GB2507467A - Operating a spatial light modulator - Google Patents

Abstract

There is provided a method of optical correlation between an input image and a reference image. The method comprising representing a correlation pattern on pixels of a spatial light modulator or SLM 201, illuminating the pixels with collimated light 220 to form an output beam comprising spatially-modulated light and zero-order light and applying the output beam to a Fourier transform lens. The correlation pattern comprises a lensing effect (on pixels 222) to move the far-field away from the focus of the zero-order.

Description

A REDUCED NOTSE OPTICAL DEVICE

Field

The present disclosure relates to a method of optical correlation. Embodiments disclosed herein relate to a method of suppressing noise in an optical correlator. More specifically, embodiments relate to a method of removing zero-order noise from an optical correlator.

Background

An optical correlator is a device for comparing a first image with a second image. An optical correlator may therefore be used to detect the presence of a reference object in a scene andlor determine the location ol [he reference object in the scene. The first image may be the reference object and the second image may be a input scene, such as a photograph, which may I S or may not contain the reference object.

There is interest in optical correlation techniques based on the use of Fourier transforms because it is possible to achieve an optical Fourier transform instantly, for example, using a simple converging lens The most commonly used proccssing architectui-e.s for optical correlation are tb.e joint transform correlator (JTC) and the Vander-Lugt correlator (VLC).

Both approaches are based on a comparison between a target image and a reference image (for example, from a database) and simple detection of a correlation peak measuring the degree of similarity between the target and the reference. Both techniques require two Fouricr tiansform (FT) Jenses (each haviig a focal length off) arranged in a so-called 4f optical con figuration resulting in a long optical path length.

Whilst the JTC is less aligmnent sensitive, its detection efficiency is lower and it takes longer to perform the correlation, particularly for multi-target recognition. Thc VLC has a higher detection efficiency but is Jess robust against misaligmnent.

Figure 1 shows an example of a. YLC including an input plane 101, a first FT lens 103, a Fourier plane 105, a second FT lens 107 and an output plane 109 respectively arranged on a common optical axis.

The VLC is based on the multiplication of the spectrum of a target image by a correlator filter made from a reference image. A linearly polarised collimated beam is used to illuminate the input plane 101 comprising of a target image displayed on a spatial light modulator (SI M). A first FT lens 103 carries out the Fourier transform of the object or target image. This results in the spectrum of the target image at the Fourier plane 105. The Fourier plane 105 is effectively the back focal plane of the first FT lens 103. A correlation lilter is placed (displayed on a SLM) in the Fourier plane 105 to coincide with the spectrum of the target image. A second FT lens 107 carries out the inverse transform of the Fourier plane 105. A CCD located at the back focal length of the second FT lens 107 is typically used to record the system output at an output plane 109. The similarity between the targc image and tile correlation filter is a.chicvcd by detection of' a correlation peak at the oulpuL plane 109.

Classical VCL architectures are prone to alignment sensitivity of the optical components. For example, a relatively small defocus crror th the Fourier plane can result in significant loss of colTelation signal.

An overview of these correlation techniques may be found in "Understanding Correlation Techniques for Face Recognition: From Basics to Applications" by A. A.lthlou and C. Brosseau, in Face Recognition, In-Tech, ISBN 978-953-7619-00-X.

Typical VIf devices usc two spatial light modulators (SLMs) and two FT lenses. However, a lensless VLC using onc phase-only SLM is disclosed in "Compact optical correlator based on one phase-only spatial light modulator" by Zeng ci a!, Optics Letters, Vol. 36, No. 8, 15 April 2011, page 1 383-1385. The system is folded by using a mirror and displaying the input and reference pattern on different halves of the same SLM. A lensless configuration is realised by adding a predetermined Fresnel lens pattern (FLP), and optionally a grating, to the input and the reference patterns.

The present disclosure provides an iniprovcd compact optical correlator based on one SLM. Summ

Aspects of the present discksure arc deflned in the appended independent claims.

There is provided a method of adding a lensing effect, such as data representative of a virtual lens, to a correlation pattern represented on a *SLM for optical correlation. Accordingly, zcro-order noise may be separated from information content to increase the signal-to-noise rati.o of the detected signal.

Further advantageous embodiments provide an improvcd device by using a combination of a physical Fourier transloirn lens and a Fresncl lens pattern to perform Fourier transforms necessary for optical correlation. Accordingly, a commercially-viable balance betwcen processing demands and physical size is struck.

Brief description of the drang

Embodiments of the present invcntion will now be described with reference to the accompanying drawings in which: Figure us an all optical correlation setup for optical correlation; Figure 2 is a Vander-Lugt correlator iii accordance with embodiments; Figure 3 is a joint transform correlator in accordance with embodiments; and Figure 4 is an example liquid crystal on silicon spatial light modulator in accordance with embodiments.

in the figures like reference numerals referred to like pans.

Detailed description

An "input image" may be a digital image of a scene, which may or may not contain a predetermined object of interest. The input image forms the input for an optical correlator.

That is, the input image is probed to determine the likelihood that the objcct of interest is in the image and/or where in the image the object is located. The input image may be a captured image such as a photograph or a sub-region of a photograph.

A "reference image" may be a digital image of a predetermined object of interest. The reference image is effectively scanned across the input image to determine if the object of interest is anywhere in the input image. The reference image may be stored in a repository such as a database of reference objects.

Embodiments relate to adding a "lensing effect" to correlation data. It can be understood from the present disclosure that a lensing effect is an effect which causes lensing of light such as locusing, converging or diverging. The term therefore encompasses adding optical or refi-active power. A lensing effect may be provided by adding (summing) lens data to the correlation data, wherein the lens data. is data representative of a virtual lens and causes incident light to behave in the same way as if a physical optica1 lens of the same optical power were used.

Figure 2 shows a Vander-Lugt optical eorrelator in accordance with an embodiment.

There is shown a SLM 201 comprising an array of pixels 203. The SLM shown in figure 2 is configured to function in a. reflectivc modc. That is, incident light is spatia.Uy modified by the SLM and output in reflection. SLM 201 may therefore be termed a reflective STM. There is also shown a mirror 205 and a charge-coupled device (CCD) 207. As shown in figui-e 2, the SLM is spatially-separated and substantially parallel to minor 205. The CCII) 207 is arranged to receive output light after two reflections off both the SLM 201 and the mirror 205.

A first subset of pixels 221 of the STIM are arranged to represent an input pattern. That is, a predetermined pixelated pattern is written to the first subset of pixels 221. The first subset of pixels 221 may he said to "display" the input pattern. The input pattern is representative of a input image. The input pattern may he a binary representation of the input image.

A second subset of pixels 222 of the SLM arc arranged to represent a filter pattern. That is, a second prcdctcrmined pixclatcc! pattern is written to the pixels in the second subset of pixels 222. The second subset of pixels 222 may be said to "display" the filter pattern. The filter pattern is representative of a reference image. In an embodiment, the filter pattern comprises a Fourier transform of the reference image. In an embodiment, the filter pattern may be a binary representation of a Fourier transform of the reference image.

In operation of the embodiment shown in figure 2, a plane wave of light 220 is incident on the first subset of pixels 221 of the SLM. The incident light is spatially modl.Tiated by the pixels in the first subset of pixels 221. The angle of incidence between the plane wave and the normal of the SLM is greater than zero. Mirror 205 reflects the spatially modulated light back towards the SLM. The skilled person will therefore understand that the angle of incidence between the plane wave and the SLM, and the spacing between the LSLM and mirror, is chosen such that the light is directed to the second subset of pixels 222. That is, a different part of the SLM. The second subset of pixels 222 ftrther spatially modulates light. More specifically, the light is modulated by Lile first subset of pixels 221 and thcn by the second subset of pixels 222. Light from the second subset of pixels 222 is directed onto a first region of CCD 207 optionally via a further reflection off mirror 205 as shown in figure 2.

The filter pattern may he any suitable filter or matched pattern for optical correlation. In an embodiment, the filter pattern is a Fourier-transform based filter pattern. That is, a filter pattern based on a Fourier transform of the refercace image. In the figure 2 ernbodii-nent, the filter pattern is a Vauder-Lugt matched pattern. The skilled person will know how to create a filter pattern from a reference image for optical correlation. The skilled person will also understand how to perform the nccessary Fourier transforms For optical correlation. Metl1ods for performing the necessary Fourier transforms for optical correlation are described below.

In an embodiment, the filter pattern is a phase-only representation of the Fourier transform of the reference image.

The skilled person will understand that using the correct filter pattern and two Fourier transforms, an optical correlation between an input image (represented in the input pattern) and a respective reference image (represented in a filter pattern) may be performed. Optical correlation maybe detected and measured using CCD 207. Correlation peaks appear at points in the received light beam where there is a correlation between the input image and the reference image. The spatial position of correlation peaks indicates points in the input image where there is a correlation with the respective reference image. The height of the corrclaticn indicates the extent of the correlation. That is, high peak indicates strong correlation.

The correlation method requires a first Fourier transform: a Fourier transform of light after spatial modulation by the input pattern. This is achieved by using a physical Fourier transform optic present after the SLM (not shown in figure 2).

The correlation method requires a second Fourier transform: a Fourier transform of the light after spatial modulation by the filter pattern. In an embodiment, this is achieved by using a second physical Fourier transform lens between the SLM and minor. In an alternative embodiment, the filter pattern further comprises information representative of a Fresnel lens pattern (FLP) arranged to perform the second Fourier transform. This ma.y be achieved by simple addition of FLP data. lhc FLP may be considered a virtual Fresnel lens. lEe skilled person will understand how to calculate the required ELI1 and add it to image data. In a frirther alternative embodiment, a physica.l Fourier transform lens and a FLP are used in conjunction to perform the respective Fourier transforms. Advantageously, a hybrid system in which the Fourier transforms are performed by a FLP in conjunction with a physical Fourier transform lens allows for the physical size of the device to be decreased whilst not requiring excessive processing power. The filter pattern may also comprise adding a grating pattern to simplify optical aligmnent during assembly.

A problem exists with pixellated SLM devices in that they exhibit the well-known problem of zero-order light. Such zero-order lighL can be regarded as "noise" and includes for example specularly reflected light off the SLM, and other light that is not modulated by the pixels on the spatial light modulator. The specularly reflected light includes light reflected off thc front surface of the SLM such as light reflected off the interpixci grid separating the pixels. This zero-order noise manifests itself as a bright spot of light in the middle of the output beam. For optical correlation, the zero-order is problematic because the bright spot affects techniques I.o analyse th.e detected signal for correlation peaks and, furthermore, correlations cannot be detected in the region occupied by the zero order light. II can be understood Uat the output light from the SLM comprises: (1) spatially-modulated light having information content; and (ii) zero-order light which is noise.

Thc present disclosure addresses this problem by spatially-separating the locus of the zero-order noise from that of the information content. This is achieved by adding a lensing effect to the input pattern. This may be considered a type of virtual lens. The skilled person will understand how to add to virtual lens data to the image data. Accordingly, the zero-order light 213 is brought to a focus at a different point in space to the information content 215 oF the output light -that is, the portion of the light that bus been spatially modulated. As shown in the figure 2 embodiment, the zero-order light 213 is brought to a focus on the minor. A carefully positioned aperture 211 ill the mirror may therefore be used to extract the zero-order light 211 from the system. In other words, the virtual lens pushes the far-field awa.y from the zero order. This improves the signal-to-noise ratio of the correlator. It also prevents the large magnitude zero order form saturating the CCD causing charge overflow, which limits the sensitivity of the device. The skilled person will understand how to calculate the necessary optical powers and distances to remove the zero-order in this manner.

The present disclosure is equally applicable to ajoint-transform correlator.

Figure 3 shows ajoint-transform correlator in accordance with the present disclosure.

SLM 401 is arranged to display the joint (Fourier) transfbi-n-i 421 of the input image and the reference image. It is known in the art how to produce a joint-transform for [he purposes of' optical correlation. A light source 41.3 and beam expanding optics 415 arc used to illuminate the SLM 401. via bcarn splitter 405. Spatially-modulated light is reflected from SLM 401 and inverse Fourier transformed by Fourier lransfonn lens 409. The correlation output is received by CCD 407.

Again, by adding virtual lens data to the joint transform pattern, thc far-field is pushed away from the zero order. Mores specifically, the zero order is brought to a foetis on bearnsplitter 405. The zero order may therefore be extracted from the system using a carefully positioned extraction point 417, for example. Tn an embodiment, the zero order cxti-action point also acts as the light insertion point. Whilst figure 3 shows a reflective SLM, thc present disclosure is equally app! icable to a transmissive SLM.

The input pattern in the Yander-Lugt correlalor embodiment (figure 2 embodiment) may be termed a "correlation patterii". The pattern produced by the joint transform of the input image and the reference image in the Joint-transform correlator embodiment (figure 3 embodiment) may also be termed a "correlation pattern". In embodiments, a lensing effect is added to thc "correlation pattern" to shift the zero-order in space away from the far-field containing the correlation information content of the light.

There is therefore provided a method of optical correlation between an input image and a reference image. the method comprising: representing a correlation pattern on pixels of a spatial light modulator; illuininatiug the pixels with collimated light to lbrnt an output beam comprising spatially-modulated light and zero-order light; and applying the output beam to a Fourier transform lens, wherein the correlation pattern comprises a Jensin.g effect arranged to move the far-field away from the focus of the zero-order.

In embodiments, the method further comprises extracting zero-order light at the focus of the Fourier transform lcns. Optionally, the zero-order light is brought to a focus before the detector so it can be removed from the system with minimal disruption to the information content portion of the output beam, It can be understood that the term "extracting" means removing the zero-order noise from the system. This includes separating the zero-order noise from the information content. In this context, the term includes absorbing, tnuismitting away, reflecting away or otherwise taking the zero-order light out the optical system such that is does not reach the CCD.

The number of pixels in the input pattern and reference patteni dcfmes the "resolution" of' the colTelatlon. The more pixels, the more accurate the correlation result or the larger the area scanned. The skilled person will know how to represent the patterns on the SLM. The skilled person will also know how to detect correlation peaks.

In an embodiment, the patterns represented on the SLM comprise phase-only information. In an embodiment, the patterns are binary (that is, each pixel takes one of only two possible values). In such embodiments, the SLM may therefore be a binary phase-only SLM. That is, a SIM which spatially modulates only the phase (and not amplitude) of an incoming light beam. Advantageously, phase-only SLMs are more light efficient because light is nol lost from the system by niodulatin.g die amplitude (intensity).

Whilst embodiments relate to a mirror, the skilled person will understand that other reflective components may be ejually suitable. Likcwise, whilst embodiments relate to a CCII), it can he understood that any spatial light detector is suitable.

Analysis of the result of the optical correlation may include, for example, determining ii there is at least one correlation peak having a height greater than a predetermined threshold. For example, if there is at least one correlation peak having a height greater than a predetermined threshold, it may he determined that an object (represented in the corresponding reference image) is present.

S

In an embodiment, the SLM is a liquid crystal on silicon (TCOS) SLM. However, the skilled person will understand that other types of SLM may he equally compatible with the present

disclosure.

1 he structure of a suitable LCOS device is shown in Figure 4.

A JICOS device is formed using a single crystal silicon substrate (302). It has a 2D array of square planar aluniiniuni electrodes (301), spaced apart by a gap (301 a), arranged on the upper surface of the substrate. Each of the electrodes (301) can be addressed via circuitry (302a) buried in the substrate (302). Each of the electrodes forms a respective planar minor.

An alignment layer (303) is disposed on the anay of electrodes, and a liquid crystal layer (304) is disposed on the alignment layer (303). A second aligmnent layer (305) is disposed on the liquid crystal layer (404) and a planar transparent layer (306), e.g. of glass, is disposed on the second aliiment layer (305). A single transparent electrode (307) e.g. of ITO is disposed between the transparent layer (306) and the second alignment layer (305).

Each of the square electrodes (301) defines, together with the overlying region of the transparent electrode (307) and the intervening liquid crystal material, a controllable phase-modulating element (308), oflen ieferred to as a pixel. The effective pixel area, or fill factor, is the percentage of the total pixel which is optically active, taking into account the space between pixels (3Ola). By control of the voltage applied to each clectrode (301) with respect to the transparent cJcetrode (307), the properties of the liquid crystal material of the respective phase modulating element may be varied, thereby to provide a variable delay to light incident thereon. Ihe effect is to provide phase-only modulation to the wavefront, i.e. no amplitude effect occurs.

An advantage of using a reflective LCOS spatial light modulator is that [lie liquid crystal layer can be half the thickness than would he necessary if a transmissive device were used.

This greatly improves the switching speed ol the liquid crystal (a key point for projection of moving video images). A TCOS device is also uniquely capable of displaying large arrays of phase only elements in a small aperture. Small elements (typically approximately 10 microns or smaller) result in a practical diffraction angle (a few degrees) so that the optical system does not require a very long optical path.

It is easier to adequately illumirrnte the small aperture (a few square centimetres) of a LCOS SLM than it would be for the aperture of a larger liquid crystal device. LCOS SLMs also have a large aperture ratio, there being very little dead space between the pixels (as the circuitry to drive them is buried under the mirrors). This is an important issue to lowering the

optical noise iii the replay field.

The above device typically operates within a tcmperatuTe range of 10°C to around 30°C, with the optimum device operating temperature being arouid 40°C to 50°C, depending however on the LC composition used..

Using a silicon backplane has the advaniage that the pixels are optically flat, which is important for a phase modulating device.

Whilst embodiments relate 10 a reflective LCOS SLM, ihe skilled person will understand that any S'LM can be used including transmissive SLMs.

The invention is not restricted to the described embodiments hut extends to the frill scope of the appended claims.

Claims (13)

1. Claims 1. A method of optical correlation between an input image and a reference image, the method comprising: representing a correlation pattern on pixels of a spatial light modulator "SLM"; illuminating the pixels with collimated light to form an output beam comprising spatially-modulated light and zero-order light; and applying the output beam to a Fourier transform lens.wherei.n the correlation pattern comprises a lcnsing effect to move the far-field away from the focus of the zero-order.
2. 2. The method of claim 1 wherein die correlation pattern is ajoint-Irarisforni matched filter.
3. 3. The method of claim 1 or 2 wherein the correlation pattern is a joint-trurisformn of an input image and a reference image.
4. 4. The method of claim 1 wherein the filter pattern is a Vander Jugt matched filter.
5. 5. The method of claim 1 or 4 wherein the correlation pattern is an input pattern representative of an input image.
6. 6. The method of claim 4 or 5 ftrther comprising: illuminating a filter pattern, representative of die reference image, to perform an optical correlation between the input image and the reference image.
7. 7. The method of any of claims 4 to 6 wherein the lilter pattern further comprises: adding a Fresnel lens pattern; andlor adding a grating pattern.
8. 8. The method of any preceding claim wherein the lensing effect is a negative lensing effect.
9. 9. The method of any preceding claim further comprising extracting the zero-order light at the focus of the Fourier transform lens.
10. 10. The method of any preceding claim wherein the correlation pattern further comprises adding a Fresnel lens pattern
11. 1 1. The method of any preceding claim wherein the correlation pattern comprises phase-only inlbrniation.
12. 12. A.n opticai correlator for performing the method of any preceding claim.
13. 13. A method of operating a spatial light modulator or an optical correlator substantially as hereinbefore described with reference to the accompanying drawings.