Some Studies On Invoice Anti-counterfeiting Based On Binary Image Watermarking Technology

With the rapid development of computer technology and digital devices，thesecurity of digital media content, confidentiality, authenticity and integrity problemare getting more and more people’s attention, and that also promoted the developmentof the information hiding and digital watermarking technology. With the rapiddevelopment of national financial the invoice safety has become an important part ofnational security. Invoice is an indispensable part of People’s Daily life, includingcash, checks, bonds, invoices, stocks, securities, stamps and all kinds of tickets. Toensure the safety of the invoice is important means to maintain the normal operationof the national economy, but invoice anti-counterfeiting faces a serious shortage andbottlenecks. The means of the anti-counterfeiting of traditional invoice includingprinting ink anti-counterfeiting, anti-counterfeiting printing technology and platemaking anti-counterfeiting. These traditional means of anti-counterfeiting can havethe purpose of anti-counterfeiting, but there are also many shortcomings. How tousing the digital method to protect the security of the invoice has become a problemwhich demanding prompt solution. Invoice printing just has two-color printing, soinvoice font image can be seen as binary image which brings to the invoiceanti-counterfeiting a new breakthrough. We can see the invoice image as the binaryimage, so the watermarks embed to invoice can base on the binary imagewatermarking algorithm. So the binary image watermarking technology has been thecore of the invoice watermarking technology, this requires research image digitalwatermarking technology which can resist printing and scanning.Binary image is different from the grayscale or color images grayscale or colorimages because the binary image just have two values that black and white, therefore,binary image pixel values have no redundant spaces, This make it difficulties forbinary image watermarking technology. According to human visual system model,few changes to the pixels on a color image or a grey image would not be perceived byhuman eyes because of the large pixel value range and high redundancy. However,binary images only have two colors (black and white) which means little redundancy.When changing the black (white) pixels of a binary image into white (black) ones,human eyes can catch the change easily. Therefore, it is more difficult to embedwatermark into binary images than into color images and grey images. This paper studied the binary image watermarking technology which robust toprinting and scanning process base on the topic of invoice anti-counterfeiting. Thestudies including the following contents:1. The multiple invariant based on the convolution process model and by thismodel derived to the invariant relationship before and after printing and scanningprocess. But this model is not accurate, so the invariants are vulnerable to noises. Thispaper gives the vulnerability and the causes of vulnerable to noises of the multipleinvariant model. Deduced and analyzed the multiple invariant model and base on themultiple invariant at before and after printing and scanning process we gave thewatermark embed and extract algorithms. After that, we gave the algorithm whichinvoice anti-counterfeiting base on the invariant.2. This paper proposed the additive invariant model before and after printing andscanning process. The watermarking algorithms base on additive model is universality,and it suitable for all kinds of binary images. The stability of the extracted watermarkis higher than the multiple model because the additive model is accurate. The binaryimage watermarking algorithm includes the following steps. The multiple invariantjust suitable for document binary images because the ordinary binary images cannotbe segmented. In this paper, an image sampling operation is used to convert a binaryimage into thumbnails. The multiple invariant model is an inaccurate model，so whenembedding watermarks it must be under certain intensity otherwise the embeddedwatermarks cannot be extracted accurately due to the imprecision of the multipleinvariant model after the printing and scanning process. First, a binary image is splitinto multiple thumbnails using a perimeter expansion and sampling operation. Then,the watermarking information is embedded by flipping pixels in the thumbnails tochange the number of black pixels. Finally, the watermarked binary image isproduced by inverse sampling the thumbnails. The binary image expansion andsampling operation can also be used on a printed and scanned binary image, where thewatermarking information is extracted by comparing the difference between thenumber of black pixels in the thumbnail and the mean value. The experimental resultsshowed that the proposed method is robust to the printing and scanning process.3. Invoice printing just has two-color printing, so invoice font image can be seenas binary image. To embed watermarks into invoice image, the pixels need to beflipped. The huge the watermark is, the more pixels need to be flipped. We proposed anew pixels flipping method in invoice image for huge watermarking capacity. The pixels flipping method includes one novel interpolation method for binary image,one flippable pixels evaluation mechanism and one denoising method based ongravity center and chaos degree.The proposed interpolation method ensures that theinvoice image keeps features well after scaling. The flippable pixels evaluationmechanism ensures that the pixels keep better connectivity and smoothness and thepattern has highest structural similarity after flipping. The proposed denoising methodmakes invoice font image more smooth and fit for human vision. Experiments showthat the proposed flipping method not only keeps the invoice font structure well butalso improves watermarking capacity.