Teaching-Learning Based Optimization of Dual Watermarking of Color Images

Digital watermarking has proven to be a promising solution for copyright protection. However, several issues relating to robustness and tampered area detection remain unresolved. In this article, two watermarks are embedded in a color image. Using the Stationary Wavelet Transform (SWT) and Singular Value Decomposition (SVD), a gray-scale image is embedded as one of the watermarks for ownership identification. The least significant bit approach is used to embed a binary watermark to locate the tampers. Teaching Learning Based Optimization (TLBO) is employed to boost the watermark's robustness. A new fitness function is proposed for color image watermarking. This technique is tested on both standard and medical images. The performance of this method is evaluated in terms of imperceptibility, robustness, and embedding capacity.