**2. Proposed techniques**

We have proposed a blind invisible watermarking technique for copyright protection of the colored images.

In blind techniques, during the extraction process original image is not required. Watermarking systems which involve marking imperceptible alteration on the cover data to convey the hidden information, is called invisible watermarking. Here 512\*512 grayscale image of 'peppers' is taken as host image and 32\*32 binary image is taken as watermark image.

Then implement second level wavelet transform on host image using wavelet function 'haar' and extracted middle level components (HL2, LH2) for embedding watermark. Middle level components are selected for embedding watermark as much of the signal energy lies at low-frequencies sub-band which contains the most important visual parts of the image and high frequency components of the image are usually removed through compression and noise attacks.

We have divided the HL2 and LH2 bands in to 4x4 blocks and applied DFT in these blocks and used to two highly uncorrelated pseudo random sequences (treated as key: key1) to embed watermarking message according to template matrix. We reshaped watermark image of 32x32 into a row vector of size 1024, called watermark message.

A template matrix is such a matrix whose size is 4x4 and elements are 0 and 1. Watermarking message is embedded into image blocks only in the position where the template matrix's element is 1. Through amounts of experiments, we found when the template matrix is set to [1,1,1,1;1,0,0,1;0,0,0,1;0,0,0,1]T,imperceptibility and robustness of the algorithm can get better balance. Here template matrix is used as a key2.

We have embedded watermark according to the (9) given below.

$$\mathbf{I}\_{\mathbf{w}}(\mathbf{x}, \mathbf{y}) = \mathbf{I}(\mathbf{x}, \mathbf{y}) + \mathbf{k} \ast \mathbf{W}(\mathbf{x}, \mathbf{y}) \tag{9}$$

In (9), *k* denotes a gain factor, and IW the resulting watermarked image, *I* the cover image and W the watermark to be embedded. Increasing *k* increases the robustness of the watermark at the expense of the quality of the watermarked image.

The algorithm for the proposed method is given below:

The watermark embedding steps are as follows:


We have proposed a blind invisible watermarking technique for copyright protection of the

In blind techniques, during the extraction process original image is not required. Watermarking systems which involve marking imperceptible alteration on the cover data to convey the hidden information, is called invisible watermarking. Here 512\*512 grayscale image of 'peppers' is taken as host image and 32\*32 binary image is taken as watermark

Then implement second level wavelet transform on host image using wavelet function 'haar' and extracted middle level components (HL2, LH2) for embedding watermark. Middle level components are selected for embedding watermark as much of the signal energy lies at low-frequencies sub-band which contains the most important visual parts of the image and high frequency components of the image are usually removed through compression and

We have divided the HL2 and LH2 bands in to 4x4 blocks and applied DFT in these blocks and used to two highly uncorrelated pseudo random sequences (treated as key: key1) to embed watermarking message according to template matrix. We reshaped watermark

A template matrix is such a matrix whose size is 4x4 and elements are 0 and 1. Watermarking message is embedded into image blocks only in the position where the template matrix's element is 1. Through amounts of experiments, we found when the template matrix is set to [1,1,1,1;1,0,0,1;0,0,0,1;0,0,0,1]T,imperceptibility and robustness of the

In (9), *k* denotes a gain factor, and IW the resulting watermarked image, *I* the cover image and W the watermark to be embedded. Increasing *k* increases the robustness of the

Implement second level wavelet transform on host Image H using wavelet function

Divide the HL2, LH2 components in several blocks of size 4x4 and DFT is applied to

Perform search to find highly uncorrelated pseudo random (PN) sequences (seq\_zero

Set gain factor K and embed the watermark to the cover image under the following

I x, y I x, y k W x, y <sup>w</sup> (9)

image of 32x32 into a row vector of size 1024, called watermark message.

algorithm can get better balance. Here template matrix is used as a key2.

We have embedded watermark according to the (9) given below.

watermark at the expense of the quality of the watermarked image.

'Haar' and Extract middle frequency components (LH2, HL2).

The algorithm for the proposed method is given below:

The watermark embedding steps are as follows:

and seq\_one) and use these as a key1. Defines the template matrix of an 4x4.

these blocks.

If wa(i,j) ==0 then

rule:

**2. Proposed techniques** 

colored images.

image.

noise attacks.

```
If template(m,n)==1 then 
I(m,n)=I(m,n) + K*seq_zero(m,n) 
End 
Else 
if template(m,n)==1 then 
I(m,n)=I(m,n)+K*seq_one(m,n) 
End 
End 
Where 1≤ i ≤ M, 1≤ j ≤ N, and 1≤ m,n ≤ 4 
Here I denotes to 4x4 DFT blocks.
```

The watermark extraction steps of this technique are as follows:


```
If corr_zero(i) > corr_one(i) then 
watermark_detected(i)=0; 
Else 
watermark_detected(i)=1
```
End

