EP0389483A1 - Fingerprint sensing device - Google Patents

Abstract

Method for generating an electrical output signal as a function of the peaks (4) and the valleys (3) of a finger (5) placed against a contact surface (10) of a transducer, said contact surface (10) comprising an elastomer (13) provided with a reflecting layer (15), by directing on said layer the beam coming from a light source (17) of the transducer and by detecting the light scattered by said reflecting layer (15). This is achieved by involving the representation of an image plane (24) close to but behind the plane (25) of the reflective layer of said transducer in order to avoid a doubling of spectral frequency. The light source (17) of the transducer includes a pulsed light source (LED) coupled to an electrical circuit (MCC) intended to control and scan the transducer.

Description

Fingerprint sensing device

The invention relates to a method according to the preamble of claim 1.

A number of systems have been proposed for processing identifi¬ cation information based on the unique configuration of ridges and valleys on the skin of an individual's finger. When such information is taken as an ink impression of an individual's finger, it is normally called a fingerprint. The more sophisti¬ cated techniques which employ optical means to evaluate the finger surface tend to provide a more refined, discriminate and accurate identification image. In these optical systems, the finger of the subject individual is placed against the surface of a transparent plate covered by a reflecting layer. This plate structure shall further be called a "lozenge". Light transmitted through the lozenge and scattered from the re- fleeting layer (mirror plane) ^'is imaged onto receiving or pro¬ cessing equipment. This receiving equipment may be a camera screen or an array of photocells.

Such a scattering-mode system is described in US-Patent 4,322,163.

Substantially flat and unscattering regions, as well as highly scattering regions are formed in the reflective layer, accord¬ ing to the pattern of ridges and valleys of the finger of a subject individual. US-Patent 4,322,163 incorrectly describes large, flat, and unscattering regions as corresponding to the finger valleys. In fact, it has been found that there are small, highly curved and highly scattering regions there. Con¬ sequently, it is the finger ridges that cause the larger, re- latively flat regions. 1 The prior known techniques further teach imaging of the re¬ flecting layer or mirror plane. It has been found that imaging the mirror plane is disadvantageous, because spatial frequency doubling of the image is observed. This effect arises because

5 there are two flat and two angled regions of the mirror plane of the elastomer lozenge for every ridge and valley period of the fingerprint. Such frequency doubling is highly undesirable.

US^'-Pfa±eπt- 4^428^:,-6;76 discloses an improvement obtained by im- 0 agir g; a^:.plane of" foci formed by the transducer surface rather tha imaging-the surface itself. The patent teaches that regions o ttie contact surface (reflecting layer) associated with the fingerprint valley are not in contact with the finger and form small concave focusing mirrors, having their focal points along 5 a focusing plane which is parallel to and at a focal distance of said -focusing mirrors away from the surface itself. Conse¬ quently, it was recommended to image said plane of foci formed by the valleys. The -existence of another focusing plane located behind the surface, defined as the virtual plane of the ridges,

20 is not previously disclosed. According to the present inven¬ tion, this virtual plane can be used for imaging. It has been found that, the relatively flat elastomer regions associated with the ridges of a finger, do not substantially carry high spatial frequency information and that consequently the finger-

25. print; image. deriVad from the ridge focussing plane does not substantially carry information associated with pores or minor scars: of the finger, improving image quality.

European patent application, serial no. 86 10 62 31.3, filed 30 May 1986, incorporated by reference herein, describes an appa¬ ratus and procedure for determining the authorization of indi¬ viduals by verifying their fingerprints. The apparatus uses a portable data carrier e. g. a plastic card, which has magnetic tracks to store data. Stored data can consist of the degree of 35 correlation between a fingerprint of an authorized individual and a selected; reference image, and the code number of the se¬ lected reference image. A fingerprint detection terminal with a sensor includes a memory in which the selected reference image is stored. The terminal compares the actual fingerprint of an individual to be checked with the corresponding reference image identified on the plastic card and stored in the fingerprint detection terminal. The determined degree of correlation is in turn compared to the degree of correlation stored on the plas¬ tic card. As a function of the result of this comparison, an appropriate decision signal is generated. The data are written on the plastic card in such a way that, based on the data alone, only ambiguous conclusions concerning the identity of the individual are possible.

Fingerprint sensors can incorporate a light source which emits a light beam for sensing the fingerprint. The light source may be a light emitting diode (LED). Light energy emitted by LED's varies with temperature and time. Consequently, light receiving equipment may deliver a different output signal as the LED fades over time and thereby may distort the evaluation proce¬ dure. ^'

It is therefore necessary to provide means for controlling ac¬ tual parameters of light emitting elements within the finger¬ print verification system.

The impression force of an individual's finger on the trans¬ ducer's sensor surface may differ for different individuals or for the same individual on different occasions on which his fingerprint is placed on the lozenge. Change of fingerprint pressure may result in a change of the position of the ridge focussing plane and valley focussing plane. The pitch of the fingerprint pattern is defined as a local wavelength, or the spacing distance between adjacent ridges or valleys.

It has been found by the inventor, that whatever the pitch, the location of the ridge or valley plane should be at a constant separation from the lozenge surface; otherwise the image will be in focus for one pitch and increasingly out of focus for larger or smaller pitches. It is therefore an object of the invention to provide a method and apparatus for generating an electrical output signal which characterises the ridges and valleys of a finger placed on a contact surface of a finger sensor transducer, which provides high sensitivity and resolution, affords a high reliability, and is to a large extent free from distortion.

The abovenamed objects of the invention are achieved by a method as defined above, in the introduction of this text.

Imaging of a plane close to, but behind, the reflecting layer plane of the transducer avoids spatial frequency-doubling in the image. Thus, optical contrast of typical fingerprints will be improved. Contrast may be further improved by a light stop designed to control the aperture of the imaging lens and consequently the range of angles scattered from the reflecting layer and imaged onto the detector.

Imaging the virtual plane associated with the ridges of a fin- ger avoids very high spatial frequency information' derived from pores or scars which may result in an image distortion.

The problems caused by varying fingerprint pitch are solved by providing a lozenge structure with a light scattering angle proportional to the pitch.

In a preferred embodiment of the invention, the lozenge struc¬ ture comprises a silicone type elastomer layer (type GE 615 or GE 670) of 400 microns thickness and a protective layer of a polyester type of 12 microns thickness. A lozenge of this con¬ struction maintains the required relationship, that scattering angle is proportional to fingerprint pitch.

Thus, image quality of the transducer will be substantially in- dependent of fingerprint pitch (normally in the range of 0,4 mm to 0,8 mm), fingerprint ridge profile, finger hardness, finger¬ print hardness, and ridge height. In another embodiment of the invention, a bright-field imaging transducer is utilized. When no finger presses against the sur¬ face of the lozenge, the transducer surface is a flat mirror and light converging from it comes to a focus, as an image of the light source, in or close to the plane of the imaging lens. In bright-field mode, the light forming an image of the source in the imaging lens is allowed to continue through the lens to illuminate a CCD photo-sensor. Any stray or scattered light, for example light reflected from steep gradient areas of the reflective layer when pressure is applied to the contact surface, is blocked by an appropriate annular light stop.

When a finger is pressed against the contact surface of the transducer in bright-field mode, when imaging the mirror surface, an image is formed, consisting of bright fingerprint lines where the elastomer surface remains substantially parallel to the undisturbed surface, and dark lines where the surface is locally non-parallel to the undisturbed surface (i. e. where there is a gradient).

When imaging the ridge plane it consists of bright fingerprint lines at positions corresponding to the ridges and dark lines where the light is not focused corresponding to the valleys, and grey where the surface remains parallel to the undisturbed surface.

Dark-field mode operation is also possible. In dark-field mode, the central axial light, which is reflected from undisturbed areas of the mirror plane, is blocked by an appropriate light stop. The light scattered by gradient areas of the mirror layer passes around the light stop to be imaged by the imaging lens onto the CCD when imaging the mirror surface.

Generally, operation in dark-field mode gives better image con- trast, but an overall light-level/contrast combination proved to be better in bright-field mode. 1 ^"Furthermore, bright-field imaging can be used when no finger Is present, for controlling image quality by varying the light ^energy transmitted by a pulsed light source coupled to a microprocessor control system. ^{~ :}5

In a further embodiment of the invention, this microprocessor control system is designed to control and diagnose the trans¬ ducer.

ϋD The microprocessor system according to the invention uses a oσk-up table or any other memory means to correct for varia¬ tions in the output of the pulsing light source - preferably an LED - due to temperature shifting, degredation, dust, or any other distortion in the system.

15

The life of the transducer system including the LED will be extended by switching it off in times of prolonged non-use.

The microprocessor can diagnose problems in the transducer -sys- 2O tem. For example, if the mirror layer (reflecting layer) is damaged, external light can penetrate the transducer and con¬ sequently deteriorate the image quality. This mirror layer failure can be detected by looking at the pulse width on the comparator associated with the central processor unit with the 25 LED turned off.

The system is designed to detect LED-failure by comparing the pulse-width of the signal derived from the comparator with the pulse width from the LED. ^•30

Advantageously, the microprocessor system detects abnormal de¬ gradation in the response of the finger transducer.

The invention will be more closely described in the following, 5 with reference to embodiments which are shown in the drawings in which Fig. 1 is a general schematic view of the optical design of a fingerprint sensor according to the invention,

Fig. 2 is a schematic view showing the three imageable infor- mation planes, namely the valley image plane, the ridge image plane, and the reflecting mirror plane,

Fig. 3 shows the arrangement of optical components of the fin¬ gerprint sensor for operation in bright-field mode,

Fig. 4 is a schematic view similar to that of Fig. 3, but for- operation in dark-field mode,

Fig. 5 is a schematic demonstrating that only the valley image plane is well-defined for a soft elastomeric lozenge,

Fig. 6 is a schematic similar to that of Fig. 5, but showing definition of both ridge and valley image planes for a hard elastomeric lozenge,

Fig. 7 is a schematic representation of the requirement that the elastomer scattering angle is proportional to the finger¬ print pitch,

Fig. 8 is a general side view of a fingerprint sensor trans¬ ducer lozenge structure according to the invention,

Fig. 9 is a partial cross-section through a combined colli- mating lens and fingerprint transducer according to the in- vention,

Fig. lOa-d shows various steps in the manufacture of the struc¬ ture of Fig. 9, and

Fig. 11 is a block circuit diagram of a computer control system for the fingerprint sensor according to the invention. Fig. 1 shows the optical system .design of a fingerprint termi¬ nal used in a fingerprint verification system as described in European Patent Application No. 86106231.3. The optical system 1 comprises - a fingerprint transducer structure or lozenge 2 including a finger transducer or multi-layer contact surface 10 having a reflecting layer 15 and a colli ating lens 11

- a light source 17 designed as a light emitting diode (LED)

- a beamsplitter 18 - an imaging lens 19

- a photo receiver 20 including a CCD-Device 21 and a field- flattening lens 22

- and a microprocessor controlled syste ^'MCC (Fig. 11).

major feature of the optical design is to obtain high uni¬ formity with high optical efficiency.

Most light sources possess a fairly uniform distribution of in¬ tensity across their range of emission angles, but have a rela- tively poor uniformity of intensity^' across their emitting sur¬ face due to their physical structure. This applies for example to LED structures.

Actually, according to the invention the CCD 21 images the ridge plane close to, but not exactly coincident with, the fin¬ ger transducer plane. The discrepancy is very small and has no noticeable effect upon uniformity.

The aim of obtaining high uniformity is achieved by the illu i- nation scheme shown in Fig. 1. Light 7 emitted by the source 17 reflects from a beamsplitter 18 and then from the finger trans¬ ducer 10. When the transducer surface is undistorted the reflected beam then passes through a collimating lens 11, and the beamsplitter 18 to form an image of the source 17 in the centre of an imaging lens 19. The imaging lens 19 images the finger transducer 10. When the finger transducer is distorted by an applied fingerprint, the relatively undistorted regions give rise to a fingerprint image upon the CCD-Device 21 that remains of substantially uniform brightness. The collimating lens 11 may, for example, be a planovex lens with a center thickness of 4.0 mm, a diameter of 22.5 mm, and f = 60 mm, while the imaging lens 19 may be a bivex lens with a center thickness of 1.8 mm, a diameter of 5 mm, and f = 20 mm. The finger transducer 10 is placed as close as possible to the col¬ limating lens 11, and is therefore as far removed as possible from any image of the light source 17 in which the spatial in¬ tensity distribution of the source 17 could be produced. The source may, for example, be a 10 mW, 880 nm LED.

If the finger transducer 10 is sufficiently removed from the light source 17, then all spatial - nonuniformity of the light source is blurred out, and the uniformity of illumination at the finger transducer surface is governed by the angular dis¬ tribution of the source alone.

The collimating lens 11 forms an image of the light source 17 in or close to the plane of the imaging lens 19. Therefore, the spatial distribution of the light source 17 appears in or close to this plane.

Thus, the imaging lens 19 is incapable of relaying any signifi¬ cant information about the light source spatial distribution onto the CCD 21. The light forming the (light source) image in the imaging lens 19 continues through the lens 19 to spread out to a uniform illumination at the CCD plane. This ensures excellent uniformity.

The beamsplitter 18 arranged in the beam path between the light source 17 and the collimating lens 11 is used to produce a com¬ pact design employing partially - common light paths from the LED to the lozenge and from the lozenge to the CCD - Device 21. The beamsplitter 18 comprises a glass substrate of about 1 mm thickness, coated with a semi-reflecting metal film. For example, the beamsplitter 18 may comprise a titania front sur¬ face coating and a single layer antireflection coating on the glass rear surface. The beamsplitter 18 preferably should not introduce imaging aberrations which would limit the image quality available from a singlet imaging lens 19.

Field curvature from the singlet imaging lens 19 tightly restricts the allowable scattering angles, 9 (Fig. 7), from the lozenge 2 to obtain a good quality image. In practice, this severely limits the freedom of lozenge construction, so that other demands on lozenge performance (0 proportional to finger- print pitch, avoiding overall curvature of the lozenge surface, etc.) can not simply be met. These further performance parameters will be discussed in detail with reference to lozenge construction (Fig. 10). To obtain, better image quality and to avoid overconstraint, a field flattening lens 22 is introduced adjacent to the CCD 21.

The lens 22 ideally has f = -20 mm, when used with the f = 20 mm imaging lens 19, to achieve a perfectly flat field.

However, to minimize total aberrations, the lens 22 chosen has f = _16 mm, is planoconcave, has centre thickness of 1.8 mm, and is mounted with the piano face toward the CCD 21. Thereby the curvature at the outer edge of the object plane is limited to 0.2 mm. With no field flattening lens 22, the outer edge field curvature would be 1.4 mm.

The above described combination of imaging and field-flattening lenses 19, 22 achieves good resolution (MTF 0.4) of finger¬ print features of 0.1 mm size, with the beamsplitter 18 and collimating lens 11 as described above. This resolution has been demonstrated experimentally by the visibility of finger pores in the image, when imaging the valley plane particularly.

The imaging technique used in this optical system 1 is gener¬ ally known as scattering mode operation. An individual's fin- ger 5 is placed from above against contact surface 10 com¬ prising a reflecting layer 15. Light 7 is transmitted against the reflecting layer 15 from below, and the light scattering from the reflecting layer 15 is imaged onto receiving equipment 20.

As shown in Fig. 2, ridges 4 and valleys 3 of a finger 5 deform the reflecting layer 15 surface, further called the mirror plane 25. Depending on the specific deformation, the surface 25 forms small parabolic focusing mirrors, which focus the light 7 onto focusing planes. These planes are parallel to the mirror plane 25, spaced at a distance determined by the focal length of the parabolic regions. The focusing plane associated with the valleys.3 of a finger 5 is called the valley image plane 23, and the focusing plane associated with the ridges 4 of a finger 5 is called the ridge image plane 24.

in scattering-mode operation, one can image any of: the valley plane 23, the mirror plane 25, or the ridge plane 24.

Both the valley plane 23 and the ridge plane 24 provide one bright region (focus point) and one dark region for every fin- gerprint period or "pitch". Thus a "true" ima e can be formed' in these planes. However, in the mirror plane 25, potentially two scattering regions and two non-scattering regions are ob¬ served for every fingerprint period. This arises because for every ridge 4 and valley 3 there are two flat and two angled regions of the reflecting layer surface 15. This leads to unde¬ sirable "frequency doubling" of the image in the mirror plane 25.

Mirror plane imaging has proved inadequate, because spatial frequency-doubling distorts or blurs the image.

According to the invention, the finger valleys 3 cause small, highly curved, and highly scattering regions in the reflecting layer, whereby the valley focusing plane 23 is precisely de- fined and the depth of field is very small. Therefore, sophisticated apparatus are needed, to bring the imaging sys¬ tems in focus again when the focusing plane shifts. Finger ridges 4 cause larger, relatively flat regions in the reflecting layer. Therefore, the ridge focusing plane 24 is not as well defined, and the depth of field is greater. Consequent¬ ly, spatial frequency information associated with pores or scars of the ridges 4, is not very critical in the ridge focus¬ ing plane 24. The imaging system can therefore be less compli¬ cated and of simpler optical construction.

Scattering-mode operation of the optical system shown in Fig. 1 can be used in either "dark field" or "bright-field" modes, as shown in Figs. 4 and 3, respectively. ^"

When no finger presses against the transducer surface, it is a flat mirror and the light converging from it comes to a focus as an image of the light source in the plane of the imaging lens.

In "bright-field" mode (Fig. 3) the light 7a forming an image of the source in the imaging lens 19 is allowed to continue through the lens" to illuminate the CCD 21. Scattered light 7b which results when a finger 5 presses the transducer surface, is blocked by an annular stop 26 so that the scattered light 7b does not pass through the imaging lens 19 and so does not reach the CCD 21. The image when imaging the mirror plane therefore consists of bright fingerprint lines of highly-uniform brightness where the reflecting layer 15 remains parallel to the undisturbed surface, and dark lines where the layer 15 is locally non-parallel to the undisturbed surface .(i. e. where there is a gradient).

When imaging the ridge plane it consists of bright fingerprint lines at positions corresponding to the ridges and dark lines where the light is not focused corresponding to the valleys, and grey where the surface remains parallel to the undisturbed surface.

In "dark field" mode (Fig. 4), an opaque stop 27 is placed co- axially in front of the imaging lens 19 to block the light 7a reflected from the undisturbed reflecting layer 15. When imaging the mirror plane the CCD 21 therefore receives no illumination when the mirror plane 25 is undisturbed. When a finger 5 presses the transducer surface, some of the scattered light 7b is deflected past the stop 27 and passes through the outer regions of the imaging lens 19 to be imaged onto the CCD 21. In this case, dark fingerprint lines appear where the layer 15.^"remains undisturbed, and bright lines appear where there is a -surface gradient. -

In-general, dark-field mode gives better contrast than bright- field mode. In bright-field operation it is necessary to scat¬ ter almost all the light 7a out of the aperture of the imaging lens 19 to obtain best contrast. On the other hand, in dark- - field mode, it is only necessary to scatter a small fraction of light 7b into the outer parts of the lens 19 to achieve high contrast.

However, dark-field operation requires a larger lens aperture than bright-field operation. Therefore, bright-field equipment can employ a simpler, lower-cost lens 19 for any given level of performance and total image brightness. By using the field flattening lens 22 (Fig. 1), it is only necessary that the depth of field of the imaging lens 19 is smaller than the se- ^" paratiαn of the ridge or valley planes 24, 23 (Fig. 2) from the mirror plane 25, in order to form high contrast images of these planes.

Furthermore, bright-field imaging can be used for the control ^" and diagnosis of the optical system 1, as described in greater ^• detail below, with reference to the microprocessor controlled diagnosing system MCC, Fig. 11.

In order to form well-defined ridge and valley foci, the defor- 5 mation^* of the lozenge mirror surface 15 should be closely para¬ bolic,, especially near the peaks and troughs of the deforma¬ tion. 1 In order that the ridge and valley planes 24, 23are adequately removed or spaced from the mirror plane 25 for unambiguous images to be formed, the lozenge deformation must be small. Peak-trough deflections of approximately 2 - 5 μm are required.

5

Two fingerprint transducer compliances can be considered:

- relatively soft, but hard enough that, where the lozenge sur¬ faces relaxes into a fingerprint valley 3, a roughly parabolic surface is produced. Where the surface contacts a ridge 4, 0- the mirror surface 15 could remain influenced by the overall rr.dqje profile, but not exactly follow the fine details ^'of the ridge 4 (Fig. 5)

- harder, so that in both fingerprint ridge 4 and valley 3 zones a roughly sinusoidal or parabolic surface 15 is pro-

15^" duced (Fig. 6).

In practice, the second i.e. harder surface offers the greater flexibility of imaging either ridge or valley planes 24, 23, and is more easily realised with available materials. 20

It is important that, whatever the fingerprint pitch, p, (Fig. 7) the location of the ridge or valley plane should be at a constant separation from the lozenge surface; otherwise the imagE will be in focus for one pitch and increasingly out of 25.:- focu for larger or smaller pitches. To achieve this, the scat¬ tering angle G should be proportional to p, over the likely range, of pitches p (Fig. 7).

i.e. 0 p or — ⁸1 = —^Q2

30: for 0.3 mm ^ p -( 1.0 mm.

Specifying a finger transducer, which effectively achieves a proportionality between scattering angle 0 and pitch p over a 35 wide range, is the main objective of this transducer design, but other objectives must be taken into consideration as well. These other objectives are: - The transducer, as well as the entire system device, must be sufficiently robust to perform consistently for longterm, remote installation.

- The deformation characteristics of the transducer should be substantially independent of finger attributes (other than the fingerprint pattern itself).

- The transducer should have a high-reflectivity surface to allow adequate illumination of the image on the CCD.

- The reflective layer should be deformable so as to be capable ^' of deforming to the fingerprint pattern with high fidelity.

These requirements are to ensure that an image of repeatable brightness, contrast, and ridge:valley mark:space ratio is ob¬ tained for all users of the verification system. Thereby simple threshold comparison, can reliably extract and evaluate the fin- gerprint pattern, independently of the wide variations in fin¬ ger attributes such as size, softness, moisture, etc.

An optimized finger transducer should have, for example, a scattering angle proportional to possible fingerprint pitch over the range of 0.3 to 1.0 mm; while having the correct over¬ all stiffness, so that scattering angles are in the range of 0 - 3 degrees for applied forces of 0 - 2 kg.

Fig. 8 shows a transducer structure 10 designed to achieve these objectives. The transducer 10 comprises a silicon-type elastomer layer 13 of approx. 0.1 - 0.3 mm thickness covered by an optically flat, metal, reflecting layer 15 of approx 200 nm thickness, and a polyester protective layer 16 of approx. 7 to 15 μm thickness.

These layers are fixed on an optical glass support substrate 12 of approx. 1 mm thickness. This structure is particularly suited to obtaining scattering angle proportional to pitch, be¬ cause the polyester top layer 16 will act as a beam under load, of which the angular deflection will naturally be an increasing function of pitch. The following embodiments of the transducer structure 10 showed the closest proportionality of scattering angle to pitch: (a) silicone thickness : 400 microns silicone type: General Electric GE 615 polyester- thickness: 12 microns (b:X silicone thickness: 400 microns silicone type: General Electric GE 670 polyester- thickness: 12 microns.

The image quality, σf these structures proved to be substan¬ tially independent σf:

- fingerprint pitch

- fingerprint ridge profile

- fingerprint valley depth - finger hardness

- fingerprint hardness.

The whole transducer 10 can be replaceably mounted within the optical system 1 by means of screws- or any other fixing.ele- ments. To avoid optical distortion, the finger transducer 10 is placed as close as posible to a collimating lens 11 (Fig. 1).

In an embodiment of the invention shown in Fig. 9, the finger transducer 10 and the collimating lens 11 together form a lens/ lozenge subassembly 2.

The lozenge subassembly 2 essentially consists of two parts:

- The collimating lens 11, which additionally acts as a base substrate for - a mirrored elastomeric surface which deforms according to the fingerprint pattern impressed upon it.

In detail, the elements of the lozenge 2 are:

- the collimating lens 11, in the form of a plastic moulded lens,

- an opaque spacer shim 14 which defines the thickness of the elastomer layer 13 is arranged on the plane surface of the collimating lens 11, - an elastomer layer 13 arranged in the cavity formed between this spacer shim 14, the lens 11 and a further layer arrange¬ ment,

- a mirror layer 15, for example in the form of a metal coat- ing,

- and a thin protective upper layer 16.

Manufacture of the lozenge structure 2 will now be described with reference to Figs. 10a to lOd.

As shown in Fig. 10a, a mould is first formed by placing a sheet of optically flat polycarbonate 28 (300 micron, manufac¬ tured by General Electric Plastics and supplied with, removable surface protection films) onto a flat glass baseplate 29. On top of this is placed a spacing shim 14 of black polycarbonate 280 micron thick, 22 mm in diameter with a 16 x 13 mm rectangu¬ lar cut-out open area. All components must be clean and dust- free (Fig. 10a).

Silicone 13, GE 615 (soft), or 670 (hard), is mixed, degassed, etc. and is poured into the mould (approx. 0.25 ml) while en¬ suring that no bubbles are formed (Fig. 10b).

As further shown in Fig. 10b, the 22 mm diameter collimating lens 11 is gently pressed piano face downwards, onto the sili¬ cone 13 which then evenly fills the mould. The slight excess silicone is extruded between the lens 11 and the spacer 14. It is helpful to press the lens 11 down at a slight angle to elim¬ inate air bubbles or air entrapment. In finally centering the lens 11 and spacer 14, care is required to avoid introduction of bubbles at the edge of the contact area.

The assembly is then cured for 2 hours at 80 C.

After curing, the assembly is allowed to cool (approx. 15 mins.) and the excess elastomer is trimmed away. The polycarbonate base sheet 28 is then carefully removed from ■ the lens 11, shim 14 and elastomer 13 assembly (Fig. 10c). The shim 14 is coated with "silicone adhesion promoter" (available from GE) and allowed to dry. An approxi atley 50 x 50 mm piece of alu inized polyester 15, 16 of 6 micron thickness is care¬ fully placed onto the lozenge (aluminized side toward the sili¬ cone) and smoothed into place (Fig. lOd). The aluminized poly¬ ester 15, 16 is available as Chamberlain Plastics Ltd 442 grade; but other polyester base material could be used as well. The polyester remains in place on the silicone 13 by surface energy effects. The aluminized polyester thin film 15, 16 is widely used for electrical capacitors. Other metallized polymer films are available, but polyester has the most attractive thermal and environmental properties. The lozenge 2 may then be fixed in appropriate optical mounting means within the optical system 1.

A microprocessor controlled circuit MCC, as shown in Fig. 11, is provided to control and diagnose the fingerprint sensor. The circuit MCC mainly uses the advantages of bright-field opera¬ tion, whereby the light which forms an image of the light source LED in the imaging lens is allowed to continue through the lens to illuminate the CCD-Sensor 21, Fig. 1. The source LED (17, Fig. 1) is pulse-driven. Light energy emitted by the source LED is controlled via the control circuit MCC by appro¬ priately adjusting or varying the pulse width of the current supplied to the LED.

The circuit MCC comprises a central processing unit CPU in- eluding a microprocessor such as an INTEL 80188, designed to communicate with a memory look-up table LT, a data store DS, and a program store PS. The outputs of the sensor CCD and a field synchronous sawtooth waveform generator SWG are both connected to a comparator CD. The output signal of the comparator CD is directed to the microprocessor central processing unit CPU. The light output emitted by the LED varies with time and tem¬ perature. To provide a constant illumination level at the mir¬ ror plane of the finger transducer, the LED output level must be controlled. Therefore, illumination of the CCD-ARRAY of 604 by 288 pixels is controlled in a line-by-line manner. The CCD integrates the light energy received per field and generates an output signal according to the light energy. This analog sig¬ nal, having a level representing the integrated light energy, is transformed into a pulse PL by means of the comparator CD. The width of the pulse represents the integrated light energy. To achieve this conversion, the comparator CD compares the in¬ tegrated line output signal with a sawtooth waveform. Depending on the level of the CCD-output signal, a pulse PI with a pulse width proportional to the light level is generated. This com- parison process, triggered by the sawtooth waveform generator, can be carried out every 25 ms to generate a rectangular pulse of about 10 ms length.

This comparator output pulse PL is then compared with reference signals stored in the look-up table LT by the microprocessor central processing unit CPU.

Previously generated reference codewords representing different LED-control signals are stored in the look-up table LT to pro- vide constant illumination at the mirror plane. The micropro¬ cessor carries out a comparison, and accordingly delivers an LED current on-off-switching signal. This signal is delivered to switch S, which controls the current supplied to the LED.

Consequently, light energy emitted by the source LED is con¬ trolled by pulsing the LED with a constant amplitude current. LED-pulses may vary in duration, for example over a range of 1 ms to 19 ms. Through this process, the circuit MCC may detect abnormal degradation in the response of the system, and may lock the terminal and/or activate a service signal on an indi¬ cation display DIS when maintenance of the terminal is neces¬ sary. The circuit MCC can be used to diagnose the fingerprint sensor as well. For example, if the mirror layer (15, Fig. 8) is dam¬ aged, external light can enter the transducer lozenge and may degrade the image. The MCC detects such a failure by checking the output pulse of the comparator when the LED is switched off. When the LED is not activated, the comparator should pro¬ vide a very short output pulse signal. In the case of light leakage at the transducer, there will be a longer comparator output pulse signal. In such a condition, an appropriate maintenance signal will also be generated.

LED-life can be extended by switching off the source LED in times of prolonged non-use of the fingerprint terminal. This function can be controlled by the MCC. The terminal is ac- tivated by inserting a magnetic card into the card reader CRE. When the fingerprint verification procedure is completed, as determined by the control circuit MCC, appropriate signals are transmitted to the indication display DIS. Then, counting means, such as a register, in the microprocessor, may be ac- tivated. When a certain delay time limit is reached, the cir¬ cuit MCC switches off the LED and puts the fingerprint terminal into a stand-by mode, awaiting the next insertion of a magnetic card.

1 Optical System

2 Fingerprint Transducer Assembly (Lozenge)

3 Valley

4 Ridge

5 Finger

7 Light (ray);

7a Parallel Light;

7b Scattered Light

10 (Multilayer Contact Surface) Finger Transducer

11 Collimating Lens

12 Glass Substrate

13 Elastomer (Silicone)

14 Opaque Spacer Shim

15 Reflecting Layer (see also Mirror Plane 25)

16 Cuter Protective Layer

17 Light Source (LED)

18 Beamsplitter

19 Imaging Lens

20 Photo-Receiver ^•

21 CCD-Device (or Array)

22 Field Flattening Lens

23 Valley Image Plane

24 Ridge Image Plane

25 Mirror Plane

26 Annular (Bright Field) Stop

27 Central (Dark Field) Stop

28 Manufacturing Base (Polycarbonate)

29 Manufacturing Table (Glass Baseplate) LED Light Source

CCD Sensor (Charge Coupled Device)

SWG Sawtooth Waveform Generator

CD Comparator

CRE Card Reader

DIS Indication Display

CPU Central Processing Unit

PS Program Store (Program Memory)

DS Data Store (Data Memory) LT Look-up Table

S Switch

PL Pulse

MCC Microprocessor Controlled Circuit

Q Light Scattering angle

P Pitch

Claims

Claim 1

Method for generating an electrical output signal according to the ridges (4) and valleys (3) of a finger (5) placed against a contact surface (10) of a finger transducer said contact sur- face (10) comprising an elastomer (13) with a reflecting layer (15), by directing the beam from a transducer light source (17) at said reflecting layer (15) and detecting the light scat¬ tering from said reflecting layer (15), achieved by employing imaging of an image plane (24) close to but behind the re- fleeting layer plane (25) of said transducer to avoid spectral frequency doubling^'.

Claim 2

Method according to Claim 1 wherein said image plane is the virtual plane associated with said ridges (24).

Claim 3

Method according to any of Claims 1 or 2 wherein said finger transducer is designed to work, as a bright-field imaging trans- dύcer, whereby the light forming a focused image of said trans¬ ducer light source (17) in an imaging lens (13) of said trans¬ ducer is allowed to continue through said imaging (19) lens to illuminate a transducer receiver (21).

Claim 4

Method according to Claim 3 wherein a (bright-field) optical stop (26) is arranged close to said imaging lens (19).

Claim 5 Method according to any of Claims 3 or 4 wherein said trans¬ ducer light source (17) comprises a pulsing light source coupled to an electrical circuit (MCC) designed to control and diagnose said transducer.

Claim 6

Method according to Claim 5 wherein said electrical circuit comprises a microprocessor controlled unit coupled to said pulsing light source so as to control said pulsing light source (17) by varying the pulse width of the activating current of said light source, thereby defining the light energy transmit¬ ted by said light source (17) according to signals received from said receiver (21).

Claim 7

Method according to Claim 6 wherein said microprocessor con¬ trolled unit comprises a memory (LT) in the form of a look-up table for storing reference signals for the width of said pulsing signals.

Claim 8

Method according to any of Claims 1 to 7 wherein said contact surface comprises a lozenge (2) which provides a light scat¬ tering angle (0) proportional to the pitch (p) of said ridges (4) and valleys (3).

Claim 9 Method according 'to any of Claims 1 to 8 wherein said contact surface is a lozenge (2) sub-assembly comprising at least a re¬ placeable contact surface (10) and a collimating lens (11).

Claim 10 Method according to Claims 8 or 9 wherein said lozenge com¬ prises a silicone-type elastomer layer (13) of 300 to 500 micron thickness covered by an optically flat metal reflecting layer (15) of a thickness of less than 200 nm and a polyester protective layer (16) of 7 to 15 microns.

Claim 11

Method according to any of Claims 1 to 10 wherein said trans¬ ducer receiver comprises a photosensitive area such as a CCD area scan device and a field flattening lens (22).