Robust Image Watermarking Using Support Vector Machine and Multi-objective Particle Swarm Optimization

Multimedia data security is paramount due to the high traffic on the internet. The internet traffic compromised with security threats such as intentional and time differential attack of multimedia data. This paper proposed an image watermarking method based on support vector machine and particle swarm optimization. The proposed algorithm works in dual mode as a selection of the embedding location of the watermark and increases the randomness factor of the watermark image. The increase randomness factor increases the value of imperceptibility and robustness factor. The support vector machine classifies the pattern of the host image and the watermark image. The processing of feature components of an image extracted by stationary wavelet transform (SWT). The stationary wavelet transform is also called redundant wavelet transform; these transforms overcome the discrete wavelet transform's limitation. The extraction of watermark image applies various attacks and measures standard parameters such as SSIM (similarity index matrix) and NC (number of correlation). The art of analysis of the proposed algorithm compares with SVM and DWT based watermarking methods. The proposed algorithm simulated in MATLAB software and tested with a reputed image dataset such as cameraman, peppers, baboon and Lena. The support vector machine applied the non-linear kernel function. The proposed algorithm decreases the 5-8% risk of geometrical attacks.