JP3401783B2 - Surface profile measuring device - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to the structure of an apparatus for measuring the surface shape of an object.

[0002]

2. Description of the Related Art Many techniques for measuring the surface shape of an object have been proposed. Among them, as a technology that can measure the surface shape at a very high speed, there is a technology that projects a stripe pattern on an object, obtains the phase of the stripe at each point from each pixel from the image, and calculates the object surface shape from the phase information. Are known. Hereinafter, this technique is referred to as a phase detection method. In the following, as a conventional technique, this phase detection method will be described first.

In principle, the surface shape measurement by the phase detection method is divided into two. One is based on the principle of triangulation, and the other is based on the principle of light wave interference.

FIG. 6 shows an example based on the principle of triangulation. Now, assume that the height of the object is smaller than the measurement range a in the figure. Considering the position of the fringe projected on the reference plane z1 as a reference, for example, the point p where the phase of the seventh fringe of the 10 fringes in the figure on the reference plane z1 is θ is p ′ depending on the shape of the object surface. Actually, it will be observed at the position of. At this time, the height h of the object surface from the reference plane at the point p ′ can be obtained as h = (p−p ′) / tan α. In other words, if the phase of the fringe at that point and the order (order of the fringe) can be determined for each pixel, the height of the object surface can be calculated for each point.

The measurement range a is a range in which the amount | p−p ′ | of one stripe having a stripe pattern deviated from the position on the reference plane z1 due to the change of the height of the object does not exceed one cycle T of the stripe. Within this range, it is not necessary to consider the order of fringes, and the shape can be calculated by simply obtaining the phase. What happens to an object with a height exceeding this range? In this case, the correct height cannot be obtained only by the phase. It is necessary to determine the order of the stripes. That is, the phase is θ = 2 at the position of point p ′.
There are various possible heights such as π | p−p ′ | / T as shown in the figure. That is, θ has an uncertainty of 2πn (n is an integer). However, the order of stripes, n, is generally not easily determined. If you do not know the order of the stripes, you will have to determine the height from the phase alone.
Even if the amount of p-p '| exceeds T, it is not known that it has exceeded T. Therefore, there is no choice but to calculate as | p-p' | -T. More generally, | p-p '| -n is applied to an object with a height that causes a positional deviation that exceeds the cycle of stripes by many cycles.
It will be calculated as T. Such an operation is said to be folded (wrapping) in this area because the height of the object does not always come out of the area a. It is called unwrapping that the folded information is converted to the unfolded state by determining the order of the stripes by some means.

FIG. 7 shows an example of a surface shape measuring device based on the principle of light wave interference. In this example, a Michelson type interferometer is used to cause the light waves to interfere. Object 73
An interference fringe that causes positive interference is observed at a position where the difference in optical path length between the reference mirror 72 and the reference mirror 72 is an integral multiple of 1 / 2λ (where λ is the wavelength of the used light). If the light source 71 is coherent light, interference fringes can be observed over a wide range in the Z direction. For example, when the plane is tilted, an interference fringe pattern such as a diffraction grating sticks to an object. The interference pattern disappears when the plane is too far from the position of the focal plane of the objective lens and is completely blurred, or when the optical path length difference exceeds the coherent length of the illumination light. In this case, the fringe phase directly represents the height of the surface shape of the object.
However, just by obtaining the phase, it is in the state of being folded in the area of 1 / 2λ as in the case of the triangulation method, and it is necessary to obtain the order of the fringes and perform unwrapping.

Next, how to obtain the phase will be described. There are roughly two ways to obtain the phase. One is a multi-step method that shifts the projected fringes horizontally and obtains the initial phase from multiple images with different phases, and the other is the one-step method that obtains the phase from one fringe image.

First, the multi-step method will be described. As shown in FIG. 8, a plurality of images in which the phases of the stripes are shifted are obtained.
Here, in order to simplify the description, a case where four images are used will be described. Expressing the resulting image as an equation, I (x, y) = a
(x, y) + b (x, y) cos (2πfx + φ (x, y)). Where I
(x, y) represents the brightness of the image at the coordinates x, y. a (x, y) represents the bias component and b (x, y) represents the amplitude component of the stripe, which change depending on the pattern of the object and the difference in reflectance. f is the spatial frequency of the stripe, and φ (x, y) is the phase to be obtained. The image obtained by shifting the stripes by π / 2 in the X direction is I _i (x, y) =
a (x, y) + b (x, y) cos (2πfx + φ (x, y) + πi / 2) (i = 0,1,2,
It is represented by 3). From these equations, φ (x, y) = atan ((I ₁ (x, y)-
The phase can be obtained as I ₃ (x, y)) / (I ₀ (x, y) -I ₂ (x, y)). Hereinafter, this method is called a phase shift method.

The one-step method is the Fourier transform method (M. Take
da and K. Mutoh, "Fourier-transform profile for
the automatic measurement of 3-D object shape, "A
ppl.Opt. 22 (24), 3977-3982 (1983)) is a method for obtaining the phase from a single image. The Fourier transform method captures the deformation of the projected fringes due to the undulations of the object as phase modulation using the projected fringes as a carrier. The phase is obtained by transforming the carrier wave component and the modulated wave component by Fourier transform and inverse Fourier transforming only the modulated wave component. It is assumed that the frequency of the modulated wave is clearly low enough to be completely separated from the frequency of the carrier wave in the frequency space. In this respect, there is a lateral resolution limitation, but since only one image is required, it is fast. Is possible. As another one-step method, an electronic moire phase shift method (for example, Arai, Yokoseki, Shiraki, Yamada, “two-dimensional spatial fringe analysis method using CCD image sampling technology”,
Optics, 25 (1), 42-47 (1995)). Moire fringes occur when the fringes are roughly sampled in the image. This moire fringe can be accurately phase-shifted by changing the sampling position. This makes it possible to generate an image in which the moire fringes are phase-shifted from one image. The phase of the moire fringes can be obtained by performing an operation using the generated image in the same manner as the phase shift method of the multi-step method.

The above phase detection method can be used regardless of the principle of the triangulation method or the light wave interference method.

Another prior art related to the present invention
There is a Shape from Focus method. Shape from Focus method is focus information (information indicating whether the focus is in or out)
This is a method for obtaining the surface shape of an object from. The principle will be briefly described. A pattern is projected on an object and an image of it is acquired. The pattern contrast is obtained for each pixel of the image by image processing. Next, the same process is performed by changing the positional relationship between the object and the objective lens using the Z table. Repeat this. Thus, when the contrast of the pattern on the object is obtained by changing the positional relationship between the object and the objective lens, the contrast changes according to the focus condition. When the pattern is in focus and the pattern looks sharp, the contrast is high, and when the image is blurred, the pattern mixes with the ground and the pattern has almost no contrast. The manner of this contrast change is shown in FIG. It is a mountain-shaped curve that gives the maximum contrast when the Z table is moved to bring it into focus. The shape of the surface of the object can be obtained by moving the Z table to obtain the contrast for each pixel, and repeating this to obtain the position of the Z table that gives the maximum contrast for each pixel.

There are several methods for obtaining the contrast. For example, in order to obtain the contrast of a certain pixel, a local region is provided around the pixel and the difference between the maximum and the minimum in that region is obtained, or the sum of the differential values in the local region is calculated. Recently, a more complete method was proposed (MAANei
l, R. Juskaitis, and T. Wilson "Method of obtaining o
ptical sectioning by using structured light in ac
onventional microscope, "Optics Letters, 22 (24), 1905
-1907 (1997)). This method is almost the same as the phase shift method described above. That is, a periodic stripe pattern is projected on the object, and the phases of the stripes are gradually shifted to obtain a plurality of images. By using these images, the fringe contrast component can be obtained in the same manner as when the phase is obtained. For example, in the case of using _three images (I ₁ , I ₂ , I ₃ ) in which the phases of the stripes differ by 2 / 3π, the stripe contrast is calculated from the following equation.
Ip can be calculated Ip ＝ [(I ₁ -I ₂ ) ² + (I ₁ -I ₃ ) ² + (I _2-
I ₃ ) ² ] ^1/2 . According to this method, it is possible to obtain the stripe contrast purely in pixel units without providing a local region or the like.

[0013]

The problem of the surface shape measurement by the phase detection method is how to perform unwrapping, as is clear from the above description. Unwrapping is generally performed by connecting phases so that adjacent phases are smoothly connected. This is effective when the surface shape of the object changes smoothly, but there are cases where there are sharp steps or areas where the phase cannot be obtained due to holes etc. Therefore, such a phase connection cannot be perfect.

In recent years, many attempts have been made that can be called a multi-wavelength method such as performing unwrapping by projecting different spatial frequency fringes and matching two or more types of phase data. However, these are more striped (rather than two) for unwrapping.
It is better to say that the area that does not need unwrapping is expanded and it is not perfect. In addition, an object having a height change exceeding the depth of focus cannot be measured at all, and there are practical problems such as having to prepare a plurality of stripe patterns or light sources having different wavelengths, and switching between them. is there.

On the other hand, the Shape from Focus method does not have any problems like the phase detection method. Basically, any step height can be measured if the Z table can be moved (however, it cannot exceed the working distance of the objective lens, because it will collide). Since it is a method of finding the in-focus point, there is no need to say that it is out of focus and measurement is not possible, and it is not necessary to prepare multiple patterns.

However, the Shape from Focus method also has a problem. The measurement accuracy of the Shape from Focus method is higher as the width of the curve shown in FIG. 9 (hereinafter referred to as an axial response curve) is narrower. In other words, the narrower the area, the higher the discriminating ability at the in-focus position. The width of this on-axis response curve is determined by the numerical aperture (hereinafter referred to as NA) of the objective lens, the frequency of the projected fringes and the pixel resolution (field size / number of pixels). Therefore, when the objective lens becomes low magnification, NA
Becomes smaller and the pixel resolution becomes rough, so that the width of this on-axis response curve becomes remarkably wide. In other words, high-accuracy measurement at low magnification is impossible or extremely difficult. When the surface shape measuring device is used as an inspection device in a factory production line or the like, it is required to be able to measure a wide field of view all at once. It is fatal that high precision measurement at low magnification is not possible.

On the contrary, in the phase detection method, the magnification of the objective lens
There is basically no relation between A and the measurement accuracy in the optical axis direction. Therefore, high-accuracy measurement at low magnification is sufficiently possible.

In general, even if the magnification is high and the NA is high, Sh
The measurement accuracy of the ape from Focus method is considerably worse than that of the phase detection method based on the principle of light wave interference.

The present invention aims to solve these problems at the same time. In other words, even with low magnification, it is possible to measure with high accuracy, there is no problem of unwrapping or it is completely solved, there is no influence of defocus, and there is no need to prepare multiple stripe patterns. It is intended to provide.

[0020]

In the present invention, it is noted that the phase detection method and the Shape from Focus method have almost the same calculation method but their mutual problems do not exist on the other side. I tried to tie them together and eliminate each other's problems.

That is, the pattern projection mechanism for projecting a periodic fringe pattern on the object to be measured, the imaging optical system for forming an optical image of the object on which the fringe pattern is projected, and the imaging optical system An image sensor that photoelectrically converts the obtained optical image of the object into an electric signal, a focus moving mechanism that displaces the object-side focal plane of the imaging optical system at least once, and a cycle obtained by the image sensor. A phase-amplitude calculating means for obtaining the phase and amplitude of the stripe pattern at that point for each pixel with respect to the image of a projected object having a typical stripe pattern, and the phase-amplitude calculating means at that position each time the focal plane is displaced. The phase and amplitude are calculated, and the maximum vibration that estimates the position where the amplitude becomes maximum based on the change information of the amplitude that changes by displacing the focal plane by interpolation calculation with a finer precision than the displacement interval of the focal plane. The position calculation means, and the phase of a certain pixel, the maximum amplitude phase detection means that uses the phase information of the pixel having the largest amplitude among the phase information obtained by the phase amplitude calculation means in each focal plane, A surface shape calculation unit that calculates accurate height information of the object surface using the maximum amplitude position information obtained by the maximum amplitude position calculation unit and the maximum amplitude phase information obtained by the maximum amplitude phase detection unit; The device is configured to include.

At this time, the optical axis of the pattern projection mechanism is projected from a direction inclined with respect to the optical axis of the imaging optical system.

Alternatively, the pattern projection mechanism projects the light wave interference fringes.

Further, the phase / amplitude calculating means has a phase shifter for changing the phase of the stripe pattern in order to calculate the phase and the amplitude, and the phase / amplitude calculating means changes the phase of the stripe pattern at least twice using the phase shifter. Then, the phase shift method calculation for calculating the phase and the amplitude is performed from the three or more images having the mutually different fringe phases.

Alternatively, the phase / amplitude calculating means calculates from a single image on which the fringe pattern is projected by using the Fourier transform method or the electronic moire phase shift method.

By configuring the apparatus as described above, the phase detection method and the Shape from Focus method complement each other's problems, and efficient calculation becomes possible. In other words, although the measurement is basically performed using the phase detection method that can measure with high accuracy even at low magnification, the contrast calculation of the stripes is also executed at the same time as the phase detection calculation, and the position of the focal plane is changed by the focus moving mechanism. If you perform the same calculation with
Since the epe from Focus method can be calculated, the rough surface shape can be measured while the measurement accuracy is poor. Complete unwrapping can be easily performed by using this rough surface shape measurement information. Sufficient unwrapping is possible because only a very low measurement accuracy of the surface shape required for unwrapping (accuracy about the width of the folded region) is required. Further, the phase detection calculation can obtain the phase calculation result when the blurred fringes are not projected by the focus moving mechanism even if the object has a very high step, so that there is no problem of out-of-focus reliability. High phase calculation is possible. Further, the stripe pattern to be projected may always be the same.

[0027]

BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 shows a first example of the embodiment of the present invention. In this example, the principle of triangulation is used as the phase detection method, and the pattern projection mechanism 1
The optical axis 7 is inclined with respect to the optical axis of the imaging optical system 6.

The structure of the device will be described. Both the pattern projection mechanism 17 and the imaging optical system 6 are optical systems that are telecentric on both sides, and the magnification does not change even when the image is blurred. The pattern projection mechanism 17 includes a light source 1, a collector lens 2 that collects light from the light source 1 and irradiates the pattern mask 3, a pattern mask 3 having a periodic stripe pattern, and a phase shifter that shifts the phase of the pattern mask 3. 4 and a telecentric projection lens 5 at an angle α with respect to the optical axis of the imaging optical system 6 and the imaging optical system 6
The image of the pattern mask 3 is formed so as to be superimposed on the focal plane of. Therefore, the pattern mask 3 is arranged so as to be inclined with respect to the optical axis of the projection lens 5. The phase shifter 4 is a uniaxial drive mechanism that shifts the pattern mask 3 in the arrow direction with high accuracy.

The image forming optical system 6 is a telecentric optical system for forming an image of the object 7 illuminated by the pattern projection mechanism 17 on the image sensor 8. Focus moving mechanism 9
In this example, is composed of a stage 10 movable along the optical axis of the imaging optical system 6 and its controller 11. With this mechanism, the positional relationship of the focal plane of the imaging optical system 6 with respect to the object 7 can be changed. The image obtained by the image sensor 8 is sent to an image processing device 16 having a phase amplitude calculating means 12, a maximum amplitude position calculating means 13, a maximum amplitude / phase detecting means 14, and a surface shape calculating means 15 to be processed.

Here, the phase shifter 4 is the pattern mask 3
It is realized by a uniaxial drive mechanism, but is not limited to this. An actuator such as a piezo element may be used for movement, or a pattern may be drawn on a liquid crystal display without using a mechanical drive mechanism and the drawing pattern may be shifted. In any case, any mechanism can be used as long as the projection pattern can be accurately shifted.

The focus moving mechanism 9 is realized by moving the position of the stage 10, but the present invention is not limited to this. The optical system side may be moved, or if it is a minute movement, only a part of the objective lens may be moved. Alternatively, instead of mechanical movement, a transparent body with a different refractive index from air is inserted between the objective lens and the object, and the refractive index is changed, or the thickness of the transparent body is changed to move the focus. May be realized. In any case, any mechanism may be used as long as the focal position of the objective lens moves relative to the object.

How the surface shape measurement is performed by this apparatus will be described step by step. Focus moving mechanism 9
It is assumed that the coordinates in the optical axis direction of are in the state of z1. At this position, the phase and amplitude of the periodic stripe pattern projected on the object 7 are first obtained for each pixel. The phase shift method is used for the calculation. That is, first one image is taken, and then the phase of the pattern mask 3 is shifted by the phase shifter 4. Then capture the image again. Similarly, the phase shifter 4 shifts the phase of the pattern mask 3 to capture an image. If the number of shifts is two or more, that is, three or more images having different phases, the phase and amplitude can be calculated. In this way, the image is taken into the image processing apparatus 16 every time the phase shift is performed a plurality of times. The phase and amplitude calculation means 12 obtains the phase and the amplitude for each pixel from the plurality of captured images, and two images of the phase image and the amplitude image are obtained as a result.

If higher speed is emphasized, it is of course possible to calculate the phase / amplitude information from one image using the Fourier transform method or the electronic moire phase shift method. In this case, the phase shifter 4 is not necessary.

Next, the focus moving mechanism 9 is used to form the image forming optical system 6.
The positional relationship between the focal plane of and the object 7 is changed. That is,
The position of the stage is moved from z1 to z2 = z1−Δz. Then, at that position, the phase and the amplitude are obtained again as described above. Next is z3 = z2-Δz, next is z4 = z3
-Δz, z5, z6. ． ． Then, the phase and the amplitude are sequentially obtained. The phase image and the amplitude image obtained at the position of each stage are sent to the maximum amplitude position calculating means 13 and the maximum amplitude phase detecting means 14.

The maximum amplitude position calculation means 13 evaluates only the amplitude image. Consider a pixel. FIG. 2 shows the obtained amplitude information in a graph with the position of the stage as the horizontal axis. As described in the explanation of the Shape from Focus method, the fringe amplitude, in other words, the contrast should be a mountain-shaped curve having a peak at the focus position, and the obtained amplitude information is obtained by sampling this mountain. Conceivable. From the sampled amplitude information, the focus position, which is the maximum amplitude position, is obtained by interpolation calculation. For example, the maximum value v _k of the obtained amplitude information and the values before and after it
It is _obtained by fitting a Gaussian function using three values of v _{k + 1} and v _k-1 . That is, the focus position z _p = z _k +
(ln (v _{k + 1} ) -ln (v _k-1 )) Δz / (2 ・ (2 ・ ln (v _k ) -ln (v _k-1 ) -ln
(v _{k + 1} ))). Here, Δz is the focus movement pitch, and ln is the natural logarithmic calculation. The interpolation calculation may be not only the fitting to the Gaussian function described above but also the fitting to other mountain-shaped functions, and the focus position can be obtained by the gravity center calculation or the correlation calculation. Such focus position estimation calculation is performed for all pixels. Since the in-focus position represents the position of the object surface, the surface shape can be measured by this calculation. This is surface shape measurement by the Shape frm Focus method. However, as described above, this result does not have sufficient accuracy when the imaging optical system has a low magnification and a low NA. In the present invention, this result is used for unwrapping the result obtained by the phase detection method. This will be described when the surface shape calculation means 15 is described.

Next, the maximum amplitude / phase detecting means 14 will be described. Since the phase amplitude calculation means 12 detects the phase each time the positional relationship between the focal plane of the imaging optical system 6 and the object 7 is changed by the focus moving mechanism 9, the phase is repeatedly detected for a certain pixel. Will be detected. Therefore, it is necessary to select when to use the phase detection result. It is the maximum amplitude / phase detection means 14 that makes this selection,
The value of the phase when the amplitude is maximum is stored. That is, a plurality of phase images are integrated and output as one image (hereinafter referred to as maximum amplitude phase image) in which the most reliable phase information is collected. Conventionally, the phase detection method generally makes measurement impossible when the focus is deviated, but even if the step on the object is so large that both the top and bottom of the step cannot be in focus at the same time, the phase detection method is reliable. It is possible to perform highly accurate phase measurement.

The surface shape calculation means 15 unwraps the maximum amplitude phase image obtained here using the rough surface shape measurement result obtained by the maximum amplitude position calculation means 13. This will be described with reference to FIG. When the height corresponding to 2π (one cycle of stripes) is a, the maximum amplitude phase image is all folded in this region. The phase, which is the value of each pixel in this image, includes an uncertain factor of 2πn (n is an integer). That is, even if the phase of a certain point p is θ, it actually means θ + 2πn, and this n is unknown.
As shown in FIG. 3, the height at which the point p is the phase θ is a
It means that there can be many in the cycle. The process of determining n is unwrapping. In order to determine this n, it suffices to have surface shape measurement data obtained with a rough accuracy higher than ± a / 2. The result obtained by the maximum amplitude position calculation means 13 as this data, that is, Sh
Use the result of the ape from Focus method. There is no comparison with the value of the adjacent pixel like the general phase connection processing, and it is purely mechanically determined from the data of that pixel only, so even if there is a large step difference or the surrounding pixels are Even if it is impossible to measure due to holes etc., it can be obtained without any problem.

Generally, the phase detection method is a relative measurement, and the absolute position cannot be determined. Consider, for example, measuring a plane parallel to the focal plane of the imaging optical system 6, and if that plane is at z1, z11, z12, z13 ... In FIG. Is the same. Therefore, it is not known at which z position the plane is actually located. The result obtained by moving the Z table and detecting the phase cannot be associated with the result before the movement. That is, it is not possible to move the Z table and integrate the results of phase detection only by phase detection. Therefore, a large step that exceeds the depth of focus cannot be measured. However, in the present invention, since unwrapping is performed using the result of the focus calculation indicating the absolute position, absolute measurement is performed, and the results obtained by moving the Z table can easily correspond to each other. That is, as long as the focus moving mechanism 9 is movable and the working distance is sufficient, it is possible to measure any large step.

In the phase detection method, generally, the higher the spatial frequency of stripes (smaller pitch of stripes), the higher the accuracy. However, if the spatial frequency of the stripes is increased, the measurable range without unwrapping becomes extremely narrow. In general, unwrapping is a time-consuming and fairly elaborate process, and it often does not work depending on the shape of the object, so the spatial frequency of stripes is widened in order to widen the measurable range in order to simplify unwrapping as much as possible. Lower to use. Therefore, there is a possibility that the accuracy is not always sufficient. Since the present invention can easily perform complete unwrapping, the spatial frequency of stripes can be increased so as to maintain sufficient accuracy. In this sense, it is possible to expect higher precision measurement than before.

Next, a second example of the embodiment of the present invention will be described with reference to FIG. This example is an example of an apparatus based on the principle of light wave interference, not the triangulation method, as the measurement principle. In this case, unlike the first example, the pattern projection mechanism 44
Is a Michelson-type interferometer in which the optical axis of (1) and the optical axis of the imaging optical system 6 are coaxial. The illumination light emitted from the light source 1 passes through the illumination optical system 45 and enters the beam splitter 41. The illumination light is split into two beams by the beam splitter 41, and illuminates the reference mirror 42 and the object 7, respectively. The light reflected by the reference mirror 42 and the light reflected by the object 7 are superimposed again by the beam splitter 41, and an image of the object 7 is formed on the imaging sensor 8 by the imaging optical system 6. At this time, the beam from the reference mirror 42 interferes with the beam from the object 7, and the optical path difference between these beams is ½λ (where λ
Is the center wavelength of the illuminating light). The light source 1 used here is a low coherent light source having a coherent length of about several tens of μm. As a light source 1 having a coherent length of this degree, there is, for example, a super luminescent diode. When such a low coherent light source is used, the interference is very local, and the interference fringes are observed only in the coherent length region centered on the position where the optical path difference between the two beams is 0. It When the intensity of one pixel of the image sensor 8 is recorded while moving the stage 10 of the focus moving mechanism 9, the intensity changes as shown in FIG. The peak of the envelope is the position where the optical path difference between the two beams is 0, and that position coincides with the peak of the maximum amplitude in the vibration waveform (interference fringe) of 1/2 λ cycle. This can be considered in the same way as in the first example by considering the envelope as focusing information and the interference fringes as a periodic fringe pattern.

That is, the object shape is calculated from the contrast information of the interference fringes giving the envelope with a rough accuracy higher than ± 1 / 4λ, and the phase of the interference fringes is obtained to obtain a more accurate peak position. You can get information. Since the phase is folded in the ½λ (2π) region, the rough shape measurement result obtained from the contrast information becomes the information for unwrapping.

The calculation method will be described more concretely. The phase shifter 43, which is a piezo actuator, moves the reference mirror 42 to perform a phase shift calculation. For example, an image is first taken and the reference mirror 42 is used for 2 / 3π (1 / 3λ).
To move and take the second image, then move 2 / 3π again and take the third image. From the obtained three images, the phase and amplitude calculating means 12 calculates the phase and the amplitude for each pixel. At this time, the phase shifter 43 may also be used as the focus moving mechanism 9 in order to shift the phase.

Alternatively, the object 7 is slightly tilted to generate spatial interference fringes, and the Fourier transform method or the phase shift moire method is used as the phase amplitude calculation means 12 without mechanical phase shift, and the phase and amplitude of each pixel are calculated. May be asked.

A fixed step (several μ
Then, the amplitude and the phase are obtained by performing the same calculation as above at each step. Shape calculation is performed using the amplitude image and the phase image obtained in each step. The following calculation method is exactly the same as in the first embodiment.

[0045]

According to the present invention, two methods in which the problems are complementary to each other by using the values of both the phase detection and the amplitude detection of a periodic pattern whose processing is almost the same (which can be performed at the same time) are the Shape from Focus method. By combining both of the phase shift methods, the conventional problems can be solved completely. This will broaden the range of applications for high-speed surface profile measurement.

[Brief description of drawings]

FIG. 1 is a diagram showing a first example of an exemplary embodiment of the present invention.

FIG. 2 is a diagram for explaining a Shape from Focus calculation.

FIG. 3 is a diagram for explaining the principle of the present invention.

FIG. 4 is a diagram showing a second example of the exemplary embodiment of the present invention.

FIG. 5 is a diagram showing an interference waveform due to a low coherence light source.

FIG. 6 is a diagram showing the principle of phase detection based on triangulation in the prior art.

FIG. 7 is a diagram showing an example of a conventional light wave interference device.

FIG. 8 is a diagram for explaining a phase shift.

FIG. 9 is a diagram for explaining the Shape from Focus method.

[Explanation of symbols]

1 light source 2 collector lens 3 pattern masks 4 phase shifter 5 Projection lens 6 Imaging optical system 7 objects 8 image sensor 9 Focus movement mechanism 10 platform 11 controller 12 Phase amplitude calculation means 13 Maximum amplitude position calculation means 14 Maximum amplitude phase detection means 15 Surface shape calculation means 16 Image processing device 17 pattern projection mechanism 41 beam splitter 42 Reference mirror 43 Phase shifter 44 pattern projection mechanism 45 Illumination optical system

Claims (5)

(57) [Claims] 1. A pattern projection mechanism for projecting a periodic fringe pattern onto an object to be measured, an imaging optical system for forming an optical image of the object on which the fringe pattern is projected, and the imaging optical system. An image sensor that photoelectrically converts the obtained optical image of the object into an electric signal, a focus moving mechanism that displaces the object-side focal plane of the imaging optical system at least once, and a cycle obtained by the image sensor. A phase-amplitude calculating means for obtaining the phase and amplitude of the stripe pattern at that point for each pixel with respect to the image of a projected object having a typical stripe pattern, and the phase-amplitude calculating means at that position each time the focal plane is displaced. Maximum amplitude position to estimate the position where the amplitude is maximum based on the change information of the amplitude, which is obtained by deriving the phase and amplitude and changing the focal plane, by the interpolation calculation with a finer precision than the displacement interval of the focal plane. Performance Calculating means, the phase of a certain pixel, the maximum amplitude phase detecting means for the phase information of the pixel when the maximum amplitude of the phase information obtained by the phase amplitude calculating means in each focal plane, Surface shape calculation means for calculating accurate height information of the object surface using the maximum amplitude position information obtained by the maximum amplitude position calculation means and the maximum amplitude phase information obtained by the maximum amplitude phase detection means. A surface shape measuring device comprising: 2. The surface shape measuring apparatus according to claim 1, wherein the optical axis of the pattern projection mechanism is projected from a direction inclined with respect to the optical axis of the imaging optical system. 3. The surface shape measuring apparatus according to claim 1, wherein the pattern projection mechanism projects light wave interference fringes. 4. The phase / amplitude calculating means has a phase shifter for changing the phase of the stripe pattern in order to calculate the phase and the amplitude, and the phase / amplitude calculating means uses the phase shifter to change the phase of the stripe pattern at least twice. 4. The surface profile measuring apparatus according to claim 1, wherein the phase shift method calculation for calculating the phase and the amplitude is performed from three or more images obtained by causing the phases of the fringes to be different from each other. . 5. The phase / amplitude calculating means calculates from a single image on which a fringe pattern is projected, using a Fourier transform method or an electronic moire phase shift method.
And the surface shape measuring apparatus according to claim 2 or claim 3.