JPH1019724A - Optical member inspection apparatus and optical member inspection method - Google Patents

Abstract

(57) [Problem] To provide an optical member inspection apparatus and an optical member inspection method capable of extracting information caused by an optical defect by removing information caused by a processing mark. Kind Code: A1 An optical member to be inspected is a knife edge.
Is located at the focal position. Therefore, the illumination light diverging from the knife edge 6a and entering the inspection target optical member A is converted into parallel light by the inspection target optical member A and is incident on the imaging device 7. Each image data imaged by the image sensor 7 at each rotation position of the knife edge 6a is subjected to differential processing by the image processing unit 14 and then combined as one image data. From this image data, extraction target image data from which only high-frequency components have been removed and image data from which only low-frequency components have been left are generated. The latter image data is slightly shifted in the high-luminance direction to become threshold data. Among the pixels constituting the extraction target image data, those having a luminance value larger than the pixel of the corresponding threshold data are extracted as indicating an optical defect.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an optical member inspection apparatus and an optical member inspection method for detecting an optical defect such as an abnormal refractive index of an optical member such as a lens. The present invention relates to an optical member inspection apparatus and an optical member inspection method that can effectively set a threshold value for determining an optical defect.

[0002]

2. Description of the Related Art An optical member such as a lens is designed such that an incident light beam is refracted regularly and travels in parallel, or converges or diverges at one point or linearly. However, if the refractive power (refractive index) is irregularly changed due to abnormal molding of the optical member, or if dust or scratches are generated on the surface of the optical member due to human handling after molding, the incident light flux may be reduced. The desired performance cannot be obtained because of the disturbance. In particular, in optical members such as lenses and prisms formed by injecting a resin into a mold and molding, in addition to the above-mentioned dust and scratches on the surface, sinking (resin is separated from the mold surface) due to molding abnormality. Depression caused by
Jetting (where the resin density is partially changed in the optical member) and flow marks (W-shaped wrinkles generated on the optical member surface due to resin shrinkage) are likely to occur. There is a need for efficient detection.

For this reason, the present inventor has filed an application for an optical member inspection apparatus capable of automatically detecting an optical defect of an optical member set as Japanese Patent Application No. 7-229242. The optical member inspection apparatus captures light transmitted through the optical member, a knife edge rotatably disposed at a focal position of the optical member, a diffuser disposed behind the knife edge to emit illuminating light. And an imaging device for performing the operation. According to the optical member inspection apparatus having such a configuration, an abnormal portion of refractive power (refractive index) on the surface or inside of the optical member appears as a rapidly changing portion of the light and dark density in a captured image. In addition, flaws, fine dust, and the like generated on the surface of the optical member are themselves imaged by the imaging device, and similarly appear as a portion where the light and dark density changes. Then, these suddenly changing portions of the light and dark densities are extracted, and their shapes or areas are quantified and used for pass / fail judgment.

[0004] Incidentally, it is inevitable that traces of cutting by a diamond tool or the like are formed on the surface of an optical member formed of glass during processing. In addition, it is inevitable that the trace of the cutting engraved on the surface of the molding die during the production of the molding die is transferred to the surface of the resin molded optical member. Although such processing marks are minute irregularities that do not affect the performance of the lens, the above-described optical member inspection apparatus extracts these processing marks together with the optical defect factors such as sink marks and scratches. As a result, accurate inspection could not be performed.

In this case, if the optical member to be inspected is a spherical lens, the above-described optical member inspection apparatus intentionally removes the position of the knife edge from the focal position of the optical member to be inspected, thereby extracting an optical defect. It is possible to reduce the sensitivity and prevent these processing traces from being extracted as an optical defect factor.

[0006]

However, when the optical member to be inspected is an aspherical lens, the focal position corresponding to each part of the lens is distributed over a certain range along the optical axis. However, the extraction sensitivity for the entire optical member to be inspected cannot be reduced uniformly.
That is, within the distribution range of the focal position, no matter where the knife edge is located, it matches the focal position corresponding to any part of the optical member to be inspected. It is because it is done. Further, if the knife edge is located outside the distribution range of the focal position, the extraction sensitivity becomes too low for the portion corresponding to the focal position farthest therefrom. This is because extraction fails.

[0007] As described above, it is not possible to extract only the optical defect while preventing the extraction of the processing trace of all kinds of optical members by merely adjusting the arrangement position of the knife edge.

In view of the above problems, an object of the present invention is to perform image processing on an image obtained by imaging an optical member to be inspected, thereby removing information caused by processing marks. An object of the present invention is to provide an optical member inspection apparatus and an optical member inspection method capable of extracting only information due to an optical defect and performing an accurate quality judgment based on the information.

[0009]

The invention described in each claim is
It has been made to solve the above problems. The invention according to claim 1 is an optical member inspection device for detecting an optical defect of an optical member, wherein the optical member inspection device is disposed at a focal position of a diffusion plate illuminated by illumination light and an optical system including the optical member. A light-shielding unit that is in contact with the diffusion plate so as to partially transmit light diffused by the diffusion plate; an imaging unit that captures light transmitted through the optical system; and an image captured by the imaging unit. A luminance change amount distribution data calculating means for calculating luminance change amount distribution data indicating a luminance change amount distribution in the image, and a luminance change amount distribution data calculated by the luminance change amount distribution data calculating means. First smoothing means for performing smoothing for removing high-frequency components, and extracting low-frequency components from the luminance variation distribution data calculated by the luminance variation distribution data calculating means. Second smoothing means for performing smoothing, and luminance change amount distribution data smoothed by the first smoothing means and luminance change amount distribution data smoothed by the second smoothing means. And extracting means for extracting only a part in which the former exceeds the latter by a predetermined amount or more.

The optical member includes a positive lens and a negative lens, a spherical lens and an aspherical lens, a glass molded lens and a resin molded lens. The optical defect of the optical member refers to a partial abnormality of the refractive index or the refractive power or a defect on the surface of the optical member.
Examples of abnormalities in the refractive index and the refractive power include sink marks, jetting, and flow marks in resin-molded optical members, and poor surface processing in optical members made of glass. The surface defects of the optical member include surface scratches, dirt, dust, and the like. However, fine irregularities that do not affect optical performance, such as processing marks, are excluded from the optical defects referred to herein.

The diffusion plate may be a translucent member illuminated from behind or a reflective member illuminated from the front side. The light-shielding means may be a light-shielding pattern directly printed on the surface of the diffusion plate, a plate-shaped opaque member may be appropriately cut out and attached to the diffusion plate, or may be separate from the diffusion plate. The light shielding pattern may be printed on the surface of the transparent member. The boundary line between the part that partially transmits light and the part that shields the light in the light shielding means may be a straight line or a curved line.
In addition, only one boundary line may be provided, or a plurality of stripe lines may be provided. Further, the boundaries may be oriented in multiple directions. This light-shielding means may be fixed when the boundary line is directed to a plurality of directions, but when only one boundary line is present or when all the boundary lines are directed to the same direction, It is desirable to rotate about a rotation axis that is tangent to the line. When the light shielding means rotates, the diffusion plate may rotate integrally, or only the light shielding means may rotate with the diffusion plate fixed.

The “optical system including an optical member” refers to the optical member itself as a positive lens or a positive lens group including an optical member as a negative lens. Each member is arranged such that the focal position of the entire optical system coincides with the position of the light shielding means.

[0013] The luminance variation distribution data may be calculated by differentiating the image captured by the imaging means, or may be calculated based on the difference between pixels separated by a predetermined distance. .

The frequency of the high-frequency component removed in the smoothing by the first smoothing means is equal to or lower than the frequency of a component caused by a processing mark formed on the surface of the optical member and higher than the frequency of a component caused by an optical defect. Is set. Also, the second
The frequency of the low frequency component extracted in the smoothing by the smoothing means is set to be equal to or higher than the frequency of the component caused by the shape of the surface of the optical member and lower than the frequency of the component caused by the optical defect. Each of these smoothing units may be a processor that executes an operation for smoothing, or may be configured as a filter circuit.

The extracting means directly compares the luminance variation distribution data smoothed by the first smoothing means with the luminance variation distribution data smoothed by the second smoothing means, and determines whether the former is the latter. A portion that exceeds the latter by a predetermined amount may be searched for. However, the luminance change amount distribution data smoothed by the second smoothing means is shifted in advance by this predetermined amount, and then the luminance thus shifted is shifted. The change amount distribution data may be compared with the luminance change amount distribution data smoothed by the first smoothing means.

According to a second aspect of the present invention, the light-shielding member of the first aspect includes a portion that partially transmits the light and a portion that partially shields the light, each of which is divided by a linear boundary line. That is what we have identified.

According to a third aspect of the present invention, in the second aspect, there is further provided a rotating means for rotating the light shielding means about a rotation axis in contact with the linear boundary in a plane of a contact surface with the diffusion plate. By having prepared, it is specified.

According to a fourth aspect of the present invention, the optical axis of the optical system including the optical member according to the third aspect is specified by being coincident with the rotation axis of the light shielding member. According to a fifth aspect of the present invention, the imaging means of the fourth aspect performs the imaging at each rotation position of the light shielding member rotated by the rotating means, and the imaging means during one rotation of the light shielding means. A plurality of brightness change amount distribution data respectively calculated by the brightness change amount distribution data calculating means based on each image captured by the image forming apparatus, further comprising a synthesizing means for synthesizing each other prior to each of the smoothing, It is specified.

According to a sixth aspect of the present invention, the luminance variation distribution data calculating means according to the first or fifth aspect divides an image taken by the image pickup device into a number of pixels and compares the luminance of each adjacent pixel with each other. The differential processing is performed by successive comparison, and the data obtained as a result of the differential processing is specified as the luminance variation distribution data.

According to a seventh aspect of the present invention, the extracting means of the first aspect shifts the entire luminance variation distribution data smoothed by the second smoothing means to a direction indicating that the luminance variation is large. And a comparing means for comparing the luminance variation distribution data shifted by the shifting means with the luminance variation distribution data smoothed by the second smoothing means. It is.

According to an eighth aspect of the present invention, the numerical means of the first aspect specifies the area by extracting the area of the part of the optical member extracted by the extracting means and defining the numerical value as the numerical value.

According to a ninth aspect of the present invention, in the first aspect, a numerical value digitizing means for digitizing the part extracted by the extracting means, and a numerical value digitized by the numerical value digitizing means determines a predetermined judgment reference value. The determination is further provided by a determination means for determining whether or not the value has been exceeded.

According to a tenth aspect of the present invention, there is provided an optical member inspection method for detecting an optical defect of an optical member, wherein the light is diverged from a diffusion plate, a part of the light is shielded by light shielding means, and a focal position of the light is adjusted. The image pickup device picks up light transmitted through an optical system including the optical member disposed so as to substantially coincide with the position of the means, and, based on an image obtained by the image pickup, shows a luminance change amount distribution in the image. Calculating the amount distribution data, performing first smoothing for removing high frequency noise on the calculated luminance change amount distribution data, and extracting a low frequency component from the calculated luminance change amount distribution data. 2, and compares the first smoothed luminance change amount distribution data with the second smoothed luminance change amount distribution data, and finds that the former exceeds the latter by a predetermined amount or more than the latter. Part By extracting the body, are those identified.

According to an eleventh aspect of the present invention, in the tenth aspect, the extracted portion is specified by digitizing and determining whether or not the digitized numerical value exceeds a predetermined determination reference value. It is.

According to a twelfth aspect of the present invention, the first smoothing means according to the first aspect is adapted to correspond to processing marks on the surface of the optical member from the luminance variation distribution data calculated by the luminance variation distribution data calculating means. The second smoothing means removes high-frequency components equal to or higher than the spatial frequency corresponding to the surface shape of the optical member from the luminance change distribution data calculated by the luminance change distribution data calculating means. It is specified by extracting only low frequency components.

[0026]

Embodiments of the present invention will be described below with reference to the drawings. <Overall Configuration of Optical Member Inspection Apparatus> FIG. 1 is a schematic configuration diagram showing an embodiment of an optical member inspection apparatus according to the present invention. As shown in FIG. 1, the illumination unit 4 and the imaging device 7 that constitute the optical member inspection device are arranged on the same device optical axis l so as to face each other.

An image pickup device 7 as an image pickup means includes an image pickup lens 8 which is a positive lens system as a whole, and an image pickup device 9 comprising a CCD area sensor for picking up an image formed by light converged by the image pickup lens 8. It is configured. The image captured by the imaging device 9 is input to the display device 10 and the image processing unit 14.

The display device 10 is an image display device used at the time of initial adjustment of the optical member inspection device (that is, at the time of adjusting the position of the illumination unit 4 described later). Image processing unit 14
Is a processor that determines whether the inspection target optical member A is a good product or a defective product, and includes first to fourth memories 14a to 14d. And the image processing unit 1
4 performs predetermined image processing on the image data input from the image sensor 9, quantifies the degree of the optical defect of the inspection target optical member A, and sets this numerical value to a predetermined determination reference value (allowable value). ) To determine whether this numerical value falls within or exceeds the determination reference value. That is, the image processing unit 14 includes a first smoothing unit, a second smoothing unit, an extracting unit (shifting unit, comparing unit), a digitizing unit,
It corresponds to the judgment means and the synthesis means. This image processing unit 1
4 also instructs the knife edge rotation control unit 15 to rotate the motor 13 in conjunction with performing the above-described determination processing.

The knife edge rotation control unit 15 rotates the motor 13 in accordance with a rotation instruction from the image processing unit 14. On the other hand, the illumination unit 4 as a whole
It can move forward and backward toward the imaging device 7. Inside the illumination unit 4, a disk-shaped diffuser plate 5 whose center is coaxial with the device optical axis 1 is held rotatably about the device optical axis 1 in a plane orthogonal to the device optical axis l. ing. A light shielding plate 6 as light shielding means is integrally attached to a surface of the diffusion plate 5 on the side of the imaging device 7. As shown in FIG. 2 which is a perspective view, the light shielding plate 6 is a semicircular plate made of an opaque member, and a radial straight line passing through the center of the diffusion plate 5 is defined as a chord (knife edge) 6a. It has an arc of the same radius as the diffusion plate 5. As a result of such a configuration, the light diffused by the diffusion plate 5 is partially shielded by the light shielding plate 6 and
Partially transmitted by the portion not covered by the

An annular gear 11 coaxial with the diffusion plate 5 is fixed to the periphery of the diffusion plate 5. This annular gear 11
Are engaged with a pinion gear 12, and the pinion gear 12 is connected to the motor 1 fixed in the lighting unit 4.
3 is attached to the rotating shaft. Therefore, the motor 13
Is rotated by the knife edge rotation control unit 15, the light shielding plate 6 and the diffusion plate 5 are rotationally driven via both gears 12 and 11, and are orthogonal to the apparatus optical axis l as shown in FIG. It is rotationally driven in the plane (that is, in the plane of the contact surface between the diffusion plate 5 and the light shielding plate 6). The rotation direction in this case is clockwise as viewed from the imaging device 7 side, as shown in FIG. As a result of this rotation, the knife edge (chord) 6a of the light shielding plate 6 also rotates about the device optical axis l. That is, the image processing unit 14, the knife edge rotation control unit 15, the motor 13, and the gears 11 and 12 constitute a rotation unit.

On the back surface side of the diffusion plate 5, the front end 3a of the optical fiber bundle 3 is fixed to the illumination unit 4. At the base end 3b of the optical fiber bundle 3, a light source device including an illumination lamp 1 and a condenser lens 2 is arranged. Then, the illumination light emitted from the illumination lamp 1 is condensed by the condenser lens 2 and enters the optical fiber bundle 3 from the base end 3b. This illumination light is transmitted through the optical fiber bundle 3, and the diffuser 5
Irradiated toward That is, the knife edge 6 of the light shielding plate 6
a is illuminated from behind. Note that the length of the optical fiber bundle 3 is set sufficiently longer than the movable distance of the illumination unit 4. Therefore, even if the illumination unit 4 moves, the optical fiber bundle 3 follows and can always illuminate the knife edge 6a of the light shielding plate 6.

The optical member A to be inspected is disposed between the imaging device 7 and the illumination unit 4 coaxially with the optical axis l of the device.
More specifically, when the optical member A to be inspected is a positive lens as shown in FIG. 1, the focal position of the optical member A coincides with the position of the knife edge 6a of the light shielding plate 6 on the device optical axis l. , Inspection target optical member A (optical system including optical member)
Is arranged. When the inspection target optical member A is a negative lens, a correction lens having a larger positive power than the inspection target optical member A in absolute value is provided between the inspection target optical member A and the illumination unit 4. Be placed. A lens group (an optical system including an optical member) including the inspection target optical member (negative lens) A and the correction lens (positive lens)
Is a lens group having a positive power as a whole, and the optical member (negative lens) A to be inspected has its composite focal point position coincident with the position of the knife edge 6a of the light shielding plate 6 on the device optical axis l. And a correction lens (positive lens). <Principle of Optical Defect Display> When the inspection target optical member A (and the correction lens) is arranged as described above, the illumination light diverging from the diffusion plate 5 and passing through the inspection target optical member A is As long as the optical member A is a good product, the light becomes parallel light.
Therefore, when viewed from the imaging device 7 side, this is equivalent to the knife edge 6a of the light shielding plate 6 being located at infinity.

By the way, if the focal position of the optical member A to be inspected (or the combined focal position of the optical system composed of the optical member A to be inspected and the correction lens, hereinafter the same), the optical axis l of the knife edge 6a If it is shifted to the imaging device 7 side from the upper position, an inverted image (real image) of the knife edge 6a is formed in the space between the inspection target optical member A and the imaging lens 8 of the imaging device 7. The inverted image (real image) of the knife edge 6a is relayed by the imaging lens 8, and an upright image (real image) of the knife edge 6a is formed in a space on the imaging element 9 side of the imaging lens 8. Conversely, when the focal position of the inspection target optical member A is shifted toward the optical fiber bundle 3 from the position of the knife edge 6a on the apparatus optical axis l, the space of the light shielding plate 6 on the side of the optical fiber bundle 3 causes the knife edge 6a to move forward. Standing image (virtual image)
Is formed. An upright image (virtual image) of this knife edge 6a
Is relayed by the imaging lens 8, and an inverted image (real image) of the knife edge 6a is formed in the space on the imaging element 9 side of the imaging lens 8. That is, the focal position of the inspection target optical member A is the object (knife edge 6a) existing at this position.
Is a boundary point between an image formed as an erect image and an image formed as an inverted image in the space on the image sensor 9 side of the image pickup lens 8, and is a position where it is in an optically unstable state.

The optical member A to be inspected and the imaging lens 8
Can be determined even if the focal position of the optical member A to be inspected is slightly shifted toward the imaging device 7 from the position of the knife edge 6a on the optical axis l of the device (to be exact, both positions). Are set as long as possible so that an inverted image (real image) of the knife edge 6a is formed between the focal positions of the knife edge 6a.
Further, the image pickup element 9 and the image formation position (average position) of the upright image and the upright image are formed so that these images can be captured to some extent clearly even if an upright image is formed or an inverted image is formed by the imaging lens 8. It is arranged at an intermediate point with the formation position (average position). This position is a position optically equivalent to the surface of the inspection target optical member A with respect to the imaging lens 8.

Therefore, a real image (inverted image) α of the outer edge of the optical member A to be inspected is always formed on the image sensor 9, and the real image α of the outer edge of the optical member to be inspected is The real image (inverted image) of the knife edge 6a that is directly visible without passing through the inspection target optical member A is slightly blurred (see FIGS. 4A to 4E).

Further, inside the real image α of the outer edge of the optical member A to be inspected, the focal position of the optical member A to be inspected (the combined focal point of the optical system composed of the optical member A to be inspected as a negative lens and the correction lens). When the position of the knife edge 6a is shifted to the imaging device 7 side from the position of the knife edge 6a on the device optical axis 1, the real image (erect image) of the knife edge 6a is slightly blurred and formed. (See FIGS. 4D and 4E). The real image (erect image) of the knife edge 6a has a larger blur amount as the shift amount of the focal position decreases (FIG. 4).
(See (d)), and the larger the shift amount, the smaller the blur amount and the more clear (see FIG. 4E).

On the contrary, when the focus position of the optical member A to be inspected is shifted toward the optical fiber bundle 3 from the position of the knife edge 6a on the optical axis 1 of the apparatus, the optical member A to be inspected is Inside the real image α of the outer edge, a knife edge 6a
The real image (inverted image) is slightly blurred (FIG. 4).
(See (b) and (a)). The real image (inverted image) of the knife edge 6a becomes larger as the shift amount of the focal position decreases (see FIG. 4B), and becomes clearer as the shift amount increases, as the blur amount decreases. FIG. 4 (a)).

When the focal position of the optical member A to be inspected coincides with the position of the knife edge 6a on the optical axis l of the apparatus,
The amount of blur inside the real image α of the outer edge of the inspection target optical member A is maximized, and light rays are irradiated with uniform brightness as a whole (see FIG. 4C).

In the image data input to the display device 10 and the image processing unit 14, the inner portion of the outer edge α of the optical member A to be inspected is such that the focal position of the optical member A to be inspected is the same as the position of the surface of the diffuser plate 5. When they match, the black part (the part where the illumination light is blocked) and the white part (the part where the illumination light is transmitted) of the knife edge 6a are completely mixed unless the optical member A to be inspected has an optical defect. Is displayed as a gray plane with uniform density (for a spherical lens). When an aspherical lens is inspected as the optical member A to be inspected, since the focal position changes not only at one point but gently, an image in which the luminance changes gently (at a low spatial frequency) is obtained. .

On the other hand, when there is a portion in the optical member A to be inspected having an abnormal refractive index or a portion having an abnormal refractive power due to a surface shape defect, only the abnormal portion is normal. This is equivalent to having a focal length different from the focal length of the part. Therefore, as shown in FIG. 5, the knife edge image γ appears only in the abnormal portion. The degree of the refractive index (refractive power) abnormality (focal length shift amount) of the abnormal portion is reflected in the appearance of the knife edge image γ. That is, the more clearly the shading of the image of the knife edge appears, the greater the degree of the refractive index (refractive power) abnormality (deviation of the focal length) becomes.

On the other hand, if dust adheres or is scratched on the surface of the optical member A to be inspected, the dust or scratch diffuses the illumination light from the illumination unit 4.
Of the illumination light diffused by dust or scratches in this manner, light incident on the imaging lens 8 of the imaging device 7 is converged by the imaging lens 8 and an image of these dusts or scratches is placed on the image sensor 9. tie.

The image data picked up by the image pickup device 7 is temporarily stored in the first memory 1 of the image processing unit 14.
After the image data is stored in the memory 4a, the image data is converted into image data (luminance change amount distribution data) indicating the distribution of the luminance change amount by a differentiation process, and is transferred to the second memory 14b. The above-described imaging and differentiation processing is performed every time the knife edge 6a rotates by 22.5 degrees until the knife edge 6a rotates once, and the image data as the differentiation result is sequentially accumulated in the second memory 14b and synthesized. You.

The reason for accumulating the differentiated image data obtained by slightly rotating the knife edge 6a in this manner is as follows. That is, when the linear knife edge 6a is inserted into the optical path, the refractive power (refractive index) abnormal component in the direction parallel to the direction of the knife edge 6a appears best in the image data, but is orthogonal to the direction of the knife edge 6a. Abnormal component of refractive power (refractive index) in the direction does not appear very well. Therefore, the knife edge 6a itself is rotated in a plane orthogonal to the optical axis l of the apparatus, and all abnormal components in all directions are detected and synthesized on the same image. As a result, the following effects can be obtained. That is, in the image in the case where the knife edge 6a is stopped, as shown in FIG. 5, in addition to the edge of the refractive power (refractive index) abnormal portion (the arc portion at the center in FIG. 5), the edge of the knife edge 6a (FIG. 5). A black-and-white boundary line extending to the left and right of the center) is also displayed as a portion where the shading changes rapidly. Since the edge of the knife edge 6a is not the edge itself of a portion having an abnormal refractive power (refractive index) which is originally required to be detected, it is desirable that the edge is not detected. Therefore, when the knife edge 6a is rotated, the edge of the knife edge 6a rotates while the position of the edge of the portion having an abnormal refractive power (refractive index) does not move. Therefore, when the image synthesis processing is performed, the edge of the abnormal portion of the refractive power (refractive index) is more and more emphasized, whereas the knife edge 6a
Are averaged in a planar manner in a closed region (portion surrounded by the edge) of a portion having an abnormal refractive power (refractive index), so that the edge is not recognized as a boundary line.

After one rotation of the knife edge 6a, the second
FIG. 6 shows an example of image data stored in the memory 14b. Since the example of FIG. 6 is image data obtained by imaging the optical member A to be inspected, which is an aspherical lens, the portion where the surface power (focal length) on the surface changes (the central portion δ, And the radially intermediate portion ε)
It is represented in a ring with high brightness. However, the circular stripes η appearing linearly at a finer pitch than the high-luminance annular portion are considered to be processing marks appearing on the surface of the optical member A to be inspected. Therefore, when the image data of FIG. 6 is obtained, only the granular region θ appearing between the central portion δ and the radially intermediate portion ε is a factor of optical defect (thin jetting). Is extracted as Then, the area of the extracted region θ is compared with a predetermined determination reference value, and a pass / fail determination is made.

The horizontal axis shown in FIG. 6 indicates the X-direction central axis extending in the X direction (horizontal direction) through the center position of the field imaged by the imaging device 7, and the vertical axis indicates the field. Through the center position of the Y direction (vertical direction)
Shows the central axis in the Y direction extending in FIG. <Principle of processing trace removal and extraction of optical defect factor> Next, based on the image data (see FIG. 6) on which the above-described differential processing and synthesis processing have been performed, only the processing trace is removed and only the optical defect factor is removed. The principle for extracting is described.

FIG. 8 shows a luminance distribution on the central axis in the X direction in the image data of FIG. Similarly, FIG. 9 shows a luminance distribution on the center axis in the Y direction in the image data of FIG. As is clear from these graphs, the high luminance portions δ and ε due to the aspherical lens which need not be extracted as an optical defect factor have a low spatial frequency as a whole. The granular high-luminance portion θ due to jetting or the like to be extracted as an optical defect factor has a higher spatial frequency than the high-luminance portions δ and ε. In addition, noise due to processing marks that do not need to be extracted as an optical defect factor (hereinafter referred to as “drawing noise”) η has a higher spatial frequency than the high luminance portion θ corresponding to the optical defect factor, It is about the same as high-frequency noise due to other factors (note that the pull noise η tends to be emphasized (amplitude increases) more in the Y direction than in the X direction. In the Y direction, the amplitude of the eye-catching noise η may be substantially equal to the amplitude of the granular high-luminance portion θ due to jetting or the like.

Here, in order to extract only the granular high-luminance portion θ due to jetting or the like without extracting the high-luminance portions δ and ε due to the aspheric surface as factors of optical defects, the following method is used. .

That is, first, for the luminance data of FIG.
Smoothing is performed so that only a spatial frequency lower than the spatial frequency of the granular high-luminance portion θ due to jetting or the like is left. This smoothing is performed by the moving average method of convolution size 2 ^4.

The outline of smoothing by the moving average method will be described below. FIG. 7 shows a data matrix (m × n) including pixels constituting image data. On this data matrix, first, the average of the luminance of an area consisting of k pixels by k pixels is calculated, and the calculation result is used as the new luminance of the pixel a located at the upper left corner of the area. Then, the area is shifted to the right by one pixel, a new luminance average is calculated, and this is set as the new luminance of the pixel b adjacent to the pixel a. The above-mentioned convolution size is the number k of vertical and horizontal pixels constituting a region to be calculated with the average luminance. Accordingly, the smoothing convolution size 2 ⁴ to the data matrix of Fig. 7, when performing, first, calculates the average luminance of 256 pixels _{P 1 ~ 32 · 1 ~ 32} , the pixel P ₁₁ New brightness, then P
And calculates the average luminance of 256 pixels _{1-32, 2} to ₃₃ as a new brightness of the pixel P _12, thereafter, repeats the same operation until the average luminance calculation target region reaches P _mn. Finally, correction for adjusting the center position of the image is performed. When such smoothing is performed, high-frequency components are removed from the original image data, and the luminance distribution is smoothed. Then, the lower limit of the frequency of the high-frequency component removed at this time becomes lower as the convolution size increases. That is, in the case of the above-mentioned convolution size 2 ^4, the high-intensity part of the aspherical late [delta], is the high component is removed than the spatial frequency of the epsilon.

[0050] The brightness distribution of the X-direction central axis in case of performing the smoothing convolution size 2 ⁴ for the luminance data of FIG. 6 shows in FIG. 8, FIG luminance distribution on the Y-direction central axis 9 Shown in Next, threshold data is generated by shifting the entire image data obtained by the smoothing in the high-luminance direction more than the amplitude of the high-luminance noise component. FIG. 8 shows a luminance distribution on the central axis in the X direction in this threshold data, and FIG. 9 shows a luminance distribution on the central axis in the Y direction. Then, for each pixel constituting the image data in FIG. 6, whether the luminance is higher or lower than that of the pixel at the corresponding position in the threshold data is compared (dynamic binarization), and the luminance corresponds to the threshold data. Only pixels larger than the location are extracted. As a result, the optical defect factor can be extracted without being affected by the low-frequency component due to the aspherical surface. (Note that the luminance data in FIG. 6 is smoothed with a larger convolution size. Is performed, even the high luminance portions δ and ε due to the aspherical surface are removed, so that dynamic binarization cannot be performed.)

However, according to such an extraction method, as described above, when the eye-catching noise η having the same amplitude as the high-luminance portion θ due to jetting or the like occurs,
It is not possible to extract only the high-luminance portion θ due to jetting or the like in distinction from the eye noise η. If the threshold data is further shifted in the high-luminance direction in order to avoid extraction of the eye noise η, the high-luminance portion θ itself due to jetting or the like cannot be extracted.

Therefore, it is necessary to effectively remove high-frequency noise such as the eye noise η. Here, it should be noted that the image of the processing mark has a higher spatial frequency than the image indicating the optical defect. Therefore, by utilizing this characteristic, the image data differentiation processing and synthesis processing is performed on the (see FIG. 6) itself, performs smoothing by small convolution size than convolution size 2 ^4.

[0053] Figure 10 is a luminance distribution of the luminance data of Figure 6 in the X-direction central axis, and a luminance distribution in a case where this smoothing convolution size 2 ¹ is made to a coordinate position x150~x380 It is a graph shown in the range of. Similarly, FIG. 11, the luminance distribution of the luminance data of Figure 6 in the X-direction central axis, and a luminance distribution in a case where this smoothing convolution size 2 ² has been made to a coordinate position x150~x38
It is a graph shown in the range of 0. Similarly, FIG. 12, the luminance distribution of the luminance data of Figure 6 in the X-direction central axis, and a luminance distribution in a case where this smoothing convolution size 2 ³ is made to the coordinate position x1
It is a graph shown in the range of 50-x380. FIG.
3, the luminance distribution of the luminance data of Figure 6 on the Y-direction central axis, and a luminance distribution in a case where this smoothing convolution size 2 ¹ is made to show a range of coordinate positions y100~x330 It is a graph.
Similarly, FIG. 14, the luminance distribution of the luminance data of Figure 6 on the Y-direction central axis, and a luminance distribution in a case where this smoothing convolution size 2 ² has been made to a coordinate position y100~x330 It is a graph shown in the range of. Similarly, FIG. 15, the luminance distribution of the luminance data of Figure 6 on the Y-direction central axis, and a luminance distribution in a case where this smoothing convolution size 2 ³ is made to a coordinate position y100~x330 It is a graph shown in the range of. These As is apparent from the respective graphs, if up to convolution size 2 ^2, even if the smoothing is performed, the amplitude of the high-intensity part θ due to the optical defects are hardly affected. However, if the smoothing convolution size 2 ¹ has been performed, the amplitude of the kerf noise η is not smoothed only about half. On the other hand, if the smoothing convolution size 2 ² is made, whereas the amplitude of the high luminance portion θ due to the optical defects are hardly affected, kerf noise η is almost completely removed You. That is, convolution size 2
According to the smoothing of ² , it is possible to remove a frequency component higher than the spatial frequency of the high luminance portion θ due to an optical defect.

Therefore, as shown in FIG. 16 showing the luminance distribution on the central axis in the X direction and FIG. 17 showing the luminance distribution on the central axis in the Y direction, First, convolution size 2 ²
The image data left to generate a extraction target image data obtained by removing high-frequency noise such as kerf noise η performs smoothing, only a component of the aspherical late performs smoothing by convolution size 2 ⁴ at Generate next,
This image data is shifted to the high luminance side (the side indicating that the luminance change amount is large) by an amount (luminance 30) including a margin more than the amplitude reduced by the smoothing,
Generate threshold data. Then, the luminance of each pixel constituting the extraction target image data is compared with the luminance of the pixel at the same position in the threshold data, and extraction is performed only when the luminance is higher than that. As a result, only the high-luminance portion θ caused by the optical defect is extracted. <Image Processing> Next, the contents of the image processing executed by the image processing unit 14 to determine whether the inspection target optical member is a non-defective product or a defective product will be described with reference to the flowchart of FIG. .

This image processing is started by pressing an inspection start button (not shown) connected to the image processing section 14. Then, loop processing of S01 to S05 is executed.

In the first step S01 after entering this loop processing, the image processing 14 converts the luminance of each pixel (pixel) constituting the image data input from the image pickup device 9 into numerical information of 256 gradations. Write to the first memory 14a.

In the next S02, the image processing section 14 sets the first
The numerical information written in the memory 14a is sequentially scanned to perform a differentiation process (corresponding to a luminance distribution data calculation unit). That is, the numerical value of each pixel is checked in order from the upper left pixel to the lower right pixel in the image. And
The numerical value of the pixel to be checked is compared with the numerical value of the pixel on the left and the numerical value of the pixel adjacent on the upper side, and the absolute value of the difference between the numerical values is calculated as the differential value [0-25
5]. In the image data (brightness change amount distribution data indicating the brightness change amount distribution in the image) converted into the differential value in this manner, the portion where the optical member A to be inspected has an optical defect, the contour of the processing mark, and the focus Only the portion where the distance is changed and the edge of the knife edge 6a form an image with high density.

In the next step S03, the image processing section 14 executes an image synthesizing process (corresponding to synthesizing means). That is, S02
Are written in the second memory 14b. At this time, as a result of S03 in the previous loop processing, the differential value with respect to the previous image data is stored in the second memory 14
b, the image processing unit 14 takes out each differential value already written in the second memory 14b, and after adding each differential value obtained in S02 in the current loop processing, Overwrites the second memory 14b.

In the next S04, the image processing section 14 checks whether or not the number of image data input after starting the processing has reached 16 images. If the number of input image data has not reached 16 images, S
At 05, a command to rotate the knife edge 6a by 22.5 degrees is given to the knife edge rotation control unit 15 (corresponding to a rotating means). The image data after the rotation is stored in the image sensor 9
, The process returns to S01, and the process for the new image data is executed.

As a result of repeating the loop processing as described above, when the number of input image data reaches 16 images,
The process exits from the loop process from S04 and proceeds to S06. In S06, the image processing unit 14 reads out the image data (image data on which the differentiation process and the synthesis process have been performed) stored in the second memory 14b at that time,
Smoothing convolution size 2 ² against the readout image data (first removing the high frequency component
, Which corresponds to the first smoothing means).

In the next S07, the image processing unit 14
The extraction target image data obtained as a result of
It is stored in the memory 14c. In the next S08, the image processing unit 14 reads out the image data (image data on which the differentiation processing and the synthesis processing have been performed) stored in the second memory 14b at that time, and performs convolution on the read out image data. smoothing of size 2 ⁴ (second smoothing for extracting a low frequency component, corresponding to the second smoothing unit).

In the next step S09, the image processing unit 14
8 is stored in the fourth memory 1
4c. In the next S10, the image processing unit 14
As a result of S09, the values of all pixels of the image data stored in the fourth memory 14c are each increased by [+30]. As a result, threshold data is generated (corresponding to a shift means for shifting the entire luminance variation distribution data to the side indicating that the luminance variation is large).

In the next step S11, the image processing section 14 executes an inter-image operation. That is, the value of the corresponding pixel (that is, located at the same position) in the threshold data stored in the fourth memory 14d is calculated from the value of each pixel constituting the extraction target image data stored in the third memory 14c. (Corresponding to comparison means for comparing the luminance change amount distribution data shifted by the shift means with the luminance change amount distribution data smoothed by the second smoothing means).

In the next step S12, the image processing section 14 executes an inspection target area extraction process. FIG. 18 is a flowchart showing the contents of the inspection target area extraction processing subroutine executed in S12. In the first step S21 after entering this subroutine, the image processing section 14 performs a binarization process. This binarization processing means that if the numerical information corresponding to each pixel of the image data in the third memory 14c exceeds a predetermined threshold, the numerical information is replaced with 255 (white),
If it does not exceed, the process is replaced with 0 (black). This threshold value is set to a value such that the outer edge α of the inspection target lens A can be left as a white (255) closed curve without interruption.

In the next S22, the image processing section 14 executes a closed area extraction process. This closed region extraction process is a process of extracting only a region surrounded by a closed white line. Specifically, of the black pixels [0] constituting the image data binarized in S21, the white pixels [25]
5] are regarded as pixels in the closed area. Then, the numerical values of all pixels regarded as being within this closed area are set to 255, and the numerical values of all other pixels are set to 0.

In the next step S23, the image processing section 14 executes a filling process. This filling process is performed by using a white pixel [2
55] is a process for erasing the black pixel [0] left in [55]. Specifically, of the black pixels [0] constituting the image data obtained in S21, white pixels [2]
55] is 255.

In the next step S24, the image processing section 14 executes an area selecting process. The region selection process is a process for validating only a region originally required and deleting other closed regions extracted based on a part of a gate or the like. Specifically, the image processing unit 14
Among the closed regions included in the image data obtained in step 3, the closed region located at the center of the screen is left as it is, and the numerical values of all the pixels constituting the other closed regions are set to 0. The image data obtained as a result of this area selection processing is hereinafter referred to as a “mask image”.

In the next step S25, the image processing section 41 executes an extraction operation. That is, the brightness of the mask image is inverted (pixels of brightness [255] are brightness [0], and pixels of brightness [0] are brightness [255]), and each pixel of the image data in the third memory 14c. Is subtracted from the value (luminance) of the corresponding pixel in the mask image after the inversion. As a result of this extraction operation, white pixels [25] in the original mask image in the image data in the third memory 14c
Only the part corresponding to the area of [5] is left as it is, and the numerical values of the pixels of the other parts are all [0].

As described above, image data used for determining a good or defective product is obtained. This subroutine ends, and the process returns to the main routine of FIG. In the next S13, the image processing unit 14 performs a dynamic binarization process based on the inspection target region extraction process result in S12.
That is, if the image calculation result has a positive polarity, the pixel value is replaced with 255, and if the image calculation result has a zero or negative polarity, the pixel value is replaced with 0.

In the next step S14, the image processing section 14 executes a pass / fail determination process. That is, the graphic feature amounts (the area, the maximum width, the center of gravity, the fillet diameter, etc., of the white portion [the portion where the luminance value is 255]) of the image composed of the pixels obtained as a result of the binarization in S13 are determined calculate. For example, the number of white pixels [having a luminance value of 255] is counted and defined as an area amount (corresponding to a digitizing unit for digitizing a part extracted by the extracting unit). Then, each of the calculated graphic feature values is compared with each of the preset pass / fail judgment reference values, and if at least one of the graphic feature values exceeds the corresponding pass / fail judgment reference value, it is rejected ( (Defective), and if all of the graphical feature values fall within the corresponding pass / fail judgment reference values, pass (non-defective) is judged (the numerical value digitized by the digitizing means is a predetermined judgment reference value). ). Which of the graphic feature amounts used for the determination is used for the pass / fail determination depends on the type of the optical member to be inspected. Upon completion of the pass / fail determination processing, the image processing ends.

As the image synthesizing process executed in S03 in the above-described image processing, a MAX operation may be performed. In the above-described addition processing, the numerical values (luminances) of the corresponding pixels between the image data to be combined are added to each other. In the MAX operation, the numerical values (luminances) of the corresponding pixels between the image data to be combined are larger. This value is adopted as the calculation result. In such a MAX operation, even when the luminance of each image data to be synthesized is high, the luminance of the synthesized image is not saturated. Therefore, there is an advantage that accurate inspection can be performed even in such a case. <Inspection Procedure by Optical Member Inspection Apparatus> When inspecting the optical member A by the optical member inspection apparatus according to the present embodiment, the inspector fixes the optical member A to be inspected to a holder (not shown) and To place. When the inspection target optical member A is a negative lens, a correction lens is inserted between the inspection target optical member A and the illumination unit 4.

Next, the inspector turns on the illumination lamp 1 to illuminate the knife edge 6 a of the light shielding plate 6. Then, an image captured by the image sensor 9 is displayed on the display device 10.

The examiner moves the lighting unit 4 while watching the image displayed on the display device 10. And
As shown in FIG. 4A or 4B, when the knife edge γ is seen inside the outer edge α of the inspection target optical member A in the same direction as the knife edge β seen outside the outer edge α of the inspection target optical member A. Since the illumination unit 4 is too close to the optical member A to be inspected, the illumination unit 4 is moved away from the optical member A to be inspected. Conversely, as shown in FIG. 4D or FIG. 4E, the knife edge in the direction opposite to the knife edge β that is visible outside the outer edge α of the inspection target optical member A inside the outer edge α of the inspection target optical member A. When γ is visible, the illumination unit 4 is too far from the optical member A to be inspected.
Approach. When the knife edge γ disappears in most of the outer edge α of the inspection target optical member A as shown in FIG. Because it is in
Stop adjustment.

Next, the examiner presses an inspection start button (not shown) to start the image processing of FIG. Then, the knife edge 6a is rotationally driven by 22.5 degrees by the knife edge rotation control unit 15 (S05), and an image formed by light passing through the inspection target optical element A at each rotation position is captured by the imaging device 7. (S01). The image processing unit 14 emphasizes the density change portion of each captured image by differential processing (S02), and adds over one rotation (S03, S04). as a result,
Regarding defect components (refractive index [refractive power] abnormality, surface defect) in any direction of the inspection target optical member, they are collected into one image data.

[0075] The image processing unit 14, the image data by performing a smoothing convolution size 2 ² against, and generates an extraction target image data to remove high frequency noise such as kerf noise (S06), After smoothing the convolution size 2 ⁴ , it is shifted in the direction of high luminance, and the surface shape (spherical or aspherical surface) of the optical member to be inspected
The threshold data reflecting only the component caused by the above is created (S08, S10).

Then, the image processing unit 14 subtracts the threshold data from the image data to be extracted (S11).
Only a portion corresponding to the image of the optical member to be inspected is extracted from the image data obtained as a result of the subtraction (S12), and a dynamic binarization process is performed (S13), whereby a portion where an optical defect occurs is obtained. Extract only Then, the area and the maximum width of the optically defective portion are quantified and compared with a predetermined criterion value, and it is objectively determined whether the product is good or defective according to the comparison result. Yes (S1
4).

[0077]

According to the optical member inspection apparatus and the optical member inspection method of the present invention configured as described above, image processing is performed on an image obtained by imaging the optical member to be inspected. Accordingly, it is possible to remove the information resulting from the processing traces, extract only the information resulting from the optical defect, and make an accurate pass / fail judgment based on this information.

[Brief description of the drawings]

FIG. 1 is a schematic diagram of an optical member inspection apparatus according to an embodiment of the present invention.

FIG. 2 is a perspective view showing a rotating state of a light shielding plate.

FIG. 3 is a flowchart showing the contents of image processing executed in the image processing unit of FIG. 1;

FIG. 4 is a diagram showing an image on a display device when the movement of the knife edge unit in FIG. 1 is adjusted.

FIG. 5 is a diagram showing an image on a display device when an optical member having a sink is inspected.

FIG. 6 is a view showing image data that has been subjected to differentiation processing and synthesis processing written in a second memory in FIG. 1;

FIG. 7 is an explanatory diagram of smoothing by a moving average method.

8 is a graph showing the image data of FIG. 6, convolution size 2 ⁴ of the image data obtained by smoothing, and the luminance distribution in the X-direction central axis on the threshold data

9 is a graph showing the image data of FIG. 6, convolution size 2 ⁴ of the image data obtained by smoothing, and the brightness distribution in the Y-direction central axis on the threshold data

[10] image data of FIG. 6, and the image data obtained by smoothing convolution size 2 ¹ X
Graph showing luminance distribution on the central axis in the direction

[11] image data of FIG. 6, and the image data obtained by smoothing convolution size 2 ² X
Graph showing luminance distribution on the central axis in the direction

[12] image data of FIG. 6, and the image data obtained by smoothing convolution size 2 ³ X
Graph showing luminance distribution on the central axis in the direction

[13] image data of FIG. 6, and the image data obtained by smoothing convolution size 2 ¹ Y
Graph showing luminance distribution on the central axis in the direction

[14] image data of FIG. 6, and the image data obtained by smoothing convolution size 2 ² Y
Graph showing luminance distribution on the central axis in the direction

[15] image data of FIG. 6, and the image data obtained by smoothing convolution size 2 ³ Y
Graph showing luminance distribution on the central axis in the direction

FIG. 16 shows the image data of FIG.
Showing the luminance distribution of the threshold data and the threshold data on the central axis in the X direction

FIG. 17 shows the image data of FIG.
Showing the luminance distribution of the threshold data on the central axis in the Y direction

FIG. 18 is a flowchart showing an inspection target area extraction processing subroutine executed in S12 of FIG. 3;

[Explanation of symbols]

 Reference Signs List 4 lighting unit 5 diffusion plate 6 light shielding plate 7 imaging device 8 imaging lens 9 imaging device 13 motor 14 image processing unit 15 knife edge rotation control unit A optical member to be inspected

 ──────────────────────────────────────────────────続 き Continued on the front page (72) Inventor Toshihiro Nakayama 2-36-9 Maenocho, Itabashi-ku, Tokyo Asahi Kogaku Kogyo Co., Ltd.

Claims (12)

[Claims] 1. An optical member inspection apparatus for detecting an optical defect of an optical member, comprising: a diffuser plate illuminated by illumination light; and a diffuser disposed at a focal position of an optical system including the optical member; A light shielding unit in contact with the diffusion plate so as to partially transmit the light diffused by the plate; an imaging unit for imaging light transmitted through the optical system; and A luminance change amount distribution data calculating means for calculating luminance change amount distribution data indicating a luminance change amount distribution in the image; and a high frequency component for the luminance change amount distribution data calculated by the luminance change amount distribution data calculating means. A first smoothing unit for performing smoothing for removing the image data, and extracting only low-frequency components from the luminance change amount distribution data calculated by the luminance change amount distribution data calculation unit. A second smoothing unit for performing smoothing, and a comparison between the brightness variation distribution data smoothed by the first smoothing unit and the brightness variation distribution data smoothed by the second smoothing unit. And an extracting means for extracting a portion in which the former exceeds the latter by a predetermined amount or more. 2. The light-shielding member according to claim 1, wherein the light-shielding member includes a part partially transmitting the light and a part partially shielding the light, each part being separated by a linear boundary line. An optical member inspection device according to claim 1. 3. The apparatus according to claim 2, further comprising rotating means for rotating said light shielding means about a rotation axis which is in contact with said linear boundary in a plane of a contact surface with said diffusion plate. Optical member inspection device. 4. The optical member inspection apparatus according to claim 3, wherein an optical axis of the optical system including the optical member coincides with a rotation axis of the light shielding member. 5. The image pickup means performs the image pickup at each rotation position of the light shielding member rotated by the rotation means, and each image picked up by the image pickup means while the light shield means makes one rotation. The plurality of brightness change amount distribution data respectively calculated by the brightness change amount distribution data calculation means based on
5. The optical member inspection apparatus according to claim 4, further comprising a synthesizing unit that synthesizes each other before each of the smoothing. 6. The luminance change amount distribution data calculating means divides an image taken by the imaging device into a number of pixels, sequentially compares the luminance of each adjacent pixel, and performs a differentiation process. The optical member inspection apparatus according to claim 1, wherein data obtained as a result is used as the luminance change amount distribution data. 7. A shift means for shifting the entire luminance change amount distribution data smoothed by the second smoothing means to a side indicating that the luminance change amount is large, 7. The shift means 2. The optical member inspection apparatus according to claim 1, further comprising comparison means for comparing the shifted luminance variation distribution data with the luminance variation distribution data smoothed by the second smoothing means. 8. The optical member inspection apparatus according to claim 1, wherein said numerical value means measures an area of a portion of said optical member extracted by said extracting means and sets the area as said numerical value. 9. Numerical means for digitizing a part extracted by said extracting means, and judging means for judging whether or not the numerical value digitized by said numerical means has exceeded a predetermined judgment reference value. The optical member inspection device according to claim 1, further comprising: 10. An optical member inspection method for detecting an optical defect of an optical member, wherein the optical member is diverged from a diffusion plate, a part of which is shielded by light shielding means, and a focal position of the optical member substantially coincides with the position of the light shielding means. Imaging the light transmitted through the optical system including the optical member disposed so as to calculate luminance variation distribution data indicating the luminance variation distribution in the image based on the image obtained by the imaging. Performing first smoothing for removing high-frequency noise on the calculated luminance variation distribution data, and performing second smoothing for extracting low-frequency components on the calculated luminance variation distribution data Comparing the first smoothed luminance variation distribution data with the second smoothed luminance variation distribution data, and extracts a portion where the former exceeds the latter by a predetermined amount or more. Specially An optical member inspection method according to. 11. The optical member inspection method according to claim 10, wherein the extracted part is digitized, and it is determined whether or not the digitized numerical value exceeds a predetermined determination reference value. 12. The first smoothing means removes high-frequency components higher than a spatial frequency corresponding to a processing mark on the surface of an optical member from the luminance change amount distribution data calculated by the luminance change amount distribution data calculating means. The second smoothing unit calculates the luminance change amount distribution data calculated by the luminance change amount distribution data calculation unit,
2. The optical member inspection apparatus according to claim 1, wherein only low frequency components lower than the spatial frequency corresponding to the surface shape of the optical member are extracted.