Adaptive and Robust Fourier-Mellin-Based Image Watermarking for Social Networking Platforms

Abstract

According to the Buckets effect, the capacity of a bucket depends on the length of the shortest board. This principle also applies to social networking platform resilient (SNPR) image watermarking, which should be comprehensive and free from significant shortcomings. In the frequency domain, the watermarked region is formed using log-polar coordinate mapping (LPM) and has a ring-like structure. However, this structure cannot be stretched or compressed, and it causes a streaking effect at the edges of the watermarked image. These issues have been addressed in the proposed method. Specifically, an adaptive optimization framework is used to adjust the embedding strength and range of the watermark, and multiple synchronization strategies are adopted to correct flip and aspect ratio. Compared with state-of-the-art works, the proposed method significantly improves the imperceptibility of the watermarked image and its robustness to various distortions and lossy transmission on social networking platforms (SNPs).