JP5413072B2 - Waveform analysis apparatus, waveform measurement apparatus, waveform analysis program, interferometer apparatus, pattern projection shape measurement apparatus, and waveform analysis method - Google Patents

Description

  The present invention relates to a waveform analysis device, a waveform measurement device, a waveform analysis program, an interferometer device, a pattern projection shape measurement device, and a waveform analysis method to which a Fourier transform method is applied.

  The fringe analysis processing by the Fourier transform method invented by Takeda et al. Is well known (see Non-Patent Document 1). In order to calculate the phase distribution of fringes from a single fringe image, interference fringes with carriers superimposed are used for the fringe analysis processing. In the fringe analysis processing, the intensity distribution of the fringe image is Fourier transformed, and only the + 1st order or −1st order spectrum is extracted from the obtained Fourier spectrum, the spectrum is inversely Fourier transformed, and the obtained complex amplitude distribution is predetermined. It is applied to By subtracting a known carrier component from the calculated phase distribution, the signal component included in the phase distribution can be made known.

Mitsuo Takeda et al. "Fourier-transform method of fringe-pattern analysis for computer-based topography and interferometry", Journal of the Optical Society of America.Vol. 72, No. 1, January 1982

  However, in this Fourier transform method that separates necessary information and unnecessary information only by spatial frequency, some of the necessary information is accidentally removed or unnecessary when the spatial frequency of the signal component covers a wide band. In some cases, some of the information could not be completely removed, which caused a large analysis error.

  Therefore, an object of the present invention is to provide a waveform analysis apparatus, a waveform measurement apparatus, a waveform analysis program, and a waveform analysis method using a Fourier transform method that can reliably suppress analysis errors. It is another object of the present invention to provide an interferometer device and a pattern projection shape measuring device with high measurement accuracy.

One aspect illustrating the waveform analysis apparatus of the present invention is a waveform analysis apparatus that extracts phase information of the input waveform by performing waveform analysis processing by a Fourier transform method on an input waveform on which carriers are superimposed. Analysis that narrows at least one of the + 1st order spectrum and the −1st order spectrum from the Fourier spectrum of the waveform and then separates the spectra, and calculates the phase information of the input waveform based on one of the separated spectra. And an estimation means for estimating an analysis error generated by the analysis means when performing a waveform analysis on the input waveform by performing the same waveform analysis as the analysis means on the input waveform model,
Correction means for correcting the phase information calculated by the analysis means based on the analysis error estimated by the estimation means.

  The waveform measuring apparatus according to the present invention is a one-dimensional or two-dimensional pixel array that records fringes on which carriers are superimposed, and the waveform analyzing apparatus according to the present invention that processes the fringes recorded by the pixel array as the input waveform. A limiting means for limiting the aperture ratio of the pixel is provided in each pixel included in the pixel array.

  An aspect of the waveform analysis program of the present invention is a waveform analysis program for extracting phase information of the input waveform by performing waveform analysis processing by a Fourier transform method on an input waveform on which a carrier is superimposed. An analysis procedure for narrowing at least one of the + 1st order spectrum and the −1st order spectrum from the Fourier spectrum of the waveform, separating the spectra, and calculating the phase information of the input waveform based on one of the separated spectra; An estimation procedure for estimating an analysis error caused by the analysis procedure when the input waveform is analyzed by performing the same waveform analysis as the analysis procedure on the input waveform model; and the estimation procedure A correction procedure for correcting the phase information calculated in the analysis procedure based on the analysis error estimated in

  One aspect illustrating the interferometer apparatus of the present invention includes one aspect illustrating the waveform analysis apparatus of the present invention.

  One aspect which illustrates the pattern projection shape measuring apparatus of this invention is equipped with one aspect which illustrates the waveform analysis apparatus of this invention.

  One aspect exemplifying the waveform analysis method of the present invention is a waveform analysis method for extracting phase information of the input waveform by performing waveform analysis processing by a Fourier transform method on an input waveform on which carriers are superimposed. An analysis procedure for narrowing at least one of the + 1st order spectrum and the −1st order spectrum from the Fourier spectrum of the waveform, separating the spectra, and calculating the phase information of the input waveform based on one of the separated spectra; An estimation procedure for estimating an analysis error caused by the analysis procedure when the input waveform is analyzed by performing the same waveform analysis as the analysis procedure on the input waveform model; and the estimation procedure And a correction procedure for correcting the phase information calculated in the analysis procedure based on the analysis error estimated in (1).

  According to the present invention, a waveform analysis apparatus, a waveform measurement apparatus, a waveform analysis program, and a waveform analysis method using a Fourier transform method that can reliably suppress analysis errors are realized. Further, according to the present invention, an interferometer device and a pattern projection shape measuring device with high measurement accuracy are realized.

It is a schematic block diagram of the interferometer apparatus of 1st Embodiment. It is a figure which shows the example of the shape of the measuring object surface 7a. It is a figure which shows the example of intensity distribution of an interference fringe. It is a flowchart of the analysis process of 1st Embodiment. It is a figure which shows the real part of Fourier spectrum h ((omega)). It is a figure which shows the imaginary part of the Fourier spectrum h ((omega)). It is a diagram showing the real part of the Fourier spectrum h _{2 (ω).} It is a diagram showing the imaginary part of the Fourier spectrum h _{2 (ω).} It is a diagram showing the real part of the Fourier spectrum h _{4 (ω).} It is a diagram showing the imaginary part of the Fourier spectrum h _{4 (ω).} It is a figure which shows the analysis error of the analysis process by the conventional Fourier-transform method. It is a figure which shows the analysis error of the analysis process of 1st Embodiment. It is a flowchart of the analysis process of 2nd Embodiment. It is a flowchart of the analysis process of 3rd Embodiment. It is a stripe image after pre-processing. It is a Fourier spectrum h ′ (ω) after removal of the zeroth-order spectrum. It is a Fourier spectrum h ₂ (ω) after narrowing the −1st order spectrum. It is a Fourier spectrum h ₄ (ω) after narrowing the + 1st order spectrum. It is a figure which shows the analysis error of 3rd Embodiment. It is a flowchart of the analysis process (analysis process with an error correction) of 4th Embodiment. It is a figure which shows the analysis error of 4th Embodiment. It is a schematic block diagram of the pattern projection shape measuring apparatus of 5th Embodiment.

[First Embodiment]
Hereinafter, a first embodiment of the present invention will be described. This embodiment is an embodiment of an interferometer apparatus.

  First, the configuration of the interferometer apparatus will be described. FIG. 1 is a schematic configuration diagram of the interferometer apparatus of the present embodiment.

  As shown in FIG. 1, the interferometer apparatus includes a laser light source 1, a beam expander 2, a polarizing beam splitter 3, a quarter wavelength plate 4, a Fizeau plate 5, a wavefront conversion lens 6, a beam diameter conversion optical system 8, A dimensional image detector 9 is provided. Although not shown, the interferometer apparatus is also provided with a computer. The measurement object 7 set in the interferometer device is an optical element such as an aspherical mirror or an aspherical lens, and is supported by a stage (not shown).

  Among these, the laser light source 1 emits a linearly polarized light beam L. The beam L expands its diameter by passing through the beam expander 2. The light beam L having an enlarged diameter enters the polarization beam splitter 3 and is reflected by the polarization separation surface 3 a of the polarization beam splitter 3. The polarization plane of the light beam L is selected in advance so as to be reflected by the polarization separation plane 3a. When the light beam L reflected by the polarization separation surface 3a enters the Fizeau surface 5a of the Fizeau plate 5 through the quarter-wave plate 4, the light beam LM passes through the Fizeau surface 5a and the light beam LR reflects the Fizeau surface 5a. To be separated. Hereinafter, the light beam LM is referred to as “measurement light beam LM”, and the light beam LR is referred to as “reference light beam LR”.

  The measurement light beam LM becomes a spherical wave by passing through the wavefront conversion lens 6. The measurement light beam LM enters the measurement target surface 7a of the measurement target 7 substantially perpendicularly. The position of the measuring object 7 in the optical axis direction is set to a position where the difference between the wavefront of the incident spherical wave and the measuring object surface 7a is small.

  The measurement light beam LM is reflected on the measurement target surface 7 a to return the optical path, and enters the polarization beam splitter 3 through the wavefront conversion lens 6, the Fizeau plate 5, and the quarter wavelength plate 4. The measurement light beam LM rotates the polarization plane by 90 ° by reciprocating the quarter-wave plate 4, so that it passes through the polarization separation surface 3 a of the polarization beam splitter 3 and enters the beam diameter conversion optical system 8. To do. The measurement light beam LM is reduced in diameter by passing through the beam diameter conversion optical system 8 and is incident on the two-dimensional image detector 9 in this state.

  On the other hand, the reference light beam LR is incident on the polarization beam splitter 3 through the quarter-wave plate 4. The reference light beam LR is rotated 90 ° by reciprocating the quarter-wave plate 4, so that it passes through the polarization separation surface 3 a of the polarization beam splitter 3 and enters the beam diameter conversion optical system 8. To do. The reference light beam LM is reduced in diameter by passing through the beam diameter converting optical system 8 and is incident on the two-dimensional image detector 9 in that state.

  Accordingly, interference fringes are generated on the two-dimensional image detector 9 due to the measurement light beam LM and the reference light beam LR. The interference fringe pattern reflects the shape of the measurement target surface 7a (see, for example, FIG. 2) (note that the horizontal axis in FIG. 2 indicates the distance from the optical axis to the number of pixels of the two-dimensional image detector 9). (The vertical axis in FIG. 2 represents the phase in terms of the light source wavelength λ.)

On the light incident side of the two-dimensional image detector 9, means for limiting the aperture ratio of each pixel (a diaphragm array or a mask pattern corresponding to the diaphragm array) is provided. The opening size of each pixel is set to about 1/8 of the pixel pitch, for example. As such a two-dimensional image detector 9, for example, the same CCD as that mounted on an interferometer apparatus (US Pat. No. 4,791,584) using sub-Nyquist interferometry can be employed. Such a two-dimensional image detector 9 can detect a fringe image having a finer fringe pitch than a normal two-dimensional image detector having a common pixel pitch and an aperture ratio that is not limited at all. it can.
The two-dimensional image detector 9 captures interference fringes and acquires a fringe image. The fringe image is input to a computer (not shown). A computer (not shown) performs analysis processing on the input fringe image. Note that the analysis processing program is preinstalled in the computer.
Here, the arrangement angle of the Fizeau plate 5 is set so that the incident angle of the light beam L with respect to the Fizeau surface 5a is a predetermined angle other than zero. In this case, there is a difference between the angle at which the measurement light beam LM is incident on the two-dimensional image detector 9 and the angle at which the reference light beam LR is incident on the two-dimensional image detector 9. For this reason, stripe-shaped carrier stripes (spatial carriers) are superimposed on the interference fringes.
For example, if the shape of the measurement target surface 7a is as shown in FIG. 2, the interference fringes are as shown in FIG. 3, for example, due to the overlapping of the carrier fringes (note that the horizontal axis in FIG. Is represented by the number of pixels of the two-dimensional image detector 9, and the vertical axis in FIG. In such interference fringes, the spatial frequency of necessary information is higher than the spatial frequency of unnecessary information, and therefore analysis processing by the Fourier transform method can be applied.
The inclination direction of the Fizeau plate 5 is selected so as to form an angle of 45 ° with respect to both the x axis and the y axis of the two-dimensional image detector 9. In this case, the carrier fringe has a common spatial frequency in both the x direction and the y direction.

Next, analysis processing by a computer will be described.
FIG. 4 is a flowchart of the analysis processing of this embodiment. Each step will be described in turn.
Step S10:
The computer inputs a fringe image. The intensity distribution g (x, y) of the fringe image is expressed by Expression (1).

In Expression (1), a (x, y) is the background unevenness (unknown) of the interference fringes, b (x, y) is the amplitude unevenness (unknown) of the interference fringes, and φ (x, y) Is the phase distribution (unknown) of the interference fringes. Among these, the phase distribution φ (x, y) is expressed by the sum of the carrier component φ _c (x, y) (known) and the signal component φ _s (x, y) (unknown) as shown in the equation (2). Is done.

Here, since the analysis target is an interference fringe, the signal component φ _s (x, y) includes the design data φ _d (x, y) (known) of the shape of the measurement target surface 7a and the shape of the measurement target surface 7a. It consists of the sum with the deviation amount δ (x, y) (unknown) from the design data. Since the deviation amount δ (x, y) is usually small, the design data φ _d (x, y) (known) is similar to the signal component φ _s (x, y) (unknown).

Step S11:
The computer obtains a Fourier spectrum h (ω) as shown in FIGS. 5 and 6 by performing a Fourier transform on the intensity distribution g (x, y) of the fringe image. FIG. 5 shows the real part of the Fourier spectrum h (ω), and FIG. 6 shows the imaginary part of the Fourier spectrum h (ω).

Step S12:
The computer removes the zero-order spectrum from the Fourier spectrum h (ω) shown in FIGS. The removal process is a bandpass filter process in which the intensity of a predetermined band centered at ω = 0 is set to 0 and the intensity of other bands is maintained. As a result, the background unevenness a (x, y) is removed from the Fourier spectrum h (ω).

Step S13:
The computer obtains a complex amplitude distribution g ₁ (x, y) by performing Fourier inverse transform on the Fourier spectrum h ′ (ω) after removal. The complex amplitude distribution g ₁ (x, y) is expressed by Expression (3).

On the right side of Equation (3), the first term corresponds to the + 1st order spectrum, and the second term corresponds to the −1st order spectrum. The term corresponding to the zeroth order spectrum has been removed.
Step S14:
The computer subjects the complex amplitude distribution g ₁ (x, y) to a -1st order spectrum narrowing process. In this narrowing process, the complex amplitude distribution g ₁ (x, y) is multiplied by a narrowing coefficient α (x, y) expressed by the equation (4).

However, φ _sm (x, y) in the equation (4) is a known model (model signal component) similar to the signal component φ _s (x, y). Here, it is assumed that design data φ _d (x, y) (known) of the signal component φ _s is used for the model signal component φ _sm (x, y).

Then, the complex amplitude distribution g ₂ (x, y) after the band narrowing process is expressed by Expression (5).

On the right side of Equation (5), the first term corresponds to the + 1st order spectrum, and the second term corresponds to the −1st order spectrum.
Here, considering that the model signal component φ _sm (x, y) is similar to the signal component φ _s (x, y), the phase component (φ _c + φ _sm + φ of the first term corresponding to the + 1st order spectrum is used. _s ) is similar to (φ _c + 2φ _s ), and the phase component (φ _c −φ _sm + φ _s ) of the second term corresponding to the −1st order spectrum is similar to φ _c . This indicates that the + 1st order spectrum is widened and the −1st order spectrum is narrowed.

Step S15:
The computer obtains a Fourier spectrum h ₂ (ω) as shown in FIGS. 7 and 8 by performing a Fourier transform on the complex amplitude distribution g ₂ (x, y). FIG. 7 shows the real part of the Fourier spectrum h ₂ (ω), and FIG. 8 shows the imaginary part of the Fourier spectrum h ₂ (ω). As is apparent from FIGS. 7 and 8, in the Fourier spectrum h ₂ (ω), the + 1st order spectrum is widened and the −1st order spectrum is narrowed.

Step S16:
The computer removes the −1st order spectrum from the Fourier spectrum h ₂ (ω) shown in FIGS. The removal process is a bandpass filter process in which the intensity of a predetermined band centered on the peak position of the −1st order spectrum is set to 0 and the intensity of other bands is maintained. Since the minus first-order spectrum is narrowed, this bandpass filter process can prevent missing of necessary information while reliably removing unnecessary information.

Step S17:
The computer obtains a complex amplitude distribution g ₃ (x, y) by performing Fourier inverse transform on the Fourier spectrum h ₂ ′ (ω) after removal. This complex amplitude distribution g ₃ (x, y) is expressed by Expression (6).

  The right side of Equation (6) represents a widened + first order spectrum.

Step S18:
The computer performs a narrowing process of the + first order spectrum on the complex amplitude distribution g ₃ (x, y). In this narrowing process, the complex amplitude distribution g ₃ (x, y) is multiplied by a narrowing coefficient β (x, y) expressed by Expression (7).

Then, the complex amplitude distribution g ₄ (x, y) after the band narrowing process is expressed by Expression (8).

  The right side of Equation (8) represents a narrowed + first order spectrum.

Here, considering that the model signal component φ _sm (x, y) is similar to the signal component φ _s (x, y), the phase component (φ _c −φ _sm + φ _s ) of this term becomes φ _c You can see that they are similar. This indicates that the + 1st order spectrum is narrowed.

Step S19:
The computer obtains a Fourier spectrum h ₄ (ω) as shown in FIGS. 9 and 10 by performing a Fourier transform on the complex amplitude distribution g ₄ (x, y). FIG. 9 shows the real part of the Fourier spectrum h ₄ (ω), and FIG. 10 shows the imaginary part of the Fourier spectrum h ₄ (ω). As is apparent from FIGS. 9 and 10, in the Fourier spectrum h ₄ (ω), the + 1st order spectrum is narrowed.

Step S20:
The computer extracts only the + first-order spectrum from the Fourier spectrum h ₄ (x, y). The extraction process is a bandpass filter process in which the intensity of a predetermined band centered on the peak position of the + 1st order spectrum is set to 0 and the intensity of other bands is maintained. Since the + 1st order spectrum is narrowed, this band pass filter process can prevent missing of necessary information while reliably removing unnecessary information.

Step S21:
The computer obtains a complex amplitude distribution g ₅ (x, y) by performing Fourier inverse transform on the extracted + 1st order spectrum.

Step S22:
The computer performs normalization processing on the complex amplitude distribution g ₅ (x, y). In this normalization process, the complex amplitude distribution g ₅ (x, y) is multiplied by the normalization coefficient γ (x, y) represented by Expression (9).

That is, the normalization process using the normalization coefficient γ is a process that cancels the narrowing process using the narrowing coefficients α and β.
Then, the complex amplitude distribution g ₆ (x, y) after the normalization process is expressed by Expression (10).

As apparent from Expression (10), according to the normalization process, the component (model signal component φ _sm (x, y)) generated by the narrowing process is deleted.

Step S23:
The computer calculates the phase distribution φ ′ (x, y) before unwrapping by applying the complex amplitude distribution g ₆ (x, y) to the following equation (11). The value of the phase distribution φ ′ (x, y) is limited to a range from −π to + π.

However, the function I [g ₆ ] indicates the imaginary part of g ₆ and the function R [g ₆ ] indicates the real part of g ₆ .

Step S24:
The computer performs unwrapping (phase connection) of the calculated phase distribution φ ′ (x, y), and obtains the phase distribution φ ″ (x, y) after unwrapping. Further, the phase distribution φ ″ ( The value of the signal component φ _s (x, y) is calculated by subtracting the carrier component φ _c (x, y) (known) from x, y). Hereinafter, this value is referred to as “calculated signal component φ _s ′ (x, y)”. This calculated signal component φ _s ′ (x, y) represents the shape of the measurement target surface 7a.
As described above, in the analysis processing of the present embodiment, the unnecessary spectrum is removed from the Fourier spectrum (step S16), and the spectrum is narrowed (step S14). Missing information can be prevented.
In the analysis processing of the present embodiment, the necessary spectrum is extracted from the Fourier spectrum (step S20), and the spectrum is narrowed (step S18). Missing information can be prevented.
FIG. 11 is a diagram showing an analysis error of the analysis process by the conventional Fourier transform method, and FIG. 12 is a diagram showing an analysis error of the analysis process of the present embodiment. 11 and 12, the horizontal axis represents the distance from the optical axis in terms of the number of pixels of the two-dimensional image detector 9, and the vertical axes in FIGS. 11 and 12 represent the analysis error (unit is phase). , Expressed by the light source wavelength λ.
As is apparent from FIGS. 11 and 12, the analysis error of the analysis processing of this embodiment is much smaller than that of the conventional one.
Therefore, according to the interferometer apparatus of this embodiment, the shape of the measurement target surface 7a can be measured with high accuracy.
In the analysis process of the present embodiment, the model signal component φ _sm (x, y) signal component φ _s (x, y) in the design data φ _d (x, y) of it was used, the signal component phi _s ( Other data similar to x, y) may be used.
For example, a calculated signal component acquired from the same fringe image by other analysis processing (for example, by a conventional Fourier transform method) may be used as the model signal component φ _sm (x, y).
Moreover, you may use the calculation signal component acquired by measuring the same measurement object surface 7a with another shape measuring apparatus (for example, coordinate measuring device) for model signal component (phi) _sm (x, y).
Alternatively, the calculated signal component may be fitted with a polynomial such as a power series or Zernike polynomial, and the calculated signal component after the fitting may be used as the model signal component φ _sm (x, y).
It is preferable that the model signal component φ _sm (x, y) used is more similar to the signal component φ _s (x, y) because the analysis error becomes smaller.

[Second Embodiment]
Hereinafter, a second embodiment of the present invention will be described. This embodiment is also an embodiment of an interferometer device. Here, only differences from the first embodiment will be described. The difference is in the content of the analysis process.

FIG. 13 is a flowchart of the analysis processing of this embodiment. This flowchart corresponds to the flowchart of FIG. 4 with steps 101, 102, and 103 added. Hereafter, each step of FIG. 13 is demonstrated in order.
Step S10:
The computer inputs a fringe image as in step S10 of the first embodiment. The intensity distribution g (x, y) of the fringe image is expressed by Expression (1) and Expression (2).
Step S101:
The computer sets the model signal component φ _sm (x, y) to an initial value. Here, design data φ _d (x, y) (known) of the signal component φ _s (x, y) is used as an initial value.

Step S11:
The computer Fourier-transforms the intensity distribution g (x, y) as in step S11 of the first embodiment to obtain a Fourier spectrum h (ω) (FIGS. 5 and 6).

Step S12:
The computer removes the zeroth-order spectrum from the Fourier spectrum h (ω) as in step S11 of the first embodiment.

Step S13:
The computer obtains a complex amplitude distribution g ₁ (x, y) by inverse Fourier transform of the Fourier spectrum h ′ (ω) after the removal, as in Step S13 of the first embodiment (Formula (3)).
Step S14:
As in step S14 of the first embodiment, the computer performs a -1st order spectrum narrowing process on the complex amplitude distribution g ₁ (x, y) (expressions (4) and (5)). The model signal component used in this narrowing process is the model signal component φ _sm (x, y) being set.
Step S15:
The computer obtains a Fourier spectrum h ₂ (ω) by performing a Fourier transform on the complex amplitude distribution g ₂ (x, y) as in step S15 of the first embodiment (FIGS. 7 and 8).

Step S16:
The computer removes the −1st order spectrum from the Fourier spectrum h ₂ (ω) as in step S16 of the first embodiment.

Step S17:
The computer obtains the complex amplitude distribution g ₃ (x, y) by inverse Fourier transform of the Fourier spectrum h ₂ ′ (ω) after the removal, as in step S17 of the first embodiment (formula (6) )).

Step S18:
As in step S18 of the first embodiment, the computer performs a + 1st order spectrum narrowing process on the complex amplitude distribution g ₃ (x, y) (expressions (7) and (8)). The model signal component used in this narrowing process is the model signal component φ _sm (x, y) being set.
Step S19:
The computer obtains a Fourier spectrum h ₄ (ω) by performing a Fourier transform on the complex amplitude distribution g ₄ (x, y) as in step S19 of the first embodiment (FIGS. 9 and 10).

Step S20:
The computer extracts only the + 1st order spectrum from the Fourier spectrum h ₄ (x, y) as in step S20 of the first embodiment.

Step S21:
The computer acquires the complex amplitude distribution g ₅ (x, y) by performing Fourier inverse transform on the extracted + 1st order spectrum, similarly to step S21 of the first embodiment.

Step S22:
The computer performs normalization processing on the complex amplitude distribution g ₅ (x, y) as in step S22 of the first embodiment (equations (9) and (10)). The model signal component used in the normalization process is the model signal component φ _sm (x, y) being set.

Step S23:
The computer calculates the phase distribution φ ′ (x, y) before unwrapping from the complex amplitude distribution g ₆ (x, y), similarly to step S23 of the first embodiment (formula (11)).

Step S24:
The computer acquires the phase distribution φ ″ (x, y) after unwrapping by unwrapping the phase distribution φ ′ (x, y) as in step S24 of the first embodiment, and the phase distribution The calculated signal component φ _s ′ (x, y) is obtained by removing the carrier component φ _c (x, y) (known) from φ ″ (x, y).
Step S102:
The computer determines whether or not the difference between the current calculated signal component φ _s ′ (x, y) and its previous value is equal to or smaller than a threshold value. If the difference is equal to or smaller than the threshold value, the computer calculates the current calculated signal component φ _s ′. The flow is terminated by regarding the value of (x, y) as the final result. On the other hand, when the difference between the two is larger than the threshold, the process proceeds to step S103.
Step S103:
The computer sets the value of the model signal component φ _sm (x, y) to the same value as the current calculated signal component φ _s ′ (x, y). As a result, the model signal component φ _sm (x, y) is updated. Thereafter, the computer returns to step S11.
As described above, the computer of this embodiment repeats the loop of FIG. 13 while updating the model signal component φ _sm (x, y). Then, the value of the model signal component φ _sm (x, y) used in each loop is set to the same value as the calculated signal component φ _s ′ (x, y) acquired in the previous loop. Then, when the calculated signal component φ _s ′ (x, y) converges, the repetition is finished.
Here, the calculated signal component φ _s ′ (x, y) acquired in the first loop is an initial value of the model signal component φ _sm (x, y) (here, the signal component φ _s (x, y)). It is considered that the design data φ _d (x, y)) is closer to the signal component φ _s (x, y). Therefore, the model signal component φ _sm (x, y) used in the second loop is more signal component φ _s (x, y) than the model signal component φ _sm (x, y) used in the first loop. Approach y).
As described above, the analysis error becomes smaller as the model signal component φ _sm (x, y) is closer to the signal component φ _s (x, y). Therefore, the second loop analysis error is the first loop analysis error. Should be smaller than.
Therefore, by repeating the loop of FIG. 13, the calculated signal component φ _s ′ (x, y) can be gradually brought closer to the signal component φ _s (x, y). Therefore, the analysis process of the present embodiment can measure the shape of the measurement target surface 7a with higher accuracy than the analysis process of the first embodiment.

In the analysis processing of the present embodiment, the end timing of loop iteration is the time when the calculated signal component φ _s ′ (x, y) has converged, but it may also be the time when the loop iteration count reaches a predetermined number. Good.
In the analysis processing of the present embodiment, the design data φ _d (x, y) of the signal component φ _s (x, y) is used as the initial value of the model signal component φ _sm (x, y). Other data similar to φ _s (x, y) may be used.
For example, a calculated signal component acquired from the same fringe image by other analysis processing (for example, a conventional Fourier transform method) may be used as the initial value of the model signal component φ _sm (x, y).
In addition, a calculated signal component obtained by measuring the same measurement target surface 7a with another shape measuring device (for example, a coordinate measuring instrument) is used as an initial value of the model signal component φ _sm (x, y). Also good.

Alternatively, the calculated signal component may be fitted with a polynomial such as a power series or Zernike polynomial, and the calculated signal component after the fitting may be used as an initial value of the model signal component φ _sm (x, y). Furthermore, the model signal component φ _{sm (x,} y) in each round of the loop even when updating, fitting the calculated signal components by a polynomial, a model signal component after the update calculation signal component after fitting phi _{sm (x,} y).

[Supplement common to the first embodiment and the second embodiment]
In any of the above-described embodiments, the extraction target of the analysis process is the signal component (excluding the carrier component), so the model signal component is used to calculate the narrowing coefficient. When the phase distribution (including the carrier component) is used, the model phase distribution may be used to calculate the narrowing coefficient. In that case, there is an advantage that the carrier component does not need to be strictly known. However, in that case, the peak position (frequency) of the narrowed spectrum shifts to zero, so it is necessary to shift the passband of the bandpass filter processing (the third embodiment and the later described). See 4 embodiments).

  In any of the above-described embodiments, the execution timing of the normalization process is before unwrapping, but it may be after unwrapping (see third and fourth embodiments described later).

[Third Embodiment]
The third embodiment of the present invention will be described below. This embodiment is also an embodiment of an interferometer device. The main difference from the second embodiment is that the extraction target of the analysis process is the phase distribution (including the carrier component), the model phase distribution is used for calculating the narrowing coefficient, and the execution timing of the normalization process is undefined. The point after wrapping.

  FIG. 14 is a flowchart of the analysis processing of this embodiment. Hereinafter, each step of FIG. 14 is demonstrated in order.

Step S10:
The computer inputs a striped image and performs preprocessing on the striped image. The preprocessing includes, for example, processing for calculating the average luminance of the effective area of the stripe image, subtracting the average luminance from the effective area, and processing for replacing the luminance of the non-effective area of the stripe image with zero. In the present embodiment, the two-dimensional image detector 9 described above is assumed, and the stripe image after the preprocessing is assumed to be the stripe image shown in FIG. The spatial frequency of the fringe image covers a wide band, and the portion with a fine fringe pitch is fine enough to cause aliasing. Such a fringe image has been difficult to analyze by the conventional Fourier transform method.

Step S101 ′:
The computer sets the model phase distribution φ _m (x, y) to an initial value. Design data of the phase distribution φ (x, y) is used as the initial value of the model phase distribution φ _m (x, y).

Step S11:
The computer obtains a Fourier spectrum h (ω) by performing a Fourier transform on the intensity distribution g (x, y) of the stripe image after the preprocessing.

Step S12:
The computer performs a removal process for removing the zeroth-order spectrum from the Fourier spectrum h (ω). The removal process is a bandpass filter process in which the intensity of a predetermined band centered at ω = 0 is set to 0 and the intensity of other bands is maintained. The Fourier spectrum h ′ (ω) after the removal process is as shown in FIG. As described above, since the spatial frequency of the fringe image of the present embodiment covers a wide band, the ± first-order spectra in the Fourier spectrum h ′ (ω) overlap each other, which is why the conventional Fourier transform method is used. It was difficult to separate.

Step S13:
The computer obtains a complex amplitude distribution g ₁ (x, y) by performing Fourier inverse transform on the Fourier spectrum h ′ (ω) after removal.

Step S14:
The computer subjects the complex amplitude distribution g ₁ (x, y) to a narrowing process of an unnecessary spectrum (hereinafter referred to as a −1st-order spectrum). In this narrowing processing, the complex amplitude distribution g ₁ (x, y) is multiplied by the narrowing coefficient α (x, y) represented by Expression (12).

Step S15:
The computer obtains a Fourier spectrum h ₂ (ω) by Fourier transforming the complex amplitude distribution g ₂ (x, y) after narrowing (see FIG. 17). In the above-described narrowing, since a model including a carrier component (here, model phase distribution φ _m (x, y)) is used, the narrowed spectrum in the Fourier spectrum h ₂ (ω). The peak position (frequency) of (−1st order spectrum) is shifted to the center (zero).

Step S16:
The computer removes an unnecessary spectrum (−1st order spectrum) located at the center of the Fourier spectrum h ₂ (ω). This removal process is a bandpass filter process in which the intensity of a predetermined band centered at ω = 0 is set to 0 and the intensity of other bands is maintained. Since the unnecessary spectrum (the above-described −1st order spectrum) is narrowed, the band-pass filter process can prevent missing of necessary information while reliably removing unnecessary information.

Step S17:
The computer obtains a complex amplitude distribution g ₃ (x, y) by performing Fourier inverse transform on the Fourier spectrum h ₂ ′ (ω) after removal.

Step S18:
The computer subjects the complex amplitude distribution g ₃ (x, y) to a necessary spectrum (+ 1st order spectrum) narrowing process. In this narrowing processing, the complex amplitude distribution g ₃ (x, y) is multiplied by the narrowing coefficient β (x, y) represented by Expression (13).

Step S19:
The computer obtains a Fourier spectrum h ₄ (ω) by performing a Fourier transform on the complex amplitude distribution g ₄ (x, y) after narrowing (see FIG. 18). In the above-described narrowing, since a model including a carrier component (here, model phase distribution φ _m (x, y)) is used, the narrowed spectrum in the Fourier spectrum h ₄ (ω) is used. The peak position (frequency) of (+ 1st order spectrum) is shifted to the center (zero).

Step S20:
The computer extracts a necessary spectrum (+ 1st order spectrum) located at the center of the Fourier spectrum h ₄ (x, y). The extraction process is a band-pass filter process that maintains the intensity of a predetermined band centered on ω = 0 and sets the intensity of other bands to 0. Since the necessary spectrum (+ 1st order spectrum) is narrowed, this bandpass filter process can prevent missing of necessary information while reliably removing unnecessary information.

Step S21: A complex amplitude distribution g ₅ (x, y) is obtained by performing inverse Fourier transform on the extracted + first order spectrum.

Step S23:
The computer calculates the phase distribution φ ′ (x, y) before unwrapping by fitting the complex amplitude distribution g ₅ (x, y) to the equation (14).

Step S24 ':
The computer unwraps the calculated phase distribution φ ′ (x, y), and obtains the unwrapped phase distribution φ ″ (x, y).

Step S241:
The computer performs a normalization process on the unwrapped phase distribution φ ″ (x, y). In this normalization process, the value τ (x, y) expressed by Equation (15) is converted into the phase distribution φ. "It is a process of adding to (x, y). Hereinafter, the normalized phase distribution φ ″ (x, y) is referred to as “calculated phase distribution φ ″ (x, y)”.

  Step S102: It is determined whether or not the difference between the current calculated phase distribution φ ″ (x, y) and the previous value is equal to or smaller than a threshold value. If the difference is equal to or smaller than the threshold value, the current calculated phase distribution φ ″ (x , Y) is regarded as the final result, and the process proceeds to step S242. On the other hand, when the difference between the two is larger than the threshold, the process proceeds to step S103.

Step S103:
The computer sets the value of the model phase distribution φ _m (x, y) to the same value as the current calculated phase distribution φ ″ (x, y). Thus, the model phase distribution φ _m (x, y) Thereafter, the computer returns to step S11, and thus the loop of FIG. 14 is repeated until the calculated phase distribution φ ″ (x, y) converges. Here, it is assumed that the number of loop iterations is three. Incidentally, the Fourier spectrum h ₂ (ω) shown in FIG. 17 and the Fourier spectrum h ₄ (ω) shown in FIG. 18 are both acquired in the third loop.

Step S242:
The computer calculates the value of the signal component φ _s (x, y) by subtracting the carrier component φ _c (x, y) (known) from the current calculated phase distribution φ ″ (x, y), and ends the flow. To do.

  FIG. 19 shows the analysis error of this embodiment. This analysis error is calculated by simulation.

  In this simulation, the Z9 term (third order spherical aberration) of the Zernike polynomial is assumed to be the design data of the signal component obtained by superimposing the aspherical shape of 15.0λP-V, and the Z9 term (third order of the Zernike polynomial). Spherical aberration of 16.5λP-V is assumed to be the true value of the signal component. As a result, the RMS value of the analysis error was suppressed to 3.1 mλ (where λ is the wavelength).

  As described above, according to the analysis processing of the present embodiment, it is possible to analyze a fringe image having a spatial frequency over a wide band with a sufficiently low analysis error. Therefore, according to the combination of the analysis processing of this embodiment and the two-dimensional image detector 9 in which the aperture ratio of the pixels is limited, it is possible to measure a measurement target surface having a complicated shape with high accuracy. In addition, the same effect as this embodiment can be acquired also by the analysis process of 2nd Embodiment.

[Fourth Embodiment]
The fourth embodiment of the present invention will be described below. This embodiment is a modification of the analysis processing of the third embodiment. The difference from the analysis process of the third embodiment is that an analysis error estimation process (steps S201 to S204) and an analysis error removal process (step S205) are added to the analysis process of the third embodiment. In the point.

  FIG. 20 is a flowchart of the analysis processing (analysis processing with error correction) of this embodiment. Hereafter, each step of FIG. 20 is demonstrated in order.

Step S100:
The computer acquires the calculated phase distribution φ ″ (x, y) by executing all the steps of the analysis process (FIG. 14) of the third embodiment except for the carrier component removal process (step S242). Here, it is assumed that the number of loop iterations is 3. Hereinafter, the processing of this step is referred to as “main analysis processing”.

Step S201:
The computer creates a model phase distribution φ _m ′ (x, y). This model phase distribution φ _m ′ (x, y) should be used to create a model fringe image. The model phase distribution φ _m ′ (x, y) may be the same as the initial value of the phase distribution φ _m (x, y) to be used for calculating the narrowing coefficient, or may be different. .

For example, the computer in this step fits the calculated phase distribution φ ″ (x, y) with a predetermined polynomial (a polynomial such as a power series or a Zernike polynomial of a predetermined order) representing a two-dimensional shape, thereby obtaining a model phase distribution. φ _m ′ (x, y) is created, and the waveform of the model phase distribution φ _m ′ (x, y) created in this way is smoother than the waveform of the calculated phase distribution φ ″ (x, y). is there.

Alternatively, the computer in this step creates the model phase distribution φ _m ′ (x, y) by smoothing the calculated phase distribution φ ″ (x, y). The model phase distribution thus created The waveform of φ _m ′ (x, y) is smoother than the waveform of the calculated phase distribution φ ″ (x, y).

Step S202:
The computer applies the value of the model phase distribution φ _m ′ (x, y) to φ (x, y) in equation (1), and a (x, y), b (in equation (1). A predetermined value is applied to each of x, y). Here, a (x, y) = 1 and b (x, y) = 1. Then, the computer calculates the value of g (x, y) by calculating the right side in the equation (1), and sets the value as the intensity distribution of the model fringe image. As a result, a model fringe image is created.

Step S203:
The computer obtains the calculated phase distribution φ ″ (x, y) by performing the same process as the main analysis process (step S100) on the created model fringe image. In order to distinguish the calculated phase distribution φ ″ (x, y) acquired in the test analysis process from the calculated phase distribution φ ″ (x, y) acquired in the present analysis process (step S100), This is referred to as “calculated phase distribution φ ′ ″ (x, y)”.

In order to correctly estimate the analysis error, the initial value of the model phase distribution φ _m (x, y), the passband range of the bandpass filter, the loop of the loop between this test analysis process and the main analysis process (step S100). The number of repetitions is also common. However, instead of making the number of loop iterations common, the iteration end condition (threshold in step S102) may be made common.

Step S204:
The computer subtracts the model phase distribution φ _m ′ (x, y) from the calculated phase distribution φ ′ ″ (x, y) acquired in the test analysis process (step S203), thereby performing the test analysis process (step S203). An analysis error Δφ (x, y) is obtained. This analysis error Δφ (x, y) is an estimated value of the analysis error in the present analysis process (step S100). Hereinafter, this analysis error Δφ (x, y) is referred to as “estimation analysis error Δφ (x, y)”.

Step S205:
The computer subtracts the estimated analysis error Δφ (x, y) from the calculated phase distribution φ ″ (x, y) obtained in the present analysis process (step S100), thereby calculating the calculated phase distribution φ ″ (x, y after error correction). y) is obtained.

Step S206:
The computer calculates the value of the signal component φ _s (x, y) by subtracting the carrier component φ _c (x, y) (known) from the calculated phase distribution φ ″ (x, y) after error correction, and the flow Exit.

  FIG. 21 shows analysis errors in the analysis processing (analysis processing with error correction) of the present embodiment. This analysis error is calculated by simulation.

  In this simulation, as in the simulation of the third embodiment, it is assumed that the Z9 term (third-order spherical aberration) of the Zernike polynomial is superimposed with an aspherical shape of 15.0λP-V as design data of the signal component, In addition, it is assumed that the true value of the signal component is obtained by superimposing an aspherical shape of 16.5λP-V on the Z9 term (third-order spherical aberration) of the Zernike polynomial. As a result, the RMS value of the analysis error of the analysis process with error correction was 1.6 mλ (where λ is the wavelength).

  When this analysis error is compared with the analysis error of the third embodiment (FIG. 19), it can be seen that the error correction of the present embodiment has a remarkable effect.

In the present embodiment, the model phase distribution φ _m ′ (x, y) to be used for creating the model fringe image is created based on the calculated phase distribution φ ″ (x, y), but the phase distribution φ (x , Y) other data similar to may be used.

For example, a calculated phase distribution acquired from the same fringe image by another analysis process (for example, a conventional Fourier transform method) may be used for the model phase distribution φ _m ′ (x, y).

In addition, a calculated phase distribution obtained by measuring the same measurement target surface 7a with another shape measuring device (for example, a coordinate measuring instrument) may be used for the model phase distribution φ _m ′ (x, y). .

[Supplementary to the third embodiment and the fourth embodiment]
In the analysis process of the third embodiment, the main analysis process, and the test analysis process of the fourth embodiment, the loop repetition end timing is set to the time when the calculated phase distribution φ ″ (x, y) converges. It may be the time when the number of loop repetitions reaches a predetermined number.

In the analysis process of the third embodiment, the main analysis process and the test analysis process of the fourth embodiment, the phase distribution is set to the initial value of the model phase distribution φ _m (x, y) to be used for calculating the narrowing coefficient. However, other data similar to the phase distribution φ (x, y) may be used.

For example, a calculated phase distribution acquired from the same fringe image by another analysis process (for example, a conventional Fourier transform method) may be used as the initial value of the model phase distribution φ _m (x, y).

In addition, a calculated phase distribution obtained by measuring the same measurement target surface 7a with another shape measuring device (for example, a coordinate measuring device) is used as an initial value of the model phase distribution φ _m (x, y). Also good.

Alternatively, the calculated phase distribution may be fitted with a polynomial such as a power series or Zernike polynomial, and the calculated phase distribution after the fitting may be used as the initial value of the model phase distribution φ _m (x, y). Furthermore, even when the model phase distribution φ _m (x, y) is updated in each loop, the calculated phase distribution may be fitted with a polynomial, and the calculated phase distribution after fitting may be used as the updated model phase distribution.

[Supplement common to the first to fourth embodiments]
In addition, although 4th Embodiment was a modification of 3rd Embodiment, you may deform | transform 2nd Embodiment similarly to it.

Further, the interferometer device of any of the above-described embodiments measures the shape of the measurement target surface 7a (that is, the spatial distribution of the height), but the height of an arbitrary point on the measurement target surface 7a. You may measure a time change.
In that case, instead of generating carrier fringes (that is, spatial carriers), a time change waveform of a pixel value corresponding to that point may be measured while generating a time carrier, and the time change waveform may be set as an analysis target.
If this measurement is performed for each pixel, the shape change of the measurement target surface 7a (including the shape change due to the movement of the measurement target surface 7a) can be measured. Such measurement is suitable for measuring an element having a variable surface shape, such as a micromirror array.
In order to generate a time carrier by the interferometer device, the Fizeau plate 5 or the measurement object 7 may be displaced in the optical axis direction by a piezo element or the like. Incidentally, when no spatial carrier is generated, it is not necessary to incline the Fizeau surface 5a.
Moreover, in the process of step S11-S20 of any embodiment mentioned above, although the spectrum finally extracted was made into the + 1st order spectrum, it is good also as a -1st order spectrum. In this case, the band narrowing target in step S14 is the + 1st order spectrum, the removal target in step S16 is the + 1st order spectrum, the band narrowing target in step S18 is the −1st order spectrum, and the extraction target in step S20. Becomes the −1st order spectrum. In addition, the sign of the imaginary number “i” in the narrowing coefficient (equations (4), (7), (12), (13)) is inverted.
In addition, the analysis processing of any of the above-described embodiments is configured by two procedures of “removing an unnecessary spectrum after narrowing and then extracting a necessary spectrum after narrowing”. , It may be constituted by one procedure of “extracting a necessary spectrum after narrowing it”. However, the former can reduce the analysis error.
Further, although the Fizeau interferometer is applied to the interferometer apparatus of any of the above-described embodiments, other types of interferometers such as a Twiman Green type may be applied. Incidentally, in order to generate a spatial carrier in a Twiman Green interferometer, the reference plane may be inclined, and in order to generate a time carrier, the reference plane may be moved in the optical axis direction.

[Fifth Embodiment]
Hereinafter, a fifth embodiment of the present invention will be described. The present embodiment is an embodiment of a pattern projection shape measuring apparatus.
FIG. 22 is a schematic configuration diagram of the pattern projection shape measuring apparatus of the present embodiment. As shown in FIG. 22, the pattern projection shape measuring apparatus includes a stage 12 that supports the measurement object 11, a projection unit 13, and an imaging unit 14. The pattern projection shape measuring apparatus also includes a computer (not shown). The surface 11a of the measurement object 11 is a measurement object surface.
The projection unit 13 includes a light source 21, an illumination optical system 22, a pattern formation unit 23, and a projection optical system 24, and illuminates the measurement target surface 11a on the stage 12 from an oblique direction.
Among these, the pattern forming unit 23 modulates the intensity of the light beam directed toward the measurement target surface 11a in a sine wave shape at a predetermined spatial frequency in the spatial direction. As a result, a stripe-shaped carrier stripe is projected onto the measurement target surface 11a. The carrier stripe is distorted according to the shape of the measurement target surface 11a.
The imaging unit 14 includes an imaging optical system 25 and an imaging element (two-dimensional image detector) 26, and images a stripe appearing on the measurement target surface 11a from the front. The fringe image (the fringe image) acquired by the imaging unit 14 through imaging is input to a computer (not shown).

  On the light incident side of the image sensor 26, means (a diaphragm array or a mask pattern corresponding to the diaphragm array) for limiting the aperture ratio of each pixel is provided. The opening size of each pixel is set to about 1/8 of the pixel pitch, for example. As such an image pickup element 26, for example, the same CCD as that mounted on an interferometer apparatus (US Pat. No. 4,791,584) based on sub-Nyquist interferometry can be employed. Such an image pickup device 26 can detect a fringe image having a finer stripe pitch than a normal image pickup device having a common pixel pitch and an aperture ratio that is not limited at all.

The computer performs an analysis process on the input fringe image. This analysis process is the same as the analysis process of any of the embodiments described above. If the signal component of the fringe phase distribution is calculated by this analysis processing, the shape of the measurement target surface 11a becomes known. Note that the analysis processing program is preinstalled in the computer.
Therefore, the pattern projection shape measuring apparatus of this embodiment can measure the measurement target surface 11a having a complicated shape with high accuracy.
Note that the pattern projection shape measurement apparatus of the present embodiment analyzes the waveform modulated in the spatial direction and measures the shape (space distribution of the height) of the measurement target surface 11a. The shape of the measurement target surface 11a may be measured after the direction is also modulated.

  Incidentally, in order to generate both a spatial carrier and a time carrier in the pattern projection shape measuring apparatus, the projection position of the carrier fringes by the projection unit 13 may be scanned in the pitch direction of the fringes.

  DESCRIPTION OF SYMBOLS 1 ... Laser light source, 2 ... Beam expander, 3 ... Polarizing beam splitter, 4 ... 1/4 wavelength plate, 5 ... Fizeau plate, 6 ... Wavefront conversion lens, 8 * ..Beam diameter conversion optical system, 9 ... Two-dimensional image detector 9

Claims (17)

A waveform analysis device for extracting phase information of the input waveform by performing a waveform analysis process by a Fourier transform method on an input waveform on which a carrier is superimposed,
After narrowing at least one of the + 1st order spectrum and the −1st order spectrum from the Fourier spectrum of the input waveform, the spectra are separated, and the phase information of the input waveform is calculated based on one of the separated spectra. Analysis means to
Estimating means for estimating an analysis error generated by the analyzing means when performing waveform analysis of the input waveform by performing the same waveform analysis as the analyzing means on the input waveform model;
Correction means for correcting the phase information calculated by the analysis means based on the analysis error estimated by the estimation means;
A waveform analysis apparatus comprising: The waveform analysis apparatus according to claim 1,
The analysis means includes
Removing means for removing a zero-order spectrum from the Fourier spectrum of the input waveform;
Narrowing means for narrowing the first-order spectrum of the Fourier spectrum of the input waveform;
Removing means for removing a narrowed first-order spectrum from the Fourier spectrum of the input waveform;
Narrowing means for narrowing the first order spectrum of the Fourier spectrum of the input waveform;
And a extracting means for extracting a narrowed first-order spectrum from the Fourier spectrum of the input waveform. The waveform analysis apparatus according to claim 1,
The analysis means includes
Removing means for removing a zero-order spectrum from the Fourier spectrum of the input waveform;
Narrowing means for narrowing the first order spectrum of the Fourier spectrum of the input waveform;
Removing means for removing a narrowed + 1st order spectrum from the Fourier spectrum of the input waveform;
Narrowing means for narrowing the first-order spectrum of the Fourier spectrum of the input waveform;
And a extracting unit for extracting a narrowed first-order spectrum from the Fourier spectrum of the input waveform. In the waveform analysis apparatus according to any one of claims 1 to 3,
The analysis means includes
The waveform analysis apparatus characterized in that the narrowing is performed by multiplying the complex amplitude distribution of the input waveform by a narrowing coefficient derived from the phase information model. The waveform analysis apparatus according to claim 4,
If the phase information model is φm,
The narrowing coefficient for narrowing the + 1st order spectrum is set to exp [−iφm] or a value corresponding thereto,
The waveform analysis apparatus, wherein the narrowing coefficient for narrowing the negative spectrum is set to exp [+ iφm] or a value corresponding thereto. In the waveform analysis apparatus according to claim 4 or 5,
The model of the phase information is
A waveform analysis apparatus characterized by being a design value of the phase information. In the waveform analysis apparatus according to claim 4 or 5,
The model of the phase information is
A waveform analysis apparatus characterized by being an actual measurement value of the phase information. The waveform analysis apparatus according to claim 7,
The measured value is
The waveform analysis apparatus is phase information calculated from the input waveform by another waveform analysis. In the waveform analysis device according to claim 7 or 8,
The model of the phase information is
A waveform analysis apparatus characterized by fitting an actual measurement value of the phase information to a predetermined function. In the waveform analysis apparatus according to any one of claims 4 to 9,
The analysis means includes
Repetitive means for repeating the calculation of the phase information until the calculation result of the phase information converges while changing the model of the phase information,
In the second and subsequent extractions,
A waveform analysis apparatus, wherein a previous calculation result is used as a model of the phase information. In the waveform analysis apparatus according to any one of claims 1 to 10,
The input waveform is
A waveform analyzer characterized by being a waveform modulated in a spatial direction. In the waveform analysis apparatus according to any one of claims 1 to 10,
The input waveform is
A waveform analyzer characterized by a waveform modulated in the time direction. A one-dimensional or two-dimensional pixel array that records superimposed stripes of carriers;
The waveform analyzer according to claim 11, wherein the fringes recorded by the pixel array are processed as the input waveform.
Each of the pixels included in the pixel array is provided with a limiting unit that limits the aperture ratio of the pixel. A waveform analysis program for extracting phase information of the input waveform by performing a waveform analysis process by a Fourier transform method on an input waveform on which a carrier is superimposed,
Analysis that narrows at least one of the + 1st order spectrum and the −1st order spectrum from the Fourier spectrum of the input waveform and then separates the spectra, and calculates phase information of the input waveform based on one of the separated spectra Procedure and
By performing the same waveform analysis as the analysis procedure on the model of the input waveform, an estimation procedure for estimating an analysis error generated by the analysis procedure when the input waveform is analyzed,
A correction procedure for correcting the phase information calculated in the analysis procedure according to the analysis error estimated in the estimation procedure,
A waveform analysis program comprising: An interferometer device comprising the waveform analysis device according to any one of claims 1 to 13. A pattern projection shape measuring apparatus comprising: the waveform analyzing apparatus according to claim 1. A waveform analysis method for extracting phase information of the input waveform by performing a waveform analysis process by a Fourier transform method on an input waveform on which a carrier is superimposed,
Analysis that narrows at least one of the + 1st order spectrum and the −1st order spectrum from the Fourier spectrum of the input waveform and then separates the spectra, and calculates phase information of the input waveform based on one of the separated spectra Procedure and
By performing the same waveform analysis as the analysis procedure on the model of the input waveform, an estimation procedure for estimating an analysis error generated by the analysis procedure when the input waveform is analyzed,
A correction procedure for correcting the phase information calculated in the analysis procedure according to the analysis error estimated in the estimation procedure,
The waveform analysis method characterized by including.