GB2189635A

GB2189635A - Optical data correlation

Info

Publication number: GB2189635A
Application number: GB08709334A
Authority: GB
Inventors: Carol David Daniel
Original assignee: General Electric Co PLC
Current assignee: General Electric Co PLC
Priority date: 1986-04-21
Filing date: 1987-04-21
Publication date: 1987-10-28
Also published as: GB8609700D0; GB8709334D0

Abstract

Data e.g. representing fingerprints and stored on an optical disc 4 is correlated in parallel against test data from a source 16. The test data and an area of the disc 4 are illuminated with coherent light and the resultant signals are combined at a frequency plane correlator 8. An intensity peak occurs in the output of the frequency plane correlator 8 when the illuminated area includes data corresponding to the test data. The position and time of occurrence of the peak are used to identify that data on the disc. Alternative embodiments use spatially or time-integrating optics and dispense with the frequency plane correlator. <IMAGE>

Description

SPECIFICATION Optical data correlation This invention relates to a method of optical data correlation and to a system for performing the method.

An object of the invention is to provide a correlation system capable of correlating data at high speed, and in particular, capable of correlating large amounts of data in parallel.

According to one aspect of the invention a method of correlating a first unit field of test data against a second field composed of a plurality of different unit fields of sample data, comprises the steps of combining respective optical representations of the test and sample data and providing an output indicative of the location in the second field of the sample data corresponding to the test data. Preferably a plurality of the unit fields of sample data are correlated in parallel against the test data. The method may comprise the step of illuminating an optical disc on which is recorded the second field of sample data, to generate an optical representation of the sample data. Different areas of the disc may be successively illuminated.

The optical representations may be respective Fourier transforms of the sample and test data, which transforms are combined to produce an interference pattern having an intensity peak at a position corresponding to said location, and the position of that intensity peak is identified, thereby to identify the test data. The Fourier transforms of the test data and/or the sample data may be generated in operation of the system.

The optical representations may be superimposed and spatially integrated to produce said output, or the disc may be illuminated with light modulated by a signal representing the test data, and the resultant optical signal integrated over time to produce the output.

According to another aspect of the invention, a data correlation system comprises means for producing a first optical signal representing sample data from a high capacity optical data store, said store holding a plurality of unit sample data fields, means for producing a second optical signal representative of a unit test data field, means for correlating the first and second optical signals instantaneously so as to provide, in the event of substantial correlation between the test data field and a particular stored data field, an output signal indicative of the location and thus the identity of that particular data field in the data store. The high capacity optical data store may comprise an optical disc arranged to be illuminated to provide the first optical signal, which signal preferably represents a plurality of unit sample data fields.

Embodiments of the invention will now be described by way of example with reference to the accompanying drawings of which: Figure 1 shows schematically an optical correlation system in accordance with the invention; Figure 2 shows a frequency plane correlator for use with the system shown in Fig. 1; Figure 3 shows an optical look-up table for use with the system of Fig. 1; and Figures 4, 5 and 6 show alternative embodiments of the optical correlation system.

Referring to the drawings, Fig. 1 shows an optical correlation system comprising an optical disc 4, a frequency plane correlator 8, an an optical look-up table 14. Also included is a read/write head 5 controlled by signals from a control unit 7 to read data from or write data to the disc 4. A test data source 16 provides one input to the frequency plane correlator 8.

The disc contains millions of sample data points recorded as pits and ridges in the disc surface, representing hundreds or thousands of images, all of which are to be correlated against a single test image. An area of the disc 4 is illuminated by coherent light from a laser 1 via a lens 2, so generating an optical signal representing a large number of the images recorded. The data can be considered as being arranged in fields of unit size corresponding to the size of the test image. The disc rotates, as indicated by the arrow, so the unit fields illuminated are changing continuously.

One example of a use to which the correlation system may be put is the identification of fingerprints in which case the 'unit field' is one fingerprint. The fingerprints of a large number of criminals are recorded on the disc, and correlated in batches of, say, a few hundred at a time (depending on the area of illumination and the density of the recording) against a single fingerprint found at the scene of a crime.

The data stored on the disc, i.e. the sample data, does not necessarily represent images as such. Various techniques are used to record data of a variety of types on an optical disc, but the particular technique used is not important to the operation of this invention, provided always that the test data is recorded in the same form. Also, although the embodiment uses a transmissive disc, a reflective disc could be used equally well.

The data-bearing optical signal from the disc 4 is input to the frequency plane correlator 8 described below with reference to Fig. 2. A second input to the frequency plane correlator is provided from the test data source 16. The output beam from the correlator 8 is transmitted to the optical look-up table 14, also described below, which measures the output beam position.

This beam position is then fed to the control unit 7.

The control unit 7 monitors the rotation of the disc 4 and on receipt of the output beam position signal controls the read/write head 5 to move to the required radial position and to read at the point in time such that the sample data read out corresponds with the test data from source 16. This data is then output in suitable form on output 26 which may lead, for example, to a VDU or a hard copy printer, or to further processing equipment.

The control unit 7 also controls the read/write head 5 to write fresh sample data onto the disc 4, updating or extending the data-base.

Referring to Fig. 2, this shows in detail the frequency plane correlator 8. Incoming beams u1, u2 respectively from the test data source 16 and the disc 4 pass through respective Fourier transform lenses L1, L2. The Fourier transform (FT) of the test data is recorded as a refractive index grating in a cell 12 of photorefractive material such as Bi12 Si 020(BSO). This cell is positioned in what may be called the Fourier plane at the focus of both lenses L1 and L2. It presents to an incoming beam an optically flat surface, so when the cell is coherently illuminated with the transform of the sample-data signal, the resultant optical signal is still coherent.An argon ion laser (,=514.5nm) is used to record the refractive index grating, as the sensitivity of BSO is greatest in the blue green part of the spectrum, whilst the sample data is read using a longer wavelength HeNe laser 1 (AR=633nm), which has little effect on the grating because BSO has very low photorefractive sensitivity in the red. Therefore the sample data beam interacts with the grating only in the sense that it is diffracted by it; the sample data beam does not modify the grating itself. The sample data can therefore be changed arbitrarily quickly. Limitation in the speed is set by the rate at which the sample data can be scanned, i.e. the rate of rotation of the disc 4, and not by the response time of the photorefractive material.

A third beam u3 is required to record the grating distribution as a uniform plane wave. This third beam u3, which is at the same wavelength, Aw, as the test data beam, has the form of a spatial Dirac J-function.

The theory of operation of the correlator is as follows: The optical processor considered here utilises the pattern analysis technique of image correlation. By the correlation theorem, the Fourier transform of the cross correlation of two functions is equivalent to a multiplication of their complex field amplitudes in the Fourier plane, i.e.

(1) Ui(x,y)#Uj(x,y) = F(Ui*(x',y') #Uj (x',y') ) where # denotes correlation and * denotes the complex conjugate.

The above operation can be performed optically utilising the Fourier transforming properties of a lens, provided that a means of multiplying the transformed fields can be found. In the system described here, use is made of the non-linear response of the photorefractive BSO cell 12 placed in the Fourier plane, which can, under certain conditions, produce the required field multiplication.

The input beam u, (x, y) representing the test data is placed on the front local plane P, of lens L,. The beam u3 (x, y) which has the form of a spatial Dirac function is also present in this plane. The lens L, produces the Fourier transforms U, (x', y') and U3 (x', y') at the photorefractive cell 12. In the cell 12 the Fourier transforms of the two input fields interfere and produce a phase grating whose complex modulation index m,3 is given by 2U1 (X',y')*. U3(X',y') m13 = (2) (1U1(x',y1)12 + 1u3(x',y')12 The form of the beam u3 (x, y) gives a resultant uniform plane wave in the Fourier plane, thus equation (2) may be rewritten as: 2u1(xI,yl)* U3 m13 = (3) (1U1(x',y1)12 + I3) where U3 is the constant field amplitude of beam U3 (x, y) and 13 is the intensity.

The area of the disc 4 currently illuminated is represented by distribution u2 (x, y). The lens L2 produces the spectrum of this distribution, Fourier transform U2 (x', y'), which is diffracted by the photo-induced grating to produce the output field U4 (x', y') given by (4) U4(xY) = A m131U2(x',y1) where m,3' is the modified modulation index in the presence of the disc data transform beam and A is a parameter dependent on the material and operating conditions.The overall effect is to produce a field multiplication given by: (5) U4(x',y') = U1*(x',y1) U2(X',Y') W(x',y') where w(xa,ys) r 2A (6) (1U1(x',y')12 + 13 + aIU2(x',y')12) The parameter 'a' reflects the degree of interaction of the reading beam, i.e. the disc data transform beam, with the grating. For a degenerate system (AR=AW) 'a' approaches unity but in the case presented here a 1, hence the last term in the denomninator may be neglected.

It can be seen from equation (5) that the photorefractive response can produce the required spectra multiplication such that after retransformation by L, the output u4 (x, y) at the plane P4 represents the two dimensional correlation of the input functions. For true correlation the weighting function must be spatially constant i.e. tU,(x', y')l2 13 and for high spatial frequency weighting (edge enhanced) correlation jU1(x', y')|2 13. By varying the intensity ratio of the writing beams u, (x, y) and u3 (x, y) the correlation function discrimination can be altered in real-time.

In an alternative arrangement, the lens L2 can be disposed of, and the data recorded onto the disc directly in its FT form. The FT is calculated by the computer and written directly onto a disc. In this case the FT will actually be in analogue format, but the operation of the system is identical. The FT may also be generated optically, as it is in the frequency plane correlator 8, but before the disc along the optical path. The data images are projected onto a spatial light modulation. The SLM is illuminated with coherent light, and the coherent image of the data is passed through a FT lens, in combination with a function distribution. The resultant spectrum is recorded on the optical disc 4.

The disc data and test data spectra are multiplied together as described above, for example by reflecting the beam U1 (x', y') off the disc, then inverse transformed to produce the correlation image. In general, it will be preferable to produce the FT of the test data as part of the correlation operation, so that it can be altered immediately at any time. However, it would equally be possible to have the FT of the test data permanently recorded, in the same way as the sample data. If both data are recorded in FT form, it is only necessary to directly image one field onto the other, before inverse transforming.

As the disc 4 rotates, successive areas containing a portion of the total sample data field are illuminated, all the unit fields in each area being correlated instantaneously in parallel against the test data. The area illuminated may not extend across the whole radius of the disc 4, in which case two or more rotations of the disc would be required in order to cover data recorded across the whole disc, the light source means 1, 2 being adjusted, by means not shown, to illuminate areas at different radial positions. The instantaneous rotational position of the disc and, where appropriate, the light source direction, are monitored by the control unit 7.The two dimensional output beam u4 (x, y) has an instantaneous intensity peak when the area currently being illuminated contains amongst many unit data fields that one (e.g. the one fingerprint) which corresponds exactly to the test data field (the culprit's fingerprint). This intensity peak is detected at the optical "look-up table" 14, shown in detail in Fig. 3.

The intensity peak is shown as beam 15. This beam is split by cube beam splitter 9 into two channel beams 15" for the x-coordinate channel and 15' for the y-coordinate channel. The two channels are treated similarly, but Fig. 3 shows the optical array for just the y-coordinate channel 15'. The beam is passed through anamorphic optics 10 which bring the beam 15' to a horizontal line focus on a binary-coded mask 12. Further anamorphic optics 11 focus the line of light onto a horizontal linear array of photodiodes 13. The mask 13 is divided into vertical strips each of which is in turn divided into sections which are alternately transmissive and nontransmissive. Each strip of sections is differently coded to provide a binary read-out of the vertical position of the horizontal line.The minimum height of a section i.e. the height of the sections in the highest resolution strip, is determined by the width of the "line" focus. Each strip has a corresponding photodiode in the detector array 13 which registers a signal when the line-focussed beam strikes a transmissive section in that strip. Each possible vertical position of the beam within the mask resolution gives a different combination of signals at the detectors 13 thus providing a unique indication of the vertical coordinate of the incoming beam, intensity peak 15.

The x-coordinate channel is processed similarly, using a mask having horizontal strips with the beam 15" brought to a vertical line focus. The corresponding detector array must of course be a vertical linear array.

As an alternative to the optical look-up table, described above, a two-dimensional array of photodetectors may be located in the plane P4 to provide the x and y coordinates. The resolution of this system would depend simply on the size and number of detectors in the array.

The x and y-coordinates of the intensity peak are fed to control unit 7 which combines the information with information on the position of the disc 4 to control the read/write head 5 to read the sample data which corresponds to the test data.

Alternative embodiments of the invention which exclude the frequency plane correlator 8 will now be described with reference to Figs. 4, 5 and 6.

These embodiments allow the correlation of one or two dimensional data on the optical disc 4 using spatially or time integrating optics. Dispensing with the frequency plane correlator also dispenses with the requirement to use a coherent light source.

Referring to Fig. 4 a light source 3 illumintes a one dimensional spatial light modulator (SLM) 16 such as a cell of photorefractive material similar to cell 12 used in the frequency plane correlator 8. A lens 6 focusses the light onto the SLM 16 which is modulated by a beam g(x) with a representation of the test data. Anamorphic optics 17 focus the modulated light to illuminate an area on the rotating optical disc 4 thereby addressing a plurality of one dimensional data channels on the disc. An array of photodetectors 19 includes one detector for each data channel in the illuminated area. Integrating optics 18 integrates each channel of light from the disc and focuses it to its respective photodiode in array 19. The effect of illuminating the sample data on the disc with light modulated by the test data at SLM 16 is to multiply together the two distributions.The lens 18 then spatially integrates (over one dimension) and for each channel the output at its respective photodiode is given by h(t) = ff(x-vt) g(x) dx where f(x-vt) is the sample data distribution, which rotates at speed v, and g(x) is the test data distribution.

The output h(t) has a sharp peak when the two data correspond. As in the embodiment of Fig. 1, a control unit (not shown in Fig. 4) monitors the disc position, and controls the read/write head 5.

The embodiment shown in Fig. 5 also uses spatial integation but correlates only one image or unit data field at a time. Lens 20 focuses light from a source 3 onto an area of the disc 4. The light then passes through a two-dimensional SLM 21, which is coded with the test data by beam g(x, y) thereby multiplying together the sample data and test data distributions. An integrating lens integrates the resultant light over the whole plane and focuses it onto a photodiode 23. Equation (1) again applies, with a peak occurring in the output from photodiode 23 when the data in the entire illuminated area corresponds to test data represented on the SLM.

In the embodiments shown in Figs. 4 and 5 it is necessary to ensure correct scaling of the two data distributions. Optical data storage is very compact so that a small area of disc contains a very large amount of data. However it may not be practicable to build an SLM at the same very small scale so that, in Fig. 4, lens 17 may also serve to reduce the image from SLM 16, and in Fig. 5, magnifying optics may be required between the disc 4 and SLM 21.

A time integrating embodiment is shown in Fig. 6. Light from a source 3' (which again need not be coherent) is modulated by a signal representing the test data. The time-varying optical signal is focused by the lens 20 onto the rotating disc 4 and then by lens 25 onto a photodetector array. The photodetectors integrate over time and may be, for example, charge coupled devices. The time integrating version of the correlation equation thus applies: h(x) = ff(x'-vt) g(t) dt A photodetector is provided for each channel in the illuminated area and the width of each detector corresponds to the length (in space) of the modulating test data signal as it would appear on the disc. The modulating signal is repeated continuously as the disc rotates, so that all the data on the disc is exposed to and correlated against the test data distribution.

Claims

1. A method of correlating a first unit field of test data against a second field composed of a plurality of different unit fields of sample data, comprising the steps of combining respective optical representative of said test and sample data and providing an output indicative of the location in the second field of the sample data corresponding to said test data.

2. A method according to Claim 1 wherein a plurality of said unit fields of sample data are correlated in parallel against said test data.

3. A method according to Claims 1 or 2 comprising the step of illuminating an optical disc on which is recorded said second field of sample data, to generate an optical representation of said sample data.

4. A method according to Claim 3 wherein different areas of said disc are successively illuminated.

5. A method according to any preceding claim wherein said optical representations are respective Fourier transforms of said sample and test data, and said Fourier transforms are combined to produce an interference pattern having an intensity peak at a position corresponding to said location, and the position of said intensity peak is identified thereby to identify the test data.

6. A method according to Claim 5 comprising the step of generating in operation the Fourier transform of said test data.

7. A method according to Claim 5 or 6 comprising the step of generating in operation the Fourier transform of said sample data.

8. A method according to any of Claims 1 to 4 wherein said optical representations are superimposed and spatially integrated to produce said output.

9. A method according to any of Claims 1 to 4 in which the disc is illuminated with light modulated by a signal representing said test data, and the resultant optical signal is integrated over time to produce said output.

10. Correlation apparatus adapted to perform a method as claimed in any preceding claim.

11. A data correlation system comprising a high capacity optical data store, means for producing a first optical signal instantaneously representative of at least one of a plurality of unit data fields in said store, means for producing a second optical signal representative of a unit test data field, means for correlating the first and second optical signals instantaneously so as to provide, in the event of substantial correlation between the test data field and a particular stored data field, an output signal indicative of the location and thus the identity of said particular data field in the data store.

12. A data correlation system according to Claim 11 wherein said high capacity optical data store comprises an optical disc arranged to be illuminated to provide said first optical signal.

13. A data correlation system according to Claim 11 or 12 wherein said first optical signal simultaneously represents a plurality of said unit sample data fields.

14. A data correlation system according to Claim 13 comprising means to generate in operation the Fourier transform of said test data.

15. A data correlator according to Claim 13 or 14 comprising means to generate in operation the Fourier transform of said sample data.

16. A data correlation system according to Claim 11, 12 or 13 wherein optical data signals from said first and second data sources are superimposed and spatially integrated to produce said output.

17. A data correlation system according to Claim 12 or 13 in which the disc is arranged to be illuminated with light modulated by a signal representing said test data, the resultant optical signal being integrated aver time to provide said output.

18. A data correlation system substantially as hereinbefore described with reference to the accompanying drawings.