Copy-Right Protection with Wavelet Based Watermarking

Copy-Right Protection with Wavelet Based Watermarking

Adnan Özdemir

Introduction

In this project a watermarking method which is working on frequency domain was generated and implemented.

In order to protect the copy-right laws, a unique QR code was inserted to the remote sensing images.

• Wavelet

The project has several targets:

Generating watermarking algorithm with existing approaches,

Generating a method which is resistant to resize,

(Minimum 1/16 resizing)

Successfully regenerating the data and image from watermarked image,

Executing an efficient useful low time consuming program.

Problems at Watermarking MethodsActually, watermarking is easy to implement any simple algorithm. Operation on

pixel values at time domain allows to programmer inserting special key in a image.

At that type of methods it is important to what type of destruction or corruption are exposed to image. In other words, if the pixels which consist watermarked values, are changed unpredictably the inverse transformation of image watermarking will

not produce any meaningful results.

For example, if the key values are located at small area of images, after deleting this area or corrupting this area will cause losing watermarking knowledge.

In order to minimize the locality, the key informations can be inserted periodical. However, at this case,

resizing operations on image will damage the watermark informations

All the watermarking methods which are operated at time domain have weakness with process at time domain.

Therefore, frequency domain watermarking operation are generated. Specially discrete time Fast Fourier Transformation discrete time and

wavelet transformation are commonly used.

Main idea is minimizing the losing information at image and generate a watermarked image which can resist most of the process at time

domain.

Due to do that approach, low frequency components are extracted and the the key components are located. This approach can resist to

the resizing and most of local operations.

Even using low frequency variables, the image loses details. Depending to image, this data losing causes destruction and

low resolute images.

In order to keep the image with high quality, low frequency bands should be selected wisely.

The key values should effect the image minimum. At this case, wavelet methods shows themselves.

If the image band processed with wavelet transformation with several times, lowest frequency elements will be separated.

The key values are placed at this frequency bands. This approach is called as "Wavelet Tree Quantization".

DWT2 and IDWT2

DWT2 and IDWT2

Landsat 8

• Each pixel is sampled by 16 bits.• OLI Spectral Bands• Spectral Band WavelengthResolution• Band 1 - Coastal / Aerosol 0.433 - 0.453 µm 30 m• Band 2 - Blue 0.450 - 0.515 µm 30 m• Band 3 - Green 0.525 - 0.600 µm 30 m• Band 4 - Red 0.630 - 0.680 µm 30 m• Band 5 - Near Infrared 0.845 - 0.885 µm 30 m• Band 6 - Short Wavelength Infrared 1.560 - 1.660 µm 30 m• Band 7 - Short Wavelength Infrared 2.100 - 2.300 µm 30 m• Band 8 - Panchromatic 0.500 - 0.680 µm 15 m• Band 9 - Cirrus 1.360 - 1.390 µm 30 m

Marking Process

DWT2Haar

kernel

Marking Process

IDWT2Haarkernel

Marking Process

PSNR = 128.2023 db

Exracting The QR

DWT2

Test

Test

Thresholding? (Otsu)

• Kullanım Alanları ÖrnekleriLig tvSpotifyNetflix

References

[1] Ruanaidh, JJK Ó., W. J. Dowling, and F. M. Boland. "Watermarking digital images for copyright protection." IEE Proceedings-Vision, Image and Signal Processing 143.4 (1996): 250-256.

[2] Wang, Shih-Hao, and Yuan-Pei Lin. "Wavelet tree quantization for copyright protection watermarking." Image Processing, IEEE Transactions on 13.2 (2004): 154-165.

[3] Lin, Shinfeng D., and Chin-Feng Chen. "A robust DCT-based watermarking for copyright protection." Consumer Electronics, IEEE Transactions on 46.3 (2000): 415-421.

[4] Iyer, Ms Aparna, et al. "IMAGE RETRIEVAL USING COLOUR AND TEXTURE ANALYSIS." IMAGE 2.5 (2013).

Questions?