SCIENCE AND TECHNOLOGY - Making wavelets in prints.
By Chang Ai-Lien.
17 September 1998
(c) 1998 Singapore Press Holdings Limited
EVEN a partial fingerprint will be enough to nail a crook in a new identification system being finalised by a team of researchers from the National University of Singapore (NUS).
A working model of the system, which is expected to be ready at the end of the year, will also be faster than existing ones.
It will be able to locate the crook's print accurately from a database of, say, 10 million, within hours.
The automated fingerprint identification system is a project of NUS' Wavelets Strategic Research Programme. Mr Tham Jo Yew, 27, a research associate with the programme, is in charge of the project.
Fingerprint or photographic identification can also be used for entering restricted areas, personal identification cards and electronic commerce.
This is the area called biometrics, in which security devices use fingerprints, voice, or other unique personal bodily characteristics such as the eyes to recognise people.
But, Mr Tham said, present devices produce fuzzy images, take too long to recognise the images, and sometimes cannot "read" the fingerprints because of variations in how hard people press the scanner. "If a person presses his finger too hard, for example, the image will be very black."
But the technology the NUS team is using, called wavelet technology, can actually clean up such patches, so the failure rate is lower.
Wavelets are a new set of mathematical formulae which let information like sound and images be represented efficiently, and are one of many ways in which information such as sounds and images can be stored in a computer, for example.
Such data needs to be compressed for transmission, say, on the Internet, and this is where wavelet technology has proven to be better than other storage methods.
With the others, images and sounds tend to lose a large amount of their tone and clarity when they are compressed. But with wavelet storage, images remain sharp and are clearer.
It can produce easily better quality images which can be transmitted up to four times faster. These also take up one-quarter the storage space of conventional methods, he said.
Compressed data can also, through a special technology called WavPRESS, be received by people who can customise the data and vary the picture size, clarity, colour and so on, according to their capabilities and needs, added Mr Tham.
This capability is being exploited in a joint project between the wavelets programme and the university's Department of Electrical Engineering.
The system being worked on would, for instance, allow a laptop or palm-top user without high-speed Internet or other network access to choose a low-resolution version of an image. The same image can be accessed in high resolution by someone with high-speed access, such as via Singapore One.
Some areas in which this can be used include video-on-demand, distance teaching and learning, video-conferencing with people in different areas, and live broadcasting. The wavelet programme is also looking into medical image enhancement and machine fault diagnosis.
Researchers from the programme won the National Science Award recently for their work on wavelets.
Professor Lee Seng Luan, Associate Professor Wayne M. Lawton and Associate Professor Shen Zuowei, all from the NUS Mathematics Department, won the award for their basic mathematical research on wavelets theory.