Research On Anti-copying 2-D Code Algorithm By Exploiting Illegal Channel Noise Characteristics

With the rapid development of two-dimensional codes(2-D bar codes),especially in the applications of mobile payment and identity authentication,there are more security challenges,such as embedding attacks,tampering attacks and copying attacks.Note that copying attacks are most challenging to defend among these attacks,because of the optical visibility of 2-D bar code.Copying attacks not only hinder the promotion of identity authentication applications,but also inevitably cause economic losses to merchants and customers.Prior approaches for anticopying attacks mainly include the use of special printing materials or processes,the use of physical non-cloning features,the use of anti-copy patterns and the use of digital watermarking techniques.Although these approaches defend copying attacks to a certain extent,they have the following drawbacks: increase the cost of producing 2-D bar codes,increase the customers' cost,and low portability.Therefore,to design a new anti-copying 2-D bar code for overcoming the above drawbacks becomes a new research hotspot.This thesis proposes two new anticopying 2-D bar code algorithms based on the noise characteristics of illegal channel.First,by exploiting the difference of noise characteristics between the legal channel and the illegal channel,we propose a low-cost anti-copying 2D barcode(LCAC 2-D bar code)algorithm.The basic idea behind this algorithm is to embed an authentication tag which cannot sacrifice the decodability of the original 2-D bar code.The legal receiver extracts the feature of the authentication tag in the scanned the 2-D bar code to identify whether it is copied.The LCAC 2-D bar code includes two embedding strategies: sequential embedding and segment embedding.We analyze the performance of two embedding strategies in terms of security,covertness,and robustness.Through actual experimental results,we prove that the segment embedding strategy has better robustness than the other one,if there is any shelter in the scanning of the 2-D bar code or some parts of the 2-D bar code itself is fuzzy.In order to improve the practicability of the LCAC 2-D bar code,besides direct-copying attack(DC attack),this thesis further considers synthesized-copying attack(SC attack).To optimize the parameters of the LCAC 2-D bar code,we use a generalized Gaussian distribution to give theoretical model of noise characteristics in the illegal channel.Through experimental results based on multiple printers and scanners,we find our theoretical modeling has good match with actual experimental results,which verifies the accuracy of the theoretical model.Based on the theoretical model,we design a prediction function.Then,we optimize the parameters of the LCAC 2-D bar code based on the prediction function,where we find the best trade-off between the production cost and the illegal copying cost.Our experimental results show that the LCAC 2-D bar code can effectively detect all DC attacks,and detect SC attacks if the number of synthesized 2-D bar codes is less than 14.Second,in order to improve the portability of a LCAC 2-D bar code,we introduce halftone visual watermarking technique(HVW)into the LCAC 2-D bar code and propose a LCAC watermarking 2-D bar code algorithm with two layers of anti-copying mechanism.The LCAC watermarking 2-D bar code overlaps halftone 2-D bar codes and observes the visuality of the embedded watermarking to identify whether it is copied.The LCAC watermarking 2-D bar code can obviously improve the portability because it does not require any electronic products for making an authentication decision.There are two parameters in the LCAC watermarking 2-D bar code: the embedding strength of the watermarking and the constraint of the decoding correctness.Through simulations,we study the relationship between two parameters and each performance metric.The performance metrics include: the accuracy of 2-D bar code demodulating and decoding,the accuracy of watermark extraction,peak signal-to-noise ratio and structural similarity.Our actual experiments show that when we set the embedding strength of the watermarking as 0.000004 and the constraint of the decoding correctness is 1,we can achieve a satisfied result,i.e.,the embedding watermarking pattern is clearly observable for legal printing whereas it totally disappears for illegal copying.