A system for identifying an anti-counterfeiting pattern based on the statistical difference in key image regions

Abstract An important aspect of identifying whether items are likely to be forged or infringe upon a copyright is the use of an anti-counterfeiting pattern because it can help in determining and detecting forged anti-counterfeiting labels and patterns. At present, most anti-counterfeiting systems require special materials or large samples of images for training. Once they are geometrically attacked or forged by printing and scanning, they will be completely invalid. Therefore, this paper proposes an anti-counterfeiting system that uses a single corresponding feature difference sequence of the key regions for statistical analysis. The aim of our technology is to use the inks to generate random subtle texture patterns, and construct a supervised and guided segmentation algorithm and bone width transformation algorithm to locate the key regions of the sample images, which is used to identify the authenticity of the inspected product. The experiment shows that the system not only has high anti-counterfeiting performance and good robustness but also provides a convenient and practical idea for anti-counterfeiting technology.