NL1024069C2 - Method and device for performing speckle interferometry. - Google Patents

Description

Method and device for performing speckle interferometry Field of the invention

The present invention relates to a method and apparatus for obtaining a reason interference patterns by speckle interferometry, comprising obtaining a new interference pattern and further performing operations on the series of interference patterns.

State of the art Such a method and device is known, for example, from British patent application GB-A-2 188 445. This publication describes an interferometric method for measuring certain properties of an object, in which a number of interference or speckle patterns are successively used. obtained with the help of an interferometer. In this method, possibly after filtering or electronic processing of the interference patterns (in which all interference patterns are processed in the same manner), summation and averaging of the speckle patterns take place, with the aim of reducing the so-called speckle noise in the final measurements.

Another interferometric technique is known from the article "Temporal Phase 20 Unwrapping In A Real-time Phase Stepped Shearing Speckle Interferometer A Comparison Of Algorithms" by P.A.A.M. Somers and H. van Brug, Proceedings International Conference on Trends in Optical Nondestructive Testing, p. 131-142. With this technique, interference patterns are detected and processed in a number of time intervals.

The method described in GB-A-2 188 445 has the disadvantage that the speckle noise is reduced only to a limited extent, and that quantitative results (phase) cannot be produced with this method. Furthermore, the known method, as well as the speckled interferometry method known from the above-mentioned article, is sensitive to temporary or permanent disturbances of the measuring object or the measuring arrangement, as a result of which these can only be used outside a laboratory environment with much effort and effort, for example in an industrial environment. .

1024069

2 I

Summary of the invention I

The present invention seeks a method and apparatus for processing I

to provide specimens of interferomctrically obtained speckle patterns with an improved I

robustness against disturbances and improved noise reduction. At I

Speckle interferometry interferes with a specimen pattern with a reference wave I

can also be a speckle pattern, or a flat or spherical wave. The result is I

an interference pattern that is also a speckle pattern. I

According to the present invention, in a first aspect, a method I

according to the type defined in the preamble, wherein the series I

10 interference patterns includes uncorrelated interference patterns and the method I

further includes calculating a quality measure with respect to the new I

interference pattern and performing further operations on the I sequence

interference patterns based on the quality measure. I

The present invention is based on the insight that an I

A speckled pattern is seen as a collection of information carriers with an I

varying reliability. In the prior art, as described above, I

only an attempt is made to reduce the noise in a speckled pattern as a whole. I

By not extracting all information obtained from every newly obtained speckle pattern I

for further monitoring, it is possible to use different types of I

20 interferometry methods to improve the quality of the I

result, for example with a reduced share of speckle noise in the final I

result or better resistance to temporary and / or permanent disruptions in the I

measurement setup. I

In a first main group of the present invention, in particular the I

25 problem of temporary and / or permanent disruptions solved In the method I

there are two phases, a detection of a disturbance and the determination of the I

nature of the disruption (temporary or permanent). Detection of a disturbance takes place I

by comparing a last recorded (new) interference pattern and the direct I

preceding interference pattern in the series. A permanent disruption is an I

30 transition between two states, each of which is stable per se. In between is I

at least one measurement interval (time between recording of two interference patterns). The I

measurement interval in which the transition is located is unusable for further I

calculations. In the event of a temporary disruption, the system returns to the I

1024069_I

3 original stable situation. Detection of temporary disruptions requires. at least two measurement intervals, one before the start of the temporary disruption and one at the end of the temporary disruption. With proper handling (see below), virtually all recorded interference patterns can be used for further processing.

In an embodiment, particularly used for detecting a disturbance in an interferometry arrangement, the method comprises obtaining and storing a reference pattern, the reference pattern being one of the reason interference patterns preceding the new in time interference pattern, and wherein calculating a quality measure and performing further operations comprises calculating a first degree of agreement between the new interference pattern and the reference pattern, if the first degree of agreement is above a predetermined comparison threshold, performing further calculations with the reference pattern and the new interference pattern, and if the first degree of agreement is below a predetermined. IS comparison threshold lies in the non-execution of further calculations with the reference pattern and the new interference pattern.

The predetermined comparison threshold must be selected depending on the measurement conditions.

In one embodiment, the reference pattern is an interference pattern preceding the new interference pattern directly in time. As a result, a speckle pattern obtained immediately after the occurrence of a disturbance is not used in further processing together with its direct predecessor. This new interference pattern can optionally be used with an interference pattern immediately following, provided that the comparison of these two patterns yields a result above the comparison threshold.

Performing further calculations in a further embodiment comprises storing the new interference pattern as a reference pattern if the first degree of correspondence is above a predetermined comparison threshold. As a result, each newly obtained interference pattern is compared with the last interference pattern included in the further operations, whereby all interference patterns that are obtained during a temporary disruption are not included in the final operations. With this embodiment, it is 1024069

I 4 I

I possible to detect a short, temporary disturbance. Only after the disappearance of the I

In the event of a disturbance, further monitoring is performed. I

In a further variant of this embodiment, after a predetermined I

I carrying out the process steps a number of times, the first degree of I

I5 agreement is always below the predetermined comparison threshold, the new I

I interference pattern saved as the reference pattern. In this case the disturbance is I

I cannot be classified as temporary and further surveillance is continued in the I

I new, stable state. In this way, when a permanent I

I disrupted a small loss of measurement data ago. I

In a particularly advantageous embodiment of the present method that I

I can be used in particular for detecting a disturbance in an I

I interferometry setup and determining the nature of the disruption, includes the I

Method: I

Obtaining and storing a reference pattern, wherein the reference pattern is an I

I 15 of the series of interference patterns that precedes the new I in time

I interference pattern, I

I where the calculation of a quality measure and the execution of further I

I surveillance includes: I

I - calculating a first degree of agreement between the new I

Interference pattern and the reference pattern; I

I - if the first degree of agreement exceeds a predetermined I

I comparison threshold is: I

Performing a further calculation with the new interference pattern

I and the reference pattern, and I

Saving the new interference pattern as a reference pattern; I

I - if the first degree of agreement falls below the predetermined I

comparison threshold is: I

Obtaining a further new interference pattern that immediately follows I

I the new interference pattern, I

· Calculating a second degree of similarity between the further I

I new interference pattern and the new interference pattern; I

If the second degree of agreement exceeds a predetermined I

I comparison threshold is: I

O performing a loss calculation with the further new interference pattern and the new interference pattern; o The saving of the otherwise new interference pattern as a reference pattern.

S This embodiment combines the aforementioned embodiments by detecting the disturbance with a comparison of the new interference pattern and previously recorded interference patterns, followed by recording a further new pattern, determining the nature of the disturbance and further handling of the disturbance. operations of the interference patterns.

If the result of a comparison of the new pattern with a last pattern with which a further processing is performed is above the comparison threshold, the disturbance was temporary and a further processing can be carried out with the new interference pattern and the reference pattern. If below the threshold, a further new interference pattern is included, making it possible to compare two immediately consecutive interference patterns. If these are almost the same (comparison above threshold), then the disturbance was permanent and a further processing can be carried out with the new pattern and the further new pattern. If the comparison is again below the threshold, apparently a stable state has not yet been reached and the process is repeated.

The degree of similarity is calculated in one embodiment by subtracting pixel values of the interference patterns from each other and determining an average of the difference over the entire image. Alternatively, the degree of similarity is calculated by subtracting pixel values of the interference patterns from each other, calculating a histogram of difference values across the entire image, and determining a statistical parameter of the histogram. In a still further alternative embodiment, the degree of similarity is determined by calculating a correlation coefficient of at least a sub-picture of the interference patterns. By making the decision whether or not to use a obtained interference pattern depend on only a part of the obtained interference pattern (for example the middle part) an even more robust method is obtained. The part may, for example, also be a part of the image where changes appear to be small, i.e. outside the area where, for example, there is a defect.

1024069 __________

I 6 I

In speckle interferometry, where correlated interference patterns are used, I

I, as in the method discussed above in GB-A-2 188 445, I

I intensities of multiple interference patterns summed to reduce

I obtain the speckle noise. Different types of interferometry are used I

I made of speckles in an interference pattern as information carriers for measuring

I of, for example, distortion of an object. The quality of the speckles depends on I

I of the intensity and the modulation, i.e. the extent to which the intensity of a speckle I

I is influenced by the presence of another speckle with which this I

I interferes. With speckle patterns that are normally obtained, the low I dominate

I 10 intensities. If two speckle patterns obtained interfere with each other, such as the I

I case is in era shearing speckle interferometer (shearography), so there are few positions I

I in the interference pattern where both speckles have an equal high intensity and there I

I therefore speak of a high modulation. At era low intensity of era of both or I

I both interfering speckles the information is unreliable, and since the low I

I dominate intensities, this results in a high degree of noise in the final I

I result. I

In a second main group of the present invention, wherein the series I

I interference patterns includes a number of mutually non - correlated subsets (while I

I do correlate the interference patterns within each subset), the calculation includes I

I 20 of the quality measure and the execution of further processing after the calculation of the I

I modulation depth of predetermined parts (up to pixel level) of the new I

Interference pattern, and further processing of the predetermined portions for which I

I the modulation depth is above a predetermined threshold value. This makes it I

I may only use parts that are actually sufficiently reliable

I 25 contain information, so that the noise level in the final result is significantly I

I becomes lower. I

The predetermined parts are parts for which calculation makes sense, for example I

I not for sub-areas from which no or insufficient light is scattered back I

I (outside the object or at the location of holes in the object). The series of interference patterns I

In this embodiment, I consists of the number of series of interference patterns, with I

I each series with regard to changes of the object under the same conditions I

I is included, and with the speckle patterns between the series themselves I

I are not correlated. This can be achieved by the object I a number of times

I 1024069 I

7 reproducibly distorted, and change the speckle pattern at every speech, or by quickly registering a number of uncorrelated speckle patterns during a change cycle and repeatable repeatably for each subsequent stage in the change cycle.

Although it is disclosed in GB-A-2 188 445 that in the event of interference between two speckle patterns the quality of the interference lines lags behind that obtained in interference between a speckle pattern and a flat or spherical beam, this publication does not address the cause thereof. . In the present invention it has been found that (parts of) speckle patterns have a low modulation (a quality measure), as a result of which the aforementioned problems arise, and that this can be solved by including only parts that have a sufficient modulation depth. Although weighted averaging based on modulation is known from US-B-6,246,447, in that case it concerns filtering within one image over several pixels (spatial filtering). The present invention extends over a plurality of (mutually non-correlated) images, and filtering takes place on a single pixel.

In the present invention, the improvement to be achieved in the end result is limited by the number of independent speckle patterns to be generated. Also in the method disclosed in GB-A-2 188 445, the reduction of speckle cleaning is dependent on the number of uncorrelated speckle patterns that are summed. The present embodiment is more effective in that the reliability of the speckles in a speckle pattern is included in the further processing, resulting in a higher signal-to-noise ratio of the resulting (calculated) phase distributions. The present invention is applicable for both static and dynamic applications.

In an advantageous embodiment, the method further comprises replacing one of the predetermined portions for which the modulation depth is below the predetermined threshold, by a corresponding portion of another set of interference patterns for which the modulation depth is above the predetermined threshold. As a result, more reliable information is added to a (part of) a speckle pattern, so that the end result contains less speckle noise. This is on the condition that the contribution from the other series is equivalent, i.e. relates to the same change of the object.

1024069

8 I

The other set of interference patterns is, in one embodiment, the most temporal

neighboring interference pattern, for example the immediately preceding or immediately following I

interference pattern. Between obtaining these two interference patterns lies I

little time, so that also the differences in the I stored in the interference patterns

5 information will be close to each other (the change of state of the object, I

measured over a certain interval, will not differ much from the I

change in state in a slightly shifted interval). This makes information I

added to a certain interference pattern that is the most reliable I

result. I

In an alternative embodiment, the one predetermined part for which I

the modulation depth is below the predetermined threshold, replaced by an I

corresponding part, which is derived from a plurality of corresponding parts I

of other interference patterns from sequences for which the modulation depth is above I

predetermined threshold For this, a (weighted) filtering I can be used

15 of (a part of) the other interference patterns are used. The weighting at the I

filtering can for example be proportional to the calculated modulation depth. I

In a special embodiment, the second I

main group can be applied, the method is used for analyzing I

of an object under load, and the set of interference patterns consists of at least I

20 a first series of interference patterns obtained during application I

of a reproducible load on the object and at least a second speech of I

interference patterns obtained during the removal of the reproducible I

load of the object. The results of further processing of the second series I

interference patterns are included with an opposite sign on the I

25 results of further monitoring of the first speeches. This makes it possible to I

obtain information about the load on an object in a reproducible manner. I

By applying the earlier embodiments of the present method I

it is achieved that the result is insensitive to temporary and / or permanent I

disturbances, and that the result contains less noise. I

In a further aspect, the present invention relates to a device I

for processing a series I obtained from an interferometry device

interference patterns, comprising processing means connected to I

detection means of the interferometry device. The processing means comprise I

1024069 | 9

Drawing means and memory means connected to the calculating means. The computing means are adapted to perform one of the embodiments of the present method.

In a particularly advantageous embodiment, the interferometry device 5 is further provided with means for influencing a speckle pattern on the detection means, the means for influencing being selected from the group of a piezoelectrically controlled diffuser for influencing the illumination of the object, a liquid crystal modulator or a spatial light modulator (spatial light modulator, SLM) for influencing the direction of observation of the interferometry device. GB-A-2 188 445 also already discloses a number of means which influence either the direction of observation or the direction of illumination. However, the means disclosed for influencing the direction of observation are all mechanical means. The means according to the present invention which influence the direction of observation (liquid crystal, SLM) provide a higher speed and a larger number of possible positions without the device becoming much more complex. The means that influence the illumination make it possible, because the wavefront of the light beam is strongly modulated after the diffuser, to obtain highly correlated speckle patterns with a small change in the orientation of the diffiisor.

20

Brief description of the drawings

The present invention will now be discussed in more detail with reference to a number of exemplary embodiments, with reference to the accompanying drawings, in which FIG. 1 shows a schematic representation of an example of an interferometer to which the present invention is applicable;

Fig. 2 shows a block diagram of an embodiment of the weathering device according to the present invention;

FIG. 3 shows a flow chart of an embodiment according to the present method;

FIG. 4 shows a flow chart of a further embodiment according to the present method.

1024069

10 I

Detailed description of exemplary embodiments I

Speckle interferometers 1 are used for non-destructive inspection of I

components, installations and means of transport, and in particular for optical inspection I

of large surfaces. Speckle interferometry hereby has the advantage that on an I

Large surfaces can be tested quickly. In a speckle interferometer 1 I

speckle patterns are obtained and processed to obtain information about the I

change in the condition of an object under test, such as a structure that loads I

to deformation, stress concentrations, stiffness distribution, weak spots, I

observe damage or defects. Other applications may be in the I

10 detection of delamination in laminated or glued structures, and loose areas in I

paintings or frescoes. I

In speckle interferometry, an object with a diffuse reflective I becomes

surface illuminated with a coherent light source. At some distance from the object, I

a camera placed At the location of the sensor of the camera a speckle distribution I

15 are built up. This is caused by interference from light beams, which are transmitted by the I

surface in the direction of the camera. I

In addition to object lighting, a second bundle I is used for interferometry

introduced, the reference bundle. This is obtained from the same coherent source, I

and can be aimed directly at the sensor, or from the object, from another I

20 position than the object beam, reach the camera. Both bundles interfere there. I

In a speckle interferometer speckle patterns are included which are a series I

represent states of an object between a start and end state. Using I

of these speckle patterns, the phase of the initial and final state of the I to be studied becomes

object determined, or the phase difference between the two states. I

Known types of interferometry are, for example, electronic I

spot pattern interferometry (ESPI), where a flat or spherical reference beam I

coincides with the object beam via a separate optical path on the sensor, ar the I

resulting interference pattern is further analyzed, or "shearo graph", where I

the object and reference bundle of the object come from substantially the same optical I

Walk 30 paths. I

In FIG. 1 is a schematic representation of an exemplary implementation I

of a speckle interferometer 1. This is a phase-step speckle interferometer 1 I

with two optical paths, which generates an interference pattern for each of both I

1024069 I

11 optical paths. In one of the optical paths a provision is provided with which a predetermined additional phase difference is made between the object and reference bundle in that path, relative to the other path. The interference patterns are detected by a single camera 2 (e.g., a CCD camera). This interferometer 1 is also referred to as "2-bucket, 1-camera, real-time phase stepping shearing speckle interferometer."

Each reference bundle, together with an object bundle from an object 3 to be examined, traverses a different optical path, whereby a predetermined phase step can be applied per optical path with the aid of, for example, polarization-sensitive optics. The interferometer 1 is equipped with a detector (camera 2) for detecting two sub-images. With the in FIG. 1, all of the phase-step speckled patterns generated and associated with a particular state of the object 3 being studied can be recorded at the same time. In FIG. 1 are two phase-stepped speckle patterns, but generalized are that n for a system with n paths.

The interferometer 1 comprises three sections, namely an object beam section 4 that supplies two beams with different polarization states, one of which has an inclination (shearing), a beam split section 5 that forms two identical paths, and a combining section 6 arranged around the phase step to be arranged in one of the paths 20.

The object beam portion 4 has a so-called Michelson configuration. Via a field lens 7 and transfer lens 8, a beam of light from the object 3 is received on an entrance surface of a polarizing beam splitter element 9. The beam splitter element 9 is provided on three sides with λ / 4 plates 10 oriented on π / 4, so that light can exit on these sides with a circular polarization. A normal mirror 12 creates a normal beam, and a tilt mirror 11 ("shearing minor") creates a beam with tilt, both of which are reflected to the polarizing beam splitting element 9. Under 45 ° linearly polarized light, or unpolarized light, that the object beam portion 4 enters, emerges as two circularly polarized beams with opposite directions of rotation.

The bundle split portion 5 splits these two bundles over two paths using a normal bundle split element, and directs them to the bundle combining portion 6.

1024069

12 I

The beam combining portion 6 comprises a 45 ° mirror 16 and a normal I

oriented mirror 17. In a path, the reference beam is subjected to one another

phase step by means of a polarization plane rotation element 13, in the other path the I

reference bundle normally. I

Two circularly polarized bundles with opposite direction of rotation will be met

interfere with each other when they are incident on a detector, such as via transfer lens 18 on I

camera 2 in FIG. 1, after they have passed a polarizer. Rotation of the I

polarization direction through polarization plane rotation element 13 through an angle φ will be an I

cause a phase step of 2φ, so that a phase step between the two paths can be I

10 introduced by placing a polarization plane rotation element 13 in one path. I

The beam combining portion 6 comprises a polarizing beam splitter 14, with I

a λ / 4 plate 15 on one side. The bundle combining section 6 receives circularly

polarized light from both paths on a different entrance surface. To an I

the polarization plane rotation element 13 is attached to the input face. After going through I

The polarizing beam splitting element 14 will be the circularly polarized beams I

be transformed into horizontally polarized bundles. I

The bundles in one path become 5 l after traversing the bundle split portion

The 45 ° mirror 16, directly transmitted to a half (image portion) of the camera 2, I

where interference occurs I

Circularly polarized light from the other path is also transmitted as I

horizontally polarized light, but falls on a mirror 17 after traversing the I

λ / 4 plate 15 that is oriented to n / 4, so that the polarization state is again I

changes to a circular polarization. It is then reflected back by I

mirror 17 and again traverses the λ / 4 plate 15 and polarizing beam splitting element I

14. By this second path through the λ / 4-plate 15 the polarization state becomes I

vertically, so that the polarizing beam splitter element 14 reflects the beam to I

the other half of the camera 2, where again interference occurs.

The two step-by-step speckle patterns are therefore available simultaneously, so that I

a good basis is created for calculating phase differences for a certain I

30 condition of the object 3. I

However, it is also possible to apply the present invention to I

types of speckle interferometers than those in FIG. 1 speckle interferometer 1 shown. 1

1024069 '| 13

Processing of the speckle patterns recorded with the camera 2 into the desired end result takes place in a processing device 21 connected to the camera 2 (see Fig. 2). The methods to be discussed below can be implemented on a daily processing device 21. For this purpose, the processing device 21 is provided with one or more processors 22, memories 23 connected to the processors in a manner known to those skilled in the art (for storing program instructions to be executed and for storing (intermediate) results of calculations, and input and output devices 24 connected to the processor 22, such as a keyboard, mouse, screen, etc.

In the following, it is assumed that calculations on speckle patterns obtained with the camera 2 are performed immediately after obtaining them. However, it is also possible to record a series of speckle patterns with the aid of the camera 2 and to store them in the memory 23, and then to retrieve the speckle patterns obtained from the memory 23 and process them according to the indicated methods 15 (off-line). line).

A disadvantage of speckle interferometry is that an installation that uses such a technique is poor or resistant to (temporary) disturbances that lead to an undesirable change in the optical path length between object and interferometer, or in the reference bundle. This occurs with ESPI, but also (although to a lesser extent) with shearography. The probability that a disturbance has a different effect on an object and reference bundle is high at ESPI, where these bundles run on different paths. With shearo graphs, the sensitivity to path length changes is much smaller, because the two bundles follow an almost common path. Temporary disruptions can make reliable measurement temporarily impossible or introduce non-recoverable errors. Permanent disruptions (for example, an unintended movement of the object) can also make reliable measurement impossible. Even with the per se known technique of 'temporal phase unwrapping', temporary or permanent disruptions can lead to an unreliable intermediate result and therefore to a lower quality end result (increase in noise).

Phaseges taped interferometry or shearo graph requires two or more instances of the reference beam that are phase-shifted relative to each other, with which two interference patterns can be obtained for the same state of the object. The different types of interferometry are (in different ways) sensitive to 1024069 ___

14 I

disruptions. In normal phase-step interferometry, where an object I

scattered beam of light is imaged on a detector and interferes with an I

reference bundle with a certain phase, the measurement setup is dependent on the I

precise configuration of the exposure system and viewing angle, sensitive to I

5 disturbances in the plane or out of the plane of the measurement object. In shearography, where the I

object beam and reference beam traversing substantially the same optical path, the I

susceptibility to disturbances that affect both paths equally - that is I

for example, displacements from the plane of the object, or changes in the I

refractive index of the medium between object and camera (for example, by motion I

10 from air) - relatively small. Like ESPI, shearography is sensitive to I

displacements perpendicular to the sensitivity vector, which is generally an I

displacement in the plane of the object, because that quickly leads to a decor relation between the I

speckle pattern before and after the move. The sensitivity to disturbances from the I

flat (common mode) is less, but can still cause I

15 unreliable (intermediate results or results of lesser quality I

In FIG. 3 is a flow chart of an embodiment of the I

present invention. A speckled pattern (from a series of speckled patterns that in I

space or time successively detected by the camera 2) is obtained in block I

51. A series can comprise two speckle patterns per measuring moment, while also an I

20 series with three or via * speckle patterns per measurement interval is possible. The two or more I

speckle patterns per measurement interval may be spatially separated, but may also be I

be recorded sequentially. Next, the correlation between the I

speckled pattern and a previous speckled pattern (block 52), and the I

results compared to a predetermined criterion that the quality of results I

25 that can be achieved with these speckle patterns (decision block 53). I

Based on the comparison, that speckle pattern is accepted (block 54) or I

rejected (block 55) for further processing. If even more speckle patterns I

must be detected (decision block 56), the flowchart returns to I

block 51. The choice of the criterion may depend on, for example, I

30 environmental conditions (laboratory environment or industrial environment) and the I

required quality of the results. I

In a first embodiment, the consecutive registered I

speckle patterns on a pixel basis are subtracted from each other, and the average of all I

1024069 _ _ | 15 difference values calculated over two consecutive patterns and compared with a threshold value. If the average difference is too large, then one of the two speckled patterns compared with each other is not used for the calculations.

In a variant, a histogram of the difference image is calculated. Certain statistical properties can be calculated from the histogram (average value, spread, maximum value, ...) and one or more of these statistical properties can be used to determine whether one of the two consecutive speckle patterns is not used in further calculations. .

In a further embodiment, a correlation coefficient is calculated for the two consecutive images. The correlation coefficient is a measure of the degree of similarity between the two patterns. In this case, a threshold value for the correlation coefficient can be taken as the criterion. If the correlation coefficient is higher than the threshold value, the speckle pattern can be included in further calculations, such as the calculation of phase or phase difference, solving the modulo 2n difference (phase unwrapping), or addition of intermediate results in an accumulator (recording of summed results in consecutive measurement intervals).

The threshold value should be selected depending on the test conditions, the speed at which the state of the object changes, and the sampling interval.

For stable environments, an average changing state of the object and sampling intervals of the order of one second, a typical value for the threshold is between 0.9 and 0.95. In test environments where more noise occurs, the state of the object changes faster, and longer sampling intervals, lower correlation coefficients are often found, and the threshold value will have to be set correspondingly lower in order to still obtain sufficiently usable measurement results. The choice of the threshold value is therefore a consideration of the desired level of rejection of unreliable or deteriorated measurement results and the prevailing environmental conditions during the measurement.

The comparison of the newly detected speckle pattern with a previously stored speckle pattern (block 52) can be implemented in various ways. In a first embodiment, the new speckle pattern is compared with the immediately preceding speckle pattern. This is a good strategy if the disruption is a permanent disruption: only the transition of the two states of the object 1024069 I 16 I (before and after the disruption) is not included in the further calculations. Because a further new pattern can be obtained without performing calculations with the speckle pattern that does not satisfy the comparison criterion, only a short period of time is created in which no information about (the change of) the object is obtained. If the disturbance is temporary and of short duration, such a strategy would also lead to the rejection of the speckle pattern following the disturbance.

In an alternative, the newly obtained speckle pattern in block 52 is compared with an earlier speckle pattern used in the further processing (block 54), i.e., a speckle pattern that can yield a reliable intermediate result. New speckle patterns are obtained (block 51) and compared with this reliable speckle pattern as a reference. If a comparison calculated in block 52 is not above the acceptance threshold, the newly obtained speckle pattern is rejected (55). Furthermore, if the first newly obtained pattern is rejected, a counter is started. For each subsequent rejected new pattern, the counter I is increased. If the number of speckled patterns that has been rejected is above a predetermined value, it is assumed that a permanent disturbance has occurred. In this case, another strategy is continued, in which the last (rejected) speckle pattern is nevertheless stored, and further speckle patterns are compared with that stored speckle pattern.

In a still further alternative, in block 52, for each newly obtained speckle pattern, two comparisons are made, one comparison with the immediately preceding speckle pattern, and one comparison with the last reliable speckle pattern (the speckle pattern used last for further processing). In block 56). If the comparison value of one of the two comparisons is above the predetermined threshold, this new speckle pattern is stored and used for further processing. This makes it possible to correctly detect both temporary and permanent disruptions, and to reject speckle patterns that would result in a less good end result as a result of the disruption. If the disturbance was temporary, a comparison of the last reliable speckle pattern with the new speckle pattern leads to an acceptable comparison result.

If the disturbance was permanent, then an acceptable comparison result is obtained for the new speckle pattern and the immediately preceding speckle pattern once the object has become stable again after the disturbance.

With this latter embodiment, it is possible to correctly eliminate both temporary and permanent disturbances of the object 3, the medium or the exposure source from the calculations. A permanent disruption will lead to a minimal loss of data. If a sufficient comparison result can be achieved between a new speckle pattern and a last reliable speckle pattern, it is possible to keep the loss of data as small as possible in the event of a temporary disruption, or it can even be zero. In this case, the requirement must be met that the change in state of the object 3 in the now longer measuring interval between the inclusion of two speckle patterns is within the maximum measuring range permitted for a measuring interval. As long as a sufficient number of speckle patterns can be found that exhibit a sufficiently high correspondence, the method will generate results. The methods according to the prior art in this case will yield erroneous results. In methods based on temporal phase unwrapping, an erroneous result will be delivered if a permanent or temporary disruption occurs. In methods based on calculating a phase difference between a final state and a first state, an erroneous result will be delivered in the case of a permanent disruption, or if a temporary disruption occurs at the beginning or end of a measurement interval.

The present method makes it possible to use speckle interferometry in situations where known methods could not be used. As long as the object is stable during a measurement interval, reliable, valid results can be obtained for that measurement interval. The object 3 does not have to remain stable during a series of measuring intervals or during the entire test. The object 3 only has to be stable in the number of measuring intervals required to measure a sufficiently large accumulated state change. Also, the measurement intervals do not have to be consecutive. There is no requirement for speckle correlation between measurement intervals since each measurement interval or number of measurement intervals can be processed independently.

In the above embodiments, it is assumed that the entire speckle pattern (all pixel data detected by the camera 2) is used in the calculations. However, in a further embodiment use 1024069 is used

I made calculations of correlation coefficients for a series of sub-images that I

I are distributed over the entire speckle pattern. The acceptance criterion in this case is I

I based on the correlation coefficients for a predetermined number of sub-images, I

For example, a majority of the number of sub-images (or even a minority). I

If there are then correlation coefficients for certain sub-images that are below I

I have a predetermined threshold, they are simply ignored. This is why I

I allow some degree of local decor relationship in a speckled pattern, I

I and the acceptance criterion can be determined more precisely. I

In an advantageous embodiment of this method, first of all an I

Correlation coefficient calculated for a first (e.g. centrally located) sub-image. I

I Only if this correlation coefficient is below the threshold value, will one or I be used

I checked more further sub-images to verify that other areas in the I

I speckle pattern are also correlated. Only when a minimum number of sub-images I

If the correlation is too low, the speckle pattern in question is rejected. I

A further aspect of the present invention relates to another I

I quality parameter that is used to assess the usability of an I

I detected interference pattern, namely the modulation depth. In the Speckle patterns the I

I speckles the information carriers, and the quality of the results depends on the I

reliability of the speckles. According to speckle statistics, the maximum of the I is

I intensity distribution of speckles at low intensities: most speckles are dark. I

I If two speckle patterns interfere with each other, a large number of I

the modulation is low, so that those places do not have reliable I

I information is available. Current systems known to those skilled in the art therefore fit I

I often apply spatial filtering, whether or not selectively on unreliable pixels, and sometimes I

I taking into account the modulation as a weighting factor. For a pixel that is too low an I

I has modulation, a replacement value is calculated from the neighboring (reliable) I

I pixels in the environment. I

I In the literature, the problem is often referred to as speckle noise

I, against which measures can be taken. Reducing I

Speckle noise in an image made with coherent light is known. Famous I

I measures are variation of the wavelength, variation of the direction of the lighting, I

Fitting a moving diffuser in the lighting, fitting an I

I moving subaperture in the imaging system. All these methods are based on I

I 1024069 I

19 non-coherent combination of a number of uncorrelated speckle distributions. This is a form of averaging.

A special embodiment for reducing speckle noise in images made with coherent light has been described by Tiziani and Lowenthal, among others. Here, two difiusors are placed a short distance apart in the lighting, and one of the two difiusors is moved. The decor relation between the speckle distributions as a function of the displacement of the diffuser increases thereby faster than with a single moving diffuser, with the result that less speckle noise in the image with equal displacement.

10 In the reduction of speckle noise in speckle interferometry, two areas of application are distinguished: interferometry on quasi-static objects, and interferometry on vibrating objects.

One of the techniques for measuring vibration patterns is the so-called time averaging method. With this method, 15 interference patterns are recorded during the vibration. The measuring time is long relative to the vibration period, whereby average values of the intensity are measured instead of instantaneous. The variations in optical path length caused by the vibration lead to periodic intensity variations in the interference pattern. These are a function of the displacement, with a zero-order Bessel function as the most important term: a fringe pattern is created in which the vibration pattern is visible, and whose modulation depth decreases with the amplitude of the vibration. Results with this method can be improved by averaging over a number of recordings, the speckle pattern being changed between recordings by one of the above-mentioned methods. The result of this is a reduction of speckle noise at the root of the number of speckle patterns involved in averaging. An improvement in fringe contrast is also achieved. The technique is described inter alia by Slettemoen as "speckle reduction" or "speckle averagmg". The in Gudmunn Slettemoen article, "Electronic speckle pattem interferometry system based on a speckle reference beam", Applied Options volume 19, No 4, pp. 616-623 (1980) means for manipulating the direction of the illumination, or the direction of observation, are mechanical. An acoustic optical modulator (AOM) is also mentioned. This form of speckle averaging is possible because the fringing pattern on observation of vibrations by means of the time averaging technique generally does not depend on the lighting.

1024069

I 20 I

I The method according to Slettemoen also relates to averaging I

I intensities, while in the present application the determination of the phase becomes I

I improved by selecting the most reliable pixels, or by one (possibly I

I weighted) averaging. I

I 5 Spot mediation for quasi-urban objects is described in EiolfVikhagen I

I and Ole J. Lokberg, "Speckle Averaging in Deformation Analysis using Video Speckle I

Interferometry ", Options and Lasers in Engineering, Volume 27, Issue 2, pp 179-190 I

I (1997). For a number of uncorrelated speckle patterns, an I is given per pixel

I phase difference calculated for an object before and after distortion. The distortion must be I

Be reproducible to some extent. Of the modulo 2π phase difference images that the I

I represent a change in shape of the object, the I becomes after some spatial filtering

I phase gradient calculated. This can be summed, making speckled noise I

I reduced. The phase distributions can also be recalculated ("unwrapped"), and I

I then summed for reduction of speckle noise. According to the method of I

I 15 Vikhagen only determines the difference between the start and end states, there are I

I have not measured any series. Determining the phase gradient is essential because I

I results cannot be summed unless the distortion is extremely accurate

I is reproduced for each phase distribution to be summed. Unwrapping of te I

I summing end results has the disadvantage that it has to be done afterwards, slowly I

I is 20, and difficult to get stable for high noise patterns. In the present I

This invention does not exist, as long as the condition that the I

I distortion per measurement interval results in a phase change that is less than +/- π. An I

I measurement covers a freely selectable number of measurement intervals, over which is summed, I

I so that both a large measuring range and real-time unwrapping are achieved. For I

The creation of uncorrelated speckle patterns are electric, hydraulic or I

Called pneumatic drives of a mechanism (for example, by I

I Slettemoen, see the above-mentioned article or GB-A-2 188 445, or the above-mentioned I

I Vikhagen article), with which the direction of the lighting can be influenced. I

As an alternative, influencing the lighting by means of an acousto-optic I

I called modulator (Slettemoen). For influencing the direction of observation I

I is driven by electric, hydraulic or pneumatic means

I called moving aperture. I

| 1024069 | 21

Another form of speckle averaging uses a complex aperture consisting of two or more slits or holes. In fact, this is the use of several observation directions. By carefully selecting the configuration of the partial apertures, it can be achieved that the information recorded with each of the apertures can be spatially separated after Fourier transformation. Each of the partial apertures generates an independent speckle distribution, from which the quantity to be measured, for example displacement, can be calculated. The results can be average.

With "shearing" (shifted) speckle interferometry, 10 speckles are used as information carriers for measuring, for example, distortion of an object. The quality of speckles as an information carrier is also dependent on their intensity and their modulation. As a result, results contain a lot of noise, especially in "shearing" speckle interferometry (shearography), where two speckle distributions interfere with each other.

The most common solution is spatial filtering, possibly selectively applied to pixels with low modulation. This solution reduces the spatial resolution of the result.

The known techniques which make use of speckle averaging have the drawback that the mechanical embodiments are relatively slow and inaccurate. For example, use is made of a stepper-controlled diaphragm, or plan parallel plate around the direction of the. lighting, or that of the observation.

If speckle averaging in combination with a temporal averaging technique is used to measure vibrations, the aforementioned disadvantages do not constitute a significant limitation. For quasi-static applications, however, the reproducibility of uncorrelated speckle patterns is a desirable property, which is difficult to achieve with the aforementioned electrical, mechanical, or hydraulic means for influencing lighting.

Other known applications of speckle averaging that use a limited number of fixed apertures are limited in the number of independent patterns available for averaging.

The present invention aims to realize a number of independent speckle distributions on the sensor quickly and reproducibly, and the results obtained with 1024069

22 I

to combine each of them in a more effective way than in I

known techniques usual averaging of results. The reproducibility I

of the speckle patterns is high. A number of I is used to influence the lighting

systems conceivable, use of a tunable laser, use of multiple lasers that I

5 are switched on in turn, displacement of the object, displacement of the I

reference bundle when applying a scattering reference bundle, as in I

shearography, application of a laterally movable diffuser in the lighting, I

wherein the object is highlighted via this diffuser. An I supplied by a diffuser

optical bundle has a strongly modulated wavefront, causing a small movement I

10 of the difiner may already lead to a significant decor relationship of the speckle pattern on I

the sensor 2. I

The latter option is attractive; the difiuser can quickly go in two directions I

be moved using two piezoelectric actuators with sufficient I

range. A large number of independent speckle patterns can hereby be used I

15 have been realized, each of which would lead to a measurement result of current quality

can lead. I

In a special embodiment thereof, a fixed with a movable I

diphiner combined, so that the correlation between speckle images decreases faster I

as a function of displacement than with a single difiner. I

The speckle pattern on the sensor can also be influenced by within the I

aperture that is palpable for speckle size and speckle distribution a lateral I

movable sub-aperture. This subaperture is marked with (non-mechanical I

resources moved. According to a possible implementation of the invention, the I

aperture realized by an LCD with a matrix structure, with which the subaperture I

25 can be placed anywhere in the aperture, under the control of an I

computer. I

According to another embodiment, a Spatial Light Modulator can be I

used to place the sub-aperture anywhere in the aperture. I

The method according to an embodiment of the present invention is in I

30 is a flow chart in FIG. 4. The method can for example be I

implemented in the embodiment shown in FIG. 2 (and discussed above) I

processing device21. I

1024069

23

With the help of the invention, multiple, independent speckle patterns are created, which complement each other, and which are stored for wilting. The data of a certain interference or speckle pattern is recovered from a memory 23 in block 61. The modulation depth of the image thus obtained, or of one or more sub-images 5 (groups of pixels or even pixels) is calculated in a manner known per se ( block 62). If the modulation depth is sufficient (decision block 63), this (part of) the image is used for further processing (block 64). Unreliable information for low modulation positions for a given speckle pattern is selectively replaced with more reliable information obtained with one or more other speckle patterns obtained with a similar change of the object (block 65). This procedure is repeated (block 66) until all interference patterns of a series of patterns from a measurement session have been assessed and processed.

For example, a calculation of the phase difference over a measurement interval involves four speckle patterns, two before and two after a change in the object to be measured. All four speckle patterns are involved in determining the (local) quality of the measurement series. In all four speckle patterns, a selective replacement (at pixel level) of corresponding pixels can take place in each of the four speckle patterns. It is possible that the replaced 20 pixels will participate over an earlier or subsequent interval, because their modulation is above the threshold, but in general bad quality pixels will retain their low modulation over the entire measuring cycle. In other systems a similar approach can be followed, for example calculation of the modulation before and after change based on three or more speckle patterns per state, and (selectively) replacement of pixels in all speckle patterns that are involved in the calculation of the phase difference.

Combination of the results achieved with independent speckle patterns now leads to an increase in the signal-to-noise ratio. The present method uses more complex data processing, such as selectively replacing unreliable pixels (or sub-images) with pixels (or sub-images) of higher quality (see preceding paragraph). The criterion for replacing an unreliable pixel is modulation, or a criterion derived therefrom. A form of filtering of the available more reliable results can be chosen per pixel, whereby the modulation is used as a weighting factor in the 1024069

24 I

determination of an average for the new value of the pixel. This form of I

filtering means weighted averaging of all results for a pixel, with I

taking into account a lower limit for the reliability of a pixel. I

in an interesting embodiment of the method according to the present I

In accordance with the invention, the deformation of the object 3 is reproducibly applied, by I

thermal or mechanical loading, e.g. air pressure differences. During I

the application of the load is measured with a certain speckle pattern. I

During the return of the object 3 to the original state, an I

a new speckle distribution, with the sign of the results being I

10 vice versa. The results are averaged or otherwise combined. This I

a number of times to be chosen by the user can be repeated. I

This method is easy to implement, for example as a program I

processor 22 present in the processing device 21. It is possible to switch on I

quickly (parts of) speckle patterns that do not meet the modulation criterion I

15 replaced by other parts, so that the final result of the processing I

of the speckle patterns becomes considerably better. I

1024069 I

Claims (19)

Method for obtaining a series of interference patterns by means of speckle interferometry, comprising obtaining a new interference pattern and further performing operations on the series of interference patterns, characterized in that the series of interference patterns comprises uncorrelated interference patterns and the method further comprises calculating a quality measure with respect to the new interference pattern and the execution of the operations on the series of interference patterns based on the quality measure 10 A method according to claim 1, used in particular for detecting a disturbance in an interferometry arrangement (1), further comprising: obtaining and storing a reference pattern, the reference pattern being one of the series of interference patterns occurring over time precedes the new interference pattern, and wherein calculating a quality measure and performing further operations comprises: - calculating a first degree of correspondence between the new interference pattern and the reference pattern; 20. if the first degree of agreement is above a predetermined comparison threshold, performing further calculations with the reference pattern and the new interference pattern; and - if the first degree of agreement is below a predetermined comparison threshold, the non-execution of further calculations with the reference pattern and the new interference pattern. The method of claim 2, wherein the reference pattern is an interference pattern immediately ahead of time for the new interference pattern. The method of claim 1 or 2, wherein performing further calculations further comprises storing the new interference pattern as a reference pattern if the first degree of correspondence is above a predetermined comparison threshold. 1024069 26 I A method according to claim 4, wherein after a predetermined number of times I perform the method steps, the first degree of correspondence always being below the predetermined comparison threshold, the new interference pattern is stored as the reference pattern. I 6. Method according to claim 1, used in particular for detecting a disturbance in an interferometry arrangement (1) and determining the nature of the disturbance, comprising: obtaining and storing a reference pattern, wherein the reference pattern is an I of the series of interference patterns that precedes the new I interference pattern in time, I wherein calculating a quality measure comprises stirring out further I operations: calculating a first degree of correspondence between the new I interference pattern and the reference pattern; If the first degree of similarity is above a predetermined I comparison threshold: I • performing a further calculation with the new interference pattern and the reference pattern, and I • storing the new interference pattern as a reference pattern; I - if the first degree of agreement is below the predetermined I comparison threshold: I • obtaining a further new interference pattern that immediately follows the new interference pattern, I • calculating a second degree of agreement between the further I new interference pattern and the new interference pattern; If the second degree of agreement is above a predetermined I comparison threshold: I 30. performing a further calculation with the further new I interference pattern and the new interference pattern; Storing the further new interference pattern as a reference pattern. I 1024069 I A method according to any of claims 2 to 6, wherein the degree of similarity is calculated by subtracting pixel values of the interference patterns from each other and determining an average of the difference over the entire image. 5 A method according to any of claims 2 to 6, wherein the degree of similarity is calculated by subtracting pixel values of the interference patterns from each other, calculating a histogram of difference values over the entire image, and a statistical parameter of the histogram to decide. 10 A method according to any of claims 2 to 6, wherein the degree of similarity is determined by calculating a correlation coefficient of at least a sub-picture of the interference patterns. The method of claim 1, wherein the series of interference patterns comprises a number of mutually non-correlated subsets, characterized in that calculating a quality measure and performing further operations comprises:. - calculating the modulation depth of predetermined parts of the new interference pattern; and 20. further processing the predetermined parts for which the modulation depth is above a predetermined threshold value. 11. Method as claimed in claim 10, further comprising: - replacing one of the predetermined parts for which the modulation depth is below the predetermined threshold, by a corresponding part of another series of interference patterns for which the modulation depth is above the predetermined threshold. 12. The method of claim 11, wherein the other set of interference patterns is a time-neighboring interference pattern. The method of claim 10, wherein the one predetermined portion for which the modulation depth is below the predetermined threshold is replaced by a 1024069 Η 28 I corresponding part, which is derived from a plurality of corresponding parts I of other interference patterns from sequences for which the modulation depth is above the I predetermined threshold I A method according to any of claims 1 to 13, wherein the method I is used to analyze an object under load, and the series I interference patterns consists of at least a first series of interference patterns I obtained during application of a reproducible load of the object and at least a second series of interference patterns obtained during the removal of the reproducible load of the object, the results of further processing of the second series of interference patterns being taken into account with an opposite note the results of further processing of the first series. I Device for obtaining a series of I interference patterns from an interferometry device, comprising processing means (21) connected to I detection means (2) of the interferometry device (1), the I processing means (21) drawing means (22) ) and I memory means (23) connected to the calculating means, and wherein the drawing means are adapted to perform the method according to one of claims 1 to 14. I 16. Device as claimed in claim 15, wherein the interferometry device (1) is further provided with means for influencing a speckle pattern on the detection means (2), the means for influencing being selected from the group of: I a piezoelectrically controlled diffuser for influencing the illumination of the object; I a liquid crystal modulator or a spatial light modulator (SLM) for influencing the direction of observation of the interferometry device. I 30 I ******** I 1024069_I