An Image Watermark Insertion and Extraction Method Based on EDA-PSO

Abstract

An innovative image watermark insertion and extraction method based on EDA-PSO is proposed to improve both the imperceptibility and robustness of the image watermark. The insertion and extraction of the watermark is performed in the discrete cosine transform domain (DCT). In the process, Watson perceptual model is first applied to find the optimal embedding position, then EDA-PSO is employed to adjust the strength of the embedded watermark and finally a new fitness value is defined as the evaluation criteria based on imperceptibility and robustness. This method embeds the watermark information of varying strengths into the low and middle frequency coefficients with the maximum visual perception threshold of each DCT block according to the different results of EDA-PSO. The simulation results proves that the proposed scheme ensures the imperceptibility of the watermark and achieves high robustness against such attacks as JPEG compaction, GLPF, addition Gaussian noise, addition impulse noise and addition product noise. Keywords—image watermarking; high-quality particle distribution; Watson visual mode; robustness; imperceptibility