Ultrasonic equipment for fingerprint recognition applied to tomographical testing of materials and biological objects
Optel, the company we represent, was established a few years ago on the basis of the idea that a fingertip structure (reflected in fingerprints) can be better analyzed with the use of ultrasound than by existing methods. It turns out that our approach was justified; however its implementation required much more effort than we had anticipated. Nevertheless, we have managed to create fully-operational prototypes of an ultrasonic camera which permit the imaging and recognition of finger ridge patterns. We are well on our way to developing a plan for mass production. The design of the current version of the camera and principles behind its operation have been presented in [1]. Its operation can be briefly summarized as follows:
A very short ultrasonic pulse is sent from a number of different directions toward a finger surface and from each direction a much longer broadband impulse response is received. This impulse response is an effect of the contact scattering of the ultrasonic wave on the surface of the fingertip. Based on a set of such impulse responses, an image of the finger surface structure is reconstructed using computer tomography methods.
In order to create our camera we had to design and make our own ultrasonic devices, mainly for two reasons: devices available on the market were too expensive and did not meet a number of requirements important in our applications. Of paramount importance was that the time skew between different impulse responses must be very small (less than 5ns). Equally important was that the ultrasonic transducer should be able to generate a very short pulse (less than 100ns) and should have a sufficiently broad bandwidth as a receiver (4-16MHz). Finally, all elements used had to be inexpensive, since the final product should be suitable for mass production at a reasonable price.
After we have created elements that meet our needs, it has become apparent that they are suitable for a number of other applications where the use of ultrasonic pulses is necessary, (e.g., classical defectoscopy, layer thickness measurement, medical applications). This paper is intended as a presentation of our solutions for the ultrasonic technology experts and a declaration of our future aims.
We have created following elements based on our own design, partially patented, which depending on the mechanics and software used can be employed to various applications.
Ultrasonic PC card, particularly well suited to automatic measurements using mechanical or electronic scanners. It can directly control such devices (it only generates control signals, step motors require an additional driver unit).
Pulser and receiver circuit with an active switch and input amplifier, the size of a matchbox, powered and controlled by the ultrasonic PC card, capable of generating very short pulses and having a broad bandwidth as a receiver.
Ultrasonic transducers capable of generating very short pulses while having broad bandwidths as receivers.
Elements, allowing to produce a very good defined gaussian ultrasonic beam.
The PC card comprises not only an input amplifier, a high-speed A/D converter (it has a sampling frequency of 80MHz, a 100 MHz version is under development) but also a control circuit. This means that all control signals required for the measurement process are generated by the card. This kind of design allows us to achieve a sufficiently small time skew between different channels (or scanner positions) which is less than 1 ns. Thanks to the independent control system, the CPU is not directly involved in the measurement process, therefore a high performance PC is not essential. The card has a simple design achieved through the use of highly integrated circuits and can be offered at a reasonable price.
Our pulser and receiver circuit also provides an inexpensive solution. It is based on a unique design: in order to generate a pulse we first charge the transducer to a required voltage (max. 600V) which takes a few microseconds, then we discharge it in a very short period of time (at the moment this time is around 20ns). This approach results in a simple and inexpensive circuit which generates very clean pulses with practically any transducer (only transducers with a parallel inductance cannot be used).
All the electronic circuits we have developed would not be sufficient if we did not have transducers capable of generating very short pulses while having broad bandwidths as receivers. We had tested a number of designs proposed by others as well as products available on the market and we were convinced that there was only one solution: to find a new way. It had to lead to a design which would not only satisfy all the technical requirements but also enable cheap production and repeatable parameters.
We have succeeded in developing a transducer satisfying the above requirements for which a patent application has been submitted. Fig. 1 shows a pulse generated by our transducer (a hydrophone made of PVDF film has been used for the measurement, 50 ns/div). Fig. 2 presents the signal received in the transceiver mode (100 ns/div). The difference between the two signals results from the fact that our transducer works differently as a transmitter and as a receiver. This property is also exhibited by other types of transducers, though the differences are usually much smaller. It is possible to develop a transducer which would have practically identical signals in both the transmitter and receiver modes (corresponding roughly to the signal shown in Fig. 1), but it would have certain drawbacks (particularly when used in the design of our camera). Therefore we have decided to use the transducers which receive slightly longer impulse response (as shown in Fig. 2). How do the signal level and sensitivity compare with other designs? The signal level generated in the transmitter mode is at least two times higher and the sensitivity in the receiver mode is slightly lower. Hence, the overall signal level in the measurement cycle is comparable. Since the development of the new transceivers has not been completed yet, we expect considerable improvement.
Our technique allows not only the observation of finger ridge patterns but also other objects. An example of such an application is a device suitable for detecting material defects in objects as well as for medical measurements. It has been created on the basis of our electronic circuits, transducers as well as mechanical elements of our ultrasonic camera. Its basic idea is similar to that behind our ultrasonic camera. Fig. 3 shows a schematic diagram of the device:
The motor M moves the transducer T around the object S under measurement, sending toward it an ultrasonic wave. Signals reflected by the object are received by the same transducer and processed in a similar way as is done in our camera. Only image reconstruction procedures need to be modified.
The device has been developed for finger imaging. We would like to see a cross-section of a finger and subsequently a three dimensional structure (using tomographic reconstruction procedures). Using the same device, it is possible to observe other types of objects, both natural and man-made, in order to find material defects.
The pictures 4 and 5 shows the cross-sections of the finger. To achieve this pictures a classical B-scan was used. Echoes, coming from 400 directions are set together to form a whole picture. In the near future we will try to make tomographic reconstruction, hoping to became much higher resolution, comparable with the resolution of our fingerprint devices (at least .1 mm).
Fig. 1 Fig. 2 Fig. 3
Pulse of the transducer, received Pulse of the transducer, received Schematic diagram
with a hydrophone (50 ns/div). with the same transducer. of the device.
Fig. 4 Fig. 5 Fig. 6
View of the device Cross-section of the finger. Cross-section of the finger.
References