查看文章

Digital watermarking methods have exceptional significance in protecting the copyright and authenticating multimedia security. An adaptive blind color image watermarking approach using a binary robust invariant scalable key (BRISK) features based on convolutional neural network and contourlet transform is designed. Arnold cat transform encrypts the watermark and is embedded in the transform domain by utilizing contourlet transform and QR decomposition. In calculating the embedding factor value, adaptively, BRISK features are used. The network is standardized based on the relationship between the watermark, cover image, and watermarked image. The proposed watermarking system is tested by applying image and signal processing attacks. From the simulation outcomes, it is clear that the proposed system has good performance, robustness, security & imperceptibility. The extracted watermark can be …