JP2018128436A - Surface roughness determination device and determination method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a surface roughness determination device capable of performing micro-roughness determination, reduction in measurement time, roughness determination of a curved surface, and accurate roughness determination of an article having a large area.SOLUTION: A surface roughness determination device includes an arithmetic processing unit 3 which acquires and calculates three-band visual sensitivity images S1i, S2i, and S3i acquired from a surface 5 by a two-dimensional colorimeter 2 having three spectral sensitivities (S1(λ), S2(λ), and S3(λ)), and calculates a color difference ΔE, creates a color space histogram distribution by integrating the number of pixels of an inspection surface and a standard surface that partition the inspection area by the lattice, calculates a surface roughness index M showing a difference of two color space histogram distributions of the inspection surface and the standard surface, and sets at least one of a first calibration function L1 showing a correlation between a surface roughness measurement value Ra and a surface roughness estimation index Est actually measured by a roughness meter, a second calibration function L2 showing the correlation between the surface roughness measurement value Ra and the color difference ΔE and a third calibration function L3 showing the correlation between the surface roughness measurement value Ra and the surface roughness index M.SELECTED DRAWING: Figure 1

Description

  The present invention relates to a surface roughness determination device and determination method, and more particularly to a determination device and method for evaluating the degree of coincidence of surface roughness.

  Various inventions have been proposed for polishing fine irregularities on a surface, for example, a metal surface (http://www.chemicoat.co.jp/column/detail_6.html).

  “Chemical polishing” in the above invention is a polishing method in which a metal is immersed in a polishing solution and the surface of the metal is corroded by the chemical reaction by the force of acid or alkali. Since the solution enters even fine parts, it is characterized in that it can process even details that cannot be reached by mechanical polishing. Chemical polishing is suitable mainly for polishing fine parts with complicated shapes and inner surface portions.

  The “electropolishing” is a polishing method in which a metal is immersed in a solution for electropolishing and a current is passed in the solution with the metal as a plus. The metal gradually dissolves and the surface becomes glossy and smooth. It is used when chemical polishing and clean polishing methods such as precision parts are necessary because it does not create a deteriorated layer by processing and does not leave dirt or burns.

  “Mechanical polishing” refers to polishing performed using a machine as its name suggests, and there are various methods. The current mainstream is a rotary type in which a polishing pad is placed on a disk-shaped surface plate, a liquid abrasive containing chemical components and fine particles is hung on it, and polished while rotating. In addition, there are various polishing methods such as polishing with a honing machine that precisely polishes the inner diameter of a metal, and buffing using a buff.

  What is pear finish (http://www.chemicoat.co.jp/knowledge/detail_169.html), a surface that has a rough surface finish with fine irregularities on the metal surface. Say the processing method. It is called this because it resembles the surface of a fruit pear. There are glossy, semi-glossy and matte satin finishes. The purpose of the satin finish is not only to improve the surface roughness and appearance, but also to prevent slipping when touched with a hand with oil, and to pretreatment when painting or anodizing. It is used in. In addition, in order to reduce the frictional resistance of the surface of a precision machine or the like, there is a case where a process of giving fine irregularities called “scraping” is performed.

  There are mainly mechanical and chemical methods for finishing the metal surface. The mechanical methods include the wire brush method in which the surface is polished with a wire brush in addition to the above-described scraping, and fine particles are compressed. There are a sand blasting method in which the surface is sprayed with air and a liquid honing method in which a processing liquid containing fine particles is sprayed on the metal processing surface. As a method of chemically finishing the satin finish, there are a method using chemical corrosion such as etching and electrolytic corrosion, and a method of performing surface treatment using electroplating such as dispersion plating.

  The surface gloss / roughness measuring method has conventionally been a roughness meter, which includes a mechanical type and an optical type such as a laser. These are indicated by numerical values such as how many mm of unevenness or how many nm, but engineers who have experience in the field judge by looking at the glossiness and shine degree of the surface as seen from these values. Yes. In addition, work is proceeding with on-site judgment based on such experience. There are the following inventions in the prior art of surface roughness.

  The inventions of Patent Documents 1 to 3 that measure the surface roughness by a scanning method using a stylus or the like have been proposed.

  The inventions of Patent Documents 4 to 6 that measure the surface roughness in a non-contact manner with an imaging device have been proposed.

  The inventions of Patent Documents 7 to 10 that measure the surface roughness with a laser beam in a non-contact manner have been proposed.

JP 2014-32177 A JP 2014-81324 A JP2015-169641A JP 2011-31370 A Japanese Patent No. 4204879 JP 2011-69688 JP 2001-4351 Japanese Patent No. 2556556 JP-A-6-307840 JP-A-2015-143666

  However, Patent Documents 1 to 10 cannot perform the determination of the surface roughness indicating the finish such as gloss, gloss, and roughness. In Patent Documents 1 to 3, since the surface is scanned with a contact, it takes time to measure the surface, and the apparatus becomes large. Although patent documents 4-6 depend on visible light etc., accuracy is not enough. In Patent Documents 7 to 10, it takes time to scan the surface with laser light and measure the surface roughness.

  Normally, when looking at a glossy object color, the object color and the light source color are recognized in a well-balanced manner. When the light is totally reflected, the brightness increases, and a color that strongly reflects the light source color is seen. On the other hand, the shade portion has low luminance and is greatly affected by the reflection of the target object. The middle of the highlight and shade appears to the human eye as an additive color mixture.

  Even in the invention based on the image of the conventional method, those having structures such as wrinkles are not merely object colors, and ΔE, ΔL, Δa, Δb, etc. Since information is lost and only an average color in a very limited area can be obtained, texture information woven with various colors and brightness due to a complicated structure has been lost. The numerical values also varied depending on the measurement location.

  Therefore, in order to finish the surface, the surface is roughened or mirror-finished. This roughness adjustment requires an expert's evaluation.

  A fine surface has a micron-order roughness and a mirror surface has a nano-order roughness determination. However, since the human eye performs both determinations, there is a problem that it takes time and cost, and accurate determination is difficult. In particular, when the unevenness of the surface roughness is fine and fine, for example, regarding the roughness of the light wavelength from the nano level (for example, 700 nm) to the micrometer level, it is considered that there is a color shift, Although there is a demand for the determination of fine roughness, it has not been realized.

  For example, when a resin is coated on a metal surface, the surface is measured so that the bond between the metal and the resin is optimal, and then the coating is performed. It is impossible to check the roughness of such small orders on a production line with a conventional roughness meter, which is a bottleneck for technological development. This is because it takes time to trace the surface with a conventional roughness meter. For example, repeating the line scan on the surface with AFM or laser to scan and measure the entire area takes too much measurement time and the line stops. Also, the surface of the product often has a gentle curve, and many industrial products have a large measurement area, so even if you want to check the surface roughness, it is impossible to measure on the product. .

  Therefore, the present invention uses an XYZ imaging device to apply light from a light source to surface irregularities and measure the reflected light distribution. Depending on the shape of the surface, the light from the light source is scattered and diffracted to produce color / roughness. It is an object of the present invention to make it possible to quantify the surface finish determination in a state close to the determination condition of the human eye.

  In addition, the present invention makes it possible to determine the roughness of the roughness of the surface roughness from the nano level to the micro level, shorten the measurement time, determine the roughness of the curved surface, and rough the product with a large area. It is an object to accurately determine the height.

  The present invention is a determination method using an XYZ imaging device, and in this method, determination that is very close to a human visual determination method can be performed. This is based on the fact that the surface roughness of the inspection surface, such as irregularities, can be quantified by the scattering and diffraction phenomenon of the illumination light, because the imaging device requires an accurate value in two or three dimensions.

  With the XYZ-based imaging device of the present invention, when the surface to which the light source is applied is imaged and the reflected light distribution is measured, the light from the light source is observed as color and surface roughness due to scattering and diffraction, depending on the shape of the surface. We focused on the fact that the surface color and surface roughness can be quantified by this roughness.

  The first invention has three spectral sensitivities acquired by an imaging apparatus having three spectral sensitivities (S1 (λ), S2 (λ), S3 (λ)) linearly converted equivalently to the CIE XYZ color matching functions. An arithmetic processing unit that acquires and calculates data obtained by converting the three-band visual sensitivity images S1i, S2i, and S3i into tristimulus values X, Y, and Z in the CIE XYZ color system, and the arithmetic processing unit calculates a color difference ΔE. The XYZ color system by dividing the inspection area of the coordinates corresponding to the color space of the XYZ color system and the color space of the XYZ color system by integrating the number of pixels of the inspection surface and the reference surface belonging to each grid. A color space histogram distribution creating unit for creating a color space histogram distribution; a surface roughness index computing unit for computing a surface roughness index indicating a difference between two color space histogram distributions of the inspection surface and the reference surface; and a roughness meter Surface roughness meter measured by A surface roughness measurement data storage unit for storing the value Ra, a first calibration curve function indicating a correlation between the surface roughness measurement value Ra and the surface roughness evaluation index Est, and a correlation between the surface roughness measurement value Ra and the color difference ΔE. A function setting unit that sets at least one of a second calibration curve function indicating a relationship or a third calibration curve function indicating a correlation between the surface roughness measurement value Ra and the surface roughness index M; A surface roughness determination apparatus comprising: a calculation unit that calculates the surface roughness evaluation index Est or the surface roughness corresponding to the surface roughness index based on at least one of the functions It is.

  In the first invention, the surface roughness evaluation index Est is a first coefficient related to the surface roughness index and a second coefficient related to the color difference so that an error between the surface roughness measurement value Ra is minimized. It is characterized by specifying. For example, an analysis method that calculates the surface roughness evaluation index Est using the surface roughness index M and the color difference ΔE by multiple regression analysis is preferable.

  The second invention having technical relevance is obtained by an image pickup apparatus having three spectral sensitivities (S1 (λ), S2 (λ), S3 (λ)) linearly converted equivalently to the CIE XYZ color matching function. Three-band visual sensitivity images S1i, S2i, and S3i having three spectral sensitivities are obtained by calculating data obtained by converting tristimulus values X, Y, and Z in the CIE XYZ color system and calculating color difference ΔE. The color space of the XYZ color system is calculated by dividing the inspection region of coordinates corresponding to the color space of the XYZ color system with a grid and integrating the number of pixels of the test surface and the reference surface belonging to each grid. A color space histogram distribution creating step for creating a histogram distribution, a surface roughness index calculating step for calculating a surface roughness index indicating a difference between two color space histogram distributions of the inspection surface and the reference surface, and a roughness meter. A surface roughness measurement data storage step for storing the measured surface roughness measurement value Ra, a first calibration curve function indicating a correlation between the surface roughness measurement value Ra and the surface roughness evaluation index Est, the surface roughness measurement value At least one of a second calibration curve function indicating the correlation between Ra and the color difference ΔE or a third calibration curve function indicating the correlation between the surface roughness measurement value Ra and the surface roughness index M is set. And a function setting step. A surface roughness determination method comprising: a function setting step;

  In the second invention, the surface roughness evaluation index Est is a first coefficient related to the surface roughness index and a second coefficient related to the color difference so that an error between the surface roughness measurement value Ra is minimized. It is characterized by specifying.

The “imaging device” divides the reference surface and the inspection surface into three channels based on the three spectral sensitivities (S ₁ (λ), S ₂ (λ), and S ₃ (λ)). The means can be used regardless of whether the optical filter, dichroic mirror, dichroic prism or the like set in order to obtain these spectral sensitivities.

The spectral sensitivity (S ₁ (λ), S ₂ (λ), S ₃ (λ)) of the “imaging device” is a mountain shape having a single peak that has no negative value from the CIE XYZ spectral characteristics. Equivalent conversion is performed under the condition that the peak values of the spectral sensitivity curves are equal and the overlapping of the spectral sensitivity curves is minimized. The curve of the spectral characteristic S ₁ has a peak wavelength of 582 nm and a half-value width. The width is 523 to 629 nm, and the 1/10 width is 491 to 663 nm. Curve of the spectral characteristics _{S 2} is a peak wavelength of 543 nm, the half value width is 506~589nm, 1/10 width is 464～632Nm. Curve of the spectral characteristics _{S 3} is a peak wavelength of 446 nm, the half value width is 423~478nm, 1/10 width is 409～508Nm.

  “Surface” includes a resin surface, a metal surface, a ceramic surface, and the like. Examples of the resin surface include a textured surface, and examples of the metal surface include a blade surface, a mold surface, and a plating surface. The surface includes a curved surface in addition to a flat surface, and also includes a surface having a complicated shape.

  “Evaluation” means evaluating the roughness of the surface. The “index” can be in various forms, not limited to a plane and a solid, such as an index, a graph, a picture, or a combination thereof.

  “Roughness meter” includes atomic force microscope (AFM), laser roughness meter (for example, measurement value of nanometer order), contact mechanical roughness meter (for example, measurement value of micron order), etc. These measurement machines measure actual values. There is a non-contact type or contact type roughness meter of an atomic force microscope AFM (Atomic Force Microscope), which detects a force acting between atoms of a sample and a probe to obtain an image, and has high accuracy. For example, a scanning probe microscope (SPM) that can measure roughness on the order of microns or nanometers is exemplified. A measurement method is appropriately selected according to the roughness level of the object to be measured.

  The “first calibration curve function, second calibration curve function, third calibration curve function” may be either a linear function or a curved function.

  It is preferable to set one calibration curve for one reference plane. This is because the standard of a calibration curve such as a mechanical type or a laser type differs depending on the degree of unevenness on the reference surface.

  In the claims, the XYZ color system means a broad meaning including other CIE color systems, and means a narrow meaning except in the claims. The XYZ color system in the narrow sense is an RGB color system defined so that negative values do not appear by simple primary conversion, and other CIE color systems such as Yxy, XYZ, Lab, Luv, etc. This is a concept including a two-dimensional chromaticity diagram or a three-dimensional color space. Therefore, the XYZ color system in a broad sense includes a XYZ color system in a narrow sense and other CIE color systems developed from the XYZ color system.

  The XYZ color system in a narrow sense is defined by CIE in 1931 at the same time as the RGB color system so that negative values do not appear in the RGB color system by simple primary conversion.

  The xyY color system (also referred to as the Yxy color system) is defined in order to express an absolute hue from the XYZ color system because the relationship between numerical values and colors is difficult to understand in the XYZ color system. .

  The Luv color system is one of the uniform color spaces established by the CIE in 1976. CIE Luv is based on the wavelength of light and improves the uniformity of the wavelength interval of the xy chromaticity diagram of the XYZ color system. It is. In Japan, it is specified in JIS Z8518.

  The Lab color system is CIELab and is derived from the XYZ color system in order to measure a color difference due to a difference between perception and apparatus. In Japan, it is specified in JIS Z 8729. A calibration curve by the Lab color system may be effective for nano-order roughness.

  The XYZ color system in a broad sense includes a color space defined by two-dimensional coordinates and three-dimensional coordinates. Typical examples of the color space include configuration examples such as an xy color space, an XYZ color space, and a Lab color space. In the case of a two-dimensional color space, for example, in the case of a Yxy color space or a Luv color space, an xy chromaticity diagram that is a two-dimensional plane (xy chromaticity value (plane) normalized in the Yxy color space), uv chromaticity diagram , U′v ′ chromaticity diagram. An xy chromaticity histogram distribution or a Luv chromaticity histogram distribution expressed as the density of pixels in a two-dimensional chromaticity diagram on a plane corresponds. In the case of a three-dimensional color space, for example, in the case of an XYZ color space or a Lab color space, an XYZ color space or a Lab color space that is a three-dimensional space may be mentioned. An XYZ color space histogram expressed as a density of pixels in a three-dimensional color space, a Lab color space histogram distribution, or the like corresponds.

  In contrast to the separation of color and surface roughness in the two-dimensional xy chromaticity plane, uv chromaticity diagram, and u'v 'chromaticity diagram, other XYZ color spaces and Lab color spaces, etc. are in a three-dimensional color space. Color and surface roughness are separated. Therefore, the terms are defined separately from the xy chromaticity histogram, such as an XYZ color space histogram and a Lab color space histogram.

  The “XYZ color space histogram” and the “Lab color space histogram” are different from each other, and the surface roughness index calculation in the Lab color space is performed by converting the XYZ color data into Lab chromaticity data. It is calculated from the obtained data.

  Since the present invention can quantify the surface roughness with an XYZ imaging device, there is an advantage that a determination very close to a human visual determination method can be performed accurately and efficiently. The surface roughness evaluation time can be shortened. The roughness on the curved surface can be evaluated.

  The present invention has an advantage of being able to accurately and efficiently evaluate nano-order or micron-order roughness.

It is a block diagram of surface roughness judgment device 1 of Embodiment 1 of the present invention. It is a function which shows the spectral sensitivity of the camera 2 of the XYZ color system in Embodiment 1 of this invention. This is a specific example of a method for acquiring image information according to three spectral sensitivities (S1 (λ), S2 (λ), S3 (λ)) in Embodiment 1 of the present invention. (A) is explanatory drawing in the case of using a dichroic mirror. (B) is explanatory drawing in the case of using a filter turret. (C) is explanatory drawing at the time of attaching optical filter 22a, 22b, 22c to the image pick-up element 23 microscopically. It is a flowchart in the camera 2 of Embodiment 1 of the present invention. It is a flowchart in the arithmetic processing part 3 of Embodiment 1 of this invention. It is a sub chart in the arithmetic processing part 3 of Embodiment 1 of this invention. (A) is explanatory drawing which shows the test | inspection area | region T in the arithmetic processing unit 1 of Embodiment 1, (b) xy chromaticity diagram which shows the test | inspection area K on the chromaticity diagram corresponding to the test | inspection area T, (c). Is an explanatory diagram of an inspection region K partitioned by a lattice G, (d) is a schematic diagram showing the state of chromaticity overlapping on an xy two-dimensional chromaticity diagram, (e) is an explanatory diagram showing a minimum distribution, (f ) Is an explanatory diagram showing an example of an xy chromaticity histogram distribution. (A) is explanatory drawing which shows the roughness of the surface, (b) is an xy chromaticity histogram distribution diagram, (c) is a three-dimensional image diagram of the xy chromaticity histogram distribution. It is a solid and plane explanatory drawing of a surface roughness index. It is a flowchart (calculation of multiple regression analysis) in the arithmetic processing apparatus 103 of the surface roughness determination apparatus 101 of Embodiment 2 of this invention. It is a block diagram which shows the structure of the surface roughness determination apparatus 101 of Embodiment 2 of this invention. It is a flowchart (XYZ color space distribution) in the arithmetic processing unit 103 of the surface roughness determination apparatus 101 of Embodiment 2 of the present invention. It is a flowchart (Lab color space distribution) in the arithmetic processing unit 103 of the surface roughness determination apparatus 101 of Embodiment 2 of the present invention. It is a block diagram which shows the structure of the surface roughness determination apparatus 201 of Embodiment 3 of this invention. It is the graph 1 of the surface roughness evaluation index Est with respect to the measured value Ra of the roughness meter of the metal sample No. 1 of Example 1. It is the graph 2 of surface roughness measured value Ra with respect to color difference (DELTA) E of the metal sample No. 1 of Example 1. FIG. It is the graph 3 of surface roughness measured value Ra with respect to the surface roughness index | exponent M of the metal sample No. 1 of Example 1. FIG. It is graph 1 of surface roughness evaluation index Est with respect to surface roughness measurement value Ra of metal sample No. 2 of Example 1. It is the graph 3 of surface roughness measurement value Ra with respect to color difference (DELTA) E of the metal sample No. 2 of Example 1. FIG. It is the graph 4 of surface roughness measured value Ra with respect to the surface roughness index | exponent M of metal sample No21 of Example 1. FIG. It is explanatory drawing which shows the mode of reflection of the surface of resin sample No. 2 of Example 2. FIG. FIG. 4 is an explanatory photograph diagram showing a resin sample No. 2 in Example 2. It is explanatory photograph figure which shows the measurement range and measurement place (within image green frame) of resin sample No. 2 of Example 2. FIG. It is a three-dimensional view showing the measurement result of AFM of metal sample No. 1 of Example 3. It is a three-dimensional view showing the AFM measurement result of metal sample No. 2 in Example 3. It is a three-dimensional view showing the AFM measurement result of metal sample No. 3 of Example 3. It is a three-dimensional view showing the measurement result of AFM of metal sample No. 4 of Example 3. It is a three-dimensional view showing the AFM measurement result of metal sample No. 5 of Example 3. It is a three-dimensional view showing the AFM measurement result of metal sample No. 6 in Example 3. It is a three-dimensional view showing the measurement result of AFM of metal sample No. 7 of Example 3. It is the graph 1 which shows the relationship between the surface roughness index | exponent M of Example 3, and surface roughness measured value Ra. It is the graph 2 of the surface roughness evaluation index Est with respect to the surface roughness measured value Ra of the metal sample No. 1-7 of Example 3.

  A surface roughness determination apparatus 1 according to a preferred embodiment 1 of the present invention will be described with reference to FIGS.

  The surface roughness determination apparatus 1 includes a two-dimensional colorimeter 2 having three spectral sensitivities (S1 (λ), S2 (λ), and S3 (λ)) linearly converted equivalent to the CIE XYZ color matching function. An arithmetic processing unit 3 that acquires and calculates data obtained by converting three-band visual sensitivity images S1i, S2i, and S3i having three spectral sensitivities acquired from 3 into tristimulus values X, Y, and Z in the CIE XYZ color system; The arithmetic processing unit 3 divides the color difference calculation unit for calculating the color difference ΔE, and the inspection area of the coordinates corresponding to the color space of the XYZ color system with the grid G, and the inspection surface belonging to each grid G and the reference surface By integrating the number of pixels, a color space histogram distribution creating unit for creating a color space histogram distribution of the XYZ color system, and a surface roughness index M indicating a difference between the two color space histogram distributions of the inspection surface and the reference surface are obtained. Surface roughness index calculation unit to calculate A surface roughness measurement data storage unit for storing a surface roughness measurement value Ra measured by a roughness meter, a first calibration curve function L1 indicating a correlation between the surface roughness measurement value Ra and the surface roughness evaluation index Est, the surface At least of the second calibration curve function L2 indicating the correlation between the measured roughness value Ra and the color difference ΔE or the third calibration curve function L3 indicating the correlation between the measured surface roughness value Ra and the surface roughness index M. And a function setting unit for setting one. The surface roughness evaluation index Est specifies the first coefficient a related to the surface roughness index M and the second coefficient b related to the color difference ΔE so that the error ERROR with the surface roughness measurement value Ra is minimized. .

The spectral sensitivity of the two-dimensional colorimeter 2 satisfies the router condition, and the spectral sensitivities (S1 (λ), S2 (λ), S3 (λ)) are obtained from XYZ color matching functions as shown in FIG. This is a mountain shape having no negative value and having a single peak, and is equivalently converted from the condition that the peak values of the respective spectral sensitivity curves are equal and the overlapping of spectral sensitivity curves is minimized. Specifically, the spectral sensitivities (S1 (λ), S2 (λ), S3 (λ)) have the following characteristics.
Peak wavelength Half width 1/10 width S1 582nm 523-629nm 491-663nm
S2 543nm 506-589nm 464-632nm
S3 446 nm 423-478 nm 409-508 nm

  The peak wavelength of the spectral characteristic S1 can be handled as 580 ± 4 nm, the peak wavelength of the spectral characteristic S2 is 543 ± 3 nm, and the peak wavelength of the spectral characteristic S3 can be handled as 446 ± 7 nm.

The three spectral sensitivities (S1 (λ), S2 (λ), S3 (λ)) are obtained using the following Equation 1. For details on the spectral characteristics themselves, refer to Japanese Patent Application Laid-Open No. 2005-257827.

  The specifications of the two-dimensional color meter 2 are, for example, the two-dimensional color meter RC-500 of the limited company Paparabo, effective frequency value of about 5 million pixels, effective area 9.93mm × 8.7mm, image size 3.45μm × 3.45μm, video Output 12Bit, camera interface GigE, number of frames (during focus adjustment) 3-7 frames / sec, shutter speed 1 / 15,600sec to 1 / 15sec, integration time up to 3 seconds, S / N ratio 60dB or more, lens mount F mount, The operating temperature is 0 ° C to 40 ° C and the operating humidity is 20% to 80%.

  As shown in FIG. 1, the two-dimensional colorimeter 2 has a photographing lens 21, three optical filters 22a, 22b, and 22c arranged behind the photographing lens 21, and a rear of the optical filters 22a, 22b, and 22c. And an image pickup device 23 (CCD, CMOS, etc.) arranged. The three spectral sensitivities (S1 (λ), S2 (λ), S3 (λ)) of the two-dimensional colorimeter 2 are obtained by multiplying the spectral transmittances of the optical filters 22a, 22b, 22c and the spectral sensitivity of the image sensor 23. Given. The arrangement relationship between the optical filters 22a, 22b, and 22c and the image sensor 23 in FIG. 1 is merely schematically shown. Specific examples of the method of acquiring the three-band visual sensitivity images S1i, S2i, and S3i according to the three spectral sensitivities (S1 (λ), S2 (λ), and S3 (λ)) will be given below. Any of these methods can be adopted, and other methods can be adopted.

  FIG. 3A shows a system using a dichroic mirror. This is because light of a specific wavelength is reflected by the dichroic mirror 22c ′, and the remaining light that has been transmitted is further reflected by another dichroic mirror 22a ′ to be spectrally separated, and the image pickup devices 23a, 23b. , 23c are read in parallel. Here, the dichroic mirror 22a ′ corresponds to the optical filters 22a and 22b, and the dichroic mirror 22c ′ corresponds to the optical filter 22c. Light incident from the photographic lens 21 is reflected by the dichroic mirror 22c ′ according to the spectral sensitivity S3, and the remaining light is transmitted. The light reflected by the dichroic mirror 22c ′ is reflected by the reflecting mirror 26, and the spectral sensitivity S3 is obtained by the imaging device 23c. On the other hand, the light transmitted through the dichroic mirror 22c ′ is reflected by the dichroic mirror 22a ′, and the light according to the spectral sensitivity S1 is reflected, and the remaining light according to the spectral sensitivity S2 is transmitted. Therefore, the light is captured by the image sensor 23a and the image sensor 23b, respectively. To obtain the spectral sensitivities S1 and S2. Instead of the dichroic mirror, a dichroic prism having the same characteristics may be used to split the light into three, and the image sensors 23a, 23b, and 23c may be bonded to the positions where each light is transmitted.

  The system shown in FIG. 3B uses a filter turret 27. Optical filters 22a, 22b, and 22c are provided on a filter turret 27 having the same direction as the incident light from the photographing lens 21 as a rotation axis, and these are mechanically rotated. S1, S2, and S3 are obtained.

  FIG. 3C shows a system in which the optical filters 22 a, 22 b, and 22 c are microscopically attached to the image sensor 23. The optical filters 22a, 22b, and 22c on the image sensor 23 are provided in a Bayer array type. In this arrangement, the optical filter 22b is provided in half of the area on the image sensor 23 divided into a grid, and the optical filter 22a and the optical filter 22c are equally arranged in the remaining half of the area. That is, the arrangement amount is optical filter 22a: optical filter 22b: optical filter 22c = 1: 2: 1. The arrangement of the optical filters 22a, 22b, and 22c other than the Bayer arrangement is not particularly hindered in the first embodiment. Each of the optical filters 22a, 22b, and 22c is very fine and is attached to the image sensor 23 by printing. However, in the present invention, this arrangement is not meaningful, but a filter having characteristics of spectral sensitivity (S1 (λ), S2 (λ), S3 (λ)) is attached to the image sensor.

  The two-dimensional colorimeter 2 transmits image information acquired by spectral sensitivity (S1 (λ), S2 (λ), S3 (λ)) to the arithmetic processing unit 3, and the arithmetic processing unit 3 uses the tristimulus in the XYZ color system. A display device 7 is provided that converts the values to X, Y, and Z, performs arithmetic processing by conversion processing on the acquired image data of the tristimulus values X, Y, and Z, and displays the visualized image.

  The arithmetic processing unit 3 calculates the degree of roughness, color difference and the like at an arbitrary position of the image acquired by the two-dimensional colorimeter 2 and performs a visualization process. Illumination is applied directly above or obliquely from the surface 5, and xy, XYZ, or Lab chromaticity distribution data are compared and indexed.

  The surface 5 is usually imaged at one place with the two-dimensional colorimeter 2, and the two-dimensional colorimeter 2 moves and images at another angle as necessary. Here, for example, it is possible to shoot at three locations (an appropriate number of locations may be used) at 45 degrees on the front and left and right.

  The illumination source of the illumination unit 6 employs a xenon lamp (pseudo sunlight). The illumination unit 6 includes a Fresnel lens assembly in addition to the xenon lamp. The xenon lamp is illuminated uniformly from above the surface 5. In addition to xenon lamps, LED artificial solar lights may be used.

  The display device 7 is connected to the arithmetic processing unit 3, receives an image signal processed by the arithmetic processing unit 3, and displays an image on the screen. The arithmetic processing unit 3 or the display device 7 includes input means (not shown) and the like as appropriate. The input means is a keyboard, a mouse, a touch panel provided on the display, or the like.

  The operation of the surface roughness determination apparatus 1 on the surface 5 will be described with a specific example. The surface roughness determination device 1 of the surface 5 operates by connecting a two-dimensional colorimeter 2, an arithmetic processing unit 3, and a display device 7 as shown in FIG. The connection method can be selected regardless of wired or wireless. A flowchart in the two-dimensional colorimeter 2 is shown in FIG. 4, and a flowchart in the arithmetic processing unit 3 is shown in FIG.

  When the power of the two-dimensional colorimeter 2 is turned on, initialization is performed as shown in FIG. 4 (initialization S1). Next, the surface 5 is imaged (imaging process S2), and then the captured 3-band visual sensitivity images S1i, S2i, and S3i are input by the image sensor 23 (input process S3), and the arithmetic processing unit 3 performs tristimulus. Conversion into values X, Y, and Z (conversion processing S4). Spectral sensitivities (S1 (λ), S2 (λ), S3 (λ)) are transmitted to the display device 7 (data transmission S5). When the image is a moving image, a series of processing from imaging processing S2 to data transmission S5 is continuously performed. The image is displayed on the display device 7.

  In the imaging process S2, an example in which various surfaces 5 are measured is given, but the surface 5 is imaged by the two-dimensional colorimeter 2 at different angles for specific regions with different imaging positions. There are a plurality of imaging locations, and the number can be selected as appropriate. Here, the measurement is made from three directions of front (0 degree), left 45 degrees, and right 45 degrees. Further, the measurement location is such that the optical axis of 0 degree of the two-dimensional colorimeter 2 is perpendicular to the surface 5. In addition, the illumination is characterized by illumination from obliquely above as with sunlight.

  Expressions for converting the tristimulus values X, Y, Z to the Y′xy color system are shown in Formulas 2 and 3, respectively. Here, a luminance meter (not shown) was used together with the two-dimensional colorimeter 2, and Y was calibrated by the luminance meter value (nt) to be Y ′. Since color space conversion formulas are commonly used, other detailed formulas are omitted.

The XYZ color system is currently the basis of each color system as the CIE standard color system. It was developed based on the principle of additive color mixing of the three primary colors of light (R = red, G = green, B = blue-violet), and the color is represented by three values of Yxy using a chromaticity diagram. Y corresponds to lightness by reflectance, and xy becomes chromaticity.

  The imaging process S2 is a process of imaging the surface 5 with the two-dimensional colorimeter 2 having three spectral sensitivities (S1 (λ), S2 (λ), S3 (λ)) (see FIGS. 1 and 4). Spectral sensitivities (S1 (λ), S2 (λ), S3 (λ)) are given according to Equation 1 above. Input processing S3 is continuously performed at the same time as imaging is performed by the imaging lens 21, the optical filters 22a, 22b, and 22c, and the imaging device 23.

  Since the input image data is a value according to the spectral sensitivity (S1 (λ), S2 (λ), S3 (λ)), the image data is captured by the conversion processing S4 in the arithmetic processing unit 3 of the two-dimensional colorimeter 2. The image data of the obtained image is converted into tristimulus values X, Y, and Z. This conversion is performed according to Equation 1. That is, the tristimulus values X, Y, and Z can be obtained by multiplying the inverse matrix of the coefficients in Equation 1. Note that the two-dimensional colorimeter 2 transmits the values in accordance with the spectral sensitivities (S1 (λ), S2 (λ), S3 (λ)) to the arithmetic processing unit 3.

  When the arithmetic processing unit 3 is turned on, initialization is performed as shown in FIG. 5 (initialization S110). The display device 7 receives the three-band visual sensitivity images S1i, S2i, S3i transmitted from the two-dimensional colorimeter 2 in a state where it is connected to the two-dimensional colorimeter 2 (data reception S120). Thereafter, the three-band visual sensitivity images S1i, S2i, and S3i are converted into tristimulus values X, Y, and Z, and the Lab average value and the xy surface roughness index M of the images of the imaged reference surface and inspection surface are calculated and visually If necessary, the color information is converted to RGB or the like (S140). The contents are transmitted to the display device 7 (display process S150). After the data reception S120 from the two-dimensional colorimeter 2, a series of processes from the conversion process S130 to the display process S150 are continuously performed.

  The display process S150 is a process of displaying the visualized surface roughness index M on the display device 7, and the process returns.

  A sub-flowchart of S140 in FIG. 6 will be described. The first image (image B) of the reference surface is imaged, the second image (image A) of the inspection surface to be compared next is imaged, and the surface roughness index M is sequentially calculated as follows. The similarity of the surface roughness is determined by the surface roughness index M obtained by separating the surface roughness.

  An inspection region K (see FIG. 7B) corresponding to a region T (see FIG. 7A) to be inspected for the captured images A and B is set (step S141). Size and place can be set freely.

  The chromaticity xy is calculated (S142).

  An xy chromaticity histogram distribution of the inspection surface of the inspection region K cut out from the image A of the inspection surface that has been imaged is created (S143). This chromaticity histogram distribution is an integrated number obtained by counting pixels belonging to the overlapping region D of the two histogram distributions shown in FIG.

  The xy chromaticity histogram distribution is a three-dimensional histogram showing the cumulative number of pixels belonging to each unit grid G, and an overlapping region D is shown in FIG.

  As shown in FIG. 7C, the color distribution of the comparison target at the position of the xy coordinates is written in a plane, the inspection area K is partitioned by the grid G, and the pixels having the xy value of the section are integrated. Then, a histogram distribution with the z axis is created. Let xy coordinates be a grid (solid grid) of a specific width, for example, a plane lattice obtained by cutting xy by 1/1000 (1000 lines). The histogram is scanned from end to end, and for each region partitioned into the grid G, the number of pixels belonging to this is scanned on the same xy plane and integrated in the z direction. Further, if only the specific range of the xy coordinates is calculated in the inspection region K, the calculation time can be shortened. If the grid cells are made finer, the accuracy will be improved, but the calculation time will be longer.

  Similarly to S143, the xy chromaticity histogram distribution of the image B on the reference plane is created (S144). In the xy chromaticity histogram distribution, the xy axis is xy chromaticity, the z axis is the integrated number of pixels, and a planar overlapping region D is shown in FIG.

For the Lab's a-axis, b-axis, and L-axis, the sum of all the pixels in the inspection region K is taken independently, and the sum of the L value, a value, and b value is divided by the number of pixels, and Lab Calculate the average L value, average a value, and average b value of the chromaticity distribution. Further, ΔE is calculated by Equation 4 (S145).

The Lab value of the Lab space converted by the following Equation 5 is calculated. The Lab color space is a kind of complementary color space, and has a dimension L indicating lightness and complementary color dimensions A and B, and is based on nonlinearly compressed coordinates of the CIEXYZ color space. The XYZ value before normalization is converted into the Lab value by Equation 5. In contrast to the distribution in the XYZ color space, the distribution in the Lab color space can be obtained in consideration of the brightness direction.

In Equation 5, the values of X, Y, and Z in the parentheses of the function f are divided by the coordinates X _n , Y _n , and Z _n of the white point, respectively, in order to make the maximum value equal to 1.

  The difference between the average values of the reference surface and the inspection surface is taken as a material for judging the difference in color.

  The surface roughness index M is calculated (S146). Thereby, the degree of roughness can be simply separated and determined, and this can be quantified. Roughness is calculated in the two-dimensional space of the xy chromaticity distribution, and the difference in the roughness can be grasped as the difference in roughness, excluding the color. Can be detected separately.

The surface roughness index M is calculated by the following formula. The xy chromaticity histogram distribution is the cumulative number of pixels, and FIG. 7D shows the overlapping region D and FIG. 7E shows the minimum distribution.
Surface roughness index M = total number of pixels belonging to overlapping region D / total number of pixels in inspection region K × 100 (%)

  The roughness histogram in the two-dimensional space of the reference surface and the inspection surface is calculated, and the minimum value at the same position in the array is the overlap frequency. Therefore, this value is the total histogram total count. Calculated by dividing.

FIGS. 7D and 7E are cross-sectional views taken along the line SS of FIG. 7C, and there is an overlap when viewed on the same line in the xy coordinates. Instead of drawing in three dimensions, it is drawn in a plane for convenience. Moreover, since it is a histogram, it has a minute staircase distribution. Cumulative number _{H 1} and the accumulated number of _{H 2} FIG. 7 (d) respectively image A, corresponding to the image B. When the two histogram distributions are compared, an overlapping region D exists.

As shown in FIG. 7E, H ₁ (x ₁ , y ₁ ) is the cumulative number of xy chromaticity histogram distributions on the reference plane, and H ₂ (x ₂ , y ₂ ) is the xy chromaticity histogram distribution on the reference plane. When the accumulated number, in the overlapping left area _H 1> _{H 2,} centrally _H 1 = _{H 2,} and the on the right side is a _H 1 _{<H 2.} If the smaller integrated number (pixel frequency) of H ₁ and H ₂ is taken, it becomes H _{1 on} the left side and H ₂ on the right side, and a minimum distribution which is a step-like histogram curve can be specified. Using this, the ratio of the overlapping area D to the entire area in two dimensions or three dimensions can be calculated.

The smaller integrated value is specified by this minimum distribution. If the smaller integrated number of H ₁ and H ₂ is added and calculated, the integrated number of the overlapping region D can be calculated, and the ratio to the total number of pixels can be specified. The total number of pixels in the inspection area K is determined, and the total number of pixels is the same for both the inspection surface and the reference surface. The calculation of this ratio may be integrated three-dimensionally for all the lattices G. For example, as shown in FIG. 7C, the inspection region K is cut along the SS axis, and y is a predetermined value. The distribution of the pixel accumulation numbers H ₁ and H ₂ when x changes from end to end is two-dimensionally integrated. FIG. 7F is a two-dimensional map on the xy coordinates of the integration result. If there is no distribution in the inspection region K and the number of pixels is zero, it is excluded from the calculation.

  Finally, display / save processing and transmission processing are performed (S147), and the processing returns.

For example, the pixels belonging to the inspection region K are assumed to be 100 vertical pixels × 100 horizontal pixels = 10,000 pixels. Since the image is cut out in the same inspection region K, the total number of pixels of the image A and the image B is 10,000 pixels. From the xy chromaticity histogram, when the number of pixels in the overlapping region D is integrated and the integrated numbers H ₁ and H ₂ are 5,000, the surface roughness index M is 50%. The difference in surface roughness increases as the surface roughness index M falls below 100%. If the distribution of the xy values is completely coincident, 100% is obtained. Thereby, when it determines with it being a numerical value beyond a fixed value, it can determine with it being a conformity surface about surface roughness.

  For the image, the color information obtained primarily is three spectral sensitivities (S1 (λ), S2 (λ), S3 (λ)) based on a function equivalent to the XYZ color matching function, and is acquired by RGB. Compared to the case, it is more accurate and sensitive to the sensitivity of the human eye. Since the overlap of spectral sensitivities (S1 (λ), S2 (λ), S3 (λ)) is small, the S / N ratio is sufficient, and the curve in the spectral sensitivity curve changes naturally, so the error in colorimetry is Minimized.

  Since the surface roughness of the image can be grasped by the histogram distribution separately from the color, the difference in roughness of the surface 5 can be accurately determined.

For example, as shown in FIGS. 8A to 8C, an example will be described in which three types are examined from those having a small roughness to those having a large roughness. A surface having a small roughness is defined as a reference surface 1, a medium roughness is defined as an inspection surface 2, and a surface having a large roughness is defined as an inspection surface 3. First, when a distribution on the xyz chromaticity diagram after performing the above processes 1 to 3 is created, as shown in the xy chromaticity diagram of FIG. . The accumulated numbers H ₁ and H ₂ are shown in light and dark, and the brighter the color, the larger the accumulated number. FIG. 8C schematically shows the integrated numbers H ₁ and H ₂ in three dimensions of the reference surface and the inspection surface. The xy axis is chromaticity, and the z axis is the integration numbers H ₁ and H ₂ . Basically, the weaker the unevenness, the lower the mountain shape, and the stronger the unevenness, the sharper the mountain shape. By comparing the two histogram distributions for the reference surface 1 and the inspection surface 2 or 3, a surface roughness index M indicating the degree of overlap is calculated. The unevenness feeling of micron or nano-order or higher can be determined by the surface roughness index M.

  In the present embodiment, color measurement is performed using a surface instead of a point, and not only an average value but also a statistical comparison can be made. Roughness information in which color information is woven intricately can be digitized. Each pixel is a colorimetric value. It is statistically processed by the histogram.

  The surface roughness index M is a numerical comparison of the surface roughness such as wrinkles, and a colorimetric image is taken with a two-dimensional colorimeter. Actually, since there is frequency information, as shown in the upper part of FIG. 9, a mountain is formed in the three-dimensional chromaticity diagram.

  In FIG. 9, (1) is a sharp mountain concentrated at one point, (2) a normal mountain, and (3) is a wide and gentle mountain. Looking at the mountain from the side, it looks like the bottom in Fig. 9. When (2) and (3) are superimposed on the basis of (1), the colored portions are the overlapping portions. The volume ratio here is the surface roughness index M.

  When the surface roughness is nano-order or micron order, the surface roughness index M and the color difference ΔE calculated as described above are used to analyze the surface roughness as shown in FIG. Est is calculated. This is because the light scattering on the surface 5 is different from the case of the roughness of nano and micron order from the case of the roughness being large. For fine roughness, a comprehensive action with the light to be applied, that is, interference occurs, and various shifts occur. Therefore, it is more accurate to step in positively. Therefore, the color shift with respect to the surface roughness index M and the color difference ΔE is positively utilized, and a correlation with the roughness meter is obtained by multiple regression analysis.

  First, when this process is started, the measurement value of the roughness meter for the determination target is acquired (S150). Here, the surface roughness measurement value Ra is obtained by a contact-type roughness meter.

The first coefficient a and the second coefficient b are specified by the calculation formula of the following mathematical expression 6 of the multiple regression analysis so that the total error Error is minimized by the least square method using Est and Ra (S151).
(Equation 6)
Est = a · M + b · ΔE
In Equation 6, Est is a surface roughness evaluation index, a is a first coefficient, M is a surface roughness index, b is a second coefficient, ΔE is a color difference, and Ra is a value measured by a roughness meter.

  The surface roughness evaluation index Est is calculated by substituting the measured values of M and ΔE into Equation 5 in which a and b are specified (S152).

In some cases, instead of Equation 6, Equation 7 may be used. This is a simpler method that does not use the parameter of the color difference ΔE, and the first coefficient is set so that the sum of the errors Error is minimized by the least square method using Est and Ra by the calculation formula of the following mathematical expression 7 of the multiple regression analysis. a, the second coefficient b is specified.
(Equation 7)
Est = a · M + b
In Equation 7, Est is a surface roughness evaluation index, a is a first coefficient, M is a surface roughness index, b is a second coefficient, and Ra is a value measured by a roughness meter.

  The surface roughness evaluation index Est is calculated by substituting the measured values of M and ΔE into Equation 5 in which a and b are specified (S152).

  From the surface roughness index M, color difference ΔE, surface roughness evaluation index Est, and surface roughness measurement value Ra measured and calculated as described above, the first calibration curve function L1, the second calibration curve function L2, and the third calibration curve function. At least one of L3 is set. Although all are set here, the function with few errors of these, or the combination thereof can be used. If it is desired to determine the degree of roughness of the surface 5 to be determined, it is possible to image the surface 5 to be determined and set each function from the calculated surface roughness index M and color difference ΔE. The surface roughness evaluation index Est ′ and the surface roughness measurement value Ra ′ (not the actual measurement value but the estimated calculation value) can be obtained. Thus, by accurately evaluating the surface roughness, it is possible to appropriately compare and determine the surface roughness with respect to the reference surface, thereby preventing the occurrence of defective products. Details will be described in Examples 1 to 3.

  Next, a surface roughness determination apparatus 101 for the surface 5 according to the second embodiment will be described with reference to FIGS. 11, 12, and 13. For the corresponding similar elements, the explanation is cited as 100 series, and the difference is mainly explained.

  Arithmetic processing unit 103 that receives a signal by connecting to a two-dimensional colorimeter 102 that images the reference surface and the inspection surface via the switch 109 and the two-dimensional colorimeter 102, and calculates the surface roughness index M; 103 and a display device 107 that is connected to 103 and displays an index.

  As illustrated in FIG. 11, the arithmetic processing device 103 includes a calculation unit 103 </ b> A that calculates a stimulus value XYZ <b> 1 acquired by imaging a reference plane, and a stimulus acquired by imaging the surface 105 of a product that is an inspection plane. A calculation unit 103B that calculates the value XYZ2, a calculation unit 103A, a calculation unit 103B, a calculation unit 103C that calculates the surface roughness index M of the surface 105, and an OK signal or an NG signal from the calculation unit 103C are displayed. It transmits to 107 or transmits to the outside. Note that the switch 109 selectively inputs the stimulus value XYZ1 and the stimulus value XYZ2.

FIG. 12 is a flowchart for calculating the surface roughness index M by comparing the chromaticity histogram distributions from the two images A and B. As shown in FIG. 12, when the program is started, the inspection area K on the inspection surface is cut out from the image A, specified, and set (S201). Next, similarly to the image A from the image B, the inspection area K on the reference plane is cut out, specified, and set (S202). The chromaticity values XYZ are calculated from the images A and B (S203). In the inspection region K, the XYZ chromaticity histogram distributions of the inspection surface and the reference surface are calculated and created (S204). An average value of the XYZ values is calculated (S205). The minimum distribution of the XYZ chromaticity histogram distribution is specified, and the integrated number of the XYZ chromaticity histogram distribution in the overlapping region D is calculated (S206). Surface roughness index M = (total number of pixels belonging to overlapping region D / total number of pixels in inspection region K) × 100 (%). The cumulative number in the overlap region D is calculated by adding the smaller cumulative number of T ₁ and T ₂ . The surface roughness index M is calculated (S207), and the process returns.

In the case of the calculation of the XYZ distribution corresponding to the inspection region K, the calculation of the surface roughness index M is performed by the distribution in the three-dimensional space of the X axis, the Y axis, and the Z axis. The XYZ values in the XYZ space coordinates on the inspection surface and the reference surface are T ₁ (L, a, b) and T ₂ (L, a, b), respectively. In the XYZ color space, the histogram distribution has a globe-like shape, and the two histogram distributions may be three-dimensionally overlapped or separated. The inspection area K in the three-dimensional space is partitioned by the grid G, and the chromaticity histogram distribution and the minimum distribution of T ₁ (X, Y, Z) and T ₂ (X, Y, Z) in three dimensions are obtained. Calculate exponents. The integration numbers H ₁ and H ₂ of the grid G may be projected on a plane, and the integration number of the overlapping region D on the grid G may be calculated by the same integration within the plane. In the case of XYZ chromaticity, since there is no brightness information, the histogram distribution does not change in the XYZ space even if the brightness of the image changes.

  When the Lab color space histogram is used for the surface roughness determination instead of the XYZ color space histogram, the flowchart of FIG. 13 is used. The description of FIG. 13 is based on the above description of FIG. In S305, the calculation of the average Lab value of the region and the average Lab value of the inspection region K of the image B are performed. In the case of Lab chromaticity, since there is brightness information, the histogram distribution changes when the brightness of the images A and B changes in the Lab space.

  Next, the surface roughness determination apparatus 201 for the surface 205 according to the third embodiment will be described with reference to FIG. About the corresponding similar element, description is used as 200 series, and a difference is mainly demonstrated.

  As shown in FIG. 14, the color / roughness determination target is a partial region of the surface 205, and the two-dimensional colorimeter 202 images the inspection region K of the surface 205. The arithmetic processing unit 203 is connected to the arithmetic unit 203A that calculates Lab from the reference stimulus value XYZ1, the arithmetic unit 203B that calculates Lab from the stimulus value XYZ2 to be determined, the arithmetic unit 203A, and the arithmetic unit 203B. A computing unit 203C that computes an average value, and a surface roughness index M computing unit 203D that computes a surface roughness index M from the reference Lab and the target Lab, and a roughness processing device that calculates the computation values from the computing units 203C and 203D. 250. According to the index value, whether or not the surface has an appropriate surface roughness is determined by looking at the screen, and further the roughness process is performed. Note that the switch 209 selectively inputs the reference Lab chromaticity and the target Lab chromaticity. Since the main processing is generally the same as the flowcharts of the first and second embodiments, the description is incorporated.

In the case of calculation of the chromaticity histogram distribution in the Lab space corresponding to the inspection area K, the XYZ value is converted to Lab. The exponent is calculated by the distribution in the three-dimensional space of the L axis, a axis, and b axis. The Lab chromaticity distribution is a solid elliptical shape. Lab values in Lab space coordinates on the inspection surface and the reference surface are U ₁ (L, a, b) and U ₂ (L, a, b), respectively. In the Lab color space, the histogram distribution has a globe-like shape, and the two histogram distributions may be three-dimensionally overlapped or separated. The inspection area K in the three-dimensional space is partitioned by the grid G, and the chromaticity histogram distribution and minimum distribution of U ₁ (L, a, b) and U ₂ (L, a, b) in three dimensions are obtained. Calculate exponents. The integrated numbers H ₁ and H ₂ of the grid G are projected on a plane, and the integrated number of the overlapping region D on the grid G is calculated by the same integration within the plane. In the case of Lab chromaticity, since there is brightness information, when the brightness of the images A and B changes in the Lab space, the L value changes, and the distributions U ₁ and U _{2 of the} matching degrees are in the Lab space. Since the position is shifted, the determination in consideration of light and dark is possible. This is because if the brightness of the images A and B is different, the position of the distribution is shifted. For example, the Lab chromaticity histogram distribution shifts downward when dark and shifts upward when bright.

  Other application examples will be described. The two images A and B acquired on the reference surface and the inspection surface are overlaid, and the respective chromaticity histogram distributions are displayed on the display device 7, and the respective chromaticity histogram distributions are displayed on one chromaticity diagram. Overlapped chromaticity diagrams can be displayed, color differences can be determined by average Lab value, while surface roughness index M calculation indicating the roughness of surface 5 can be separated and displayed as a percentage. As a result, it is possible to reliably confirm the concavo-convex sensation and the roughness sensation indicated by the deviation of the spatial spread of the chromaticity distribution of the reference surface of the chromaticity distribution of the inspection surface with numerical values. Inspection results are displayed numerically for each region K. The grid width of the grid G can be adjusted. An exponent threshold value can be arbitrarily set. The measurement results and the captured images A and B can be saved. It is possible to reduce the problems of individual differences that cannot be avoided by visual inspection, troubles in judgment standards with customers, etc., and to standardize the degree of finish of surface roughness and perform stable surface roughness management.

Although the present embodiment has been described above, the following effects are obtained. (1) Average L value, Average a value, Average b value, and (2) Two H ₁ (x, y), H ₂ (x, y), T ₁ (X, Y, Z), T ₂ Examples of the surface roughness index M regarding (X, Y, Z), U ₁ (L, a, b), and U ₂ (L, a, b) are given. The difference between color and surface roughness By separating and presenting, it is possible to perform accurate and quick evaluation, and by adjusting the surface roughness, it is possible to give an appropriate guideline for the finish.

  In addition, the height of the unevenness of the surface roughness enables determination of fine roughness from the nano level to the micrometer level, shortening measurement time, determining curved surface roughness, It is possible to accurately determine the roughness of a large product.

  In order to evaluate the roughness of the surface 5 by the surface roughness determination apparatus 1 of the present invention, measurement and evaluation were performed as in Example 1 and Example 2. The present invention will be specifically described, but the present invention is not limited thereto. In addition, the measurement and evaluation of the characteristic value in an Example were performed as follows.

(1) Judgment device The PPLB-200 from Paparabo Co., Ltd. was used. PPLB-200 is equipped with a two-dimensional color meter RC-500. Lighting was Panasonic D50 lighting.

(2) Shooting
The PPLB-200 was shot in a dark room. The two-dimensional colorimeter was a still image type, and the measurement was performed with the L value of the white plate set to 100.
(3) Measurement range Regarding the measurement range of the sample, Examples 1, 2, and 3 are the entire images A and B.

(4) About a roughness meter A FORM TALITYSURF (registered trademark) -120 (AFM contact type) roughness meter which is a roughness meter manufactured by Taylor Hobson, and a non-contact type roughness meter of Keyence VN-8010. Using.

(5) Measurement items and results The acceptance reference surface of the sample and the test sample were measured. The difference between the surface roughness index, ΔE, and average Lab of the photographed reference / inspection sample images A and B was determined. Of the degree of coincidence, xy-3D is a value obtained without shifting the center coordinates of the histogram distribution. Difference of average Lab value = average Lab value of test sample−average Lab value of reference sample. From the measurement result, the sample whose color and the surface roughness of the surface 5 are most similar to the reference sample was identified from the result of the surface roughness index and the numerical value of ΔE. For ΔE, ΔE00 was used in consideration of human visibility characteristics.

  The surface roughness indices M, xy3D, Lab, ΔL, Δa, Δb, and ΔE of samples 1 to 3 were obtained. Of the surface roughness indices, the surface roughness index M is an index calculated without shifting the center coordinates of the respective xy3D histogram distributions. Lab is the surface roughness index in the Lab coordinate, xy3D is the surface roughness index in the xy coordinate, ΔL = difference in average Lab value = average Lab value of the inspection sample−average Lab value of the reference sample. The same applies to ΔE, Δa, and Δb. From the measurement results, the sample 2 or 3 having the most similar color and surface roughness to the sample 1 was identified from the result of the surface roughness index and the numerical value of ΔE. For ΔE, ΔE00 was used in consideration of human visibility characteristics.

(6) Multiple regression analysis Using mathematical expressions 6 and 7 of multiple regression analysis, a and b were set so as to minimize the error by the method of least squares.

  In Example 1, multiple regression analysis was performed in order to evaluate the roughness on the micron order. In Example 2, the same analysis was performed but the results were omitted. In Example 3, the roughness on the nano order was evaluated. Therefore, a multiple regression analysis was performed.

  This is an evaluation of the roughness of the aluminum alloy of Example 1 treated with chemical etching.

Table 1 and FIG. 1 shows the measurement / calculation results. In FIG. 15, the horizontal axis represents the measured surface roughness Ra (unit: μ), the vertical axis represents the surface roughness evaluation index Est, and the straight line indicated by the dotted line is the first calibration curve L1. In FIG. 16, the horizontal axis indicates the color difference ΔE, the vertical axis indicates the measured surface roughness Ra (unit: μ), and the straight line indicated by the dotted line is the second calibration curve L2. In FIG. 17, the horizontal axis represents the surface roughness index M, the vertical axis represents the surface roughness measurement value Ra (unit is μ), and the straight line indicated by the dotted line is the third calibration curve L3.

  a = 0 and b = 0.105242 were obtained. The correlation coefficient between the surface roughness evaluation index Est and the surface roughness measurement value Ra is 0.887104, the correlation coefficient between the color difference ΔE and the surface roughness measurement value Ra is 0.887104, the surface roughness index M and the surface roughness measurement. The correlation coefficient of the value Ra is -0.8377, and the total error TotEro is 5.9516.

  Sample No. 2 in Example 2 2 is an aluminum alloy subjected to chemical etching.

Table 2 and FIG. The calculation result by measurement of 2 and Formula 6 etc. is shown. In FIG. 18, the horizontal axis represents the measured surface roughness Ra (unit: μ), the vertical axis represents the surface roughness evaluation index Est, and the straight line indicated by the dotted line is the first calibration curve L1. In FIG. 19, the horizontal axis indicates the color difference ΔE, the vertical axis indicates the surface roughness measurement value Ra (unit: μ), and the straight line indicated by the dotted line is the second calibration curve L2. In FIG. 20, the horizontal axis represents the surface roughness index M, the vertical axis represents the measured surface roughness Ra (unit: μ), and the straight line indicated by the dotted line is the third calibration curve L3.

  a = 0.008995 and b = 0.694288 were obtained. The correlation coefficient between the surface roughness evaluation index Est and the surface roughness measurement value Ra is 0.849651, the correlation coefficient between the color difference ΔE and the surface roughness measurement value Ra is 0.849774, the surface roughness index M and the surface roughness measurement. The correlation coefficient of the value Ra is −0.88685, and the total error TotEro is 11.60513. In FIG. 18, there is Est = 8.3352787 that deviates upward by one point, but if it is removed as a measurement error, the correlation coefficient is approximately 0.9.

  In Tables 1 and 2, when the etching time was increased from No. 21 to No. 29 and No. 31 to No. 39, the surface roughness index was compared with that in which the aluminum alloy was not chemically etched. As M decreases, ΔE, Ra, and Est increase.

  The roughness of aluminum alloy by chemical etching is that the surface 5 is flat at first, and the surface 5 becomes rough when etching is performed, and when etching is further performed, a hole is dug in the surface 5 or a dent is formed. In addition, there are various forms of roughness due to the formation of cavities therein, but it can be accurately evaluated.

  According to Example 1, the first calibration curves L1 to L3 can accurately calculate the micron-order or nano-order roughness, can grasp a proper unevenness, and can reduce defective products by comparing with a threshold value. . For example, when a resin layer is coated on an aluminum alloy that has been subjected to chemical etching, the resin layer is likely to peel from the aluminum alloy due to excessive or insufficient etching. Before coating the resin on the aluminum alloy, the surface 5 of the aluminum alloy is inspected, and the aluminum alloy and the resin layer are not peeled off. Therefore, the resin layer can be prevented from peeling off, and the industrial utility value is high.

  For sample No. 3 which is a resin part of Example 2, the measurement location, coincidence, ΔE00, ΔL, Δa, Δb values, etc. are shown. Although the multiple regression roughness analysis was performed in the same manner as in Example 1, the same results were obtained by the same method as in Example 1, and the description is omitted. The measurement ranges are within the frame lines of the images A and B, respectively.

  Sample No. 2 shown in FIG. 3 performed about quantification of the surface roughness by the difference in the embossing of two types of resin P1 and P2. The illumination color temperature is 5000K.

  Sample No. The evaluation result of 3 is shown in FIG. Differences between Lab and ΔE were detected. The surface roughness index M (roughness coincidence) is 85%. The three-dimensional xy coincidence xy3D is 93%. Based on P1, P2 had a Lab coincidence of 26%, and the surface roughness could be clearly distinguished. The color difference ΔE is 1.163, ΔL is 1.532, Δa is −0.017, and Δb is 0.108. It can be seen that the L value of S2 is high and bright. The difference in material color can be seen from Δa and Δb.

  As shown in FIG. 21, the difference in “appearance” due to the difference in wrinkle feeling shows the difference in reflection due to the difference in the surface shape of the wrinkle. The total reflection portion reflects the color of the illumination strongly, and the brightness increases. Therefore, in (1), a lot of light that has been totally reflected locally appears in the eye. On the other hand, when the number of planes is small as in (2), the total reflection direction is dispersed and no glare is felt when viewed on the plane. In addition, color interference occurs due to various reflections, and the color spread is shown. Since (3) is finely diffused as a whole, the luminance decreases. Colors are also averaged.

  According to Example 2, it is possible to grasp an appropriate unevenness feeling by the surface roughness index M, and it is possible to reduce defective products by comparing with a threshold value. For example, when the surface 5 of the resin is textured, even if there is almost no color difference ΔE, it is possible to grasp a subtle change in the glossiness of the surface 5 when the surface is deep or shallow. Since it can be used in various modes, such as avoiding defects in embossing, the industrial utility value is high.

  Example 1 is an example of evaluating the surface roughness on the micron order of the metal surface of an aluminum alloy, while Example 3 is an example of the surface roughness on the nano order. The surface roughness measurement Ra (unit: nm) was measured using a Keyence VN-8010 non-contact type roughness meter.

Table 3 shows the measurement / calculation results of the surface roughness index M, the measured surface roughness Ra, and the surface roughness evaluation index Est. The calculation of Table 3 is based on Equation 7. A three-dimensional CG showing surface irregularities of sample Nos. 1 to 7 of the surface roughness measurement value Ra is shown in FIGS. In FIG. 1 has an average unevenness of 88.33 nm. In FIG. 2 has an average unevenness of 58.83 nm. In FIG. 3 has an average unevenness of 88.33 nm, and in FIG. 4 has an average unevenness of 57.71 nm. In FIG. 5 has an average unevenness of 93.9 nm. In FIG. 6 has an average unevenness of 112.85 nm. In FIG. The average unevenness in all 7 regions is 100.18 nm. The above numerical value is an average value of the height from the reference surface, and is a value indicating the distortion of the surface averaged with unevenness. When the sample numbers of sample Nos. 1 to 7 are increased, the etching time is sequentially increased, so that the surface roughness index M is decreased and the surface roughness measurement value Ra is increased.

  a = 1.141439 and b = 143.1954 were obtained. The total error TotEro is 940.2347.

  According to the third embodiment, it is possible to grasp an appropriate feeling of unevenness by using the first calibration curve L1 and the third calibration curve L3, evaluate nano-order surface roughness, and reduce defective products.

  The embodiments of the present invention are not limited to the above-described embodiments, and modifications and the like can be made without departing from the technical idea of the present invention. Needless to say, objects and the like are also included in the technical scope of the present invention and can take various forms as long as they belong to the technical scope. For example, with respect to a method of acquiring images A and B according to three spectral sensitivities (S1 (λ), S2 (λ), and S3 (λ)), the method given in this embodiment is only a specific example. However, the technical idea of the present invention is not limited to these and may be implemented by other methods.

  Since the determination device of the present invention can quantify the surface roughness by the scattering and diffraction phenomenon of illumination light, it can make a determination very close to the human visual determination method, so that the surface that has been dependent on the human eye until now Applicable to roughness evaluation. The present invention can be applied to various surfaces such as a wide range of materials including metals, resins, etc., flat surfaces, curved surfaces, and complicated shapes.

DESCRIPTION OF SYMBOLS 1, 101, 201 ... Surface roughness determination apparatus 2, 102, 202 ... Two-dimensional color meter 21 ... Shooting lens 22a, 22b, 22c ... Optical filter 23 ... Imaging element 22a ', 22c '... Dichroic mirrors 23a, 23b, 23c ... Imaging device 26 ... Reflector 27 ... Filter turrets 3, 103, 203 ... Arithmetic processing units 5, 105, 205 ... Surface 6 106, 206 ... Illumination unit 7 ... Display device 109 ... Switch 250 ... Roughness processing device a ... First coefficient b ... Second coefficient A, B ... Image AFM ... Atomic force microscopes C1, C2 ... Center coordinates D ... Overlapping region ΔE ... Color difference Error ... Error Est ... Surface roughness evaluation index G ... Lattice M ... Surface roughness index H _1, H ₂ ··· cumulative number K · And inspection area T ··· inspection area TotErro ··· Total error

As shown in FIG. 1, the two-dimensional colorimeter 2 has a photographing lens 21, three optical filters 22a, 22b, and 22c arranged behind the photographing lens 21, and a rear of the optical filters 22a, 22b, and 22c. And an image pickup device 23 (CCD, CMOS, etc.) arranged. The three spectral sensitivities (S1 (λ), S2 (λ), S3 (λ)) of the two-dimensional colorimeter 2 are obtained by multiplying the spectral transmittances of the optical filters 22a, 22b, 22c and the spectral sensitivity of the image sensor 23. Given. The arrangement relationship between the optical filters 22a, 22b, and 22c and the image sensor 23 in FIG. 1 is merely schematically shown. Specific examples of the method of acquiring the three-band visual sensitivity images S1i, S2i, and S3i according to the three spectral sensitivities (S1 (λ), S2 (λ), and S3 (λ)) will be given below. Any of these methods can be adopted, and other methods can be adopted. Reference numeral 24 denotes a calculation unit, and 25 denotes a display unit.

FIG. 3A shows a system using a dichroic mirror. This is because light of a specific wavelength is reflected by the dichroic mirror 22c ′, and the remaining light that has been transmitted is further reflected by another dichroic mirror 22a ′ to be spectrally separated, and the image pickup devices 23a, 23b. , 23c are read in parallel. Here, the dichroic mirror 22a ′ corresponds to the optical filters 22a and 22b, and the dichroic mirror 22c ′ corresponds to the optical filter 22c. Light incident from the photographic lens 21 is reflected by the dichroic mirror 22c ′ according to the spectral sensitivity S3, and the remaining light is transmitted. The light reflected by the dichroic mirror 22c ′ is reflected by the reflecting mirror 26, and the spectral sensitivity S3 is obtained by the imaging device 23c. On the other hand, the light transmitted through the dichroic mirror 22c ′ is reflected by the dichroic mirror 22a ′, and the light according to the spectral sensitivity S1 is transmitted, and the remaining light according to the spectral sensitivity S2 is transmitted . The light transmitted through the dichroic mirror 22a ′ is imaged by the image sensor 23b to obtain the spectral sensitivity S2. The light reflected by the dichroic mirror 22a ′ is reflected by the reflecting mirror 29, and the spectral sensitivity S1 is obtained by the imaging device 23a . Instead of the dichroic mirror, a dichroic prism having the same characteristics may be used to split the light into three, and the image sensors 23a, 23b, and 23c may be bonded to the positions where each light is transmitted.

The surface gloss / roughness measuring method has conventionally been a roughness meter, which includes a mechanical type and an optical type such as a laser. These are indicated by numerical values such as how many mm of unevenness or how many nm, but engineers who have experience in the field judge by looking at the glossiness and shine degree of the surface as seen from these values. Yes. In addition, work is proceeding with on-site judgment based on such experience. There are the following inventions in the prior art for measuring the surface roughness.

With the XYZ-based imaging device of the present invention, when the surface irradiated with light from the light source is imaged and the reflected light distribution is measured, the light from the light source is scattered and diffracted as color and surface roughness due to the surface shape. It was observed that this roughness allows us to quantify the surface color and surface roughness.

The second invention having technical relevance is obtained by an image pickup apparatus having three spectral sensitivities (S1 (λ), S2 (λ), S3 (λ)) linearly converted equivalently to the CIE XYZ color matching function. Three-band visual sensitivity images S1i, S2i, and S3i having three spectral sensitivities are obtained by calculating data obtained by converting tristimulus values X, Y, and Z in the CIE XYZ color system and calculating color difference ΔE. The color space of the XYZ color system is calculated by dividing the inspection region of coordinates corresponding to the color space of the XYZ color system with a grid and integrating the number of pixels of the test surface and the reference surface belonging to each grid. A color space histogram distribution creating step for creating a histogram distribution, a surface roughness index calculating step for calculating a surface roughness index indicating a difference between the two color space histogram distributions of the inspection surface and the reference surface, and a roughness meter A surface roughness measurement data storage step for storing the measured surface roughness measurement value Ra, a first calibration curve function indicating a correlation between the surface roughness measurement value Ra and the surface roughness evaluation index Est, the surface roughness measurement value At least one of a second calibration curve function indicating the correlation between Ra and the color difference ΔE or a third calibration curve function indicating the correlation between the surface roughness measurement value Ra and the surface roughness index M is set. a surface roughness determination method characterized by comprising: a function setting step.

The XYZ color system in a broad sense includes a color space defined by two-dimensional coordinates and three-dimensional coordinates. Typical examples of the color space include configuration examples such as an xy color space, an XYZ color space, and a Lab color space. In the case of a two-dimensional color space, for example, in the case of a Yxy color space or a Luv color space, an xy chromaticity diagram that is a two-dimensional plane (xy chromaticity value (plane) normalized in the Yxy color space), uv chromaticity diagram , U′v ′ chromaticity diagram. An xy chromaticity histogram distribution or a Luv chromaticity histogram distribution expressed as the density of pixels in a two-dimensional chromaticity diagram on a plane corresponds. In the case of a three-dimensional color space, for example, in the case of an XYZ color space or a Lab color space, a Lab color space can be cited. An XYZ color space histogram distribution or a Lab color space histogram distribution expressed as the density of pixels in a three-dimensional color space corresponds.

It is a block diagram of surface roughness judgment device 1 of Embodiment 1 of the present invention. It is a function which shows the spectral sensitivity of the camera 2 of the XYZ color system in Embodiment 1 of this invention. This is a specific example of a method for acquiring image information according to three spectral sensitivities (S1 (λ), S2 (λ), S3 (λ)) in Embodiment 1 of the present invention. (A) is explanatory drawing in the case of using a dichroic mirror. (B) is explanatory drawing in the case of using a filter turret. (C) is explanatory drawing at the time of attaching optical filter 22a, 22b, 22c to the image pick-up element 23 microscopically. It is a flowchart in the two-dimensional color meter 2 of Embodiment 1 of this invention. It is a flowchart in the arithmetic processing part 3 of Embodiment 1 of this invention. It is a sub chart in the arithmetic processing part 3 of Embodiment 1 of this invention. (A) is an explanatory view showing an inspection area T in the surface roughness determination equipment 1 of the invention Embodiment 1, (b) the xy chromaticity diagram showing an inspection area K on the chromaticity diagram corresponding to the inspection area T, (C) is an explanatory diagram of the inspection region K partitioned by the lattice G, (d) is a schematic diagram showing the state of chromaticity overlapping on the xy two-dimensional chromaticity diagram, and (e) is an explanatory diagram showing a minimum distribution. (F) is explanatory drawing which shows an example of xy chromaticity histogram distribution. (A) is explanatory drawing which shows the roughness of the surface, (b) is an xy chromaticity histogram distribution diagram, (c) is a three-dimensional image diagram of the xy chromaticity histogram distribution. It is a solid and plane explanatory drawing of a surface roughness index. It is a flowchart (calculation of multiple regression analysis) in the arithmetic processing apparatus 103 of the surface roughness determination apparatus 101 of Embodiment 2 of this invention. It is a block diagram which shows the structure of the surface roughness determination apparatus 101 of Embodiment 2 of this invention. It is a flowchart (XYZ color space distribution) in the arithmetic processing unit 103 of the surface roughness determination apparatus 101 of Embodiment 2 of the present invention. It is a flowchart (Lab color space distribution) in the arithmetic processing unit 103 of the surface roughness determination apparatus 101 of Embodiment 2 of the present invention. It is a block diagram which shows the structure of the surface roughness determination apparatus 201 of Embodiment 3 of this invention. It is the graph 1 of the surface roughness evaluation index Est with respect to the measured value Ra of the roughness meter of the metal sample No. 1 of Example 1. It is a graph 2 of total Hakachi Ra of the surface roughness meter for the color difference ΔE of the metal sample No.1 Example 1. It is the graph 3 of surface roughness measured value Ra with respect to the surface roughness index | exponent M of the metal sample No. 1 of Example 1. FIG. It is graph 1 of surface roughness evaluation index Est with respect to surface roughness measurement value Ra of metal sample No. 2 of Example 1. It is the graph 2 of surface roughness measured value Ra with respect to color difference (DELTA) E of the metal sample No. 2 of Example 1. FIG. Metal sample No. 1 of Example 1 . It is a graph 3 of the surface roughness measurement Ra for 2 of the surface roughness index M. It is explanatory drawing which shows the mode of reflection of the surface of resin sample No. 2 of Example 2. FIG. FIG. 4 is an explanatory photograph diagram showing a resin sample No. 2 in Example 2. It is explanatory photograph figure which shows the measurement range and measurement place (within image green frame) of resin sample No. 2 of Example 2. FIG. It is a three-dimensional view showing the measurement result of AFM of metal sample No. 1 of Example 3. It is a three-dimensional view showing the AFM measurement result of metal sample No. 2 in Example 3. It is a three-dimensional view showing the AFM measurement result of metal sample No. 3 of Example 3. It is a three-dimensional view showing the measurement result of AFM of metal sample No. 4 of Example 3. It is a three-dimensional view showing the AFM measurement result of metal sample No. 5 of Example 3. It is a three-dimensional view showing the AFM measurement result of metal sample No. 6 in Example 3. It is a three-dimensional view showing the measurement result of AFM of metal sample No. 7 of Example 3. It is the graph 1 which shows the relationship between the surface roughness index | exponent M of Example 3, and surface roughness measured value Ra. It is the graph 2 of the surface roughness evaluation index Est with respect to the surface roughness measured value Ra of the metal sample No. 1-7 of Example 3.

The surface roughness determination apparatus 1 includes a two-dimensional colorimeter 2 having three spectral sensitivities (S1 (λ), S2 (λ), and S3 (λ)) linearly converted equivalent to the CIE XYZ color matching function. An arithmetic processing unit 3 that acquires and calculates data obtained by converting three-band visual sensitivity images S1i, S2i, and S3i having three spectral sensitivities acquired from 3 into tristimulus values X, Y, and Z in the CIE XYZ color system; The arithmetic processing unit 3 divides the color difference calculation unit for calculating the color difference ΔE, and the inspection area of the coordinates corresponding to the color space of the XYZ color system with the grid G, and the inspection surface belonging to each grid G and the reference surface By integrating the number of pixels, a color space histogram distribution creating unit for creating a color space histogram distribution of the XYZ color system, and a surface roughness index M indicating a difference between the two color space histogram distributions of the inspection surface and the reference surface are obtained. A surface roughness index calculation unit to be calculated; A surface roughness measurement data storage unit for storing a surface roughness measurement value Ra measured by a roughness meter, a first calibration curve function L1 indicating a correlation between the surface roughness measurement value Ra and the surface roughness evaluation index Est, the surface At least of the second calibration curve function L2 indicating the correlation between the measured roughness value Ra and the color difference ΔE or the third calibration curve function L3 indicating the correlation between the measured surface roughness value Ra and the surface roughness index M. And a function setting unit for setting one. The surface roughness evaluation index Est specifies the first coefficient a related to the surface roughness index M and the second coefficient b related to the color difference ΔE so that the error Error between the surface roughness measurement value Ra is minimized. .

As shown in FIG. 1, the two-dimensional colorimeter 2 has a photographing lens 21, three optical filters 22a, 22b, and 22c arranged behind the photographing lens 21, and a rear of the optical filters 22a, 22b, and 22c. And an image pickup device 23 (CCD, CMOS, etc.) arranged. The three spectral sensitivities (S1 (λ), S2 (λ), S3 (λ)) of the two-dimensional colorimeter 2 are obtained by multiplying the spectral transmittances of the optical filters 22a, 22b, 22c and the spectral sensitivity of the image sensor 23. Given. The arrangement relationship between the optical filters 22a, 22b, and 22c and the image sensor 23 in FIG. 1 is merely schematically shown. Specific examples of the method of acquiring the three-band visual sensitivity images S1i, S2i, and S3i according to the three spectral sensitivities (S1 (λ), S2 (λ), and S3 (λ)) will be given below. Any of these methods can be adopted, and other methods can be adopted. Reference numeral 24 denotes a calculation unit, and 25 denotes a display unit.

The operation of the surface roughness determination apparatus 1 on the surface 5 will be described with a specific example. The surface roughness determination device 1 of the surface 5 operates by connecting a two-dimensional colorimeter 2, an arithmetic processing unit 3, and a display device 7 as shown in FIG. The connection method can be selected regardless of wired or wireless. The flowchart of the two-dimensional colorimeter 2 in FIG. 4, FIG. 5 a flowchart in the arithmetic processing unit 3, respectively.

When the power of the two-dimensional colorimeter 2 is turned on, initialization is performed as shown in FIG. 4 (initialization S1). Next, the surface 5 is imaged (imaging process S2), and then the captured 3-band visual sensitivity images S1i, S2i, and S3i are input by the image sensor 23 (input process S3), and the arithmetic processing unit 3 performs tristimulus. Conversion into values X, Y, and Z (conversion processing S4). Spectral sensitivities (S1 (λ), S2 (λ), S3 (λ)) are transmitted to the display device 7 (data transmission S5). When the image is a moving image, a series of processes from the imaging process S2 to the data transmission S5 is continuously performed (S6) . The image is displayed on the display device 7.

Since the input image data has values according to the spectral sensitivities (S1 (λ), S2 (λ), S3 (λ)), the conversion processing S4 in the arithmetic processing unit 3 connected to the two-dimensional colorimeter 2 The image data of the captured image is converted into tristimulus values X, Y, and Z. This conversion is performed according to Equation 1. That is, the tristimulus values X, Y, and Z can be obtained by multiplying the inverse matrix of the coefficients in Equation 1. Note that the two-dimensional colorimeter 2 transmits the values in accordance with the spectral sensitivities (S1 (λ), S2 (λ), S3 (λ)) to the arithmetic processing unit 3.

When the arithmetic processing unit 3 is turned on, initialization is performed as shown in FIG. 5 (initialization S110). The display device 7 receives the three-band visual sensitivity images S1i, S2i, S3i transmitted from the two-dimensional colorimeter 2 in a state where it is connected to the two-dimensional colorimeter 2 (data reception S120). Thereafter, the three-band visual sensitivity images S1i, S2i, and S3i are converted into tristimulus values X, Y, and Z, and the Lab average value and the xy surface roughness index M of the images of the imaged reference surface and inspection surface are calculated and visually processing of (S130), if required to be displayed on the display device 7 converts processing color information in RGB or the like (S140). The contents are transmitted to the display device 7 (display process S150). After the data reception S120 from the two-dimensional colorimeter 2, a series of processing from the conversion processing S130 to the display processing S150 is continuously performed (S160) .

An inspection region K (see FIG. 7B) corresponding to a region T (see FIG. 7A) to be inspected for the captured images A and B is set ( S141). Size and place can be set freely.

As shown in FIG. 7C, the color distribution of the comparison target at the position of the xy coordinates is written in a plane, the inspection area K is partitioned by the grid G, and the pixels having the xy value of the section are integrated. Then, a histogram distribution with the z axis is created. The xy coordinates, grid of a particular width, for example, a planar grating cut xy with each 1/1000 (1000 lines). The histogram is scanned from end to end, and for each region partitioned into the grid G, the number of pixels belonging to this is scanned on the same xy plane and integrated in the z direction. Further, if only the specific range of the xy coordinates is calculated in the inspection region K, the calculation time can be shortened. If the grid cells are made finer, the accuracy will be improved, but the calculation time will be longer.

As shown in FIG. 7E, H ₁ (x ₁ , y ₁ ) is the cumulative number of xy chromaticity histogram distributions on the inspection surface , and H ₂ (x ₂ , y ₂ ) is the xy chromaticity histogram distribution on the reference surface. When the accumulated number, in the overlapping left area _H 1> _{H 2,} centrally _H 1 = _{H 2,} and the on the right side is a _H 1 _{<H 2.} If the smaller integrated number (pixel frequency) of H ₁ and H ₂ is taken, it becomes H _{1 on} the left side and H ₂ on the right side, and a minimum distribution which is a step-like histogram curve can be specified. Using this, the ratio of the overlapping area D to the entire area in two dimensions or three dimensions can be calculated.

In FIG. 9, (1) is a sharp mountain concentrated at one point, (2) a normal mountain, and (3) is a wide and gentle mountain. Looking at the mountain from the side, it looks like the bottom in Fig. 9. When (2) and (3) are superimposed on the basis of (1), the colored portions are the overlapping portions. The volume ratio here is the surface roughness index M. As the surface roughness index M, the integrated number within the range of the inspection region K is used as it is, so that the inspection surfaces 2 and 3 are 58% and 27%, respectively, with respect to the reference object 1, which is clear and simple numerically. The surface roughness can be identified.

When the surface roughness is nano-order or micron order, the surface roughness index M and the color difference ΔE calculated as described above are used to analyze the surface roughness as shown in FIG. Est is calculated. This is because the light scattering on the surface 5 is different from the case of the roughness of nano and micron order from the case of the roughness being large. For fine roughness, interaction with the light to be applied, that is, interference occurs, and various offsets occur. Therefore, it is more accurate to step in positively. Therefore, the color offset for the surface roughness index M and the color difference ΔE is positively used, and a correlation with the roughness meter is obtained by multiple regression analysis.

The following surface roughness determination device 101 of the surface 105 of the second embodiment 11, FIG. 12 will be described with reference to FIG. 13. For the corresponding similar elements, the explanation is cited as 100 series, and the difference is mainly explained.

As illustrated in FIG. 11, the arithmetic processing device 103 includes a calculation unit 103 </ b> A that calculates a stimulus value XYZ <b> 1 acquired by imaging a reference plane, and a stimulus acquired by imaging the surface 105 of a product that is an inspection plane. A calculation unit 103B that calculates the value XYZ2, and a calculation unit 103C that is connected to the calculation unit 103A and the calculation unit 103B and calculates the surface roughness index M of the surface 105 are provided. An OK signal or an NG signal from the calculation unit 103C is transmitted to the display device 107 or transmitted to the outside. Note that the switch 109 selectively inputs the stimulus value XYZ1 and the stimulus value XYZ2.

In the case of the calculation of the XYZ distribution corresponding to the inspection region K, the calculation of the surface roughness index M is performed by the distribution in the three-dimensional space of the X axis, the Y axis, and the Z axis. The XYZ values in the XYZ space coordinates on the inspection surface and the reference surface are T ₁ ( X, Y, Z ) and T ₂ ( X, Y, Z ), respectively. In the XYZ color space, the histogram distribution has a globe-like shape, and the two histogram distributions may be three-dimensionally overlapped or separated. The inspection area K in the three-dimensional space is partitioned by the grid G, and the chromaticity histogram distribution and the minimum distribution of T ₁ (X, Y, Z) and T ₂ (X, Y, Z) in three dimensions are obtained. Calculate exponents. The integration numbers H ₁ and H ₂ of the grid G may be projected on a plane, and the integration number of the overlapping region D on the grid G may be calculated by the same integration within the plane. In the case of XYZ chromaticity, since there is no brightness information, the histogram distribution does not change in the XYZ space even if the brightness of the image changes.

When the Lab color space histogram is used for the surface roughness determination instead of the XYZ color space histogram, the flowchart of FIG. 13 is used. The description of FIG. 13 is based on the above description of FIG. In S305, the average Lab value of the region K of the image A and the average Lab value of the inspection region K of the image B are calculated. In the case of Lab chromaticity, since there is brightness information, the histogram distribution changes when the brightness of the images A and B changes in the Lab space.

The surface roughness indices M, xy3D, Lab, ΔL, Δa, Δb, and ΔE of samples 1 to 3 were obtained. Of the surface roughness indices, the surface roughness index M is an index calculated without shifting the center coordinates of the respective xy3D histogram distributions. Lab surface roughness index at Lab coordinates, Xy3D surface roughness index in the xy coordinate, [Delta] L = average L value of the difference = test sample average L value - and an average L value of the reference sample. The same applies to ΔE, Δa, and Δb. From the measurement results, the sample 2 or 3 having the most similar color and surface roughness to the sample 1 was identified from the result of the surface roughness index and the numerical value of ΔE. For ΔE, ΔE00 was used in consideration of human visibility characteristics.

Claims (4)

Three-band visual sensitivity image having three spectral sensitivities acquired by an imaging device having three spectral sensitivities (S1 (λ), S2 (λ), S3 (λ)) linearly converted equivalently to the CIE XYZ color matching function An arithmetic processing unit that acquires and calculates data obtained by converting S1i, S2i, and S3i into tristimulus values X, Y, and Z in the CIE XYZ color system;
The arithmetic processing unit is
A color difference calculation unit for calculating the color difference ΔE;
Creates an XYZ color system color space histogram distribution by dividing the inspection area of coordinates corresponding to the color space of the XYZ color system with a grid and integrating the number of pixels on the test surface and the reference surface belonging to each grid. A color space histogram distribution creation unit to
A surface roughness index calculator that calculates a surface roughness index indicating a difference between two color space histogram distributions of the inspection surface and the reference surface;
A surface roughness measurement data storage unit for storing a surface roughness measurement value Ra measured by a roughness meter;
A first calibration curve function indicating the correlation between the measured surface roughness Ra and the surface roughness evaluation index Est, a second calibration curve function indicating the correlation between the measured surface roughness Ra and the color difference ΔE, or the surface roughness A function setting unit for setting at least one of the third calibration curve functions indicating the correlation between the measured thickness Ra and the surface roughness index M;
A surface roughness determination apparatus comprising:   The surface roughness evaluation index Est specifies the first coefficient related to the surface roughness index and the second coefficient related to the color difference so that the error between the surface roughness measurement value Ra and the surface roughness is minimized. The surface roughness determination apparatus according to claim 1, wherein Three-band visual sensitivity image having three spectral sensitivities acquired by an imaging device having three spectral sensitivities (S1 (λ), S2 (λ), S3 (λ)) linearly converted equivalently to the CIE XYZ color matching function A calculation processing step for acquiring and calculating data obtained by converting S1i, S2i, and S3i into tristimulus values X, Y, and Z in the CIE XYZ color system;
A color difference calculating step for calculating the color difference ΔE;
Creates an XYZ color system color space histogram distribution by dividing the inspection area of coordinates corresponding to the color space of the XYZ color system with a grid and integrating the number of pixels on the test surface and the reference surface belonging to each grid. Creating a color space histogram distribution,
A surface roughness index calculating step for calculating a surface roughness index indicating a difference between two color space histogram distributions of the inspection surface and the reference surface;
A surface roughness measurement data storage step for storing a surface roughness measurement value Ra measured by a roughness meter;
A first calibration curve function indicating the correlation between the measured surface roughness Ra and the surface roughness evaluation index Est, a second calibration curve function indicating the correlation between the measured surface roughness Ra and the color difference ΔE, or the surface roughness A function setting step for setting at least one of the third calibration curve functions indicating the correlation between the measured thickness Ra and the surface roughness index M;
A method for determining surface roughness, comprising:   The surface roughness evaluation index Est specifies the first coefficient related to the surface roughness index and the second coefficient related to the color difference so that the error between the surface roughness measurement value Ra and the surface roughness is minimized. The surface roughness determination method according to claim 3, wherein the surface roughness is determined.