•  
  •  
 

Abstract

In today's digital landscape, the need for secure transmission of processed data is more crucial than ever. Data sent over open networks is vulnerable to numerous security threats, making content protection a key aspect of reliable communication. One proven method for safeguarding data is digital image watermarking, which embeds hidden information within media files such as audio, video, or images to protect them from unauthorized access. By examining the degradation of the embedded data, potential attacks can be identified. This paper introduces an enhanced digital image watermarking technique that employs an optimal scaling factor, determined by the Firefly Optimization algorithms, and integrates Convolutional Neural Networks (CNN) to bolster security. The process starts by performing a non-subsampled contourlet transform (NSCT) on the high-entropy sub-band of the original image. Next, a redundant wavelet transform (RDWT) and multiresolution singular value decomposition (MSVD) are applied to the high-frequency coefficients obtained from the NSCT. Two watermarks are embedded into the designated frequency bands using the optimized scaling factor from the Firefly Optimization algorithm, refined through CNN. To further secure the image, the Henon Map encryption algorithm is employed. A variety of cover images were tested to assess the system's performance. The results demonstrate that this method stands out in terms of invisibility, security, robustness, and its ability to embed watermarks effectively, making it particularly suitable for copyright protection. A comprehensive evaluation based on several performance metrics confirms the method's effectiveness, setting a new benchmark for digital watermarking techniques.

Share

COinS