EP0594668A1

EP0594668A1 - Optical signal processing

Info

Publication number: EP0594668A1
Application number: EP92914231A
Authority: EP
Inventors: Herbert Alan French; Philip Sutton
Original assignee: UK Secretary of State for Defence
Current assignee: UK Secretary of State for Defence
Priority date: 1991-07-12
Filing date: 1992-06-29
Publication date: 1994-05-04
Also published as: JPH06509169A; GB9400341D0; CA2113270A1; WO1993001477A1; GB9115158D0; GB2272765B; GB2272765A

Abstract

Un système de détection optique, composé d'un dispositif de transformation modulant le rayonnement reçu de façon à sensibiliser le système à des caractéristiques sélectionnées dépendant du profil de la fonction de cohérence dans le champ optique reçu, comprend un processeur spectral. Le processeur spectral possède un filtre limiteur de bande (BLF) et un filtre modificateur (MF) ayant une bande passante recouvrant au moins partiellement la bande passante du BLF (103). Le BLF et le MF sont sélectionnés de manière à concentrer l'information clé sur le rayonnement cible dans une petite région du profil de la fonction de cohérence. De préférence, le BLF a un profil de transmission rectangulaire et le MF a un profil gaussien, correpondant à celui de la ligne d'absorption du gaz à détecter. Le centre du profil d'absorption du MF est décalé par rapport au centre du profil du BLF. Dans d'autres configurations, le MF peut être tel qu'il produit deux ou plusieurs profils d'absorption (122, 123) dans le profil du BLF. La fonction de transmission d'un des filtres peut être dépendante du temps de manière à améliorer la détection du champ optique transformé. Cette dépendance est obtenue grâce au mouvement périodique du filtre produit par la variation de l'angle d'inclinaison d'un filtre d'interférence pour moduler la fréquence centrale du profil de transmission du filtre par un système faisant en sorte que la surface du filtre périodiquement déplacé, inséré dans le chemin optique du système, varie de manière à moduler périodiquement la profondeur d'absorption du filtre.An optical detection system, composed of a transformation device modulating the received radiation so as to sensitize the system to selected characteristics depending on the profile of the coherence function in the received optical field, comprises a spectral processor. The spectral processor has a band limiting filter (BLF) and a modifying filter (MF) having a bandwidth at least partially overlapping the bandwidth of the BLF (103). The BLF and MF are selected in such a way as to concentrate the key information about the target radiation in a small region of the profile of the coherence function. Preferably, the BLF has a rectangular transmission profile and the MF has a Gaussian profile, corresponding to that of the absorption line of the gas to be detected. The center of the MF absorption profile is offset from the center of the BLF profile. In other configurations, the MF may be such that it produces two or more absorption profiles (122, 123) in the profile of the BLF. The transmission function of one of the filters can be time-dependent so as to improve the detection of the transformed optical field. This dependence is obtained thanks to the periodic movement of the filter produced by the variation of the angle of inclination of an interference filter to modulate the center frequency of the transmission profile of the filter by a system ensuring that the surface of the filter periodically moved, inserted in the optical path of the system, varies so as to periodically modulate the depth of absorption of the filter.

Description

OPTTCAL SIGNAL PROCESSING

The invention relates to optical detection systems and in particular to selective processing of an image signal field in the optical domain before signal detection so as to enhance the signal to noise ratio in the system.

European Patent No 0155142 describes a remote gas detection system relying on signal processing in the optical domain using principles of optical transform image modulation (OTIM).

Changes in the optical field resulting from such modulation techniques can then be measured by making use of a detection arrangement as described in UK Patent Application 9200074.

The object of the present invention is to provide a spectral pre-detection processing arrangement to enhance the performance of optical detection systems such as the above-mentioned remote gas sensor.

The invention provides an optical detection system comprising a) means to receive radiation from a target within a field of view b) transform means to modulate the received radiation to sensitise the system to selected features dependent on the coherence function profile within the received optical field; and

c) detection means responsive to the modulated received radiation;

characterised in that the transform means includes a spectral processor having: (i) a band limiting filter (BLF) ; and

(ii) a modifying filter (MF) having a pass band at least partly overlapping the passband of the BLF;

the characteristics of the BLF and the MF being selected to concentrate key information on the target radiation in a small region of the coherence function profile.

In one arrangement the BLF and MF have Gaussian transmission profiles and the MF profile is centred in the BLF profile. Alternatively the centre of the MF profile may be offset with respect to the centre of the BLF profile.

In an arrangement for detecting a target gas, the MF is arranged to have the same profile as an absorption band of the gas but centred at a different wavenumber. This arrangement leads to improved detection.

In a further arrangement the MF can have a transmission characteristic so as to produce a periodicity in the signal modulation of the coherence function.

In practice it may prove better to choose a BLF with a non-Gaussian profile, for example a square characteristic.

Detection of the transformed optical field may be enhanced by making one of the filters time dependent, for example by periodic movement of the filter. The invention will now be descibed by way of example only in relation to the accompanying Drawings of which:

Figures 1 to 3 show spectral profiles of a band limiting filter (BLF), received radiation with a single target gas absorption line within the band of the BLF, and the spectrally processed signal after transmission through the BLF;

Figure 4 is a graph of the coherence envelope function y(L);

Figure 5 shows graphs of measurable parameters from changes in the coherence envelope with gas concentration;

Figures 6-8 show graphs illustrating the effect of use of a BNF whose centre frequency is offset relative to the peak in a target gas absorption line;

Figure 9 is a graph of the real part of the coherence envelope;

Figure 10 is a table giving developments of different pre-detection spectral processing;

Figures 11a and lib show an alternative filter approach, with and without gas absorption; and

Figure 12 shows how a comb filter may be used for spectral sampling;

As is well known a Fourier transform relationship exists between spectral profile and the coherence function. Various manipulations of the received spectral profile can lead to enhanced performance. That is, by using our a priori knowledge of the spectral characteristics of a target such as a wanted gas, the coherence profile/function can be made sensitive/specific to the gas over a particular (coherence) region, so that it is not necessary (as in FT spectroscopy) to scan the whole function in order to extract the wanted information. Throughout this specification the following terms are used:

The Band-Limiting Filter (BLF): This is the relatively broad-band filter which sets the spectral acceptance range of the system. Strictly speaking it also includes the spectral response of the detector/detectors and the spectral transmission of the optics; and some aspects of atmospheric transmission.

The Modifying Filter or Filters (MFs): These are additional spectral filters with their significant profile within the band-pass of the BLF.

Band-Pass : Region over which spectral transmission is approximately 50% of the peak transmission.

Centre-Wavelength/Wavenumbers: This is the characteristic wavelength/wavenumber of the filters which can be defined in many ways for different filter types. For example, it can be the wavelength/wavenumber at peak transmission, or the point midway between the 3dB points. The exact definition will be explained in each case that is considered.

Spectral Profile: Description of the shape of the filters e.g. Gaussian, rectangular or periodic.

Peak Transmission: Highest transmission of the filter across its band-pass. Spectral Processing will include the manipulation of:

(i) BLF

- band-pass

- centre wavelength/wavenumber

- spectral profile

- peak transmission

(ii) MF

- same list as for BLF

(iii) Path difference scanning

All of these can be made time and space dependent. For example: for time dependence the MF centre wavelength can be scanned in time; and particular filters need not cover the entire field-of-view (space) and therefore can be made to scan in space. Furthermore these manipulations can be employed singly or in a variety of combinations.

A fundamental of the optical transform image modulation (OTIM) technique is that it manipulates and preserves the full complex nature of the multi-dimensional optical signal (eg polarisation spatial coherence, temporal coherence, spatial geometry, instantaneous amplitude and phase.... and statistics). Consequently it is important to note that these properties can be preserved and employed after detection (in spite of the passage of the information through the square-law detector) due to the OTIM selective modulation process. For example, in the remote gas detection system described in European Patent 0155142 the real and imaginary part of the coherence function are available for manipulation.

The Fourier transform relationship between the received spectral profile and the coherence function can be represented:

or

where Γ (L,τ) is the Mutual Coherence Function

Γ (κ,ʋ) is the received spectral profile

L = path difference

k = wavenumber

τ = time difference = L/C; C = speed of light

γ = frequency

Remembering that we are building a detection system and not a gas analysis instrument, the aim is to concentrate the key information about the presence of the wanted gas to a small region of the coherence function profile. For example, it has previously been shown (European Patent No 0155142) that a movement of a specific null in the coherence function, or the value of the visibility at a particular path difference can be employed, together with high performance post-detector processing (e.g. correlation or matched filtering). Hence the objective is to maximise the change in some aspect (or feature) of the coherence function Γ (L;T) so that the smallest possible concentration of the wanted gas, causes a significant change in the electronic output.

Clearly there is no universal analytic solution to all problems/applications since the optimum approach is dependent upon such issues as the exact gas and background characteristics, and purpose of the sensor and constraints in its design (e.g size, cost, operating environment, permitted instrumental complexity). Hence the examples chosen are illustrative of the first steps in a possible design and are based on simple analytic forms e.g. Gaussian and rectangular spectral profiles. When adapting them for "real" problems it may be necessary to employ small adjustments e.g. additional minor off-sets in path differences or phase references.

CASE I

It will be useful to define a basic initial arrangement against which more advance techniques can be compared.

By using the BLF 10 shown in Figure 1 to isolate the spectral feature 20 shown in Figure 2, the resultant intensity profile I(k), 30, shown in Figure 3 is obtained. (Note all profiles are Gaussian).

Where I(κ) = I₀( κ) - i₁ ( κ) (3) for I₀(κ ) = background spectral profile without gas absorption

Hence :

and I₁(κ ) = gas absorption

for κ₀ = centre wavenumber

δ κ₀ = BLF Band Pass (NB. at κ = κ₀ + δκ₀/₂, I₀( κ)/Ip₀≈ 0.605) δ κ₁ = Gas Band Width

Substituting eqns (4) and (5) into (3) and evaluating eqn (1) gives

where y(L) is the envelope of the Mutual Coherence Function/ (π/2)^½

Two of the main features of this function are the zero-point 40 and the negative peak 41 shown in Figure 4. It can be shown that the zero-point 40 occurs at a path-difference (Lz) given by

and the path-difference corresponding to the negative peak (Lp), 41, is:

and the value of the negative peak (yp) is given by;

Plots of equations (7), (8) and (9), shown respectively by 50, 51 and 52 in Figure 5, show that these values are all dependent upon gas absorption strength Ip₁/Ip₀; and therefore the gas could be detected by a change in any, or all, of these characteristic features.

CASE 2

An extension of Case 1 is to off-set the centre-wavelength 21 of the BLF with respect to the gas absorption line 10 to give the intensity profile 31 as shown respectively in Figures 6, 7 and 8. In this case the spectral profile I(k) is given by

and the coherence function envelope

A "typical" plot of the real part of y(L) is given by 90 at Figure 9.

From eqn (11) it can be seen that the real and imaginary parts of the coherence function are: yr(L) = Ip₀· A(L) - Ip₁· B(L) Cos (2πLΔκ) (12) yi(L) = Ip₁· B(L) Sin (2πLΔκ) (13)

Where for convenience , the substitutions

have been made.

With Case 1 the structure associated with the gas being present occurred at moderately large values of L, where the coherence function visibility was decreasing fairly rapidly. In Case 2 however it is seen to be possible to bring the part of the curve which is of main interest to smaller values of L, where the visibility is significantly greater. This is the region where the depth of signal-modulation is greater. This can be illustrated numerically by reference to equation 13 where Sin (2πLΔk) becomes unity, i.e. at

At this value of L

Substituting the values used for Figure 9

and using

yi(L) = 48.3

This number should be compared with the value obtained for Case 1 as given in Figure 5 where at the ratio Ip₁/Ip₀ = 0.05 the negative peak of the coherence function is 45.0. While this is only an increase of 7 .4% it demonstrates that by simply off-setting the BLF centre wavenumber (a very simple form of pre-detector spectral processing) an improvement in signal modulation can be obtained. It is evident therefore, that by selecting more effective changes in the pre-detector spectral processing, further enhancement can be gained.

Continuing this principle of manipulating the spectral profile of the received optical radiation towards more advanced spectral configurations will bring about further improvements where the coherence function changes can be made more sensitive to the presence of the gas. In doing so, the signal formation can be made to influence a wider range of measurable quantities. For example, phase changes, changes in zero or null points, in both the real and imaginary domains.

Figure 10 illustrates the principle of the present invention. Light 101 101 received from a field of view in which a target gas might be present is received by the optical detector system 102. The received light passes through a band limiting filter BLF 103 and then a modifying filter MF 104. Further optical processing may then take place in a pre-detector optical processor 105 before detection (106). The signal from the detector 108 is then connected to an electronic processor 107

This approach is illustrated in the table Figure 11 where two further cases (Cases 3 and 4) are summarized. These cases illustrate the use of a modifying filter (MF) whose profile is the same as that of the gas, but whose centre wavenumber is different. Note the depth of absorption for the MF is chosen to be at the detection threshold for the gas. For each case in Figure 10, the spectral profile and the real and imaginary parts of the corresponding coherence function are given. Also the coherence functions are shown for both the situation where the gas is absent and present.

In both of the Cases 3 and 4 it can be seen that the enhanced modulation of the signal visibility is possible as with Case 2, but they also enable gas detection by measuring changes in a null (or other feature) of the visibility.

In Case 3. it can be seen that the presence of the gas adds an additional term Ip_f·H(L)·Cos 2πL .Δk to the function Ip_f·H(L) in the real part of the coherence function.

However, in Case 4 the effect of the presence of the gas is to double the term Ip_f H(L) ·Cos(2π L .Δk). Clearly by the judicious choice of the modifying filter (MF) with respect to the Band-Limiting filter (BLF), the modulation depth either side of a chosen null can be arranged to be sufficiently large so that the position of the null can be very clearly defined. For example, by the electronic circuit arrangements described in GB Patent Application No 9100127.

Development of the Table

Only four cases are given here as an illustration of how the technique can be evolved to meet particular applications. Clearly further stages in the table could include manipulating filters (MF) of greater complexity.

For example, benefits are to be gained by arrangements which lead to a periodicity in the signal modulation of the coherence function which can be made to change when the gas to be detected is present. The basics of such an arrangement are sketched in Figure 12.

Figure 12(a) shows an intensity profile 120 with two filter lines of δk separation and with no gas present, and it can be deduced that there will be a periodic structure in the coherence function proportional to 1/δk.

In Figure 12(b), the gas absorption line 121 is seen to appear midway between the two modifying filter lines 122, 123. In this case the coherence function will now have a periodicity proportional to 2/δk. This, in effect re-distributes the modulation efficiency, so the sensitivity of detection at a particular path difference is enhanced. The concept of double filter absorption lines can be extended to include multiple lines where the gas feature is in effect sampled at a number of specific wavenumbers. In this case, the MF could appear as a spectral comb function consisting of many absorption bands 130, periodically spaced at wavenumbers intervals Δκ where Δ <<δκ₁ (the gas feature (121) bandwidth). This situation is sketched in Figure 13. The spectral profile of the light after passage through the filter is in effect the product of the gas absorption profile and the filter transmission. Thus the resulting coherence function (the Fourier transform of the spectral profile) is a convolution of the absorption line coherence function with the Fourier transform of the filter transmission, ie a replication of the absorption line coherence function with period 1/ Δκ. It is therefore possible to sample various features of the gas coherence profile in a simultaneous fashion (say by multiple folding of one arm in the interferometer).

It should be noted that further benefits can be obtained in some cases, if the multiple line profile of the MF is not periodic, e.g. if the separation of the lines 130 increases or decreases with increasing wavenumber.

The examples given so far have all been dependent upon a simple Gaussian absorption line for the wanted gas. Many practical applications will involve gases with multiple absorption lines, each of which may have a profile quite unlike the Gaussian form so far considered. It will be clear therefore, that the starting point for the design of the filters (BLF and MFs) will be heavily dependent upon the spectral characteristics of the gas to be detected. There is no general solution which can be formulated in order to optimise the design. However, an heuristic approach coupled with experience in pre-detector signal processing, optical filtering and Fourier transform techniques will permit a practical and optimized solution to be reached.

Research to date has shown that it is usually better to choose a BLF with non-Gaussian profile. A better option is to employ a filter which has a profile closer to a rectangular transmission profile. There are 3 principle reasons for this.

Firstly, if the gas absorption is to one side of the BLF centre wavenumber, then the light at the absorption wavenumber is hardly alternated by the filter. Secondly, as can be seen from the table in Figure 11, by selecting A(κ) to be rectangular, the structure of the coherence envelope is more dependent on the gas absorption line because G(L) (the Fourier transform of A(κ)) decreases more quickly with respect to L than with a Gaussian line of notionally the same spectral width. Finally, the rectangular spectral profile introduces elements of π phase shift into the coherence profile which can further advantageously modify the coherence function.

MODULATION SCHEMES

Considering a range of modulation options, the original OTIM gas detection patent (European Patent No. 155142) was principally concerned with modulations achieved by varying phase or path-difference. Not only is this a very convenient method, but it has the advantage that it is possible to derive a reference signal to aid the post-detector electronic signal processing. The techniques described here can still use this approach but in addition, other modulation schemes are possible.

The principle of the modulation techniques is to ensure that one or more unique features of the coherence characteristic are selectively modulated prior to detection, thereby maximising the immunity of the system to interferant species and the effects of background duties.

Modulation of one or both of the spectral filters (BLF or MF) could be advantageous.

For example, the filter centre wavelength, λ_m can be modulated; in the case of interference filters by angular tilt ( λ_m≈ λ_oCos θ; λ_o= centre wavelength at normal incidence, θ = tilt angle from normal). When using Fabry-Perot interference filters λ_m (centre wavelength) can be modulated by varying the etalon thickness. Another variable filter is the Tunable Acousto-Optic Filter (TAOF) where λ_m can be varied by changing the frequency of the TAOF electrical driving signal. Also, the depth of absorption of the TAOF at λ_m can be varied by changing the amplitude of the electrical driving signal.

The depth of absorption can be varied using "conventional" (interference or dye) filters by changing the area of filter inserted into the optical path. Also, gas cell filters can be amplitude modulated by controlling the amount of absorptive gas in the active region of the cell. Centre frequency or depth of absorption modulation can be applied to any or all of the filters employed (broad band-pass or narrow band pre-processing filters). Furthermore, the use of such filter modulation does not preclude the other forms of modulation such as phase modulation via variation of interferometer path difference. Note also that pressure modulation of gas cells will lead to variable spectral width; modulating the width of the TAOF has the same effect. In all cases it is important to remember that both the real and imaginary components are available. There are many advantages following from this, not least is that variations in background spectra (e.g. caused by scanning from a blue sky to a brown corn-field) can be regarded as a change in spectral slope. Since the real part is immune to slope (odd function multiplied by the even Cosine term and integrated in the Fourier Transform(FT) process) whilst the imaginary part is very sensitive to slope, the system can operate with minimal adverse effect caused by background spectral changes.

Active Illumination

Before describing the detail it is important to define "active" in the context of this technique. Three cases need to be considered:

(i) where the gas is emitting instead of absorbing

(ii) when the gas is illuminated by a source under the control of the system designer

(iii) exploiting the inherent linearity of the technique so that the spectral processing can be applied at any one or a variety of points between the active illumination source and the receiver. Emitting Gases

The discussion in previous sections applies equally well when the gas emits its unique spectral profile rather than absorbs selectively. The only difference being a π phase shift due to the amplitude inversion. Clearly further flexibility exists by the combination of the emitting feature from the gas and the absorbing feature of the MF within the BLF bandwidth.

Light Source Under Control of System Designer

The types of active illumination sources can include lasers, band-limited white light, fluorescence stimulating, spectral discharge lamps (e.g. sodium or xenon), solid state devices (e.g. LEDs) etc. In this case, a greater variety of spectral profiles are available by combining the natural line shape of the emitted light and the transmission profile of the spectral filters (BLF or MF). Note that in general, the sensor can employ any of the sources individually or in any combination that is desirable. Note in this context, solar radiation can be regarded as one of the possible sources.

Linearity of the Spectral Processing

Since the system is essentially linear in its spectral processing (ie the filters can be placed at any point in the system) it can be advantageous for certain purposes to position some (or all) of the spectral processing at or near the source and not (as so far discussed) in the receiver. This may well be beneficial where a number of low-cost receivers are needed and therefore only one unit (the transmitter) needs contain the relatively complex manipulation. Similarly, the same philosophy applies if a number of low cost transmitters are needed with only one or a few higher value receivers; in this case the receivers contain the spectral processing.

Other Areas of Application

Thus far the invention has been based on the application of spectral pre-detector processing to gas detection. The principles can however be applied to a wider class of problems. For example, if it is required to remotely measure tilt or more generally movement of an object (e.g. rotation of a shaft), angular dependent spectral changes can be detected. This could be noted by observing spectral variations from an interference filter fitted to the object or even by exploiting the wavelength dependence of reflectivity (ie Fresnel reflectance which depends on object refractive index which is a function of wavelength).

These small variations in wavelength or temporal coherence could be exploited for a secure, passive communication system.

Similarly, the spectral changes associated with thin films or animal/vegetable layers could lead to applications in the maritime (oil pollution) environment or more generally resource and health monitoring.

Claims

1. An optical detection system comprising a) means (102) to receive radiation from a target within a field of view b) transform means (105) to modulate the received radiation to sensitise the system to selected features dependent on the coherence function profile within the received optical field; and

c) detection means (106) responsive to the modulated received radiation; characterised in that the transform means (105) includes. a spectral processor having:

(i) a band limiting filter (BLF) (103); and

(ii) a modifying filter (MF) (104) having a pass band at least partly overlapping the passband of the BLF (103);

2. An optical detection system as claimed in claim 1 characterised in that at least one of the BLF or MF filters (10, 121) has a Gaussian profile.

3. An optical detection system as claimed in claim 1 or 2 characterised in that the BLF has a rectangular transmission profile.

4. An optical detection system as claimed in any one of claims 1 to 3 characterised in that the centre of the MF absorption profile is offset with respect to the centre of the BLF profile.

5. An optical detection system for detecting a target gas as claimed in any one preceding claim characterised in that the MF is arranged to have the same profile as an absorption band of the gas but centred at a different wavenumber.

6. An optical detector system as claimed in any one of claims 1 to 4 characterised in that the MF has a transmission characteristic arranged to produce two absorption profiles (122, 123) within the BLF profile.

7. An optical detection system as claimed in any one of claims 1 to 4 characterised in that the MF has a transmission characteristic arranged to produce a plurality of absorption profiles (130) within the BLF profile so as to produce a periodicity in the signal modulation of the coherence function.

8. An optical detection system as claimed in any one preceding claim characterised in that one of the transmission function of one of the filters is made time dependent so as to enhance detection of the transformed optical field.

9. An optical detection system as claimed in claim 8 characterised in that the time dependence of the filter function is achieved by periodic movement of the filter.

10. An optical detection system as claimed in claim 9 characterised in that the periodically moved filter is an interference filter and means is provided to periodically vary the angle of tilt of the filter so as to modulate the centre frequency of the transmission profile of the filter.

11. An optical detection system as claimed in claim 9 characterised in that the area of the periodically moved filter inserted in the optical path of the system is varied to periodically modulate the depth of absorption of the filter.