WT-DCT domain based image watermarking technique using SVM

Abstract

Digital watermarking technique can be utilized for the protection of digital images from unauthorized use and distribution. In this paper a robust image watermarking technique has been developed based on the combination of discrete wavelet transform (DWT) and discrete cosine transform (DCT). To achieve the improved robustness under various attack conditions, support vector machine (SVM) has been incorporated during watermark extraction. The 3-level DWT transformed DCT coefficients are modified to insert a particular binary watermark bit in the host image. The watermarking system provides an average imperceptibility of around 42.45 dB over 300 images. The algorithm also offers improved robustness against different types attacks including both geometrical and non-geometrical attacks. The imperceptibility has been observed in terms of peak signal to noise ration (PSNR) while robustness is carried out using normalized cross-correlation (NC) and bit error rate (BER). Experimental observation suggests that the developed technique may be suitable for the copyright protection applications of digital images.