Multi image watermarking using Lagrangian support vector regression

Watermarking has been used widely for passing the confidential information over Internet, a good watermark will be imperceptible, robust over the various geometric and non-geometric attacks. The watermark shouldn't be easily extracted by the hacker, to achieve this several watermarking methodologies were developed for different domains namely spatial, frequency and wavelet domains using DCT, DWT and fourier transforms. Here we perform wavelet domain watermarking using Lagrangian Support Vector Regression method. Lagrangian coefficient is to find the minimum point in a line, which uses support vector machine and regression concept for data classification. Arnold Transform is used for image scrambling of the watermark image. The advantage of Arnold transform is the frequency response is same even after scrambling. We purpose a multi watermarking in the host cell using LSVR for a colour image, which showed good PSNR value. We see that LSVR supports multi image watermarking techniques.