Watermarking Algorithm for Vector Point Clusters

This paper presents a digital watermarking algorithm for protecting vector point data from illegal use. Three rules of watermarking, i.e. usability, invisibility, and robustness, are taken into account, and corresponding measures, including shuffling image before embedding, selecting feature points, and utilizing the least significant bits (LSBs) as insertion positions, are employed to implement the rules. Based on the measures, an approach to watermark embedding for point data is developed. Firstly, the original watermark (an image) is shuffled; secondly the feature points of the data are obtained by an iterative process. then the watermark is embedded in the LSBs of the coordinates of the feature points. To detect the watermark, a similar process for obtaining the feature points in watermark embedding is carried out first; then the LBSs of the coordinates of the feature points are assigned to a one-dimensional array to form a bit chain; after this, a reverse process of watermark shuffling is done to form a two-dimensional image; finally the similarity degree between the image and the original watermark is calculated for judging whether the data contains the watermark or not. Our experiments show the technique can resist the attacks from data compressing and format changes.