Application Research on Support Vector Machine in Image Watermarking

As the good learning ability and generalization capacity, support vector machine (SVM) which based on the statistical learning theory is increasingly noticed by many researchers. Although SVM has many prominent theoretical advantages, its application still drops behind. In this paper, A new application of SVM is studied. Firstly, the strategy of embedding and extracting watermark using support vector regression (SVR) from a digital image is given. Then, the influences of SVR-learning parameters on the watermarking performance are analyzed, and the ideal values range of SVR-learning parameters for different images is given respectively. Finally, the results are validated with other images and compared with the similar method. Experimental results show that SVM can be successfully fused with conventional watermarking technique to improve the robustness and imperceptibility of watermark with the help of sound SVR-learning parameters.