Inspection system, image processing method, and defect inspection device

Abstract

To provide an inspection system, an image processing method and a defect inspection device which can effectively suppress noise even when measuring a periodical change in a state.SOLUTION: An inspection system 100 includes a control unit 4 which suppresses noise in a complex number image D showing each pixel with a complex number indicating a periodical change in a vibration state of an inspection object P. The control unit 4 acquires a similarity between a pixel included in an object image region z(xi) defined in the complex number image D and a pixel included in a plurality of reference image regions z(xj) defined in the complex number image D separately from the object image region z(xi) by comparing the complex numbers indicating the respective pixels with each other, and executes noise suppression processing of the object image region z(xi) by using weight based on the acquired similarity.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to an inspection system, an image processing method, and a defect inspection apparatus. [Background technology]

[0002] BACKGROUND ART Conventionally, there is known a defect inspection device that measures the vibration state of an inspection object (see, for example, Patent Document 1).

[0003] The defect inspection device described in Patent Document 1 above includes an excitation unit that excites elastic waves in the object to be inspected, an illumination unit that irradiates a measurement area on the surface of the object to be inspected with strobe lighting, and a displacement measurement unit. The displacement measurement unit is configured to measure the longitudinal displacement of each point in the measurement area at at least three different phases of the elastic wave by controlling the phase of the elastic wave and the timing of the strobe lighting. This defect inspection device then measures the longitudinal displacement of each point in the measurement area, thereby measuring the vibration state (amplitude and phase) of each point in the measurement area. Patent Document 1 above also discloses a configuration that generates an image based on the measured vibration state (amplitude and phase) of each point in the measurement area. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Japanese Patent Application Laid-Open No. 2017-219318 Summary of the Invention [Problem to be solved by the invention]

[0005] Although not described in Patent Document 1, when an image is generated by measuring a measurement area, the generated image may contain noise. For example, when an image is generated by measuring the vibration state using laser interferometry, as in the defect inspection device described in Patent Document 1, noise occurs in the generated image due to thermal noise caused by the irregular movement of electrons in a conductor and vibration of the device. In such cases, a noise suppression process is generally performed on the generated image to suppress the noise contained in the image. However, when an image is generated by measuring the periodically changing vibration state of an inspected object (inspection target) in which elastic waves are excited, as in the defect inspection device described in Patent Document 1, the noise state of the generated image may change depending on the periodically changing vibration state. In such cases, when noise suppression is performed on the generated image, it is possible that the noise that changes depending on the vibration state is not sufficiently suppressed. Therefore, it is desirable to effectively suppress noise even when measuring periodic state changes.

[0006] The present invention has been made to solve the above-mentioned problems, and one object of the present invention is to provide an inspection system, an image processing method, and a defect inspection device that can effectively suppress noise even when measuring periodic state changes. [Means for solving the problem]

[0007] In order to achieve the above object, an inspection system according to a first aspect of the present invention includes an excitation unit that excites elastic waves in an object to be inspected, a measurement unit that measures a vibration state of the object to be inspected in which elastic waves have been excited by the excitation unit, and an image processing unit that is configured to suppress noise in a complex image in which each pixel is represented by a complex number that indicates a periodic change in the vibration state of the object to be inspected, based on the measurement results of the measurement unit, wherein the image processing unit obtains similarities between pixels included in an object image area defined in the complex image and pixels included in a plurality of reference image areas defined in the complex image separately from the object image area by comparing the complex numbers that represent each pixel, and performs noise suppression processing on the object image area using weights based on the obtained similarities. The image processing unit is configured to approximate a noise distribution of signal points on a complex plane of complex numbers representing each pixel in the target image region so that the noise distribution is distributed according to a two-dimensional normal distribution that spreads unevenly on the complex plane, and to obtain a similarity based on the approximated distribution. . An inspection system according to a second aspect of the present invention includes an excitation unit that excites elastic waves in an object to be inspected, a measurement unit that measures a vibration state of the object to be inspected in which elastic waves have been excited by the excitation unit, and an image processing unit that is configured to suppress noise in a complex image in which each pixel is represented by a complex number that indicates a periodic change in the vibration state of the object to be inspected based on the measurement results of the measurement unit, wherein the image processing unit obtains a similarity between a pixel included in an object image area defined in the complex image and a pixel included in a plurality of reference image areas defined in the complex image separately from the object image area by comparing the complex numbers that represent each pixel, and performs noise suppression processing on the object image area using a weight based on the obtained similarity, and the image processing unit performs a noise suppression processing on the object image area by comparing the complex numbers that represent each pixel with each other. The noise distribution of signal points on a complex plane is approximated to be distributed according to a two-dimensional normal distribution that spreads biasedly from an origin toward the signal points on a complex plane, depending on the magnitude of the amplitude of the periodic change in the vibration state of the test object, and a similarity is obtained based on the approximated distribution. The apparatus further includes an operation unit that accepts an operation to change a parameter that indicates the degree of spread of the distribution when approximating the noise distribution. The image processing unit is configured to store a comparison result obtained by comparing complex numbers that represent pixels included in each of the target image region and a plurality of reference image regions, and, when an operation to change the parameter that indicates the degree of spread of the distribution is accepted, to obtain a similarity using the changed parameter that indicates the degree of spread of the distribution and the stored comparison result. An inspection system according to a third aspect of the present invention comprises an excitation unit that excites elastic waves in an object to be inspected, a measurement unit that measures a vibration state of the object to be inspected in which elastic waves have been excited by the excitation unit, and an image processing unit that is configured to suppress noise in a complex image in which each pixel is represented by a complex number that indicates a periodic change in the vibration state of the object to be inspected based on the measurement results of the measurement unit, wherein the image processing unit obtains a similarity between a pixel included in a target image area defined in the complex image and a pixel included in a plurality of reference image areas defined in the complex image separately from the target image area by comparing the complex numbers that represent each pixel, and The image processing unit is configured to perform noise suppression processing of the target image region using weights based on the acquired similarities, and the image processing unit is configured to perform non-local means processing as noise suppression processing to suppress noise in the target image region by averaging using weights based on the acquired similarities for multiple reference image regions, and the image processing unit is configured to perform non-local means processing as noise suppression processing to suppress noise in the target image region by removing low-frequency components from the multiple reference image regions and averaging using weights based on the acquired similarities for the multiple reference image regions from which the low-frequency components have been removed.

[0008] This invention Fourth The image processing method in this aspect includes the steps of: acquiring a complex number image in which each pixel is represented by a complex number indicating a periodic change in state; acquiring a target image region defined in the acquired complex number image and a plurality of reference image regions defined in the complex number image separately from the target image region; acquiring similarities between pixels included in the acquired target image region and pixels included in the plurality of reference image regions by comparing the complex numbers representing each pixel; and performing noise suppression processing on the target image region using weights based on the acquired similarities. The step of acquiring the similarity includes a step of approximating a noise distribution of signal points on a complex plane of complex numbers representing each pixel of the target image region so that the noise distribution is distributed according to a two-dimensional normal distribution that spreads unevenly on the complex plane, and acquiring the similarity based on the approximated distribution. .

[0009] This invention No. 5The defect inspection device in this aspect includes an excitation unit that excites elastic waves in an object to be inspected, an irradiation unit that irradiates laser light onto the object to be inspected in which elastic waves have been excited by the excitation unit, a measurement unit that causes reflected laser light, which is laser light irradiated by the irradiation unit and reflected from the object to interfere with reference laser light, which is laser light irradiated by the irradiation unit, and captures and measures the interference light resulting from the interference of the reflected laser light and the reference laser light, thereby measuring the vibration state of the object to be inspected in which elastic waves have been excited by the excitation unit, and a control unit that is configured to suppress noise in a complex image for defect inspection, in which each pixel is represented by a complex number that indicates a periodic change in the vibration state of the object to be inspected, based on the measurement result of the measurement unit, wherein the control unit obtains similarities between pixels included in an object image area defined in the complex image and pixels included in a plurality of reference image areas defined in the complex image separately from the object image area by comparing the complex numbers that represent each pixel, and performs noise suppression processing on the object image area using weights based on the obtained similarities. The control unit is configured to approximate a noise distribution of signal points on a complex plane of complex numbers representing each pixel in the target image region so that the noise distribution is distributed according to a two-dimensional normal distribution that spreads unevenly on the complex plane, and to obtain a similarity based on the approximated distribution. . [Effects of the Invention]

[0010] the above 1st~3rd Inspection system in the above aspect, Fourth The image processing method according to the aspect of the present invention, No. 5In this defect inspection device, similarities between pixels included in a target image region defined in a complex image and pixels included in multiple reference image regions defined in the complex image separately from the target image region are obtained by comparing the complex numbers representing the respective pixels. Then, noise suppression processing is performed on the target image region using weights based on the obtained similarities. Thus, by performing noise suppression processing on the target image region defined in the complex image, noise in the complex image can be suppressed. Therefore, since the complex image, in which each pixel is represented by a complex number indicating a periodic state change, reflects the periodic state change of vibration, suppressing noise in the complex image can effectively suppress noise even when the noise state changes according to the state of vibration, unlike when noise suppression processing is performed on an image represented by real values. As a result, noise can be effectively suppressed even when measuring periodic state changes. [Brief explanation of the drawings]

[0011] [Figure 1] FIG. 1 is a diagram for explaining the configuration of an inspection system according to a first embodiment. [Figure 2] FIG. 3 is a schematic diagram showing an example of a vibration state image according to the first embodiment. [Figure 3] FIG. 4 is a diagram for explaining the timing of measurement according to the first embodiment. [Figure 4] 10A and 10B are diagrams for explaining measurement of interference light and acquisition and correction of a complex image. [Figure 5] 10A and 10B are diagrams for explaining extraction of a target image region and a reference image region from a complex image. [Figure 6] FIG. 1 is a diagram showing noise distribution on a complex plane. [Figure 7] FIG. 10 is a diagram for explaining approximation of noise distribution of signal points. [Figure 8] FIG. 2 is a schematic diagram showing a display on a display unit according to the first embodiment. [Figure 9]FIG. 2 is a diagram (flowchart) for explaining an image processing method according to the first embodiment. [Figure 10] FIG. 10 is a block diagram showing the configuration of an inspection system according to a second embodiment. [Figure 11] FIG. 10 is a diagram showing a vibration state image in which low frequency components have been removed and noise has been suppressed. [Figure 12] FIG. 10 is a block diagram showing the configuration of an inspection system according to a third embodiment. [Figure 13] 10A and 10B are diagrams for explaining acquisition of the average value of pixels in each of a target image region and a reference image region according to the third embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0012] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, specific embodiments of the present invention will be described with reference to the accompanying drawings.

[0013] [First embodiment] (Overall configuration of the inspection system) An inspection system 100 according to a first embodiment of the present invention will be described with reference to FIGS.

[0014] As shown in Fig. 1, the inspection system 100 according to the first embodiment includes a transducer 1, an irradiation unit 2, a speckle shearing interferometer 3, a control unit 4, a signal generator 5, a display unit 6, and an operation unit 7. The transducer 1 is an example of an "excitation unit" in the claims, and the speckle shearing interferometer 3 is an example of a "measurement unit" in the claims. The control unit 4 is an example of an "image processing unit" in the claims.

[0015] The oscillator 1 and the irradiation unit 2 are connected to a signal generator 5 via cables. In addition, the speckle shearing interferometer 3, the signal generator 5, the display unit 6, and the operation unit 7 are connected to a control unit 4 via cables.

[0016] The vibrator 1 excites elastic waves in the measurement area Pa of the test object P. Specifically, the vibrator 1 is placed so as to be in contact with the test object P, converts an AC signal from the signal generator 5 into mechanical vibration, and generates excitation vibration to excite elastic waves in the measurement area Pa. The frequency of the elastic waves excited in the test object P by the vibrator 1 is equal to the frequency of vibration (excitation vibration) of the vibrator 1. In other words, the frequency of the AC signal input to the vibrator 1 and the frequency of the elastic waves excited in the test object P by the vibrator 1 are equal to each other.

[0017] The irradiation unit 2 irradiates a measurement area Pa of the inspection object P in which elastic waves are excited by the vibrator 1 with laser light. The irradiation unit 2 includes a laser light source (not shown). The laser light irradiated from the laser light source is expanded by an illumination lens 21 and irradiated over the entire measurement area Pa on the surface of the inspection object P. The irradiation unit 2 also irradiates the laser light at a predetermined timing based on an electrical signal from a signal generator 5. That is, the irradiation unit 2 irradiates the inspection object P with laser light in response to the elastic waves generated by the vibrator 1. The laser light source is, for example, a laser diode, and irradiates laser light (near-infrared light) with a wavelength of 785 nm.

[0018] In the first embodiment, the speckle shearing interferometer 3 measures the vibration state in a measurement area Pa of the inspection object P in which elastic waves are excited by the oscillator 1. Specifically, the speckle shearing interferometer 3 causes laser light irradiated by the irradiation unit 2 and reflected at the measurement area Pa of the inspection object P to interfere by laser interferometry, and captures an image of the interfered laser light (interference light). In more detail, the speckle shearing interferometer 3 causes laser light (reflected laser light and reference laser light) reflected at two different points in the measurement area Pa of the inspection object P excited by the oscillator 1 to interfere by laser interferometry. The speckle shearing interferometer 3 also includes a beam splitter 31, a first reflecting mirror 32a, a second reflecting mirror 32b, a condenser lens 33, an image sensor 34, and a phase shifter 35.

[0019] The beam splitter 31 includes a half mirror. The beam splitter 31 is disposed at a position where the laser light reflected from the measurement area Pa of the test object P is incident. The beam splitter 31 reflects the incident laser light toward the first reflecting mirror 32a as shown by the optical path indicated by the straight line L1 in FIG. 1, and transmits the incident laser light toward the second reflecting mirror 32b as shown by the optical path indicated by the dashed line L2 in FIG. 1. The beam splitter 31 transmits the incident laser light reflected by the first reflecting mirror 32a toward the image sensor 34 as shown by the optical path indicated by the straight line L1 in FIG. 1, and reflects the laser light reflected by the second reflecting mirror 32b toward the image sensor 34 as shown by the dashed line L2 in FIG. 1.

[0020] The first reflecting mirror 32a is disposed on the optical path of the laser light reflected by the beam splitter 31 at an angle of 45 degrees with respect to the reflecting surface of the beam splitter 31. The first reflecting mirror 32a reflects the laser light reflected by the beam splitter 31 towards the beam splitter 31.

[0021] The second reflecting mirror 32b is disposed on the optical path of the laser light passing through the beam splitter 31, at an angle slightly tilted from 45 degrees with respect to the reflecting surface of the beam splitter 31. The second reflecting mirror 32b reflects the incident laser light passing through the beam splitter 31 towards the beam splitter 31.

[0022] The condenser lens 33 is disposed between the beam splitter 31 and the image sensor 34, and condenses the laser light that has passed through the beam splitter 31 (the straight line L1 in FIG. 1) and the laser light that has been reflected by the beam splitter 31 (the dashed line L2 in FIG. 1).

[0023] Image sensor 34 captures an image of the interference light, which is the interfered laser light. Image sensor 34 has a large number of detection elements arranged in a plane, and is disposed on the optical paths of the laser light (straight line L1 in FIG. 1) that is reflected by beam splitter 31, then reflected by first reflecting mirror 32a, and transmitted through beam splitter 31, and the laser light (dashed line L2 in FIG. 1) that is transmitted through beam splitter 31, then reflected by second reflecting mirror 32b, and then reflected back by beam splitter 31. Image sensor 34 includes, for example, a CMOS image sensor or a CCD image sensor.

[0024] The phase shifter 35 is disposed between the beam splitter 31 and the first reflecting mirror 32a, and changes (shifts) the phase of the transmitting laser light (straight line L1 in FIG. 1) under the control of the control unit 4. Specifically, the phase shifter 35 is configured to change the optical path length of the transmitting laser light.

[0025] In the speckle shearing interferometer 3, for example, laser light reflected by position Pa1 on the surface of the measurement area Pa and the first reflecting mirror 32a (straight line L1 in FIG. 1) and laser light reflected by position Pa2 on the surface of the measurement area Pa and the second reflecting mirror 32b (dashed line L2 in FIG. 1) interfere with each other and enter the same location on the image sensor 34 (detected by the same element). Positions Pa1 and Pa2 are positions separated from each other by a small distance. The laser light (reflected laser light and reference laser light) reflected from different positions (Pa1 and Pa2) in each region of the measurement area Pa is guided by the speckle shearing interferometer 3 to become interference light, which then enters the image sensor 34.

[0026] The control unit 4 controls each part of the speckle sharing interferometer 3 and the operation of the signal generator 5. The control unit 4 is, for example, a computer including a CPU (Central Processing Unit). The control unit 4 also includes a storage device that stores various parameters, programs, etc. Details of the control by the control unit 4 will be described later.

[0027] The signal generator 5 outputs an AC signal for controlling the vibration of the vibrator 1 and the timing of the laser light irradiation by the irradiation unit 2, based on the control by the control unit 4. The signal generator 5 includes, for example, an FPGA (field-programmable gate array).

[0028] The display unit 6 displays an image generated by the control unit 4. Specifically, the display unit 6 displays a vibration state image E (see FIG. 2), which will be described later, generated by the control unit 4. The display unit 6 includes a liquid crystal display, an organic EL (Electro-Luminescence) display, or the like.

[0029] The operation unit 7 receives input operations from an inspector. The operation unit 7 is, for example, a keyboard and a pointing device such as a mouse. The operation unit 7 also outputs an operation signal to the control unit 4 based on the received input operations.

[0030] The inspection object P is, for example, a painted steel plate with a paint film applied to it. The defects Q are defective areas occurring inside (surface layer / surface) of the measurement area Pa, and include cracks, peeling, etc.

[0031] (Controlled by the control unit) As shown in Fig. 2, the control unit 4 generates a vibration state image E for inspecting a defect Q based on the interference light measured by the speckle shearing interferometer 3. The vibration state image E is an image that allows visual recognition of the vibration state of the elastic waves excited in the measurement area Pa of the generated inspection object P. For example, the vibration state image E allows recognition of the position of a defect Q (defective portion) included in the inspection object P by visually recognizing discontinuities in the vibration state.

[0032] In the first embodiment, the control unit 4 executes processing to suppress noise in the vibration state image E to be generated. Specifically, in the first embodiment, the control unit 4 generates a complex number image D (see FIG. 4) for defect inspection, in which each pixel is represented by a complex number indicating a periodic change in the vibration state of the inspection object P in the measurement area Pa, based on the measurement results of the speckle sharing interferometer 3. The control unit 4 is configured to suppress noise in the vibration state image E by executing processing to suppress noise in the generated complex number image D. Note that FIG. 2 illustrates a vibration state image E in a noisy state in which noise suppression processing has not been executed. Details of the control processing by the control unit 4 will be described below.

[0033] <Measurement of interference light by the control unit> The control unit 4 measures the vibration state of the test object P by irradiating the measurement area Pa of the test object P in which elastic waves are excited with laser light. Specifically, the control unit 4 acquires (captures) an interference light image that measures the vibration state in the measurement area Pa of the test object P, based on the intensity pattern of the interference light captured by each detection element of the image sensor 34, while changing the phase of the laser light using the phase shifter 35. The interference light image is an image in which the speckle pattern of the interference light is captured.

[0034] Specifically, the control unit 4 operates the phase shifter 35 arranged in the speckle sharing interferometer 3 by an actuator (not shown) to change the phase of the transmitted laser light. position The laser light reflected at Pa1 and position The phase difference of the laser light reflected at Pa2 changes. The intensity of the interference light produced by the interference of these two laser lights is detected by a plurality of detection elements of the image sensor .

[0035] As shown in FIGS. 3 and 4, the control unit 4 controls the vibration of the vibrator 1 and the timing of the laser light emitted by the irradiation unit 2 via the signal generator 5, and captures the interference light at each detection element of the image sensor 34 while changing the phase shift amount. The control unit 4 changes the phase shift amount by λ / 4 increments, and captures a total of 37 interference light images, including 32 images for timings t (t=0-7) at which the phases of the elastic waves differ by T / 8 for each phase shift amount (0, λ / 4, λ / 2, 3λ / 4) and five images with the light turned off before and after each phase shift amount (0, λ / 4, λ / 2, 3λ / 4). λ is the wavelength of the laser light. T is the period of the elastic waves excited in the test object P.

[0036] Then, based on the captured interference light image, the control unit 4 calculates the optical phase (phase difference between the two optical paths when the phase shift amount is zero) Φ from the brightness values ​​I0 to I3 of the four images that have the same elastic wave phase timing t (t=0 to 7) and the laser light phase shift amount that differs by λ / 4, using equation (1). t Ask for. Φ t =-arctan{(I3-I1) / (I2-I0)}···(1) The control unit 4 also controls the optical phase Φ t is approximated to a sine wave by the least squares method to find the approximation coefficients A, φ, and C in equation (2). Φ t =Acos(φ+tπ / 4)+C=Bexp(tπ / 4)+C...(2) Here, B is a complex amplitude and is expressed as in equation (3). B = Aexp(iφ): complex amplitude (3) Furthermore, the amount of phase shift and the step of the timing of irradiating the laser light are not limited to these. In this case, the calculation formulas will be different from the above formulas (1) to (3).

[0037] The complex amplitude expressed by the above formula (3) can be expressed as I(x) = a(x) + ib(x) according to Euler's theorem. I(x) is a complex number representing the state of the optical phase (vibration state of the inspection object P) based on the interference light detected by the detection element corresponding to the coordinate x of the measurement area Pa in the image sensor 34. In this way, in the first embodiment, the control unit 4 generates a complex image D based on the measurement results of the interference light captured by the image sensor 34 of the speckle-sharing interferometer 3. The complex image D has multiple pixels arranged in a plane so as to correspond to each detection element of the image sensor 34. In the complex image D, each pixel is represented by the complex number I(x) obtained by the above formulas (1) to (3). This complex number I(x) represents the measured vibration state (the phase and amplitude of the periodically changing optical path difference). The complex number I(x) represents the complex number of the pixel at the coordinate x of the complex image D corresponding to the coordinate of the measurement area Pa.

[0038] <Interference degree correction> 4, the control unit 4 is configured to correct the complex image D in accordance with the degree of interference coh(x). Specifically, in the first embodiment, the control unit 4 is configured to acquire the degree of interference coh(x) of the measured interference light. The control unit 4 is then configured to correct the complex number I(x) representing each pixel of the complex image D by weighting the complex number I(x) representing each pixel in accordance with the acquired degree of interference coh(x). Note that coh(x) indicates the degree of interference at the coordinate x.

[0039] Specifically, the control unit 4 measures the degree of interference coh(x) at each pixel of the complex image D based on the amount of change in the luminance values ​​I0 to I3 of four images that have the same elastic wave phase timing t (t=0 to 7) and laser light phase shift amounts that differ by λ / 4. That is, even when the shift amount is changed by the phase shifter 35, pixels with small changes in the luminance values ​​I0 to I3 have small degrees of interference coh(x). On the other hand, when the shift amount is changed by the phase shifter 35, pixels with large changes in the luminance values ​​I0 to I3 have large degrees of interference coh(x).

[0040] Then, the control unit 4 corrects the degree of interference coh(x) of each pixel in the complex image D using the following equation (4).

number

[0041] <Noise suppression processing> In the first embodiment, the control unit 4 is configured to execute a process of suppressing noise in the complex image D after the correction of the degree of interference coh(x). In the following description, the complex image D and the complex number I(x) indicate values ​​after the correction using the degree of interference coh(x) has been executed.

[0042] As shown in FIG. 5, in the first embodiment, the control unit 4 calculates the target image region z(xi ) and the target image region z(x i ) are defined in the complex image D separately from the reference image region z(x j ) (j: 1 to N). Specifically, the control unit 4 extracts a predetermined area from the complex image D to obtain the target image area z(x i ) is acquired. Then, the control unit 4 acquires the target image region z(x i ), a target image region z(x i ) to obtain multiple (N) reference image regions z(x j ) to get the

[0043] Target image region z(x i ) is the coordinate x i For example, the target image region z(x i ) is the coordinate x i The target image area z(x i ) is the complex number I(x i ) is a set of 49 pixels centered on the pixel represented by the target image region z(x i ) is the area where noise suppression processing is performed. j ) is the target image region z(x i ) are N distinct regions around the reference image region z(x j ) is coordinate x1 to coordinate x N That is, the reference image area z(x j ) are each complex numbers I(x j )(Complex number I(x1)~Complex number I(x N The reference image region z(x) is a set of 49 pixels centered around the pixel represented by j ) is the coordinate x i A given region S(x i ) (for example, a region containing 21 × 21 pixels). j) may be extracted from the entire complex image D. Also, the reference image region z(x j ) is the target image region z(x i ) may share some pixels. Similarly, multiple reference image regions z(x j ) among different reference image regions z(x j ) may have common pixels.

[0044] Then, the control unit 4 calculates the reference image area z(x j ) to find the target image region z(x i ) and perform noise suppression processing.

[0045] Specifically, in the first embodiment, the control unit 4 calculates the target image area z(x i ) and multiple reference image regions z(x j The control unit 4 then obtains a weight based on the obtained similarity. The control unit 4 also obtains a weight based on the obtained weight for the plurality of reference image regions z(x j ) are averaged using weights to obtain the target image region z(x i ) is subjected to non-local means processing to suppress noise as noise suppression processing.

[0046] <Non-local means processing> Here, the control unit 4 is configured to perform optimal noise suppression processing for the inspection system 100 by using the signal distribution (noise distribution) as a weight in the non-local means processing. This weight is calculated by dividing the signal distribution (noise distribution) by the weight of the multiple reference image regions z(x j ) and the target image region z(x i ) represents the similarity to

[0047] As shown in Fig. 6, noise contained in complex image D has a specific noise distribution. For example, if complex image D does not contain noise, the phase and amplitude of vibration in each of the 7 x 7 pixels in a region including 7 x 7 pixels will be approximately equal, and therefore, when the complex numbers I(x) of each extracted pixel are plotted on a complex plane, the plotted signal points will be clustered at approximately the same point. On the other hand, if complex image D contains noise, when the complex numbers I(x) of each extracted pixel are plotted on a complex plane, the plotted signal points will appear to be spread out according to a spread distribution (noise distribution).

[0048] Furthermore, this noise distribution spreads significantly depending on the strength of the detected signal. That is, this noise distribution increases depending on the magnitude of the amplitude of the periodic change in the vibration state of the inspection object P in the measurement area Pa. For example, where the vibration state of the vibration state image E is at an antinode, the amplitude increases, and the noise distribution spreads significantly. Where the vibration state is at a node, the amplitude decreases, and the noise distribution spreads less.

[0049] As shown in FIG. 7, in the first embodiment, the control unit 4 calculates the target image area z(x i The noise distribution of the signal points on the complex plane of the complex number I(x) representing each pixel of the signal points is approximated to be distributed according to a two-dimensional normal distribution that spreads biased in the direction of the signal points from the origin on the complex plane, depending on the magnitude of the amplitude of the periodic change in the vibration state of the test object P.

[0050] Specifically, when the true signal point in a certain pixel of complex image D when there is no noise is set to u=a+ib, control unit 4 approximates that the distribution of actual signal points in complex image D (noise distribution including noise) follows a normal distribution N(u|Σ). If the degree of spread from the origin in the signal point direction (θ) is set to λ1 and the degree of spread in the orthogonal direction orthogonal to the signal point direction is set to λ2, then considering an expansion matrix that is multiplied by λ1 in the angle θ direction (signal point direction) and by λ2 in the orthogonal direction, u can be expressed as u=|u|cosθ+i|u|sinθ, and therefore the covariance matrix Σ of the normal distribution N(u|Σ) can be expressed as shown in the following equation (5).

number

[0051] Then, the control unit 4 determines λ1 and λ2 as in the following equation (6), thereby approximating the noise distribution to a two-dimensional normal distribution that spreads biased toward the signal point according to the magnitude of the amplitude.

number

[0052] In the first embodiment, the control unit 4 performs the following calculation process to calculate the noise distribution of the target image region z(x i ) and the pixels in the reference image area z(x j ) and executes noise suppression processing.

[0053] Specifically, the target image region z(x i ) noise suppression in the target image region z(x i ) is the region consisting of the true values ​​when there is no noise in the pixels in u(xi ) and the target image region z(x i ) is the area where noise is estimated to be suppressed (area after noise suppression) i ), the optimal estimate is obtained by minimizing the posterior expected loss, as shown in equation (7).

number

number

number

[0054] And p(z(x i )|u(x i )) and p(u(x i Since it is not possible to measure an accurate value of the target image region z(x i ) is extracted from the surrounding area of ​​the reference image region z(x j ) is a complex number I(x j ) and obtain the prior distribution p(u(x i)) follows a uniform distribution, then u´(x i ) can be approximated as shown in equation (9) and equation (10).

number

[0055] Then, the target image region z(x i ) are expressed as z(x i,k ), (k:1 to n), and the reference image region z(x j ) are expressed as z(x j,k ), (k:1~n), z(x i,k )|z(x j,k ) is a normal distribution N(z(x i,k )|Σ) and p(z(x i,k )|z(x j,k )) is assumed to be as shown in equation (11).

number

[0056] By the above-described calculation process, the control unit 4 approximates the noise distribution of the signal points to follow a normal distribution with a covariance matrix Σ shown in equation (5), and by the calculation process shown in equations (12) to (14), subject Image area z(x i) and perform noise suppression processing.

number

[0057] The control unit 4 calculates the weight p(z(x i,k )|z(x j,k )) to find multiple reference image regions z(x j ) are weighted and averaged to obtain the region where noise is estimated to be suppressed (region after noise suppression), u'(x i The control unit 4 obtains the obtained u'(x i ) into the noise-suppressed target image region z(x i ) to obtain the target image region z(x i ) noise suppression processing is performed.

[0058] Then, the control unit 4 calculates the target image region z(x i ) are extracted (defined) sequentially, and the extracted target image region z(x i ), thereby suppressing the noise in the entire complex image D.

[0059] <Generation of vibration state image> As shown in FIG. 8 , in the first embodiment, the control unit 4 is configured to generate a vibration state image E that allows visual recognition of the vibration state of the inspection object P in the measurement area Pa, based on a complex image D in which noise has been suppressed by executing a noise suppression process. For example, the control unit 4 generates the vibration state image E as a moving image by rotating the phase of the complex number I(x) of each pixel of the noise-suppressed complex image D and converting the real part into a time-series image. The control unit 4 may also acquire the amplitude or phase of the complex number I(x) of each pixel of the noise-suppressed complex image D as a real value, and generate the vibration state image E as a single still image using the acquired amplitude or phase as a luminance value. The control unit 4 then displays the vibration state image E on the display unit 6 so that discontinuous areas of the vibration state can be visually recognized as defects Q (defective portions) occurring inside the inspection object P. Note that FIG. 8 illustrates a state in which noise has been suppressed from the vibration state image E of FIG. 2.

[0060] (Parameter change) In the first embodiment, the control unit 4 also uses a parameter n c , parameter n r , and parameter n of the parameter γ c and n r Specifically, the parameter n c and n rThe control unit 4 receives an input operation to change the parameter n c and n r By changing the value of the normal distribution, it is possible to adjust the extent to which the two-dimensional normal distribution spreads when approximating the noise distribution in the noise suppression process.

[0061] In the first embodiment, the control unit 4 also calculates the target image region z(x i ) and multiple reference image regions z(x j ) and the complex numbers I(x) representing the pixels included in each of the target image region z(x) are compared and the comparison results are stored. For example, as shown in equation (13), i ) and the reference image region z(x j The difference between the real parts a(x) and the imaginary parts b(x) of the complex numbers I(x) representing the pixels included in each of the c and n r When the control unit 4 first executes the process of suppressing noise in the complex image D, the parameter n c and n r The calculation results such as the above comparison results that do not change even when the value of the variable is changed are stored in the storage device.

[0062] In the first embodiment, the control unit 4 executes the noise suppression process once, and then the operation unit 7 is used to set the parameter n c and n r When an operation that changes parameter n is accepted, c and n r Using the stored comparison results, the spread of the two-dimensional normal distribution is adjusted, and the similarity is re-obtained by approximating the noise distribution again, and the weight p(z(x i,k )|z(x j,k )) is obtained again by the above equation (13). Then, the obtained weight p(z(x i,k )|z(xj,k )) and the noise suppression process is performed again.

[0063] (About defect inspection methods) Next, an image processing method performed by the inspection system 100 according to the first embodiment will be described with reference to Fig. 9. The image processing method according to the first embodiment is executed by control processing performed by the control unit 4.

[0064] First, in step 401, based on the measurement results from the speckle shearing interferometer 3, a complex image D is acquired in which each pixel is represented by a complex number I(x) that indicates the periodic change in the vibration state of the object P in the measurement area Pa.

[0065] Next, in step 402, a target image region z(x) defined in the acquired complex image D is calculated. i ) and the target image region z(x i ) are divided into multiple reference image regions z(x j ) is obtained.

[0066] Next, in step 403, the acquired target image region z(x i ) and multiple reference image regions z(x j ) is obtained by comparing the complex numbers I(x) representing each pixel. Specifically, i ) and the reference image region z(x j ), the difference between the real parts a(x) and the imaginary parts b(x) of the complex numbers I(x) that represent pixels with the same relative coordinates is obtained as a comparison result, and the target image region z(x i ) and the reference image region z(x j ) is acquired, where j=1 to N. The acquired comparison result is stored.

[0067] Next, in step 404, a plurality of reference image regions z(x j) weight p(z(x i,k )|z(x j,k )) is obtained.

[0068] Next, in step 405, the obtained weights p(z(x i,k )|z(x j,k )) based on the target image region z(x i ) is subjected to noise suppression processing. Specifically, the obtained weight p(z(x i,k )|z(x j,k )) to find multiple reference image regions z(x j ) are weighted and averaged to obtain the noise suppressed target image region z(x i )(u´(x in Eq. (12) i )) is obtained.

[0069] Next, in step 406, the target image region z(x i ) is performed, the noise in the complex image D is suppressed.

[0070] Next, in step 407, a vibration state image E that allows visual recognition of the vibration state of the inspection object P in the measurement area Pa is generated based on the noise-suppressed complex image D. Then, the generated vibration state image E is displayed on the display unit 6.

[0071] Next, in step 408, a parameter n c , parameter n r , and parameter n of the parameter γ c and n r It is determined whether or not an input operation to change the parameter n has been accepted by the operation unit 7. c and n r If it is determined that an input operation to change the parameter n has been accepted, the process proceeds to step 409. c and n r If it is not determined that an input operation to change the value has been accepted, the control process is terminated.

[0072] In step 409, the parameter n c and n r is changed. And the changed parameter n c and n r and the comparison result stored in step 403, the similarity is calculated again, and the process returns to step 404. Then, steps 404 to 407 are executed again, and the similarity is calculated again using the changed parameter n c and n r The noise suppression process for the complex number image D using the above is executed again, and the vibration state image E is generated again and displayed on the display unit 6.

[0073] (Effects of the first embodiment) In the first embodiment, the following effects can be obtained.

[0074] In the inspection system 100 of the first embodiment, as described above, the target image region z(x i ) and the pixels in the target image region z(x i ) are defined in the complex image D separately from the reference image region z(x j The similarity between the pixels included in the image and the complex numbers I(x) representing each pixel is obtained by comparing them. Then, a weight p(z(x i,k )|z(x j,k )) to find the target image region z(x i ) is subjected to noise suppression processing. i) can suppress noise in the complex image D. Therefore, since the complex image D, in which each pixel is represented by a complex number I(x) indicating a periodic state change, reflects the periodic state change of the vibration, unlike when noise suppression processing is performed on an image represented by real values, by suppressing noise in the complex image D, noise can be effectively suppressed even when the noise state changes according to the state of the vibration. As a result, noise can be effectively suppressed even when measuring a periodic state change.

[0075] Furthermore, in the first embodiment, further effects can be obtained by configuring as follows.

[0076] That is, in the first embodiment, as described above, the control unit 4 (image processing unit) calculates the target image region z(x i ) on the complex plane of the complex number I(x) representing each pixel of the signal point of the complex number I(x) is approximated to be distributed according to a two-dimensional normal distribution that is biased in spreading from the origin in the complex plane toward the signal point, depending on the magnitude of the amplitude of the periodic change in the vibration state of the test object P, and similarity is obtained based on the approximated distribution. With this configuration, even if the spread of the noise distribution on the complex plane is biased, it is possible to approximate the noise distribution to correspond to the biased spread of the actual noise by approximating the distribution according to a two-dimensional normal distribution that is biased in spreading from the origin in the complex plane toward the signal point. Therefore, when the spread of the noise distribution on the complex plane is biased, it is possible to more appropriately calculate the similarity. As a result, the weight p(z(x)) based on the similarity is calculated. i,k )|z(x j,k )) can suppress noise more effectively.

[0077] In the first embodiment, as described above, the inspection system 100 uses a parameter n c and n rThe apparatus is provided with an operation unit 7 that accepts an operation to change the noise distribution. With this configuration, the inspector can easily perform an operation to adjust the degree of noise distribution spread. Therefore, the degree of noise suppression in the complex image D can be easily adjusted.

[0078] In the first embodiment, as described above, the control unit 4 (image processing unit) calculates the target image region z(x i ) and multiple reference image regions z(x j ) and the complex numbers I(x) representing the pixels included in each of them are compared and the comparison results are stored, and a parameter n c and n r When an operation to change the distribution is accepted, the parameter n c and n r and the stored comparison result are used to obtain the similarity. i ) and multiple reference image regions z(x j The value of the comparison result of comparing the complex numbers I(x) representing the pixels contained in each of the c and n r Therefore, by storing the comparison results once calculated, the parameter n c and n r When the value of the parameter is changed, the time required to perform the process of suppressing noise in the complex image D again can be reduced.

[0079] In the first embodiment, as described above, the control unit 4 (image processing unit) calculates a plurality of reference image regions z(x j ) based on the similarity obtained for i,k )|z(x j,k )) to obtain the target image region z(x i ) is configured to be executed as noise suppression processing. With this configuration, the weight p(z(x i,k)|z(x j,k )) to find the reference image region z(x j ) is simply averaged over the non-local means process, so the obtained weights p(z(x i,k )|z(x j,k )), the target image area z(x i ) noise can be easily suppressed.

[0080] Furthermore, in the first embodiment, as described above, the inspection system 100 includes an irradiation unit 2 that irradiates laser light onto the inspection object P in which elastic waves are excited by the oscillator 1 (excitation unit), and the speckle shearing interferometer 3 (measurement unit) is configured to cause interference between reflected laser light, which is laser light irradiated by the irradiation unit 2 and reflected by the inspection object P, and reference laser light, which is laser light irradiated by the irradiation unit 2, and to capture and measure the interference light caused by interference between the reflected laser light and the reference laser light, and the control unit 4 (image processing unit) is configured to suppress noise in a complex image D generated based on a measurement result of the captured interference light. With this configuration, the interference light of the measured laser light changes periodically due to the vibration of the elastic waves, so that the noise component included in the measurement result can be effectively suppressed by representing the measurement result using the complex image D and suppressing the noise.

[0081] Furthermore, when laser light is irradiated onto the inspection object P in which elastic waves are excited, the vibration of the elastic waves causes a change in the irradiation angle of the laser light with respect to the surface of the inspection object P, or a change in the speckle pattern due to displacement of the surface of the inspection object P, resulting in a distribution such that the noise distribution of the signal points on the complex plane of the complex number I(x) representing each pixel of the complex image D spreads unevenly. Therefore, as in the first embodiment, by generating the complex image D based on the measurement results of the interference light and approximating the noise distribution of the signal points to be distributed according to a two-dimensional normal distribution that spreads unevenly in the signal point direction depending on the magnitude of the amplitude, it is possible to suppress noise so as to correspond to the noise distribution specific to the complex image D generated based on the measurement results of the interference light.

[0082] Furthermore, in the first embodiment, as described above, the control unit 4 (image processing unit) is configured to acquire the measured degree of interference coh(x) of the interference light and to weight the complex number I(x) representing each pixel of the complex image D according to the acquired degree of interference coh(x), thereby correcting the complex number I(x) representing each pixel. Here, for pixels with a small degree of interference coh(x), the measured vibration state will have an inaccurate value. Therefore, if the similarity is calculated while including the complex number I(x) representing a pixel with a small degree of interference coh(x), the accuracy of the noise suppression process will decrease. In contrast, in the first embodiment, the control unit 4 is configured to acquire the measured degree of interference coh(x) of the interference light and to weight the complex number I(x) representing each pixel of the complex image D according to the acquired degree of interference coh(x), thereby correcting the complex number I(x) representing each pixel. With this configuration, by correcting the weighting of the complex number I(x) representing a pixel with a small degree of interference coh(x) to be smaller, it is possible to prevent the noise suppression process from being performed using an inaccurate complex number I(x).As a result, even when the complex image D includes pixels with a small degree of interference coh(x), it is possible to prevent the accuracy of the noise suppression process from decreasing.

[0083] Furthermore, in the first embodiment, as described above, the control unit 4 (image processing unit) is configured to generate a vibration state image E that allows visual recognition of the vibration state of the inspection object P, based on the complex image D in which noise has been suppressed by executing noise suppression processing. With this configuration, noise in the generated vibration state image E can be suppressed, and therefore, by visually recognizing the generated vibration state image E, the inspection operator can easily recognize the distribution of the vibration state of the inspection object P. Therefore, when inspecting a defect Q of the inspection object P, the continuity of the vibration state in the inspection object P can be recognized by visually recognizing the vibration state image E, and therefore the defect Q in the inspection object P can be easily recognized.

[0084] (Effects of the image processing method according to the first embodiment) The image processing method of the first embodiment can provide the following effects.

[0085] In the image processing method of the first embodiment, by configuring as described above, the target image region z(x i ) and the pixels in the target image region z(x i ) are defined in the complex image D separately from the reference image region z(x j The similarity between the pixels in the image and the image is obtained by comparing the complex numbers I(x) representing each pixel. Then, a weight p(z(x i,k )|z(x j,k )) to find the target image region z(x i ) is subjected to noise suppression processing. i ) can suppress noise in the complex image D. Therefore, the complex image D, in which each pixel is represented by a complex number I(x) indicating a periodic state change, reflects the periodic state change of the vibration. Therefore, unlike when noise suppression processing is performed on an image represented by real values, by suppressing noise in the complex image D, noise can be effectively suppressed even when the noise state changes according to the state of the vibration. As a result, it is possible to provide an image processing method that can effectively suppress noise even when measuring a periodic state change.

[0086] [Second embodiment] Next, the configuration of an inspection system 200 according to a second embodiment of the present invention will be described with reference to Figures 10 and 11. This second embodiment is configured to suppress background components of the vibration state image E by removing low-frequency components. Note that the same components as those in the first embodiment are denoted by the same reference numerals in the figures, and their description will be omitted.

[0087] As shown in Fig. 10, an inspection system 200 according to the second embodiment includes a control unit 204. Incidentally, the measurement of interference light by the speckle sharing interferometer 3 according to the second embodiment is similar to that of the first embodiment. The control unit 204 generates a complex image D based on the measurement results by the speckle sharing interferometer 3 by the same control processing as that of the control unit 4 according to the first embodiment. Incidentally, the control unit 204 is an example of an "image processing unit" in the claims.

[0088] 11, in the vibration state image E generated based on the complex number image D, the vibration state changes sharply (high frequency) near the defect Q, but the vibration state changes relatively gently (low frequency) in the background part other than the defect Q. Therefore, in the second embodiment, the control unit 204 is configured to generate a vibration state image E200 in which low frequency components are removed and noise is suppressed.

[0089] Specifically, the control unit 204, like the control unit 4 in the first embodiment, extracts the target image region z(x i ) and multiple reference image regions z(x j ) (j: 1 to N). In addition, the control unit 204 acquires the target image region z(x i ) to suppress noise in multiple reference image regions z(x j ) weight p(z(x i,k )|z(x j,k )) (see equation (13)).

[0090] Then, the control unit 204 calculates the reference image area z(x j ) to remove low-frequency components. For example, the control unit 204 may perform a low-cut filter process using a Fourier transform on the reference image region z(x j )

[0091] In the second embodiment, the control unit 204 then calculates the plurality of reference image regions z from which the low-frequency components have been removed.lowcut (x j ) based on the similarity obtained for i,k )|z(x j,k )) to obtain the target image region z(x i ) is configured to execute non-local means processing to suppress noise as noise suppression processing.

[0092] That is, the control unit 204 calculates the multiple reference image regions z(x j ) and the target image region z(x i ) and the weight p(z(x i,k )|z(x j,k )) to obtain the multiple reference image regions z after the low-frequency components have been removed. lowcut (x j ) and average it to suppress noise and remove low-frequency components. i ) is u lowcut ´(x i ) to get u lowcut ´(x i ) is added to equation (12) by z lowcut (x j ) is substituted, it is expressed as the following equation (15).

number

[0093] Then, the control unit 204 calculates the target image region z(x i ), noise in the complex image D is suppressed and background components (low frequency components) are removed. Then, the control unit 204 generates a vibration state image E200 from which low frequency components have been removed and noise has been suppressed, based on the complex image D from which low frequency components have been removed and noise has been suppressed, by processing similar to that in the first embodiment. Note that other configurations of the second embodiment are similar to those of the first embodiment.

[0094] [Effects of the second embodiment] In the second embodiment, the following effects can be obtained.

[0095] In the second embodiment, as described above, the control unit 204 (image processing unit) calculates the reference image area z(x j ) and a plurality of reference image regions z lowcut (x j ) based on the similarity obtained for i,k )|z(x j,k )) to obtain the target image region z(x i ) as noise suppression processing. Here, when low-frequency components are included as background components in the complex image D, if similarity is acquired and weights are calculated with the low-frequency components removed in advance, an accurate similarity cannot be acquired, and the noise suppression effect is reduced. Furthermore, when low-frequency components are removed after performing noise suppression processing, harmonic components (artifacts) generated by the non-local means processing that were not noticeable before the removal of the low-frequency components are emphasized. In contrast, in the second embodiment, the control unit 204 is configured to perform non-local means processing for suppressing noise in a plurality of reference image regions z(x j ) and a plurality of reference image regions z lowcut (x j ) based on the similarity obtained for i,k )|z(x j,k )) to obtain the target image region z(x i ) is configured to be executed as noise suppression processing. With this configuration, when low-frequency components are included as background components in the complex image D, the weight p(z(x i,k )|z(x j,k When calculating the weight p(z(x i,k )|z(x j,k When averaging is performed using the reference image region z lowcut(x j ) can be weighted and averaged. Therefore, the weight p(z(x i,k )|z(x j,k ) can be calculated, the low-frequency components can be removed while suppressing a decrease in accuracy, and the emphasis of harmonic components generated by non-local means processing can be suppressed. As a result, when low-frequency components are included as background components in the complex image D, noise can be effectively suppressed while removing the background components.

[0096] Other effects of the second embodiment are the same as those of the first embodiment.

[0097] [Third embodiment] Next, the configuration of an inspection system 300 according to a third embodiment of the present invention will be described with reference to Figs. 12 and 13. In this third embodiment, a target image region z(x i ) and each pixel in the reference image region z(x j ) and uses the average value when comparing each pixel. Note that the same components as those in the first embodiment are denoted by the same reference numerals in the drawings, and the description thereof will be omitted.

[0098] As shown in Fig. 12, an inspection system 300 according to the third embodiment includes a control unit 304. Incidentally, the measurement of interference light by the speckle sharing interferometer 3 according to the third embodiment is similar to that of the first embodiment. The control unit 304 generates a complex image D based on the measurement results by the speckle sharing interferometer 3 by the same control processing as that of the control unit 4 according to the first embodiment. Incidentally, the control unit 304 is an example of an "image processing unit" in the claims.

[0099] Furthermore, the control unit 304, like the control unit 4 in the first embodiment, calculates the target image region z(x i ) and multiple reference image regions z(x j )(j:1~N) is obtained.

[0100] As shown in FIG. 13, in the third embodiment, the control unit 304 calculates the target image region z(x i ) for each pixel (z(x i,k )) is a complex number I(x i,k ) average value I avr (x i ) and the reference image region z(x j ) for each pixel (z(x j,k )) is a complex number I(x j,k ) average value I avr (x j ) to obtain the similarity.

[0101] Specifically, the control unit 304 controls the target image region z(x i ) is a complex number I(x i,k ) average value I avr (x i ) is calculated. avr (x i ) is I avr (x i )=a avr (x i )+ib avr (x i ) Similarly, the control unit 304 controls the target image region z(x j ) is a complex number I(x j,k ) average value I avr (x j ) is calculated. avr (x j ) is I avr (x j )=a avr (x j )+ib avr (x j ) where k=1 to n.

[0102] The control unit 304 calculates the average value I avr (x i ) and the average value I avr (x j ) and calculate the difference between the target image area z(x i ) and the reference image region z(x j) to obtain the similarity and the average value I avr (x i ) and the average value I avr (x j ) and the reference image region z(x j ) weight p(z(x i )|z(x j That is, the control unit 304 obtains a(x i,k ) and b(x i,k ) instead of a avr (x i ) and b avr (x i ) and a(x j,k ) and b(x j,k ) instead of a avr (x j ) and b avr (x j ) to obtain the weight p(z(x i )|z(x j )) to get the

[0103] That is, in the third embodiment, the control unit 304 controls the reference image area z(x j Unlike the control unit 4 of the first embodiment, which is configured to weight each pixel in the reference image region z(x j ) are weighted based on a single similarity and averaged.

[0104] Then, the control unit 304 calculates the reference image area z(x j ) are weighted and averaged to perform noise suppression processing on the target image region z(x i ) to suppress noise in the complex image D.

[0105] The control unit 304 determines the target image area z(x i ) average value I avr (x i ) and the reference image region z(x j ) average value I avr (x j) and the control unit 304 stores the parameter n c and n r If is changed, the changed parameter n c and n r and the stored average value I avr (x i ) and mean value I avr (x j ) to calculate the weight.

[0106] The other configurations of the third embodiment are the same as those of the first embodiment.

[0107] [Effects of the third embodiment] In the third embodiment, the following effects can be obtained.

[0108] In the third embodiment, as described above, the control unit 304 (image processing unit) calculates the target image region z(x i ) is a complex number I(x i,k ) average value I avr (x i ) and the reference image region z(x j ) is a complex number I(x j,k ) average value I avr (x j ) to obtain the similarity. i ) and multiple reference image regions z(x j ) can be compared to comparing each pixel included in the complex image D. Therefore, the time required for the calculation process to calculate the similarity can be shortened, and the time required for noise suppression in the complex image D can be shortened.

[0109] Other effects of the third embodiment are similar to those of the first and second embodiments.

[0110] [Variations] The embodiments disclosed herein should be considered to be illustrative and not restrictive in all respects. The scope of the present invention is defined by the claims rather than the above description of the embodiments, and further includes all modifications (variations) within the meaning and scope of the claims.

[0111] For example, in the first to third embodiments described above, an example of an inspection system 100 (200, 300) in which the transducer 1 (exciter), irradiation unit 2, speckle shearing interferometer 3 (measurement unit), and control unit 4, 204, 304 (image processing unit) are each provided separately has been shown, but the present invention is not limited to this. In the present invention, the excitation unit, irradiation unit, measurement unit, and control unit may be integrally configured as a defect inspection apparatus. In this case, the control unit of the defect inspection apparatus has the same configuration as the control unit 4, 204, 304 (image processing unit) of the first to third embodiments.

[0112] Furthermore, in the above first to third embodiments, an example was shown in which the noise distribution of the signal points is approximated to be distributed according to a two-dimensional normal distribution that spreads biasedly from the origin in the signal point direction on the complex plane in accordance with the magnitude of the amplitude of the periodic change in the vibration state of the test object P, but the present invention is not limited to this. In the present invention, the noise distribution of the signal points may be approximated to have a constant spread regardless of the magnitude of the amplitude of the periodic change in the vibration state. Alternatively, the noise distribution of the signal points may be approximated to be distributed according to a normal distribution that spreads isotropically without bias. Alternatively, the noise distribution of the signal points may be approximated to spread not only in the signal point direction but also in an orthogonal direction orthogonal to the signal point direction in accordance with the magnitude of the amplitude of the periodic change in the vibration state.

[0113] In the first to third embodiments, the parameter n c and n r However, the present invention is not limited to this. For example, three parameters n c , n r , and γ may all be configured to be variable, or the three parameters n c, n r , and γ may be configured to be variable. For example, three appropriate parameters n c , n r The values ​​of γ and γ may be set to be selectable in advance. c The optimal value of varies depending on the lighting and other factors in the shooting environment, while the parameter n r , and γ may not change depending on the material of the inspection object P. In that case, by performing a calibration work in advance, it is possible to set the parameter n r , and γ are set, and the parameter n c The noise suppression may be adjusted while changing the

[0114] In the first to third embodiments, the target image region z(x i ) and multiple reference image regions z(x j ) and the complex numbers I(x) representing the pixels included in each of the parameters n c and n r The comparison result may be recalculated each time the value is changed.

[0115] In the first to third embodiments, the weighted reference image area z(x j ) to obtain the target image region z(x i ) is an example of a non-local means process for performing noise suppression processing, but the present invention is not limited to this. For example, i,k )|z(x j,k )) is equal to or greater than a predetermined threshold value. j ) is used to suppress noise in the target image region z(x i ) may be calculated. j ) may be calculated by taking the square mean instead of the simple mean.

[0116] In the second embodiment, the plurality of reference image regions z(x j ) and the target image region z(x i ) and the weight p(z(x i,k )|z(x j,k )) to obtain the multiple reference image regions z after the low-frequency components have been removed. lowcut (x j ) and average it to suppress noise and remove low-frequency components. i ) is obtained based on the similarity after removing the low frequency components. i,k )|z(x j,k )) and the reference image region z lowcut (x j ) may be weighted and averaged. i,k )|z(x j,k )) to find multiple reference image regions z(x j ) may be weighted and averaged, and then low frequency components may be removed.

[0117] In the third embodiment, the target image region z(x i ) and the reference image region z(x j ) the average value of each pixel (I avr (x i ) and I avr (x j However, the present invention is not limited to this. For example, instead of the average value, the median or the mode may be acquired.

[0118] In the first to third embodiments, the speckle shearing interferometer 3 is used to measure the interference light, which is the interfered laser light. However, the present invention is not limited to this. For example, the interference light may be measured using a Michelson interferometer. The vibration state on the outer surface may also be measured using an optical measurement method such as a digital image correlation method or a moire sampling method. The vibration state of the inspection object P may also be measured using a ToF (time of flight) camera.

[0119] In the first to third embodiments, each pixel in the entire complex image D is corrected according to the degree of interference coh(x), but the present invention is not limited to this. For example, only pixels in the complex image D with a small degree of interference coh(x) may be extracted and corrected.

[0120] Furthermore, in the above first to third embodiments, an example was shown in which a vibration state image E was generated that allows visual recognition of the vibration state of the inspection object P in the measurement area Pa, but the present invention is not limited to this. For example, an area estimated to be a defect Q may be detected by detecting the continuity of vibration. In this case, an image that allows recognition of the area estimated to be a defect Q, or numerical values ​​such as coordinates, may be output.

[0121] In addition, in the first to third embodiments, examples have been shown in which the vibrator 1 is used as an example of an excitation unit that excites elastic waves in the test object P, but the present invention is not limited to this. For example, elastic waves may be excited by a loudspeaker that emits sound.

[0122] [Aspect] It will be appreciated by those skilled in the art that the exemplary embodiments described above are examples of the following aspects.

[0123] (Item 1) an excitation unit that excites elastic waves in the test object; a measurement unit that measures a vibration state of the test object in which the elastic waves are excited by the excitation unit; and an image processing unit configured to suppress noise in a complex number image in which each pixel is represented by a complex number indicating a periodic change in the vibration state of the test object based on the measurement result of the measurement unit, The image processing unit The method obtains a similarity between pixels included in a target image area defined in the complex image and pixels included in a plurality of reference image areas defined in the complex image separately from the target image area by comparing complex numbers representing the respective pixels; An inspection system that performs noise suppression processing on the target image region using the obtained weights based on the similarity.

[0124] (Item 2) Item 1. The inspection system according to item 1, wherein the image processing unit is configured to approximate a noise distribution of signal points on a complex plane of complex numbers representing each pixel of the target image region so that the noise distribution follows a two-dimensional normal distribution that spreads biasedly from an origin toward the signal points on the complex plane, depending on the magnitude of the amplitude of the periodic change in the vibration state of the object to be inspected, and to acquire the similarity based on the approximated distribution.

[0125] (Item 3) 3. The inspection system according to item 2, further comprising an operation unit that accepts an operation to change a parameter that indicates the degree of spread of the distribution when approximating the noise distribution.

[0126] (Item 4) Item 4. The inspection system according to item 3, wherein the image processing unit is configured to store a comparison result obtained by comparing complex numbers representing pixels included in the target image area and each of the plurality of reference image areas, and, when an operation to change a parameter indicating a degree of spread of the distribution is accepted, to obtain the similarity using the changed parameter indicating the degree of spread of the distribution and the stored comparison result.

[0127] (Item 5) 5. The inspection system according to any one of items 1 to 4, wherein the image processing unit is configured to perform, as the noise suppression processing, a non-local means processing that suppresses noise in the target image region by averaging using weights based on the similarities obtained for the multiple reference image regions.

[0128] (Item 6) Item 6. The inspection system according to item 5, wherein the image processing unit is configured to perform, as the noise suppression processing, non-local means processing that suppresses noise in the target image region by removing low-frequency components from the plurality of reference image regions and averaging the plurality of reference image regions from which the low-frequency components have been removed using weights based on the similarity obtained.

[0129] (Item 7) 7. The inspection system according to any one of items 1 to 6, wherein the image processing unit is configured to acquire the similarity by comparing an average value of complex numbers representing each pixel of the target image region with an average value of complex numbers representing each pixel of the reference image region.

[0130] (Item 8) an irradiation unit that irradiates the test object in which the elastic waves are excited by the excitation unit with laser light; the measurement unit is configured to cause interference between a reflected laser light, which is the laser light irradiated by the irradiation unit and reflected by the test object, and a reference laser light, which is the laser light irradiated by the irradiation unit, and to capture and measure interference light caused by interference between the reflected laser light and the reference laser light, 8. The inspection system according to any one of items 1 to 7, wherein the image processing unit is configured to suppress noise in the complex image generated based on the measurement result of the captured interference light.

[0131] (Item 9) Item 9. The inspection system of item 8, wherein the image processing unit is configured to acquire the degree of interference of the measured interference light, and is configured to correct the complex numbers representing each pixel of the complex number image by weighting the complex numbers representing each pixel in accordance with the acquired degree of interference.

[0132] (Item 10) The inspection system according to any one of items 1 to 9, wherein the image processing unit is configured to generate a vibration state image that allows visual recognition of the vibration state of the inspection object based on the complex image in which noise has been suppressed by executing the noise suppression processing.

[0133] (Item 11) obtaining a complex image in which each pixel is represented by a complex number that indicates a periodic change in state; acquiring a target image region defined in the acquired complex image and a plurality of reference image regions defined in the complex image separately from the target image region; obtaining a similarity between pixels included in the obtained target image area and pixels included in the plurality of reference image areas by comparing complex numbers representing the respective pixels; and performing noise suppression processing on the target image region using the weight based on the obtained similarity.

[0134] (Item 12) an excitation unit that excites elastic waves in the test object; an irradiation unit that irradiates the test object with laser light in which the elastic waves are excited by the excitation unit; a measurement unit that causes interference between a reflected laser light, which is the laser light irradiated by the irradiation unit and reflected by the test object, and a reference laser light, which is the laser light irradiated by the irradiation unit, and captures and measures interference light caused by interference between the reflected laser light and the reference laser light, thereby measuring a vibration state of the test object in which the elastic wave is excited by the excitation unit; and a control unit configured to suppress noise in a complex number image for defect inspection, in which each pixel is represented by a complex number indicating a periodic change in the vibration state of the inspection object, based on the measurement result of the measurement unit; The control unit The method obtains a similarity between pixels included in a target image area defined in the complex image and pixels included in a plurality of reference image areas defined in the complex image separately from the target image area by comparing complex numbers representing the respective pixels; A defect inspection device that performs noise suppression processing on the target image region using weights based on the acquired similarity. [Explanation of symbols]

[0135] 1. Vibrator (excitation part) 2. Irradiation unit 3 Speckle shearing interferometer (measurement unit) 4, 204, 304 Control section 100, 200, 300 Inspection Systems

Claims

1. an excitation unit that excites elastic waves in the test object; a measurement unit that measures a vibration state of the test object in which the elastic waves are excited by the excitation unit; and an image processing unit configured to suppress noise in a complex number image in which each pixel is represented by a complex number indicating a periodic change in the vibration state of the test object based on the measurement result of the measurement unit, The image processing unit The method obtains a similarity between pixels included in a target image area defined in the complex image and pixels included in a plurality of reference image areas defined in the complex image separately from the target image area by comparing complex numbers representing the respective pixels; performing noise suppression processing on the target image region using the weight based on the acquired similarity; the image processing unit is configured to approximate a noise distribution of signal points on a complex plane of complex numbers representing each pixel of the target image region so that the noise distribution follows a two-dimensional normal distribution that spreads unevenly on the complex plane, and to obtain the similarity based on the approximated distribution.

2. 2. The inspection system according to claim 1, wherein the image processing unit is configured to approximate the noise distribution of signal points on a complex plane of complex numbers representing each pixel of the target image region so that the noise distribution follows a two-dimensional normal distribution that spreads biasedly from an origin toward the signal points on the complex plane, depending on the magnitude of the amplitude of the periodic change in the vibration state of the object to be inspected, and to acquire the similarity based on the approximated distribution.

3. The inspection system according to claim 2 , further comprising an operation unit that accepts an operation to change a parameter that indicates a degree of spread of the distribution when approximating the noise distribution.

4. An excitation unit that excites elastic waves in an object to be inspected; a measurement unit that measures a vibration state of the test object in which the elastic waves are excited by the excitation unit; and an image processing unit configured to suppress noise in a complex number image in which each pixel is represented by a complex number indicating a periodic change in the vibration state of the test object based on the measurement result of the measurement unit, The image processing unit The method obtains a similarity between pixels included in a target image area defined in the complex image and pixels included in a plurality of reference image areas defined in the complex image separately from the target image area by comparing complex numbers representing the respective pixels; performing noise suppression processing on the target image region using the weight based on the acquired similarity; the image processing unit is configured to approximate a noise distribution of signal points on a complex plane of complex numbers representing each pixel of the target image region so that the noise distribution follows a two-dimensional normal distribution that spreads biasedly from an origin toward the signal points on the complex plane, depending on the magnitude of the amplitude of the periodic change in the vibration state of the object to be inspected, and to acquire the similarity based on the approximated distribution; an operation unit that accepts an operation to change a parameter that indicates a degree of spread of the distribution when approximating the noise distribution; the image processing unit is configured to store a comparison result obtained by comparing complex numbers representing pixels included in the target image area and each of the plurality of reference image areas, and, when an operation to change a parameter indicating the degree of spread of the distribution is accepted, to obtain the similarity using the changed parameter indicating the degree of spread of the distribution and the stored comparison result.

5. The inspection system according to any one of claims 1 to 4, wherein the image processing unit is configured to perform non-local means processing as the noise suppression processing, which suppresses noise in the target image area by averaging using weights based on the similarities obtained for the multiple reference image areas.

6. An excitation unit that excites elastic waves in an object to be inspected; a measurement unit that measures a vibration state of the test object in which the elastic waves are excited by the excitation unit; and an image processing unit configured to suppress noise in a complex number image in which each pixel is represented by a complex number indicating a periodic change in the vibration state of the test object based on the measurement result of the measurement unit, The image processing unit The method obtains a similarity between pixels included in a target image area defined in the complex image and pixels included in a plurality of reference image areas defined in the complex image separately from the target image area by comparing complex numbers representing the respective pixels; performing noise suppression processing on the target image region using the weight based on the acquired similarity; the image processing unit is configured to perform, as the noise suppression processing, a non-local means processing that suppresses noise in the target image region by averaging using weights based on the similarities acquired for the plurality of reference image regions; The image processing unit is configured to perform non-local means processing as the noise suppression processing, which suppresses noise in the target image region by removing low-frequency components from the multiple reference image regions and averaging them using weights based on the similarity obtained for the multiple reference image regions from which the low-frequency components have been removed.

7. The inspection system according to any one of claims 1 to 6, wherein the image processing unit is configured to obtain the similarity by comparing an average value of complex numbers representing each pixel of the target image region with an average value of complex numbers representing each pixel of the reference image region.

8. an irradiation unit that irradiates the test object in which the elastic waves are excited by the excitation unit with laser light; the measurement unit is configured to cause interference between a reflected laser light, which is the laser light irradiated by the irradiation unit and reflected by the test object, and a reference laser light, which is the laser light irradiated by the irradiation unit, and to capture and measure interference light caused by interference between the reflected laser light and the reference laser light, The inspection system according to any one of claims 1 to 7, wherein the image processing unit is configured to suppress noise in the complex image generated based on the measurement results of the captured interference light.

9. 9. The inspection system according to claim 8, wherein the image processing unit is configured to acquire a degree of interference of the measured interference light, and to correct the complex number representing each pixel of the complex number image by weighting the complex number representing each pixel in accordance with the acquired degree of interference.

10. The inspection system according to any one of claims 1 to 9, wherein the image processing unit is configured to generate a vibration state image that allows visual recognition of the vibration state of the object to be inspected, based on the complex image in which noise has been suppressed by executing the noise suppression processing.

11. obtaining a complex image in which each pixel is represented by a complex number that indicates a periodic change in state; acquiring a target image region defined in the acquired complex image and a plurality of reference image regions defined in the complex image separately from the target image region; obtaining a similarity between pixels included in the obtained target image area and pixels included in the plurality of reference image areas by comparing complex numbers representing the respective pixels; and performing noise suppression processing on the target image region using the weight based on the acquired similarity, the step of acquiring the similarity includes a step of approximating a noise distribution of signal points on a complex plane of complex numbers representing each pixel of the target image region so that the noise distribution follows a two-dimensional normal distribution that spreads unevenly on the complex plane, and acquiring the similarity based on the approximated distribution.

12. an excitation unit that excites elastic waves in the test object; an irradiation unit that irradiates the test object with laser light in which the elastic waves are excited by the excitation unit; a measurement unit that causes interference between a reflected laser light, which is the laser light irradiated by the irradiation unit and reflected by the test object, and a reference laser light, which is the laser light irradiated by the irradiation unit, and captures and measures interference light caused by interference between the reflected laser light and the reference laser light, thereby measuring a vibration state of the test object in which the elastic wave is excited by the excitation unit; and a control unit configured to suppress noise in a complex number image for defect inspection, in which each pixel is represented by a complex number indicating a periodic change in the vibration state of the inspection object, based on the measurement result of the measurement unit; The control unit The method obtains a similarity between pixels included in a target image area defined in the complex image and pixels included in a plurality of reference image areas defined in the complex image separately from the target image area by comparing complex numbers representing the respective pixels; performing noise suppression processing on the target image region using the weight based on the acquired similarity; the control unit is configured to approximate a noise distribution of signal points on a complex plane of complex numbers representing each pixel of the target image region so that the noise distribution follows a two-dimensional normal distribution that spreads unevenly on the complex plane, and to acquire the similarity based on the approximated distribution.

Images