JP2000065751A - Surface inspection apparatus - Google Patents

Abstract

PROBLEM TO BE SOLVED: To obtain a surface inspection apparatus by which a defect is detected with high accuracy by detecting a pattern-like scab defect having no remarkable uneven property such as a crack, a twist or a burr on a surface on a face to be inspected surely. SOLUTION: In a signal processing part 31, an output image signal from a linear array camera which is arranged so as to be at 0 deg., 45 deg. and -45 deg. is shading-corrected, it is normalized and flattened in such a way that a normal part becomes a center density in the whole gradation and that it is converted into an image signal which indicates a relative change with reference to the normal part. A flaw candidate region which indicates a change larger than a density level expressing a normal state with respect to the normal part is extracted. Feature amounts such as length, width, area, density peak value, the integrated value of density and the like are calculated on the basis of three kinds of image signals inside the extracted flaw candidate region. The kind of a flaw is judged on the basis of the relative ratio of the integrated amount of the three kinds of integrated image signals out of the calculated feature amounts. In addition, the density peak value and the integrated value are compared with a value corresponding to a predetermined grade, and the grade of the flaw is judged.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a surface defect inspection apparatus for irradiating a non-inspection surface such as a thin steel sheet surface with light to optically detect surface defects on a surface to be inspected.

[0002]

2. Description of the Related Art Surface flaw inspection for optically detecting surface flaws present on a surface to be inspected by irradiating light to a surface to be inspected such as a thin steel sheet surface and analyzing reflected light from the surface to be inspected. Various conventional methods have been proposed and implemented.

For example, light is incident on the surface of a subject,
Japanese Patent Application Laid-Open No. 58-2043 discloses a method of detecting a surface of a metal object by detecting specularly reflected light and diffusely reflected light from the surface of a test object with a camera.
No. 53 has been proposed. In this surface flaw detection method, light is incident on the surface of the subject at an angle of 35 to 75 degrees, and the reflected light from the surface of the subject is reflected at an angle within 20 degrees from the specular reflection direction and the incident direction or the specular reflection direction. Light is received by two cameras installed in different directions. The light receiving signals of the two cameras are compared, and, for example, the logical sum of the two is calculated. Then, only when two cameras simultaneously detect abnormal values, the abnormal values are regarded as flaws, thereby realizing a surface flaw detection method that is not affected by noise.

A method for inspecting a flaw on the surface of an object by receiving backscattered light from the object is disclosed in Japanese Patent Laid-Open No. 60-22 / 1985.
No. 8943 has proposed this. In this flaw inspection method, light is incident on the stainless steel plate at a large incident angle, and reflected light returning to the incident side, that is, backscattered light is detected, thereby detecting a barbed flaw on the surface of the stainless steel plate.

Further, a flat steel hot flaw detector by detecting a plurality of backscattered reflected lights is disclosed in Japanese Patent Laid-Open No. Hei 8-17867.
No. 1993. This flat steel hot flaw detector detects a scratch on a hot-rolled flat steel. In this flaw detector, the angle of the flaw slope of the flaw is 10 degrees to 40 degrees, and a plurality of cameras are arranged in the backward diffuse reflection direction so as to cover all the specularly reflected light from the flaw slope in this range. ing.

A surface measuring apparatus utilizing polarized light is disclosed in Japanese Patent Application Laid-Open Nos. Sho 57-166533 and Hei 9-1665.
No. 52 has proposed this. JP-A-57-16653
In the measuring device proposed in Japanese Patent Publication No. 3 (1993), a 45-degree direction deflection is incident on a measurement object, and reflected light is received by a polarization camera. In a polarizing camera, the reflected light is split into three using a beam splitter inside the camera, and the reflected light is received through polarizing filters having different azimuth angles. A technique is disclosed in which three signals from a polarization camera are displayed on a monitor by signal processing similar to that of a color TV system, and the polarization state is visualized. This technology uses the technology of ellipsometry, it is desirable that the light source is parallel light,
For example, laser light is used.

In the surface inspection apparatus proposed in Japanese Patent Application Laid-Open No. 9-166552, Japanese Patent Application Laid-Open No.
Similar to the technique described in JP-A-533, an ellipsometry is used to inspect the surface of the steel sheet for flaws.

[0008]

However, each of the measuring techniques proposed in each of the above-mentioned publications detects a flaw having remarkable unevenness or detects a flaw having a foreign substance such as an oxide film. Therefore, it is difficult to reliably capture all flaws with respect to pattern-shaped scab defects and the like having no noticeable unevenness.

For example, the flaw detection method described in Japanese Patent Application Laid-Open No. 58-204353 has two cameras for receiving specularly reflected light and scattered reflected light. This is to remove the influence of noise due to the logical sum of the signals. Therefore, flaws having remarkable unevenness, that is, flaws having cracks, gouging or curling up on the surface, can be applied since both flaw signals can be captured by both cameras. However, in the case of a flaw such as a pattern-shaped scab defect having no noticeable unevenness such that only one of the cameras can capture the flaw signal, it is not possible to detect all the flaws.

The surface condition inspection method disclosed in Japanese Patent Application Laid-Open No. Sho 60-228943 is directed to a raised barbed flaw that is manifested in a stainless steel sheet having a small surface roughness.
Therefore, it is not possible to apply a flaw having no raised portion that has not been exposed or a flaw-free part to a steel plate having a rough surface that reflects light returning to the incident side.

The flat steel hot flaw detector disclosed in Japanese Patent Application Laid-Open No. Hei 8-17867 is intended for scratches and has no noticeable unevenness because it is based on capturing specularly reflected light on the slopes of the scratches. In the case of a flaw such as a patterned scab, there are some which cannot be caught by the backscattered reflected light, and there has been a problem that a detection leak occurs. Also, once the camera is installed and the angle of the reflected component to be received is determined, the camera position cannot be easily changed.

Further, the measuring device disclosed in Japanese Patent Application Laid-Open No. 57-166533 and the surface inspection device disclosed in Japanese Patent Application Laid-Open No. 9-166552 use ellipsometry technology, and the thickness and refractive index of a thin transparent layer are determined. Unevenness in physical property values can be detected. However, for example, like a surface-treated steel sheet,
Even if the flaw originally had a property value different from that of the base material, the problem that the effectiveness was reduced for an object covered by something with the same property value from above there were.

In the ellipsometry, it is necessary to receive reflected light from the same point at a corresponding pixel of each CCD and calculate an ellipsometric parameter for each pixel. For this reason, in JP-A-57-166533, the reflected light is divided into three by a beam splitter and detected by three CCDs, and there is a problem that the amount of light is reduced and it is difficult to align pixels between the CCDs. .

In Japanese Patent Application Laid-Open No. Hei 7-28633,
Three cameras are arranged in the traveling direction of the steel plate, arranged vertically or horizontally, and the inclination of the three cameras is changed so that the same area is viewed. However, there is a problem that processing when the speed of the steel sheet changes is complicated. Also, since the angles of the cameras are different, the optical conditions are not the same. Therefore, there is a problem that it is difficult to perform pixel alignment.

Further, in Japanese Patent Application Laid-Open Nos. 58-204353 and 8-17867, since the optical axes of a plurality of cameras are not common but have different emission angles, the fields of view of the corresponding pixels of two obtained images are different. In addition to the difference in size, there is a problem that a positional shift occurs in the visual field when there is flapping of the surface to be inspected or a change in distance due to a change in thickness of the object. Particularly, in Japanese Patent Application Laid-Open No. 58-204353, the problem is serious because it is required that two cameras take a logical sum for the same field of view.

Further, there has been no method for effectively discriminating pattern-like and stain-like scabble defects. In particular, the problem of stain-like scabble defects is sometimes evaluated because the signal is large even though the degree of flaws is low.

In a surface inspection apparatus incorporated in a product quality inspection line, from the viewpoint of quality assurance of a manufactured product, it is an absolute condition that there is no omission of flaw detection. However, a surface flaw inspection apparatus for inspecting even a surface-treated steel sheet or the like has not been put to practical use.

The present invention has been made in view of such circumstances, and distinguishes and detects a specular reflection component and a specular diffuse reflection component contained in light reflected from a surface to be inspected, thereby obtaining an image on the surface to be inspected. A surface that can reliably detect pattern-like scab defects that do not have noticeable irregularities such as surface cracks, twists, or curls, can exhibit high defect detection accuracy, and can be fully incorporated into product quality inspection lines. It is an object to provide an inspection device.

[0019]

A surface inspection apparatus according to the present invention has a light projecting section, a light receiving section, and a signal processing section, wherein the light projecting section enters polarized light on a surface to be inspected, and the light receiving section has at least a light receiving section. It has a plurality of light receiving optical systems that receive polarized light with different angles in three directions, detects reflected light reflected on the surface to be inspected, converts it into an image signal, and the signal processing unit outputs an image output from each light receiving optical system. The signal is normalized such that the background signal of the surface to be inspected becomes a reference, and a flaw candidate area is extracted from the variation with respect to the reference value, and the signal intensity of the signal reference value from each light receiving optical system in the extracted flaw candidate area is extracted. A surface inspection apparatus that integrates a change amount and determines a type of a surface flaw from a relative ratio of an integrated signal intensity integration amount between respective light receiving optical systems.

A surface inspection apparatus according to a second aspect of the present invention has a light projecting unit, a light receiving unit, and a signal processing unit. It has multiple light receiving optical systems that receive polarized light of an angle, detects reflected light reflected on the surface to be inspected, converts it into image signals, and the signal processing unit performs inspection based on the image signals output from each light receiving optical system. Normalization is performed so that the background signal of the surface becomes a reference, a flaw candidate region is extracted from the amount of change with respect to the reference value, and the amount of change in signal intensity with respect to the signal reference value from each light receiving optical system in the extracted flaw candidate region is integrated. Then, determine the type of surface flaw from the relative ratio of the integration amount of the signal intensity change between each light receiving optical system,
The class of the flaw is determined by comparing the integral of the signal intensity change with a predetermined pattern.

A surface inspection apparatus according to a third aspect of the present invention has a light projecting unit, a light receiving unit, and a signal processing unit, and the light projecting unit emits polarized light having a polarization angle of 45 degrees with respect to the incident surface of the inspection surface. The light enters the surface to be inspected, and the light receiving portion has a polarization angle of 0 degree with respect to the incident surface of the surface to be inspected.
It has three light receiving optical systems for receiving polarized light of 45 degrees and -45 degrees, detects reflected light reflected on the surface to be inspected and changes the image signals into image signals, and a signal processing unit is output from each light receiving optical system. The background signal of the surface to be inspected is standardized from the extracted image signal so as to be a reference, a flaw candidate area is extracted from a change amount with respect to the reference value, and a signal intensity change amount from each light receiving optical system in the extracted flaw candidate area is determined. A stain that causes a defect in the cold-rolled steel sheet in which the ratio of the signal intensity integral of the light receiving optical system at a polarization angle of 45 degrees to the maximum value of the integral of the signal intensity change of the three light receiving optical systems is less than 1/4. A defect in which the ratio of the signal intensity integral of the light receiving optical system at a polarization angle of 45 ° to the maximum value of the integral of the signal intensity change of the three light receiving optical systems is 1/4 or more is regarded as a barge defect. It is characterized by judging and determining the grade of the judged flaw type Surface inspection apparatus of the cold-rolled steel sheet to be.

[0022]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS First, the form of optical reflection on the surface of a steel sheet to be inspected by the surface flaw inspection apparatus of the present invention will be described in relation to the microscopic unevenness of the steel sheet surface. For example, when the inspection target is an alloyed galvanized steel sheet, FIG.
As shown in (a), the cold rolled steel sheet of the base passes through an alloying furnace after being hot-dip galvanized. During this time, the iron element of the base steel sheet 1 diffuses into the zinc of the plating layer 2 and usually, as shown in FIG.
As shown in (c), columnar crystals 3 of the alloy are formed. The plated steel sheet 4 is then temper rolled on rolls 5a and 5b. Then, as shown in FIG.
Are particularly flattened by the rolls 5a and 5b, and the other portions maintain the original shape of the columnar crystal 3. The portions flattened by the temper rolling rolls 5a and 5b are referred to as a tempered portion 6, and the remaining portions of the other temper rolling rolls 5a and 5b which are not in contact with the original uneven shape are non-tempered. This is referred to as a unit 7.

FIG. 2 is a schematic sectional view modeling what kind of optical reflection occurs on the surface of the steel plate 4 having such a tempered portion 6 and the non-tempered portion 7. The surface of the steel plate 4 is composed of countless minute plane elements 13 oriented in various directions when viewed microscopically. Rolls 5a and 5b for temper rolling
The incident light 8 incident on the tempered portion 6 crushed by the light is specularly reflected in the regular reflection direction of the steel plate 4 to become specular reflected light 9. On the other hand, the incident light 8 that has entered the non-tempered portion 7 that leaves the original structure of the columnar crystal 3 where the temper rolling rolls 5a and 5b do not come into contact with each other is one microscopic element on each surface of the columnar crystal 3 when viewed microscopically. Although the light is reflected specularly by one, the direction of the reflection is the specular diffused reflected light 10 which does not always coincide with the regular reflection direction of the steel plate 4. Therefore, the angular distribution of each reflected light of the tempered portion 6 and the non-tempered portion 7 on the surface of the steel plate 4 is as shown in FIGS. 3 (a) and 3 (b) when viewed macroscopically. That is, a sharp specular reflection occurs in the regular reflection direction of the steel sheet in the tempered portion 6, and reflected light having a spread corresponding to the angular distribution of the microplane on the surface of the columnar crystal 3 in the non-tempered portion 7. As described above, the reflected light from the tempered portion 6 is referred to as specular reflected light 9, and the reflected light from the non-tempered portion 7 is referred to as specular diffused reflected light 10. Since the tempered portion 6 and the non-tempered portion 7 are macroscopically mixed, the angular distribution of the reflected light observed by an optical measuring instrument such as a camera has a specular reflection as shown in FIG. The angular distributions of the light 9 and the specular diffuse reflection light 10 are added according to the respective area ratios of the tempered portion 6 and the non-tempered portion 7.

As described above, the tempered portion 6 and the non-tempered portion 7 have been described using an alloyed galvanized steel plate as an example. However, the present invention is generally applicable to other steel plates in which a flat portion is formed by temper rolling.

Next, a description will be given of the optical reflection characteristic of a defect called a pattern-shaped scab defect having no noticeable unevenness to be detected in the present invention. As shown in FIG. 4, barge defects (barge portions) 11 found in the galvannealed steel sheet are:
The barge defects exist in the cold-rolled steel sheet before plating, and the plating layer 2 rides thereon, and further, alloying of the barge defects due to the diffusion of the iron element of the base steel sheet 1 progresses.

In general, the barbed portion 11 has a difference in plating thickness or a degree of alloying, for example, as compared with the base material 12 indicating a normal portion of the steel plate 4. As a result, for example, when the plating thickness of the barbed portion 11 is large and is convex with respect to the base material 12, the area of the tempered portion 6 becomes larger than that of the non-tempered portion 7 by applying the temper rolling. Conversely, hege part 1
In the case where the plating thickness of 1 is thinner than that of the base material 12 and is concave, the non-tempered portion 7 occupies most of the barb portion 11 because the rolls 5a and 5b of the temper rolling do not abut. Further, when the alloying of the barbed portion 11 is shallow, the angular distribution of the minute surface element is strong in the direction of the steel plate, and the diffusivity is small.

Next, the scab 11 and the base 12
A description will be given of how the pattern-like scab defect looks due to the difference in the surface properties. Hege part 1 based on the model described above
1 and the base material 12 are generally classified as follows:
Divided into types.

(A) The area ratio of the tempered portion 6 in the barbed portion 11 and the angular distribution of the micro-surface element in the non-tempered portion 7 are determined by the area ratio of the tempered portion 6 in the base material portion 12 and the minute surface of the non-tempered portion 7. It differs from the elementary angular distribution (FIG. 6A, FIG.
(A)).

(B) Although the area ratio of the tempered portion 6 in the barbed portion 11 is different from the area ratio of the tempered portion 6 in the base material portion 12, the angle distribution of the small surface element of the non-tempered portion 7 in the barbed portion 11 is It is not different from the angular distribution of the micro-surface element of the non-tempered part 7 in the base material part 12 (FIGS. 6B and 5B).

(C) The angular distribution of the micro-surface elements of the non-tempered portion 7 in the barbed portion 11 is different from the angular distribution of the micro-plane elements of the non-tempered portion 7 of the base material portion 12. 6 is the same as the area ratio of the temper portion 6 in the base material portion 12 (FIGS. 6C and 5C).

As shown in FIG. 7, the inclination angle of the normal direction of the small surface element 13 with which the incident light 8 comes into contact with the steel plate normal direction of the steel plate 4 is defined as the normal angle ξ of the small surface element 13. FIG. 6 shows the relationship between the angle 率 and the area ratio S (テ ン) of the temper portion 6 in the three cases (a), (b), and (c) described above.
(A), (b) and (c) show.

The difference between the area ratio S (ξ) of the temper portion 6 and the angular distribution of the micro-surface element 13 is shown in FIGS.
This is observed as a difference in the angular distribution of the reflected light amount as shown in FIG. In the drawing, the angle distribution indicated by the solid line is the barb portion angle distribution 11a corresponding to the barge portion 11, and the angle distribution indicated by the dotted line is the base metal portion angle distribution 12a corresponding to the base material portion 12.

That is, FIG.
FIG. 5B shows a case where there is a difference between the specular reflection component and the specular diffuse reflection component between 1a and the base material part angle distribution 12a, and FIG. 5B shows a case where there is a difference only in the specular reflection component. FIG. 5C shows a case where there is a difference only in the specular diffuse reflection component. If there is a difference in the area ratio S (パ) of the tempered portion 6 between the barge angle distribution 11a and the base metal angle distribution 12a, as shown in FIGS. 5 (a) and 5 (b),
The difference is observed from the regular reflection direction. Specifically, the area ratio S (ξ) of the tempered portion 6 of the barbed portion 11 is determined when the reflected light of the barbed portion 11 is measured from the regular reflection direction and when the reflected light of the base material portion 12 is measured. When the area ratio S (ξ) of the tempered portion 6 of the base material portion 12 is larger than that of the base material portion 12, the barb portion 11 looks relatively bright. Conversely, when the tempering ratio 6 of the barb portion 11 is smaller than the base material portion 12, the barge portion 11 is observed to be relatively darker than the base material portion 12.

Heavy part angle distribution 11a and base metal part angle distribution 1
In the case where there is no difference in the area ratio S (パ) of the temper portion 6 from that in the case of 2a, as shown in FIG. I cannot observe its existence. However, when there is a difference in the diffusivity (angle distribution) of the specular diffuse reflection component, a defect is observed from a diffusion direction other than the regular reflection direction as shown in FIG.

When the diffusivity (angular distribution) of the specular diffuse reflection component of the barb portion 11 is small, the barb portion 11 is generally observed brightly from a diffusion direction relatively close to the regular reflection direction,
The brightness decreases as the distance from the specular reflection direction increases, and observation becomes impossible at a certain angle. As the distance from the specular reflection direction further increases, the barbed portion 11 is observed darker.

In order to reliably detect such a stub 11 in distinction from the base material 12, the reflected light from the microscopic element 13 at any angle (normal angle ξ) is extracted in FIG. It is necessary to consider whether or not. For example, FIG.
As shown in the examples of FIGS.
And detecting the difference between the base material 12 and the base material 12 means extracting the normal angle ξ = 0 of the micro surface element 13 from the angle distribution of the micro surface element 13 shown in FIG. Part 12
That is, the difference from the above is detected.

Here, extracting the reflected light at a normal angle ξ = 0 of the microscopic element 13 is mathematically expressed as follows. Each of the characteristics (area ratio S (ξ)) shown in FIG. This corresponds to integration by multiplying by a function indicating an extraction characteristic represented by a delta function δ (ξ) shown in a) (hereinafter, this function is referred to as a weight function Ι (ξ)).

For example, measuring the reflected light at an angle position of 40 degrees shifted from the regular reflection direction by 20 degrees at an incident angle of 60 degrees requires a delta function δ (ξ) as shown in FIG.
+10) is equivalent to the calculation using the weight function Ι (Ι).

As shown in FIG. 7, the relationship between the reflection angle θ degrees, the normal angle ξ of the microscopic element 13 and the incident angle θ of the incident light 8 is expressed by Equation (1) by simple geometric considerations. I get it. θ degree = −θ + 2ξ (1) In other words, what angle (normal angle ξ) is the small plane element 13
It can be understood that extraction of the reflected light from is equivalent to what kind of weighting function Ι (ξ) is designed.

From this point of view, FIG.
Considering the weight function 重 み (ξ) for discriminating and detecting each of the barbed portions 11 represented by (b) and (c) from the base material portion 12, FIG. Delta function δ (ξ),
δ (ξ + 10) is also one of the effective weight functions I (ξ). Note that the weight function Ι (必 ず し も) does not necessarily have to be an infinitesimal delta function δ (が) in which only the specific normal angle shown in FIG. 8 is extracted, and may have a certain signal width. It is possible.

However, in such a discrimination method, the two optical systems cannot have the same field of view. Further, once a camera is installed for measuring diffuse reflection light, it is not easy to change the weight function Ι (ξ) since it is necessary to change the installation position of the camera.

For the former problem, measurement on the same optical axis is required. That is, it is desirable to capture both the specular reflection component and the specular diffuse reflection component only by measuring from the specular reflection direction of the steel plate 4 instead of capturing the diffuse reflection light. For the latter problem, it is desirable that the weighting function Ι (ξ) can be set with a certain degree of freedom.

Therefore, in the present invention, a linear light source having a diffusion characteristic, that is, a linear diffusion light source, is used as a light source instead of a parallel light source such as a laser. Also,
Since it is necessary to separate and extract the specular reflection component and the specular diffuse reflection component from the specular reflection direction of the steel plate 4, polarized light is used. To explain the effect of this linear diffused light source, FIG.
As shown in (a) and (b), the linear diffused light source 14 is arranged in parallel to the surface of the steel plate 4, and is located in a plane perpendicular to the light source and whose incident angle is the direction coincident with the outgoing angle. Consider a reflection characteristic when one point on the steel plate 4 is observed from the reflection direction.

As shown in FIG. 9A, the linear diffused light source 1
In the case of incident light 8 emitted from the center of
Is reflected specularly, and is all captured in the steel plate regular reflection direction. On the other hand, the light incident on the non-tempered portion 7 is specularly reflected, and only the light reflected by the minute surface element 13 that happens to be oriented in the same direction as the normal direction of the steel sheet is captured. Micro-plane element 1 facing in such a direction
Since the number 3 is very small, the specular reflection light from the tempering portion 6 is dominant among the reflection light captured by the light receiving camera disposed in the steel plate regular reflection direction.

On the other hand, as shown in FIG. 9B, in the case of the incident light 8 radiated from a position outside the central portion of the linear diffused light source 14, the light incident on the tempering portion 6 is specularly reflected. Then, the light is reflected in a direction different from the regular reflection direction of the steel sheet. Therefore, the specularly reflected light cannot be captured in the steel plate regular reflection direction. On the other hand, the light incident on the non-tempered portion 7 is specularly diffusely reflected, of which the light reflected in the steel plate regular reflection direction is captured by the light receiving camera. Therefore, all the reflected light captured by the light receiving camera disposed in the steel plate regular reflection direction is the specular diffuse reflection light reflected by the non-tempered portion 7.

When the above two cases are combined, the observation from the regular reflection direction of the steel plate out of all the incident light 8 emitted from the entire longitudinal direction of the linear diffusion light source 14 is the mirror surface from the tempering portion 6. This is the sum of the reflected light and the specularly diffused reflected light from the non-tempered portion 7.

Next, how the polarization characteristics change when observed using the linear diffused light source 14 from the regular reflection direction of the steel plate 4 will be described. Generally, in reflection from a mirror-like metal surface, the direction of the electric field is parallel to the incident surface (p
In the case of (polarized light) or light perpendicular to the plane of incidence (s-polarized light), the polarization characteristics are preserved even by reflection. That is,
The light is emitted as p-polarized light or s-polarized light. When linearly polarized light having an arbitrary polarization angle having both p-polarized light component and s-polarized light component at the same time is reflected, the reflectance of p and s-polarized light is not tan.
The light is emitted as elliptically polarized light according to Ψ and the phase difference △.

A linear diffusion light source 14 is applied to an alloyed galvanized steel sheet.
The case in which light is irradiated from FIG. 10 will be described with reference to FIGS. As shown in FIG. 10A, the light emitted from the central portion of the linear diffused light source 14 is specularly reflected by the tempering portion 6 of the steel plate 4 and is observed in the regular reflection direction of the steel plate. In this regard, reflection on the above-mentioned general mirror-like metal surface is established as it is.

On the other hand, as shown in FIG. 10B, the light emitted from a position outside the central portion of the linear diffused light source 14 is generated by the inclined minute surface element 13 on the crystal surface of the non-tempered portion 7 of the steel plate 4. Specularly reflected and observed in the direction of regular reflection of the steel sheet. In this case, even if the p-polarized light parallel to the incident surface of the steel plate 4 is incident, the incident surface is not parallel to the minute surface element 13 in consideration of the tilted minute surface element 13 that actually reflects. No, p, s
Since it is linearly polarized light having both polarization components, it is emitted as elliptically polarized light. The same applies to the case where s-polarized light is incident from the linear diffusion light source 14.

When linearly polarized light having an arbitrary polarization angle α having both p and s polarization components is incident on the steel plate 4 from the linear diffused light source 14, minute linear light inclined from a position other than the center of the linear diffused light source 14. Since the light incident on the surface element 13 acts with the polarization angle α inclined, the shape of the elliptically polarized light emitted in the steel plate regular reflection direction is
This is different from the light that enters from the central part of the linear diffusion light source 14 and is specularly reflected by the tempering unit 6.

Hereinafter, the case where linearly polarized light having p and s amphoteric components is incident on the steel plate 4 from the linear diffusion light source 14 will be described in detail. First, as shown in FIG.
After the incident light 8 from 4 is linearly polarized by a polarizing plate 15 having an azimuth angle (polarization angle) α, the light is made incident on a horizontally disposed steel plate 4, and the specularly reflected light is received by a light receiving camera 16. As described above, with respect to the incident light 8 emitted from the point C on the linear diffused light source 14, the component specularly reflected by the tempered portion 6 in the steel plate 4 and the normal of the non-tempered portion 7 Normal angle ξ = 0 facing the vertical direction
The component which is specularly diffusely reflected from the microscopic surface element 13 contributes to light reflected from the zero point on the steel plate 4 toward the light receiving camera 16.

On the other hand, as shown in FIG. 12, for the incident light 8 from the point A which is shifted from the point O of the steel plate 4 on the linear diffused light source 14 by an angle φ, the specular reflection component is
Since the light is reflected in directions different from the six directions, only the specular diffuse reflection component by the micro-plane element 13 having the normal angle ξ contributes.

Here, the angle φ indicating the incident direction of the incident light 8
And the normal angle ξ of the microscopic element 13 is given by Expression (2) by simple geometrical consideration using the incident angle θ of the incident light 8 with respect to the steel plate 4.

[0054]

(Equation 1)

Next, the polarization state of the light reflected as described above will be considered. The incident light 8 emitted from the point C is
O passing through the polarizing plate 15 having the azimuth angle (polarization angle) α
The polarization state Ec after specular reflection at a point is expressed as Ec = T · Ein (3) using a Jones matrix generally used in polarization optics. Here, Ein indicates the orthogonal polarization vector of the azimuthal angle (polarization angle) α of the polarizing plate 15, and T indicates the reflection characteristic matrix of the steel plate 4. Then, the linear polarization vector Ein and the reflection characteristic matrix T express the amplitude reflectance ratio of p and s polarizations as tanΨ,
Assuming that the phase difference between the reflectances of the p- and s-polarized light is △ and the amplitude reflectance of the s-polarized light is rs, they are given by equations (4) and (5), respectively.

[0056]

(Equation 2)

Similarly, the polarization state Ea of the incident light 8 emitted from the point A on the linear diffused light source 14 and reflected by the minute pixels 13 having the normal angle に in the direction of the light receiver 16 is represented by Assuming that it is orthogonal to the polarizer 15 and the analyzer of the light receiving camera 16, it is given by equation (6). In the equation (6), R is a rotation matrix, which is given by the equation (7).

[0058]

(Equation 3)

Equation (3) is obtained by adding small area element 1 in equation (6).
This is a special case where the normal angle ξ of 3 is = 0, and both the specular reflection component and the specular diffuse reflection component can be considered in unified manner using equation (6). (6) is calculated, and the elliptically polarized state of the reflected light from the microscopic surface element 13 at the normal angle ξ is illustrated in FIG. Here, the azimuth angle (polarization angle) α of the incident polarized light is 45 degrees, and the incident angle θ is 60.
And the reflection characteristic of the steel plate 4 as the inverse tangent of the amplitude reflectance ratio of the p and s polarized light Ψ = 28 degrees, and the phase difference of the reflectance of the p and s polarized light △ =
It can be understood from FIG. 13 that the ellipse is inclined as the normal angle ξ = 0, that is, the value of the normal angle に 対 し て changes with respect to the ellipse in the case of specular reflection. Therefore, for example, by inserting the analyzer 17 in front of the light-receiving camera 16 and setting the analysis angle β, it is possible to select which normal angle す る to extract more reflected light from the microscopic element 13. be able to.

In order to quantify this, as shown in FIG. 12, the polarization after the insertion of the analyzer 17 having the analysis angle β with respect to the reflected light having the polarization state Ea represented by the equation (3) is shown. When the state Eo is obtained, the equation (8) is obtained.

[0061]

(Equation 4)

In the equation (8), A is a matrix representing the analyzer 17 and is represented by the equation (9).

[0063]

(Equation 5)

Next, from this equation (8), the light intensity of the reflected light from the microscopic element 13 at the normal angle ξ detected by the light receiving camera 16 is obtained. As described above, when the area ratio of the microscopic element 13 is S (ξ), the following equation (10) is satisfied.

[0065]

(Equation 6)

Ι (式, β) in the above equation is a weighting function indicating how much the reflected light from the microscopic surface element 13 having the normal angle ξ can be extracted, as described above. Depends on the polarization characteristics of Then, the product of the reflectance rs ^{2 of the} steel plate 4, the amount of incident light Ep ^2, and the area ratio S (ξ) is the detected light intensity.

When considering a target having a uniform surface material such as a surface-treated steel sheet, the value of the reflectance rs ² is considered to be constant. Further, the incident light quantity Ep ² good as well constant value if uniform regardless of the position of the incident light amount is a light source. Therefore, in order to obtain the light intensity detected by the light receiving camera 16, the area ratio S (ξ) of the small plane element 13 having the normal angle ξ is obtained.
And the weighting function Ι (ξ, β).

Now, consider the weight function Ι (ξ, β). If an attempt is made to select an analysis angle βo of the analyzer 17 that maximizes the contribution of the normal angle ξ from the microscopic surface element 13, the candidate is given by solving the following equation (11) for β. .

[0069]

(Equation 7)

When the normal angle ξ = 0, that is, the analysis angle β that maximizes the contribution of the specular reflection component is obtained from the equation (11), the analysis angle β is about −45 degrees. However, also in this case, the reflection tangent Ψ = 28 degrees and the phase difference △ = 120 degrees of the reflectance ratio described above are adopted as the reflection characteristics of the steel plate 4, and the azimuth of the polarizing plate 15 with respect to the incident light 8 from the linear diffusion light source 14. Angle (polarization angle) α = 45 degrees was adopted.

FIG. 14 shows that the detection angle β of the analyzer 17 is -45.
The relationship between the normal angle ξ of the small plane element 13 and the weighting function Ι (ξ, −45) in the case of degrees is shown. However, the maximum value of the weight function Ι (Ι, -45) is normalized to [1] for easy viewing. From the characteristics of FIG. 14, the normal angle ξ = 0 degrees, that is, the specular reflection component is the most dominant, and conversely, the normal angle ξ =
It can be understood that the specular diffuse reflection light from the micro-plane element 13 near ± 35 degrees is least extracted.

On the other hand, the analysis angle β of the analyzer 17 that best extracts the reflected light having the normal angle ξ = ± 35 ° is set to (1
According to the equations (0) and (11), β = 45 degrees. . FIG. 15 shows the relationship between the normal angle ξ of the microscopic element 13 and the weighting function Ι (ξ, 45) with respect to the analysis angle β of the analyzer 17 = 45 degrees. Here, the weight function I (ξ,
β) is not bilaterally symmetric because the entrance surface (small surface element 1)
Considering the plane defined by the incident light 8 and the reflected light with respect to 3), when the normal angle ξ of the microscopic element 13 is positive,
This is because apparently the azimuth angle (polarization angle) α of the polarized light of the incident light 8 is reduced (approaching to p-polarized light) and the p-polarized light reflectance of the steel plate 4 is smaller than the s-polarized light reflectance.

The weighting functions Ι (ξ, 0), Ι (ξ, β) calculated also for β = 0 ° and 90 °, which are characteristics intermediate between the detection angle β = −45 degrees and 45 degrees of the analyzer 17. β) is also Fig. 1.
5 is shown. Although Ι (ξ, 0) has a peak near −50 degrees, the influence near ξ = 15 degrees is often the largest depending on the area ratio of the measurement target. As shown in the equation (10), the intensity of the reflected light from the microscopic surface element 13 at the normal angle ξ is given by the product of the weight function I (ξ, β) and the area ratio S (ξ).
The light intensity finally received by the light receiving camera 16 is [S (ξ)
[Ι (ξ, β)] is integrated with respect to the normal angle ξ. For example, the reflected light from the steel plate 4 having the reflection characteristics as shown in FIG.
, The area ratio S shown in FIG.
The value obtained by integrating (ξ) with a weight represented by a weight function Ι (ξ, β) shown in FIG. 14 is the actually received light intensity.

Therefore, the surface of the steel plate 4 is provided on the surface of FIG.
Let us consider a case where there is a barbed portion 11 having characteristics as shown in FIGS. In that case, each area ratio S (ξ) is
6 (a), 6 (b) and 6 (c) respectively.

First, consider the case where there is a difference only in the specular reflection component as shown in FIGS. 5 (b) and 6 (b). The light intensity when such a flaw is received through the analyzer 17 having the analysis angle β = −45 degrees is expressed by a weighting function expressed in FIG. 14 with respect to the area ratio S (ξ) shown in FIG. Since it corresponds to the value obtained by multiplying by Ι (ξ, β), it is possible to detect the difference in the amount of reflected light between the base material portion 12 and the stub 11.

FIG. 6 shows the light intensity when the same flaw is received through the analyzer 17 having the analysis angle β = 45 degrees.
As shown in (b), since there is no difference in the specular diffuse reflection component, the weight function Ι (ξ, β) of the analysis angle β = 45 degrees in FIG.
When considering the integration by multiplying, the difference between the base material portion 12 and the barb portion 11 cannot be detected.

When there is a difference only in the specular diffuse reflection component as shown in FIGS. 5 (c) and 6 (c), on the contrary, the light passes through the analyzer 17 having an analysis angle β = −45 degrees. In this case, the light cannot be detected, and can be detected when the light passes through a degree analyzer 17 having an analysis angle β = 45 degrees. However, the normal angle ξ at which the difference between the specular diffuse reflection component of the base material portion 12 and the specular diffuse reflection component of the barge portion 11 disappears is as shown in FIG.
In (c), the normal angle ξ is around ± 20 degrees, but if there is a flaw whose angle happens to be around ± 30 degrees, it is also detected through the analyzer 17 with the analysis angle β = 45 degrees. become unable. In that case, another weighting function such as Ι (ξ, 9
0), the analyzer 1 having an analysis angle β (eg, 90 °)
7 may be prepared separately, and the third light receiving camera 16 may receive light.

Generally, the reflection characteristics of the base material portion 12 and the barge portion 11 on the surface of the steel plate 4 are any of FIGS. 5A, 5B, and 5C. In order to eliminate the problem, it is necessary to use three analyzers 17 having different analysis angles β to extract and receive the reflected light from the micro surface element 13 having the corresponding three normal angles ξ. is there. When there is a difference between the specular reflection component and the specular diffuse reflection component as shown in FIGS. 5A and 6A, basically, for example, any one of -45 degrees and +45 degrees is used. The difference between the base material portion 12 and the stub 11 can be detected even by the reflected light passing through the portion 17. Therefore, in the present invention, the linear diffusion light source 14 is used,
The first light receiving unit extracts and receives more specular reflection components than the specular diffuse reflection components among the specular reflection components and the specular diffuse reflection components included in the specular reflection light from the surface to be inspected, and receives the second light. The light receiving means extracts more of the specular reflection component and the specular reflection component included in the specular reflection light from the surface to be inspected, as compared with the specular reflection component.

Therefore, for example, even with the first and second light receiving means for receiving only the specularly reflected light from the surface to be inspected, FIG.
In each of the reflection characteristics of the surface of the steel plate 4 shown in (a), (b), and (c), the presence of the barbed portion 11 can be reliably detected by comparison with the base material portion 12.

Since the measurement is performed on the common optical axis from the specular reflection direction by such an optical system, there are two types of specular reflection and specular diffuse reflection which are not affected by the fluctuation of the steel plate distance and the speed. A signal can be obtained, and a surface flaw inspection apparatus capable of detecting a pattern-shaped barbed flaw having no remarkable unevenness without causing leakage can be realized.

Further, since the weighting function representing the extraction characteristic is different for each polarization angle, it is possible to discriminate defects having different normal angle distributions of the area ratio. Particularly, in the case of the stain-like defect, it is considered that some substance adheres to the small surface element having a small normal angle, and the area ratio of the small surface element having a large normal angle is almost the same as that of a normal portion. When polarized light of a certain degree is received, the difference between the defective part and the normal part tends to be small. Therefore, it is possible to discriminate stain-like defects by paying attention to this polarization angle.

Therefore, in the present invention, the light projecting portion is arranged so that polarized light is incident on the whole surface of the surface to be inspected at a constant angle of incidence with respect to the surface to be inspected, and light reflected from the surface to be inspected is received. The light receiving unit to be used is arranged at a predetermined position. The light receiving unit includes a beam splitter that separates the incident light into, for example, three beams, three linear array sensors including, for example, a CCD sensor that separately enters the three separated beams and outputs an image signal, and a beam. An analyzer is provided between the splitter and each of the linear array sensors to convert reflected light from the surface to be inspected into polarized light having different vibration planes. The three analyzers are arranged so that different azimuth angles, that is, angles formed by the transmission axis and the incident surface of the surface to be inspected are, for example, 0 degree, 45 degrees, and -45 degrees.

The signal processing section performs shading correction on the output image signal from each linear array sensor, normalizes the normal section so that the normal section has the central density of all gradations, flattens the image, and shows an image indicating a change relative to the normal section. Convert to a signal. A flaw candidate area showing a change larger than the density level representing a normal state is extracted from the normal part, and the length, width, area, density peak value and density are integrated from three types of image signals in the extracted flaw candidate area. Calculate feature values such as values. The type of flaw is determined from the relative ratio of the three integrated image signal amounts among the calculated characteristic amounts. Further, the grade of the flaw is determined by comparing the concentration peak value or the integral value with a value corresponding to a predetermined grade.

[0084]

FIG. 17 is an arrangement diagram showing an optical system according to an embodiment of the present invention. As shown in the figure, the optical system 21 is
It has a two- and three-plate polarization linear array camera 23. The light projecting unit 22 emits polarized light at a predetermined incident angle on the surface of the inspection object, for example, the steel plate 4, and includes a light source 24 and a polarizer 25 provided on the front surface of the light source 24. The light source 24 is composed of a rod-shaped light emitting light source and a cylindrical lens extending in the width direction of the steel plate 4 and irradiates light having a uniform intensity distribution over the entire width direction of the steel plate 4. The polarizer 25 is made of, for example, a polarizing plate or a polarizing filter, and is arranged such that the angle α between the transmission axis P and the incident surface of the steel plate 4 is 45 degrees, as shown in the layout explanatory diagram of FIG. The three-plate polarization linear array camera 23 includes a beam splitter 26, three analyzers 27a, 27b, 27c, and three linear array sensors 28a, 28b, 28c, as shown in the configuration diagram of FIG. The beam splitter 26 includes three prisms,
Two semi-transmissive reflecting surfaces on which a dielectric multilayer film is deposited are provided on the incident surface, and a first reflecting surface from which the reflected light from the steel plate 4 is incident.
The reflection surface 26a has a transmittance and a reflection ratio of about 2: 1, and the second reflection surface 26b on which light transmitted through the first reflection surface 26a has a transmission and a reflection ratio of 1: 1. The ratio is 1, and the light reflected from the steel plate 4 is separated into three beams of the same light amount. Further, the optical path lengths from the entrance surface of the beam splitter 26 to the exit surfaces of the three separated beams are the same. The analyzer 25a is provided in the optical path of the transmitted light of the second reflection surface 26b, and is arranged such that the azimuth angle, that is, the angle β formed between the transmission axis and the incident surface of the steel plate 4 is 0 degrees, as shown in FIG. , The analyzer 27b is provided on the optical path of the reflected light from the second reflecting surface 26b, and is disposed so that the azimuth angle β becomes 45 degrees, and the analyzer 27c is provided on the optical path of the reflected light from the first reflecting surface 26a. And the azimuth angle β is -45 degrees. Linear array sensors 28a, 28b, 2
8c is composed of, for example, a CCD sensor,
7a, 27b, and 27c are arranged at the subsequent stage. Also,
Beam splitter 26 and analyzers 27a, 27b, 27c
The slits 29a, 29b, and 29c for cutting multiple reflected light and unnecessary scattered light in the beam splitter 26 therebetween.
Is provided, and a lens group 30 is provided in front of the beam splitter 26. In addition, the linear array sensor 28
The gains of a, 28b, and 28c are adjusted so that the same signal is output when light having the same light intensity enters.

Since the analyzers 27a to 27c and the linear array sensors 28a to 28c are integrally provided on the optical paths of the three beams obtained by separating the incident light, the linear array sensors 28a to 28c are When detecting the reflected light from the steel plate 4 by arranging it near the conveyance path of No. 4, the position adjustment of the linear array sensors 28a to 28c is not required, and the reflected light from the same position of the steel plate 4 is detected at the same timing. can do. Also, three sets of linear array sensors 2 are provided in a three-plate polarization linear array camera 23.
Since 8a to 28c are collectively stored and miniaturized, the three-plate polarization linear array camera 23 can be easily arranged on the optical path of the reflected light of the steel plate 4, and the arrangement position can be arbitrarily selected. The degree of freedom of arrangement of the optical system 1 can be improved.

The linear array sensors 28a to 28c of the three-plate polarization linear array camera 23 are connected to a signal processing unit 31 as shown in the block diagram of FIG. The signal processing unit 31 includes a signal pre-processing unit 32a, 32b, 32c, an edge detection unit 33, a brightness unevenness correction unit 34, and a frame memory 3
5a, 35b, 35c, a binarization processing unit 36, a binary memory 37a, 37b, 37c, an OR processing unit 38, a binary memory 39, a flaw candidate area extraction unit 40, a feature amount calculation unit 41, and a flaw determination unit 42 Have. The signal preprocessors 32a to 32c average and average the image signals indicating the light intensities I1, I2, and I3 of the reflected lights output from the linear array sensors 28a to 28c, detect the movement amount of the line of the steel plate 4, and Is moved by the length of one line in the signal processing, an averaged signal is sent to the edge detection unit 33 as one-line data, and the length of one line in the signal processing is kept constant even when the line speed changes. The edge detection unit 33 detects an edge portion of the steel plate 4 in the I1, I2, and I3 images. The brightness unevenness correction unit 34 corrects the unevenness in intensity in the width direction of the I1, I2, and I3 images due to the unevenness in the intensity of the light source 24 and the unevenness in the reflectivity of the steel plate, and the unevenness in the sensitivity, and stores them in the frame memories 35a to 35c. Frame memory 35a
35c is constituted by, for example, 1024 horizontal pixels × 200 vertical lines. One line data of 1024 pixels is sampled at the same timing between the frame memories 35a to 35c, and is sequentially stored until reaching 200 lines to obtain a two-dimensional I1. , I2, I3 polarization images are formed. Binarization processing unit 3
6, I1, stored in the frame memories 35a to 35c.
Binarize the I2 and I3 polarized images and store them in binary memories 37a-3
7c. The OR processing unit 38 includes a binary memory 37a-
The binary image of I1, I2 and I3 stored in 37c is OR-processed and stored in the binary memory 39. The flaw candidate area extraction unit 40 specifies the position of the flaw candidate area from the density of each pixel of the binary image stored in the binary memory 39. Feature value calculation unit 4
1 is the light intensity I1, I in the extracted flaw candidate area.
2 and I3 are integrated from the normal part to obtain a concentration integrated value I.
S1, IS2, and IS3 are calculated to clarify the feature values.
The flaw determination unit 42 determines the light intensity I1, I2,
The type of the flaw is determined from the relative ratio of the concentration integrated values IS1, IS2 and IS3 of I3. Further, the flaw determining unit 42 obtains the maximum value of the integrated density values IS1, IS2, and IS3, compares it with a predetermined pattern, determines the flaw grade, and displays the determined flaw type and flaw grade on a display device (not shown). Or to a recording device.

Next, the operation when the surface of the steel plate 4 is inspected by the surface inspection apparatus configured as described above will be described. The three-plate polarization linear array camera 23 converts the polarized light emitted from the light projecting unit 22 and reflected on the surface of the steel plate 4 moving at a constant speed.
To receive light. The reflected light of the steel plate 4 that has entered the three-plate polarization linear array camera 23 is split by the beam splitter 26 and passes through the analyzers 27a, 27b, 27c and enters the linear array sensors 28a to 28c. When detecting the light intensity of the reflected light with the linear array sensors 28a to 28c,
Since the analyzers 27a to 27c having different azimuth angles are provided on the front surfaces of the linear array sensors 28a to 28c, the linear array sensors 28a to 28c have different light intensities I of different polarizations.
1, I2 and I3 are detected and sent to the signal processing unit 31.

The image signals I1, I2, and I3 of the respective polarized lights sent to the signal processing section 31 are respectively processed by the signal preprocessing sections 32a to 32a.
Even if the moving speed of the steel sheet changes at 32c, the length of one line in the signal processing is sent to the edge detecting unit 33 with the line length being fixed. The image signals I1, I2, sent to the edge detection unit 33
I3, as shown in FIG. 21, is that the signal level is high in the region of the steel plate 4 and the signal level is low in the background region other than the steel plate 4. And the signal processing area is determined. 1 of I1, I2 and I3 in this signal processing area
The signal intensity of the line has large unevenness in the width direction, for example, as shown in FIG. Therefore, the brightness unevenness correction unit 34
Is the number 1 on the left and right around the reference point in the width direction
Moving average is performed at the zero point to create moving averaged signals I1m, I2m, and I3m as shown in FIG. 21B. Then, as shown in FIG. 21C, the signal I before the moving average is calculated.
1, I2, I3 and the signals I1m, I2m, I
The correction signal I1c, I2c,
I3c is calculated and stored in the frame memories 35a to 35c. In the expression (12), A is a constant.

[0089]

(Equation 8)

The signals I1c and I2 corrected for the luminance unevenness
At c and I3c, as shown in FIG. 22C, the signal level of the flaw 51a that looks bright with respect to the normal part which is the background of the steel plate 4 is higher than the reference level C of the normal part 52 and darker than the normal part. The signal level of the visible flaw 51b is lower than the reference level C. The corrected signals I1c and I2
c and I3c are binarized by the binarization processing unit 36 to obtain I1c and I3c.
The binary images of 2c and I3c are respectively stored in a binary memory 37a.
To 37c. The binarization level at the time of binarization is determined according to the surface roughness of the steel plate 4 and the state of oiling of the surface. However, the binarization level is automatically obtained from the peak value or variation of the measured data to obtain a noise level. May be set. Also, the flaw may be higher or lower than the level of the normal part 52 depending on the type,
As shown in the figure, both the plus and minus binarization levels 53a and 53b are set for the normal level, and the binarization is performed.
The range other than the positive binarization level 53a and the negative binarization level 53b is “1”, and the range of the binarization levels 53a and 53b is “0”, which is a normal part.

The binarized image is composed of I3c, I2c and I3c.
There is an image, for example, as shown in FIG.
a and 51b are not necessarily detected as abnormal values in common for the three images. Therefore, the OR processing unit 38 uses the binary memory 37
As shown in FIG. 24B, the binary images I1c, I2c, and I3c stored in a to 37c are OR-processed for each pixel, and the OR-processed image 54 is stored in the binary memory 39. The flaw candidate area extraction unit 40 finds the position of the white portion of the value “1” shown in the flaw parts 51 a and 51 b of the OR processing image 54 stored in the binary memory 39, and as shown in FIG. The rectangular area circumscribing the part is the flaw candidate area 5
5a and 55b, and the extracted flaw candidate area 55
a and 55b, for example, P1 and P3 at the upper right and P at the lower left
The flaw candidate areas 55a and 55b are specified from the coordinates of the points P2 and P4, and are sent to the feature value calculation unit 41. The feature amount calculation unit 41 calculates the correction signal I for each pixel in the flaw candidate areas 55a and 55b.
1c, I2c and I3c are integrated, and the integrated density values IS1, IS2 and IS3 are obtained from the equation (13).

[0092]

(Equation 9)

The feature value calculating section 41 calculates a relative ratio IS2 of the integrated density values IS1, IS2, IS3 as a secondary feature value.
/ IS1, IS2 / IS3 and integrated concentration values IS1, IS
2, IS3, which is the maximum value among IS3, and the integrated values IS1, IS for the maximum integrated value IS.
2, ratio of IS3 IS1 / IS, IS2 / IS, IS3 /
The IS is obtained, and the length, width, area, and signal peak value of the flaw are also obtained from the coordinates of the flaw candidate areas 55a and 55b specified by the flaw candidate area extraction unit 40, and are sent to the flaw determination unit 42.

The flaw determining section 42 sends the flaw candidate area 55
The numerical pattern of the relative ratio of the flaw features IS1, IS2, IS3, and IS of a and 55b is compared with the numerical pattern of the flaw features previously obtained by experiments for a plurality of flaw types, and a match is selected. To determine the type of flaw. For example, water adhering to a minor defect portion of a cold-rolled steel sheet may rust and extend in the longitudinal direction, and a stain defect having an appearance similar to a scab defect which is a serious defect may occur. The stain-like defect looks dark to the ground like a barbed flaw when viewed by a human and has a high density and is easy to detect. However, since there are almost no irregularities, the harmfulness is lower than that of the barbed flaw. Therefore, if a grade is determined by recognizing a barbed flaw, it is overestimated, and it is necessary to discriminate it from a barbed flaw. Therefore, the relation between the relative ratios IS2 / IS1 and IS2 / IS3 of the concentration integrated values IS1, IS2, and IS3 for the stain-like defects and the bark defects,
Ratio IS of concentration integrated value IS2 to maximum concentration integrated value IS
When 100 or more flaws were collected and measured for 2 / IS, the results were as shown in FIGS. 25 and 26, respectively. FIG. 25 shows the relationship between the ratios IS2 / IS1 and IS2 / IS3, where (a) shows the case of a barbed flaw and (b) shows the case of a stain-like defect. FIG. 26 shows the distribution characteristics of the ratio IS2 / IS, where (a) shows the case of a barbed flaw and (b) shows the case of a stain defect. As shown in the figure, a stain-like defect whose surface is rusted and in the form of a film has a characteristic that the integrated value IS2 is smaller than the integrated values IS1 and IS3, and the ratio IS2 / IS is 1/4 or less. . On the other hand, in the case of a barbed flaw, the integrated density value IS2 is equal to the integrated density values IS1 and IS3.
And the ratio IS2 / IS also increases.
Therefore, the flaw determination unit 42 determines a flaw candidate area where the ratio IS2 / IS of the integrated density value IS2 to the maximum integrated density value IS is 1/4 to be a stain-like defect, and determines a flawed area above that. In this way, it is possible to discriminate between the scab defect and the stain defect. In addition to stain-like defects, defects called stains such as oil stains that have formed on the surface of the steel plate 4 in the form of a film can be similarly discriminated from scab defects using the ratio of IS1, IS2, and IS3. it can.

The flaw determining section 42 determines the grade of the flaw after determining the type of the flaw. The grade of the flaw is determined by judging whether the characteristic amount such as the maximum concentration integrated value IS or the signal peak value is a value corresponding to the grade determined in advance for the flaw type. For example, as a result of examining the relationship between the visual rating of the barbed flaw and the stain-like defect and the maximum concentration integrated value IS, the distribution shown in FIG. 27 was shown. In FIG. 27, (a) is a case of a scab,
(B) shows the case of a stain-like defect, in which grade D is a heavy defect than grade B. As shown in the figure, in the same grade, the stain-like defect has a larger maximum density integrated value IS. For example, when the grade determination is determined by the maximum concentration integrated value ls, if the stain-like defect cannot be discriminated from the barbed flaw, the grade determination of the stain-like defect will be overestimated. Since the discrimination can be made surely, the threshold level of the grade judgment can be set to be different between the barbed flaw and the stain-like defect, so that overestimation can be reduced.
FIG. 28 shows the results of the visual inspection grade, the case where the stain-like defect is not discriminated from the barbed flaw, and the number of the grade judgment flaws when the stain-like defect and the barbed flaw are discriminated at the ratio IS2 / IS <1/4. The following shows a specific example. In FIG. 28, (a) shows the case where the stain-like defect is not discriminated from the barb defect, and (b) shows the ratio IS2 / IS.
<1/4> shows the results of the grade determination in the case of discriminating between the stain-like defect and the scab defect. As shown in the figure, the ratio IS2 / IS <
When the grade of the flaw was determined after discriminating the stain-like defect from the barb flaw in 1/4, the same grade determination result as that in the case of the visual inspection could be obtained.

The optical system 1a shown in the side view of FIG. 29A and the top view of FIG. 29B may be used. This optical system 1
The light receiving portion 61 of FIG.
Analyzers 62a and 62 set at 45 degrees, 45 degrees, and 0 degrees
b, 62c, three linear array cameras 63a,
63b and 63c. The optical axes of the linear array cameras 63a to 63c are maintained parallel to each other. In addition, each linear array camera 63a-63
The shift in the visual field of c is corrected by the signal processing unit 31. The signal processing unit 31 performs binarization, flaw candidate area extraction, and feature value calculation for each signal from each of the linear array cameras 63a to 63c, and compares the representative coordinates of each flaw candidate area to thereby obtain a linear array. The flaw candidate areas of the cameras 63a to 63c are associated with each other.

In this embodiment, the same result as in the above embodiment can be obtained. When the optical axes of the linear array cameras 63a to 63c are maintained parallel to each other, the three linear array cameras 63a to 63c
The optical conditions of c are exactly the same, and each pixel has the same size. In addition, three linear array cameras 63a to 63c
Are arranged, the loss of the light amount is reduced as compared with the case where the beam splitter is used, and the measurement can be performed more efficiently. Further, by performing such signal processing, it is also possible to omit the position alignment for each pixel between the CCDs.

[0098]

As described above, according to the present invention, a polarized light is incident on a surface to be inspected at a constant incident angle, the light intensity distribution of a plurality of polarized lights having different reflected lights is detected, and the light intensity distribution of a plurality of polarized lights is detected. , A flaw candidate area is extracted from the flaw candidate area, an integral amount of light intensity signals of a plurality of polarizations having different flaw parts with respect to a normal part in the flaw candidate area is calculated, and a type of the flaw is determined from a feature amount indicating a ratio of the calculated integral amounts Thus, surface flaws that could not be discriminated by scattered light or diffraction can be discriminated. In addition, since the grade of the flaw is determined after the type of the flaw is determined, the detection accuracy of the flaw can be improved.

When determining the grade of a flaw, a change in the amount of light equivalent to visual observation is calculated from the image signal output from each light receiving optical system, and the calculated change in the amount of light is compared with a predetermined pattern to determine the grade of the flaw. From the determination, the grade of the flaw can be determined with high accuracy.

Further, when discriminating a stain-like defect and a scab defect generated in a cold-rolled steel sheet, the signal intensity integral of the light receiving optical system at a polarization angle of 45 degrees with respect to the maximum value of the signal intensity integral of the three light receiving optical systems. A defect having a ratio of less than 1/4 is determined as a stain defect, and the ratio of the signal intensity integral of the light receiving optical system at a polarization angle of 45 degrees to the maximum value of the signal intensity integral of the three light receiving optical systems is 1 Since a defect of / 4 or more is determined as a barbed flaw, it is difficult to determine the defect and a stain-like defect and a barbed flaw can be accurately determined. Therefore, the grade of the flaw can be determined in the same manner as in the case of the visual inspection.

[Brief description of the drawings]

FIG. 1 is an explanatory view showing a micro uneven shape on the surface of a steel sheet.

FIG. 2 is a schematic sectional view showing optical reflection on the surface of a steel plate.

FIG. 3 is an explanatory diagram showing an angular distribution of reflected light on a steel plate surface.

FIG. 4 is an explanatory view showing a scab defect.

FIG. 5 is an explanatory diagram showing a difference in an angular distribution of a reflected light amount on a steel plate surface.

FIG. 6 is an explanatory diagram showing a relationship between a normal angle and an area ratio of a temper portion.

FIG. 7 is an explanatory diagram showing a normal angle of a micro surface element.

FIG. 8 is an explanatory diagram showing a weight function.

FIG. 9 is an explanatory diagram showing reflection characteristics of light from a linear diffusion light source on a steel sheet surface.

FIG. 10 is an explanatory diagram showing reflection of light from a linear diffusion light source on a steel plate surface.

FIG. 11 is an explanatory diagram showing reflected light when linearly polarized light is incident on the surface of a steel sheet.

FIG. 12 is another explanatory diagram showing reflected light when linearly polarized light is incident on the surface of a steel sheet.

FIG. 13 is an explanatory diagram showing an elliptically polarized light state of reflected light from a minute surface element.

FIG. 14 is an explanatory diagram showing a relationship between a normal line angle of a small surface element and a weighting function.

FIG. 15 is another explanatory diagram illustrating a relationship between a normal angle of a microscopic element and a weight function.

FIG. 16 is an explanatory diagram showing reflection characteristics of a steel plate.

FIG. 17 is an arrangement diagram showing an optical system according to an embodiment of the present invention.

FIG. 18 is an explanatory view of the arrangement showing the operation of the optical system.

FIG. 19 is a configuration diagram of a three-plate polarization linear array camera.

FIG. 20 is a block diagram illustrating a configuration of a signal processing unit.

FIG. 21 is a signal intensity distribution diagram showing an edge detection operation.

FIG. 22 is an explanatory diagram showing a luminance unevenness correction operation.

FIG. 23 is a density characteristic diagram showing a binarization level.

FIG. 24 is an image diagram showing an operation of detecting a flaw candidate region.

FIG. 25 is a distribution characteristic diagram showing a relationship between flaw ratios IS2 / IS1 and IS2 / IS3.

FIG. 26 is a distribution characteristic diagram with respect to the flaw ratio IS2 / IS.

FIG. 27 is a distribution diagram showing a relationship between a visual rating of a barbed flaw and a stain-like defect and a maximum density integrated value IS.

FIG. 28 is an explanatory diagram showing a specific example of a result of determining the grade of a flaw.

FIGS. 29A and 29B show an optical system according to another embodiment, in which FIG. 29A is a side view and FIG. 29B is a top view.

[Explanation of symbols]

 Reference Signs List 4 steel plate 21 optical system 22 light projecting unit 23 three-plate polarizing linear array camera 24 light source 25 polarizer 26 beam splitter 27 analyzer 28 linear array sensor 31 signal processing unit 32 preprocessing unit 33 edge detection unit 34 brightness unevenness correction unit 35 frame Memory 36 Binarization processing unit 37 Binary memory 38 OR processing unit 39 Binary memory 40 Flaw candidate area extraction unit 41 Feature amount calculation unit 42 Flaw determination unit

Continued on the front page (72) Inventor Yuji Matoba 1-2-1, Marunouchi, Chiyoda-ku, Tokyo Nihon Kokan Co., Ltd. (72) Masakazu Inomata 1-2-1, Marunouchi, Chiyoda-ku, Tokyo Sun (72) Inventor Shoji Yoshikawa 1-2-1, Marunouchi, Chiyoda-ku, Tokyo Japan In-tube (72) Inventor Yoshiro Yamada 1-2-1, Marunouchi, Chiyoda-ku, Tokyo Japan Within Kokan Co., Ltd. (72) Inventor Takahiko Oshige 1-2-1, Marunouchi, Chiyoda-ku, Tokyo Nihon Kokan Co., Ltd. (72) Inventor Hiroyuki Sugiura 1-2-1, Marunouchi, Chiyoda-ku, Tokyo Nihon Kokan F term in the company (reference) 2G051 AA37 AB02 AB07 BA11 CA03 CA07 CB01 EA09 EA11 EB01 EC01

Claims (3)

[Claims] A light-emitting unit, a light-receiving unit, and a signal processing unit;
The light projecting unit receives polarized light on the surface to be inspected, and the light receiving unit has a plurality of light receiving optical systems for receiving polarized light at different angles in at least three directions. The signal processing unit normalizes the background signal of the surface to be inspected from the image signal output from each light receiving optical system so as to be a reference, and extracts a flaw candidate area from the variation with respect to the reference value,
Integrate the amount of change in signal intensity with respect to the signal reference value from each light receiving optical system in the extracted flaw candidate region, and determine the type of surface flaw from the relative ratio of the integrated signal intensity integral between each light receiving optical system. A surface inspection device characterized by determining. 2. A light-emitting device comprising: a light projecting unit, a light receiving unit, and a signal processing unit;
The light projecting unit receives polarized light on the surface to be inspected, and the light receiving unit has a plurality of light receiving optical systems for receiving polarized light at different angles in at least three directions. The signal processing unit normalizes the background signal of the surface to be inspected from the image signal output from each light receiving optical system so as to be a reference, and extracts a flaw candidate area from the variation with respect to the reference value,
Integrate the amount of change in signal intensity with respect to the signal reference value from each light receiving optical system in the extracted flaw candidate region, and determine the type of surface flaw from the relative ratio of the amount of integration of the signal intensity change between each light receiving optical system A surface inspection device for comparing the integral of the signal intensity change with a predetermined pattern to determine the grade of the flaw. 3. It has a light emitting unit, a light receiving unit, and a signal processing unit,
The light projecting unit enters a polarized light having a polarization angle of 45 degrees with respect to the incident surface of the surface to be inspected, and the light receiving unit has a polarization angle of 0 degrees, 45 degrees, with respect to the incident surface of the surface to be inspected. It has three light receiving optical systems for receiving polarized light of -45 degrees, detects reflected light reflected on the surface to be inspected, changes it into an image signal, and the signal processing unit outputs the image signal output from each light receiving optical system. Normalized so that the background signal of the surface to be inspected becomes a reference, a flaw candidate area is extracted from the variation with respect to the reference value, and the signal intensity variation from each light receiving optical system in the extracted flaw candidate area is integrated, The ratio of the signal intensity integral of the light receiving optical system at the polarization angle of 45 degrees to the maximum value of the integral of the signal intensity change of the three light receiving optical systems is 1 /
Defects less than 4 are determined as stain-like defects generated in the cold-rolled steel sheet, and the ratio of the signal intensity integral of the light receiving optical system at a polarization angle of 45 degrees to the maximum value of the integral of the signal intensity change of the three light receiving optical systems. Is determined to be a dent, and
A surface inspection device for a cold-rolled steel sheet, wherein a grade of the determined flaw type is determined.