Sep 1, 2020
Publication date	Mar 17, 2026
Grant date	Mar 17, 2026

Abstract

Disclosed in this application are a method for embedding a watermark in video data and apparatus, a method for extracting a watermark in video data and apparatus, a device, and a storage medium. The method for embedding the watermark includes: acquiring a target image frame in video data; performing time-frequency transformation on the target image frame to obtain target frequency domain data, the target frequency domain data comprising a matrix formed by frequency domain coefficients; changing the frequency domain coefficients in the target frequency domain data according to watermark data to obtain watermarked frequency domain data; performing inverse time-frequency transformation on the watermarked frequency domain data to obtain a watermarked image frame; and synthesizing watermarked video data according to the watermarked image frame.

First claim

What is claimed is:

1 . A method for embedding a watermark in video data, applied to a computer device, the method comprising:
converting a format of an original image frame in video data to obtain a YCrCb original image frame in a YCrCb format; converting image data of a Cb channel in the YCrCb original image frame from an integer type to a floating-point type; acquiring a target image frame comprising the image data of the Cb channel in the YCrCb original image frame in the floating-point type; performing time-frequency transformation on the target image frame to obtain target frequency domain data, the target frequency domain data comprising a matrix formed by frequency domain coefficients; changing the frequency domain coefficients in the target frequency domain data according to watermark data to obtain watermarked frequency domain data; performing inverse time-frequency transformation on the watermarked frequency domain data to obtain a watermarked image frame; synthesizing a YCrCb watermarked image frame from the watermarked image frame and image data of a Y channel and a Cr channel in the YCrCb original image frame; converting a format of the YCrCb watermarked image frame to obtain an RGB watermarked image frame in an RGB format; and synthesizing watermarked video data according to the RGB watermarked image frame.

2 . The method according to claim 1 , wherein the watermark data occupies k bits, and k is a positive integer; and
the changing the frequency domain coefficients in the target frequency domain data according to watermark data to obtain watermarked frequency domain data comprises: changing a frequency domain coefficient at an i th target position in the frequency domain data according to a value of the watermark data at an i th bit, the i th target position being a position in the matrix corresponding to the i th bit, i being a positive integer less than or equal to k; and repeating the foregoing operation to change frequency domain coefficients at k target positions in the frequency domain data.

3 . The method according to claim 2 , wherein the watermark data is binary data; and
the changing a frequency domain coefficient at an i th target position in the frequency domain data according to a value of the watermark data at an i th bit comprises: determining a first value range of frequency domain coefficients corresponding to 0 in the binary data and a second value range of frequency domain coefficients corresponding to 1 in the binary data according to a value range step size; and changing the frequency domain coefficient at the i th target position in the frequency domain data to a value in the first value range in response to the value of the watermark data at the i th bit being 0, i being a positive integer less than or equal to k; or changing the frequency domain coefficient at the i th target position in the frequency domain data to a value in the second value range in response to the value of the watermark data at the i th bit being 1.

4 . The method according to claim 3 , wherein the first value range comprises at least one first value interval, the second value range comprises at least one second value interval, and interval lengths of the first value interval and the second value interval are equal to the value range step size;
changing the frequency domain coefficient at the i th target position in the frequency domain data to a value in the first value range in response to the value of the watermark data at the i th bit being 0 comprises: changing the frequency domain coefficient at the i th target position in the frequency domain data to a midpoint value of a first value interval closest to the frequency domain coefficient in response to the value of the watermark data at the i th bit being 0, i being a positive integer less than or equal to k; and the changing the frequency domain coefficient at the i th target position in the frequency domain data to a value in the second value range in response to the value of the watermark data at the i th bit being 1 comprises: changing the frequency domain coefficient at the i th target position in the frequency domain data to a midpoint value of a second value interval closest to the frequency domain coefficient in response to the value of the watermark data at the i th bit being 1.

5 . The method according claim 2 , the method further comprising:
scrambling the watermarked video data to obtain scrambled watermarked video data; acquiring a scrambled watermarked image frame in the scrambled watermarked video data; performing time-frequency transformation on the scrambled watermarked image frame to obtain scrambled frequency domain data, the scrambled frequency domain data comprising a matrix formed by scrambled frequency domain coefficients; generating a scrambled watermark vector based on the scrambled frequency domain data; inputting the scrambled watermark vector into a watermark extraction model to output scrambled watermark data; calculating a loss value between the watermark data and the scrambled watermark data based on the watermark data; and training the watermark extraction model based on the loss value, wherein the watermark extraction model is configured to output the watermark data based on the inputted vector.

6 . The method according to claim 5 , wherein the generating a scrambled watermark vector based on the scrambled frequency domain data comprises:
generating a scrambled watermark vector at the i th target position based on scramble information and watermark information at the i th target position in the scrambled frequency domain data; and repeating the foregoing operation to generate k scrambled watermark vectors at the k target positions, wherein the watermark information comprises a scrambled frequency domain coefficient at the i th target position in the scrambled frequency domain data, scrambled frequency domain coefficients at n target positions around the i th target position, at least one of step sizes of the i th target position and an (i+1) th target position, n being a positive integer, i being a positive integer less than or equal to k.

7 . The method according to claim 6 , wherein the inputting the scrambled watermark vector into a watermark extraction model to output scrambled watermark data comprises:
inputting the scrambled watermark vector at the i th target position into the watermark extraction model to output a value the scrambled watermark data at the i th bit; and repeating the foregoing operation to obtain k values of the scrambled watermark data at the k bits.

8 . The method according to claim 6 , wherein the scrambling the watermarked video data to obtain scrambled watermarked video data comprises:
scrambling the watermarked video data to obtain the scrambled watermarked video data, the scrambling comprising at least one of compression, offsetting, or cropping, wherein the scramble information comprises at least one of compression information, offsetting information, or cropping information, the compression information comprises at least one of a compression ratio or a compression mode, the offsetting information comprises at least one of an offset amount or an offset position, and the cropping information comprises at least one of a cropping position or a cropping size.

9 . A method for extracting a watermark in video data, applied to a computer device, the method comprising:
converting format of an RGB watermarked image frame in watermarked video data to obtain a YCrCb watermarked image frame in a YCrCb format; converting image data of a Cb channel in the YCrCb watermarked image frame from an integer type to a floating-point type;

Assignees

• Tencent Tech Shenzhen Co Ltd

Inventors

• Ma Huaguan
• Chen Zhaohong
• Zhao Jie
• Cui Xiaodong
• Wang Lifu
• Huang Wei
• Liu Yicong
• Li Xin
• Pan Xiaoxia

Classifications

• G06T2207/20132

  Image cropping · CPC title

• G06T2207/20081

  Training; Learning · CPC title

• G06T2207/10024

  Color image · CPC title

• G06T2207/10016

  Video; Image sequence · CPC title

• G06T9/00

  Image coding (bandwidth or redundancy reduction for static pictures H04N1/41; coding or decoding of static colour picture signals H04N1/64; methods or arrangements for coding, decoding, compressing or decompressing digital video signals H04N19/00) · CPC title