Multiobjective Image Data Hiding Based on Neural Networks and Memetic Optimization

This paper presents a hybridization of neural networks and multiobjective memetic optimization for an adaptive, robust, and perceptual data hiding method for colour images. The multiobjective optimization problem of a robust and perceptual image data hiding is introduced. In particular, trade-off factors in designing an optimal image data hiding to maximize the quality of watermarked images and the robusteness of watermark are investigated. With the fixed size of a logo watermark, there is a conflict between these two objectives, thus a multiobjective optimization problem is introduced. We propose to use a hybrid between general regression neural networks (GRNN) and multiobjective memetic algorithms (MOMA) to solve this challenging problem. Specifically, a GRNN is used for the efficient watermark embedding and extraction in the wavelet domain. Optimal watermark embedding factors and the smooth parameter of GRNN are searched by a MOMA. The experimental results show that the propsed approach achieves adaptation, robustness, and imperceptibility in image data hiding. Key–Words: Information hiding; image data hiding; image watermarking; multiobjective optimization; memetic optimization; general regression neural networks; wavelet transforms; human visual system; quality metrics.