High Capacity Analog Channels for Smart Documents
Widely-used valuable hardcopy documents such as passports, visas, driving licenses, educational certificates, entrance-passes for entertainment events etc. are conventionally protected against counterfeiting and data tampering attacks by applying analog security technologies (e.g. KINEGRAMS®, holograms, micro-printing, UV/IR inks etc.). How-ever, easy access to high quality, low price modern desktop publishing technology has left most of these technologies ineffective, giving rise to high quality false documents. The higher price and restricted usage are other drawbacks of the analog document pro-tection techniques. Digital watermarking and high capacity storage media such as IC-chips, optical data stripes etc. are the modern technologies being used in new machine-readable identity verification documents to ensure contents integrity; however, these technologies are either expensive or do not satisfy the application needs and demand to look for more efficient document protection technologies. In this research three different high capacity analog channels: high density data stripe (HD-DataStripe), data hiding in printed halftone images (watermarking), and super-posed constant background grayscale image (CBGI) are investigated for hidden com-munication along with their applications in smart documents. On way to develop high capacity analog channels, noise encountered from printing and scanning (PS) process is investigated with the objective to recover the digital information encoded at nearly maximum channel utilization. By utilizing noise behaviour, countermeasures against the noise are taken accordingly in data recovery process. HD-DataStripe is a printed binary image similar to the conventional 2-D barcodes (e.g. PDF417), but it offers much higher data storage capacity and is intended for machine-readable identity verification documents. The capacity offered by the HD-DataStripe is sufficient to store high quality biometric characteristics rather than extracted templates, in addition to the conventional bearer related data contained in a smart ID-card. It also eliminates the need for central database system (except for backup record) and other ex-pensive storage media, currently being used. While developing novel data-reading tech-nique for HD-DataStripe, to count for the unavoidable geometrical distortions, registra-tion marks pattern is chosen in such a way so that it results in accurate sampling points (a necessary condition for reliable data recovery at higher data encoding-rate). For more sophisticated distortions caused by the physical dot gain effects (intersymbol interfer-ence), the countermeasures such as application of sampling theorem, adaptive binariza-tion and post-data processing, each one of these providing only a necessary condition for reliable data recovery, are given. Finally, combining the various filters correspond-ing to these countermeasures, a novel Data-Reading technique for HD-DataStripe is given. The novel data-reading technique results in superior performance than the exist-ing techniques, intended for data recovery from printed media. In another scenario a small-size HD-DataStripe with maximum entropy is used as a copy detection pattern by utilizing information loss encountered at nearly maximum channel capacity. While considering the application of HD-DataStripe in hardcopy documents (contracts, official letters etc.), unlike existing work [Zha04], it allows one-to-one contents matching and does not depend on hash functions and OCR technology, constraints mainly imposed by the low data storage capacity offered by the existing analog media. For printed halftone images carrying hidden information higher capacity is mainly attributed to data-reading technique for HD-DataStripe that allows data recovery at higher printing resolution, a key requirement for a high quality watermarking technique in spatial domain. Digital halftoning and data encoding techniques are the other factors that contribute to data hiding technique given in this research. While considering security aspects, the new technique allows contents integrity and authenticity verification in the present scenario in which certain amount of errors are unavoidable, restricting the usage of existing techniques given for digital contents. Finally, a superposed constant background grayscale image, obtained by the repeated application of a specially designed small binary pattern, is used as channel for hidden communication and it allows up to 33 pages of A-4 size foreground text to be encoded in one CBGI. The higher capacity is contributed from data encoding symbols and data reading technique.