Source: - Bo Xiaochen Shen Lin Cheng Changwensen
---- Financial Security is an important part of national security, regardless of traditional business or e-commerce,
The importance of various paper notes and electronic tickets is self-evident. Along with high quality, cheap replication equipment
The emergence and the rise of e-commerce, and the banned technology is constantly developing. Digital watermarks will play an important
character of.
Challenges from color copiers
---- 1970 The birth of the first commercial color copier brought a tricky question: the pseudo of important notes
It is no longer necessary special equipment and specialization technology, and users of each color copier have become potential bills forgery.
By. In order to prevent the use of color copiers from fake bills, many countries are printed very refined, making photocopying
A mixed distortion occurred. In addition, special anti-counterfeiting techniques such as plastic stripes, discoloration inks are also used.
---- Some government machines represented by the US Treasury In addition to the various anti-counterfeiting measures to the ticket itself
It is concluded that a certain anti-counterfeiting function is also added to the color copier is also a way to solve the problem. This point of view
Two basic solutions: one is based on "prevention first", that is, through the copy of the bill to monitor the circuit
The test file is measured in order to do the copying machine to automatically refuse to work during counterfeit bills. This solution needs real-time fast.
Pattern recognition technology, it is difficult to implement at the time; its second is Chairman MiChael, Chairman of the US House Monetary Policy Committee
"Afterward Tracking" scheme proposed in Castle, that is, embedding unique, not available in each output image of the color copier
See the identification information for easy tracking. The latter is more realistic relative to the technical level at the time.
---- If the appearance of color copiers only brings a certain degree of panic, then, in recent years, high
The trouble brought about the precision cheap scanner and color printer is not as small. According to the report of the US Treasury,
In the counterfeit coins collected in the United States in 1997, 19% came from a color printer. The 1998 figures were 43%, and 1995
This figure is just 2.36%. At present, small-size currency and most commonly used bills lack of complicated processes such as watermarks.
Protection, and cheap scanners and printers make the counterfeit can assume forged costs and are advantageous.
---- In order to solve the anti-counterfeiting problem of the printing bill, the Massachusetts Institute of Technology, IBM, etc.
Some research institutions representing a digital watermarking technology based on the support of the US Treasury / Printing Ticketing
Anti-counterfeiting research, these research ideas are basically previously derived from two solutions for the anti-counterfeiting.
Print image hidden identification with printer tracking
---- In order to track printers for fake tickets when needed, you can embed in the printer output image.
Enter the serial number capable of identifying the printer as a trail of the forged track. In order to achieve this goal, MIT Media Experiment
Professor W. Bender Professor of Room proposed PatchWork -Patch Track method. PatchWork is an embedded embedded in anti-counterfeiting watermark
Procedure, it integrates in the printer driver. Patch Track is the corresponding interpretation process (see Figure 1).
---- PatchWork algorithm is embedded in a number of data, very low visibility, strong robustness
Watermarking, can be resistant to image tailoring, blurring, and color jitter. The word "patchwork" is originally referring to a variety of colors
The fabric of the shape of the straw film stitching, its image stated the core idea of the algorithm, that is, on the image domain
Excessive mode redundancy to achieve robust digital watermarks. Unlike most image domain digital water printers,
PatchWork is not hidden in the minimum effective bit (LSB) of the image data, but hidden in image data.
Statistical properties.
---- To hide 1bit data as an example, the PatchWork algorithm first generates two random data sequences through the key.
The column is scaled according to the size of the image, and is a random point coordinate sequence. The pixel luminance value corresponding to one of the coordinate sequences is then lowered, and the pixel brightness corresponding to the other coordinate sequence is also raised. Due to the magnitude of the brightness changes
Small, and random scattered, not concentrated, so it will not significantly affect image quality. In order to improve robustness,
Change the brightness of the pixel brightness in the random neighborhood, which forms bright, dark mode (ie, called patch) on the image domain.
Paving.
---- There are many factors affecting the effect of the PatchWork algorithm, mainly:
---- · The depth of PATCH
---- PATCH depth refers to the amplitude of the grayscale value of the random neighborhood, the larger the depth, the robust to the watermark
The stronger sex, but it will also affect concealment and improve visibility.
---- · · Size of Patch
---- Large-size PATCH can better resist rotation, displacement, etc., but the size of the size will inevitably cause
The amount of watermark information is reduced, causing PATCH overlapping each other. The size and quantity of the patches must be in the specific application.
Comfort is compromised.
---- · The contour of patch
---- Patch with steep edge will increase the high-frequency energy of the image, although this is conducive to the hidden of watermarking,
However, the watermark is also easily destroyed by loss of compression. In contrast, the PATCH with smooth edges can be well resistant to loss.
It is easy to cause visual attention. The reasonable solution should be determined after considering the attacks that may suffer.
If there is a damaged attack, it should be used with a smooth edge, so that the watermark energy is concentrated in low frequency;
Conversely, if the attack adjustment is facing, the PATCH with steep edge should be used to concentrate the watermark energy.
At high frequency. If there is no accurate estimate of the attack, the energy of the watermark should be scattered on the entire spectrum.
---- · · Arrangement of PATCH
---- The arrangement of patches should try not to form a clear boundary because the human eye is very sensitive to the grayscale border.
W. Abnder recommends using a random hexagonal arrangement.
---- · The quantity of patches
---- The more the number of patches, the more reliable decoding, but this will also sacrifice the quality of the image.
---- In addition to these factors, you can also fuse a lot of image filtering techniques in the PatchWork Watermark Algorithm.
Surgery, such as visual mask technology, etc., to improve the concealedness or robustness of watermarking.
---- Watermark Decoding Program Patch Track is actually a statistical signal detector. Patch TRACK
First, the scanned ticket image is corrected, and the watermark characteristics of rotation, breakage are overcome. Follow
After the PATCH TRACK generates a two-dimensional random point coordinate sequence using a decryption key to form a decoding window. Should construct
When the pixel grayscale statistic, it can be judged whether or not the PatchWork watermark is included in the decoding window. Digital hidden lines and fast
Speed water print decoding
---- In order to realize the automatic ticket identification and bill of tickets, the Massachusetts Institute of Technology hides
The team proposed a linear digital watermark-digital hidden line (Tartan Thread) technology. Method with hidden identification
Similarly, Tartan Thread is a active protection technology that must be cooperated with the bill producer, in the real bill of paper.
In the case, anti-counterfeiting watermark is added, and this linear digital watermark can exist in the scanned ticket image, in the printout
At the time, the watermark decoding module in the printer driver can quickly interpret the watermark, once the ticket is found to be anti-counterfeiting,
That is, the print output is refused.
---- The biggest difficulty in the digital hidden line anti-counterfeiting program is the problem of decoding space. In general, the printer drive
The moving sequence only caches a few lines of pixels, during the printing process, there is no complete print image in the middle of the memory,
The decoding space with digital hidden lines is very small. In addition, the interpretation process of digital hidden lines must be very rapid. If it is too much to affect printing efficiency, whether it is printer manufacturer or user is unacceptable.
---- Tartan thread digital hidden line core technology is a one-dimensional expansion modulation, so that the watermark information is spread
The code is adjusted into a signal having noise properties, superimposed on the ticket image. The decoder uses the same spreading code by despreading
Read the digital hidden line. Figure 2 depicts the basic process of the Tartan Thread watermark algorithm.
Bills in e-commerce security
---- Whether it is traditional business or e-commerce, buyers and sellers need a certain bill as a transaction
Certificate, quotation, payment, and electronic tickets such as electronic notes occupy an important role in e-commerce. Therefore, electronic ticket
According to the security issues, it is not only directly related to the interests of merchants and consumers, but also related to the economic order of the whole country.
A problem that cannot be ignored.
---- Currently using the most electronic bill files are the PS and PDF files initiated by Adobe, it integrates
Various media formats such as text, images can be independent of the platform. In the early 1990s, Brassil et al.
People began to study the hidden identification method for the PS file, the so-called "document structure fine-tuning algorithm". Digital identification
Information is converted into a series of slight adjustment actions for document structures, including vertical movement line spacing, horizontal adjustment
The word and adjustment font, etc. This algorithm can resist some standard file operations such as photograph and scanning copy.
---- PS / PDF file format multimedia nature determines the anti-counterfeiting watermark of electronic notes must be a variety of waterprints
Comprehensive, including image watermark, text watermark, etc. With the rapid development of e-commerce and the research of digital watermarking technology
The continuous in-depth, PS / PDF document watermark research will increasingly attract extensive attention.
