A Complex Network Analysis of Image Watermarking Scheme Based on SVD and DWT

An investigation of the impact that watermarking has on image statistics is presented here. This paper provides a fresh methodology that is based on the degrees of visibility graphs, in contrast to other methods that investigate variations in intensity levels. The experiment utilizes the use of twelve standard images in addition to three watermarks. The findings indicate that the conventional method of intensity difference analysis produces distributions that are inconsistent across images. As an alternative, the method that has been offered, which makes use of degrees of visibility graphs, displays normal distributions that are more consistent across all images, regardless of the watermark. In light of this, it appears that the degrees of visibility graph provides a promising independent metric for watermarking analysis that is independent of both the image and the watermark. According to the findings, the difference in degrees between the original and watermarked images occurs according to a normal distribution, with the mean falling somewhere between \(-\) $-$ 0.98 and \(-\) $-$ 25.93. This range appears to be quite broad, and the reason for this is the outlier. After the outliers have been removed, this range is now \(-\) $-$ 0.98 to \(-\) $-$ 3.54 degrees. Overall, the study indicates that analyzing the degrees of visibility graph is a more robust and informative method of evaluating the effectiveness of watermarking than directly comparing intensity levels.