# Introduction

igital watermarking technology has taken an important position in the industries where copyright, identification, protection and authenticity of a digital content are crucial and priority. Subsequently it has drawn attention to the researchers for developing and improving various techniques of using watermarking in different applications. In a traditional watermarking the watermark is intended to be visible due to identifying the manufacturing company. In contrast to traditional one, the digital watermarking is intended to be imperceptible in the noise tolerant signal of image, audio and video file in such way that the host data does not distort. The multimedia data on the internet has been integrated in the global economy in such a way that it become subject to attack with various readily available tools found in the internet. As a result the watermark in the data requires resistance against various attacks. Watermarking need to be within an acceptable limit and robust when different types of processing is applied in digital content namely adding noise, cropping, compression, resizing [1]. Some attacks are considered as the state-of-the art watermarking attacks which are geometric attacks, cryptographic attacks, removal attacks, protocol attacks and removal attacks [2]. There are mainly two general methods for watermarking which are commonly known as spatial domain technique and frequency domain technique. In the spatial domain approach the information is embedded by altering the pixel value of the cover image. On the other hand the frequency domain method uses the technique of domain transformation. The later one is more robust than the former one. In this method several transformation techniques is available such as Discrete Cosine Transformation (DCT), Discrete Fourier Transformation (DFT) and Discrete Wavelet Transformation (DWT). The obtained frequency component is then modified to hide the watermark. The RGB, YIQ and YUV color space are used for embedding watermark in an host image. The Discrete Wavelet Transformation (DWT) function is combined with the Singular Value Decomposition (SVD) for embedding the watermark in the YUV color space. The RGB color space can be converted into YUV color space and the watermark is embedded into its channels [3]. In the proposed method, the R planes of the watermark is converted using thresholds. Then combined with SVD and DWT to embed data in frequency domain of cover image. The review of related works is given in section II. In section III proposed algorithm is given. The result and analysis of the proposed result is given in section IV.


# II.


# Review of Related Works

Some researchers have used DCT, DWT and SVD technique on the R, G and B components of a host and watermark image. The watermarking technique used with SVD overcomes the weakness that was found in other methods [4]. The technique of converting the RGB color components first into the YIQ color component and then embedding the watermark image into Y and Q color space is shown by Sun and Yu [5]. Gunjal and Mali proposed embedding watermark in all YIQ, RGB and YUV color channels [6].


# III.


# Proposed Method a) Algorithm for Embedding process

Step 1: The R, G and B planes of host image and watermark image are separated first. Then the R plane is used for embedding. 


# b) Extraction Process

The watermark and the cover picture are extracted applying the reverse process on the watermarked image. The extraction procedure is described below:

Step 1: Separate R wmi , G wmi and B wmi planes from the watermarked image.

Step 2: Decompose R wmi planes four times to receive W LL3 band.

Step 3: Apply SVD on W LL3 band and perform the equation S EWM = S WMI -S ORI to receive watermark. The SEWM is the extracted watermark. SWMI has come from watermarked image and SORI from cover image.

Step 4: Apply inversed SVD, inversed DWT and threshold technique to re-construct the watermark image. The extraction formula is exposed through the figure 2.

Step 5: Normalized Correlation is calculated between watermark and extracted watermark image using the following equation: (2) Where W is Watermark image and EW is extracted watermark image. Fig. 3 Step 6: Pick signal to noise ratio is calculated using the following equation:
(3) (4)
Where Wmi is Watermarked image and A is attacked image.


# IV.


# Experiment result

In our experiment we have used a watermark to embed into a cover image to construct the watermarked image. Figure -3 shows the watermarked image derived with Cover image and watermark image. The performance of the proposed algorithm is measured through the obtained values. The peak signal to noise ratio (PSNR) and normalized correlation (NC) are used as the performance criteria.

The table-1 describes PSNR values between watermarked image and attacked image. It also demonstrates the NC values of original watermark and the extracted watermark from attacked image.

The PNSR value shows the intensity of noise added on the watermarked image through different types of noise attacks. The lower is PSNR, the higher is the noise added in watermarked image. The higher is normalized correlation (NC) the better is the similarity between original and extracted watermark image.  From the statistic of the tables it shows that some of the attacks like Sobel, Prewitt, Log, Laplacian, Cropping and Gaussian High pass add higher noise values in the watermarked image, though the extracted watermark from it produce higher NC values. The higher NC values mean the better perceptibility of extracted watermark image. The NC values of this table indicate that the proposed algorithm in this paper is more robust against the above mentioned attacks.
N C= p i=1 q j=1 W r i,j EW r i,j +W g i,j EW g i,j +W b i,j EW b i,j p i=1 q j=1 W r i,j +W g i,j +W b i,j EW r i,j +EW g i,j +EW b i,j M SE = p i=1 q j=1 W mi r i,j ?A r i,j +W mi g i,j ?A g i,j +W mi b i,j A b i,j
The figure-4 demonstrate watermarked image after diverse attacks.

V.


# Conclusion

In this paper we have proposed an algorithm for embedding and extracting digital watermark on color image by adopting the technique of DWT and SVD. The color components of the host and watermark image are separated into R, G and B channels separately. Then the value of the watermark image is embedded into R planes. A reverse system is applied to retrieve the watermark.

The experimental results are analyzed with PSNR and NC values. The result shows the robustness of the algorithm against malicious and unintentional attacks. The implanted method has received a highest PSNR and NC values which are 8.035 and 0.9924 respectively. This indicate the strength of robustness of the applied algorithm in this paper.

In future experiment we will concentrate on the new technique to improve the perceptibility of the cover image.
23112![Fig. 1 : Embedding Process](image-2.png "Step 2 :DStep 3 : 1 )Fig. 1 :Fig. 2 :")
![DWT-SVD based Watermarking Technique for RGB Image decomposition increases the PNSR by reducing the effect of noise on the cover image.](image-3.png "F")
4![Fig. 4 : (a)Cropping,(b)Deleting,(c)Highpass,(d)Prewitt](image-4.png "Fig. 4 :")
			Non-Blind DWT-SVD based Watermarking Technique for RGB Image
		
		
			
			
			

				


* 
	
		DWT-SVD Combined Full Band Robust watermarking technique for color images in YUV color space
		
			ASanthi
		
		
			Thangavelu
		
	
	
		International Journal of Computer Theory and Engineering
		
			1
			4
			
			October2009
		
	




* 
	
		Attack modelling: Towards a second generation watermarking benchmark
		
			SVoloshynovskiy
		
		
			SPereira
		
		
			VIquise
		
		
			TPun
		
	
	
		Signal Processing, Special Issue on Information Theoretic Issues in Digital Watermarking
		
			2001
		
	




* 
	
		Strongly Robust And Highly Secured DWTSVD Based Color Imane Watermarking: Embedding Data In All Y,U,V Color Spaces
		
			BGunjal
		
		
			SMali
		
	
	
		I.J. Information Technology and Computer Science
		
			2012
		
	




* 
	
		Content based Image watermarking scheme using block SVD and Arnold transform
		
			SRadharani
		
		
			MValarmathi
		
	
	
		Electronics and Communication Systems (ICECS)
				
			2014
		
	




* 
	
		DWT based watermarking algorithm of color images
		
			GSun
		
		
			YYu
		
	
	
		Industrial Electronics and Applications
				
			2007, ICIEA 2007. May 2007
			
		
	




* 
	
		Comparative performance analysis of DWT-SVD based color image watermarking technique in YUV, RGB and YIQ color spaces
		
			BLGunjal
		
		
			SNMali
		
	
	
		International journal of computer theory and engineering
		
			3
			6
			December 2011