Digital Watermarking Using 2-DCT

: Digital Watermarking is the process of hiding a message into the digital file like image, audio, video, etc. for the protection and authentication of the content. Sometimes it is related with steganography, as both are used for hiding the message but difference is that in steganography there is no relation between digital file and message, it is only used to hide the message existence, on the other hand with watermarking one hides a message related to the digital content. Digital Watermarking consist of several characteristics like Invisibility, Robustness, Readability and Security from any unauthorized user.


Spatial Domain Method
Spatial Domain Method directly transforms the raw data into the original image. It can also implemented by using color separation so that watermark will appear in one of the color bands which will be difficult to detect by naked eye. But it can be made visible during printing by separating colors. This technique basically changes the image representation of object to enhance image for different kind of applications. This approach is mostly used by the journalists for inspecting the digital pictures. Spatial domain consists of following algorithms:

SSM Modulation Based Technique
Spectral Spectrum techniques generate energy over discrete frequencies which is distributed at time. This modulation technique helps in increasing robustness against natural interference, jamming and watermark detection, establish a secure communication. When SSM based watermarking is applied to the image, it embed the information by linearly combining the image with the embedded watermark modulated small noise signal.

Least Significant Bit(LSB)
In this technique, watermark is embedded in the LSB of the pixels. As we know pixel is represented by 8-bit sequence, we embed the watermark into the least significant bit in the selected pixels of the image. It is easier to implement and doesn't cause any major distortion in the image but it is not used commonly as it is not robust against different kind of attacks.
II. FREQUENCY DOMAIN TECHNIQUES Frequency-domain methods are most widely used as compared to spatial-domain methods. This technique aims to embed the watermarks in the spectral coefficients of the image. Discrete Cosine Transform(DCT),Discrete Wavelet Transform(DWT),Discrete Fourier Transform(DFT) are most commonly used transforms under this technique. Frequency Domain method use the property that Human Visual system (HVS) are better captured by the spectral coefficients. For example, HVS is less sensitive to the high frequency and more sensitive to low frequency coefficients. Hence, alterations to the low frequency components may cause distortion to the original image and high frequency coefficients are considered insignificant, hence processing techniques, such as image compression and watermarking tend to remove high frequency coefficients. Due to this reason, most of the algorithms embed watermarks in the midrange frequencies to maintain the balance between robustness and imperceptibility.

Discrete Cosine Transformation (DCT)
Just like Fourier transform, DCT represents data in terms of frequency space instead of amplitude space.This technique is more robust to image processing operations like blurring, brightness, low pass filtering,etc as compared to spatial domain techniques but it'sweak against geometric attacks like scaling, rotation, cropping etc. However, these are difficult to implement and computationally more expensive. It embeds the watermark into significant portion of the image as most of the compression schemes remove the insignificant portion of the image. It can be classified into two type's i.e Global DCT watermarking and Block based DCT watermarking.

Discrete Wavelet Transformation (DWT)
The Discrete Wavelet Transform decomposes the image into sub-image of different spatial domain and independent frequencies. It is used in different kind of signal processing applications, such as removing noise in audio, audio and video compression, etc. Wavelets consist of their energy concentrated in time and are very well suited for analysis of transient, time-varying signals. One of the major challenge watermarking faces is to balance between perceptivity and robustness. If we increase the strength of the embedded watermark, robustness increases but it may increase visible distortion. However, DWT is preferred as it provides both spatial localization and a frequency spread of the watermark within the given image [6].
III. METHOD: Most of the lossy compression uses transform compression like JPEG (Joint Photographers Experts Group) for the image encoding. Lossy Compression involves sampling and/or quantization by reducing number of bits per sample or ignoring some of the samples as a result the size of data file may reduce. The DCT uses cosine waves to present a signal unlike Fourier transform which uses both sine and cosine waves. DCT is taken in 8*8 group form which results in an 8*8 spectrum. As DCT is designed to work over the pixel values of range -128 to 127.After quantization, the low frequencies are present in the upper-left corner of the spectrum and high frequencies reside in the lower right. Later, the remaining coefficients will only be considered for reconstructed image.

Waterma
Algo Step Step Step Step Step Step Wate Algo Step Step Step Step Step Expe VI.

CONCLUSION
In this paper we watermarked the images by our choice using 2 dimensional discrete cosine transform. We calculated MSE and PSNR to compare original input image and reconstructed image. This watermarking algorithm provides a better quality picture as we are reduce the coefficient of watermark. The above algorithm can be used to watermark the image that is used in the web applications and where we need different copyright over an image as it need less processing time. Furthermore in future we can analyze different image transform algorithms for improvement of different parameters.