A hybrid digital image watermarking technique based on fuzzy-BPNN and shark smell optimization

Digital Image Watermarking (DIW) has recently gained attention due to the increase in copyright issues. The robustness, imperceptibility, and security in the prevailing techniques are not efficient because of the poor selection of embedding parameters. Therefore, we make use of the benefits from some techniques like Human Visual System (HVS), Fuzzy Inference Systems (FIS), Back Propagation Neural Network (BPNN), and Shark Smell Optimization (SSO) to develop a robust as well as secure watermarking system. HVS helps to identify the areas with noise that may not be identified by the human eye. Therefore, by using this technique, the information can be made invisible to the human eye by adjusting the values of the pixel. The FIS and BPNN are then used to get the weight factor. Moreover, the introduction of the SSO algorithm helps to attain the ideal embedding parameter. The SSO is adopted based on the hunting behavior of the shark. Sharks have the capability to identify the prey very quickly even in large search spaces. This makes the shark a superior hunter. Therefore, the SSO can obtain optimal embedding parameters compared to other existing techniques such as Particle Swarm Optimization (PSO), Artificial Bee Colony (ABC) optimization, etc. At the end of this process, the watermarked image will be obtained. The attacks are then applied to the watermarked images and finally, the watermark extraction process takes place in which the watermarks are extracted from the attacked image. The conducted performance analysis reveals that the proposed hybrid Fuzzy-BPNN with SSO approach outperforms other recent techniques.

Automated summary

Digital Image Watermarking (DIW) has gained attention due to copyright issues. This study proposes a robust and secure watermarking system using Human Visual System (HVS), Fuzzy Inference Systems (FIS), Back Propagation Neural Network (BPNN), and Shark Smell Optimization (SSO). The results show that the proposed hybrid approach outperforms other recent techniques.