Вы находитесь на странице: 1из 20

DWT BASED VIDEO WATERMARKING USING FRAME SEPARATION

TECHNIQUE

Department of Computer Science and Engineering


VSSUT Burla, Odisha

Guided by: Submitted by:


Mrs. Alina Dash Shakti Prasad Mishra
Asst. Professor (1602040025)
Content

● Introduction
● Objective
● Literature Review
● Work Done
● Conclusion
● References
Introduction
● Due to advanced image and signal processing technologies illegal reproductions and
redistributions of digital media has become a serious issue.
● Digital watermarking is a data masking technique used for asserting copyright of
media.
● Digital watermarking is divided into various types and domains. Domains are named
as spatial domain and frequency domain.
● Spatial domain: Changes/alters pixels of some random frames of the video.
● Frequency domain: Certain frequencies/signals are changed in the original frame.
Objective

● Developing an algorithm, that can be used for securing video


sequences from piracy.
● Algorithm extracts frames of the original video channelize some
specific ones and alters the blue component.
● DWT used for extraction of mid frequency signal coefficients for
alteration.
Literature Review
Year Authors Contribution

2015 Priya Chandrakar, Used single key to randomize frames from a video before DWT
Mrs. Shahana Gajala making it secure. Various factors like PSNR, MSE and
Qureshi Normalization coefficients was calculated.

2017 T. Shankar, Used an algorithm called “Oppositional Krill Herd” to find the
G. Yamuna optimal wavelet before embedding watermark. For more
security the watermarked images were also encrypted.

2017 Yassine Himeur, Used an algorithm that encrypts randomly selected frames of a
Abdelkrim Boukabou video using chaotic encryption before watermark embedding.
Used shot detection process to avoid redundancy and
meaningless frames.
Work Done

In this project with the help of various libraries like cv2, os and math etc.
of python programming language the algorithm is carried out to add
watermark in the video. Generally the methodology is divided into two
parts that is of encryption or embedding and extraction.
Encryption Flow Diagram
Continued...

For encryption process the following steps were followed:

Step 1: The video of .mp4 format was taken as input for watermarking process.

Step 2: The video was divided into frames using opencv libraries and using KEY1

the frames were selected for channeling process.

Step 3: The frames were split into RG and B components and those frames were

stored on different directories.

.
Continued...
Step 4: 3 - DWT was applied on the B components of the images to get 4 signal

arrays out of which the mid frequency arrays were selected for encryption.

Step 5: Using KEY2 pseudo random numbers were generated for encryption

according to the formula:


Continued...
Step 6: Next using inverse DWT the signal was merged and the watermarked blue

image was merged to its corresponding RG components.

Step 7: The newly made frames merged to the previously unselected frames to

make the final watermarked video


Results

In this project an unmarked and unsafe video sequence was digitally


watermarked using the above algorithm which added an invisible
watermark to the project directly on the mid frequency signals of the blue
channel of the frames.
Outputs

First of all frames were extracted from the video.


Continued...
Next random frames were selected to be used for channelization using the KEY1 i.e. the frame number is
divisible by 15.
Continued...

After channeling the blue images and reg-green images were kept separately.
Continued...
Continued...
3 - DWT implemented on a single blue frame gave the following coefficients.
Continued...
Pseudorandom numbers were generated using the key: X n+1 = (aX n + b) mod m

where value of a = 35, b = 528, m = 28 and X 0 = 1073.


Conclusion

This project throws light upon the watermarking techniques to prevent data
theft so that the data ownership is preserved and theft doesn’t occur. The
watermarking here is done on a video to hide relevant information in the
blue component of specific frames in a given video. Those frames are later
used to denote secure frames or watermarked frames to prevent the data
from being pirated. This can also be implemented in a further study where
the various frames might have different components watermarked and
secured for end to end encryption.
References

1. Balamurugan, Dr. K.B Jayarraman, Arulalan. V (2014) - A Survey on Medical Image Watermarking
Techniques.
2. Priya Chandrakar, Mrs. Shahana Gajala Qureshi (2015) - A DWT Based Video Watermarking Using
Random Frame Selection.
3. Operations with images.
(Online)https://docs.opencv.org/3.4/d5/d98/tutorial_mat_operations.html
4. T.Shankar, G.Yamuna (2017) - An optimised DWT based approach in video watermarking with
multiple watermarks using oppositional Krill Herd algorithm.
5. Yassine Himeur, Abdelkrim Boukabou (2017) - A robust and secure key-frames based video
watermarking system using chaotic encryption.
6. PyWavelets - Wavelet Transforms in Python. (Online)https://pywavelets.readthedocs.io/en/latest/

Вам также может понравиться