JP6500432B2 - Defect detection apparatus, inspection apparatus, cleaning apparatus, defect detection method, and defect detection program - Google Patents

Description

  The present invention relates to a defect detection apparatus, an inspection apparatus, a cleaning apparatus, a defect detection method, and a defect detection program.

Conventionally, there has been known a method of detecting as a defect a dense object of non-removable minute defects such as foreign particles or flaws fixed on the surface of a subject to be inspected or foreign particles or bubbles existing inside the object (eg, patent) Reference 1 and 2).
In the method described in Patent Document 1, first, based on the normal sample and the representative defect sample, a distance threshold according to the type of the inspection object and the content of the defect is set. Next, among the defect candidates in the captured image of the inspection object, the defect candidate having a distance less than the distance threshold is combined, and the area of the combined defect candidate (hereinafter sometimes referred to as “combined defect candidate”) At the same time, the type of joint defect candidate is determined. Then, an area threshold is set according to the type of bonding defect candidate, and when the area exceeds the area threshold, the object to be inspected is determined to be a defective product, and when less than the area threshold, it is determined to be a non-defective product. .
In the method described in Patent Document 2, defect candidates having a high possibility of micro defects are extracted based on an inspection image, and a combination process is performed to detect a state in which the defect candidates are concentrated. In this combining process, using extraction rods, defect candidates having an inter-edge distance shorter than a predetermined value are combined as one combined defect candidate. Then, based on the number of pixels constituting the joint defect candidate, the joint defect candidate is detected as a defect of the inspection pattern.

  That is, according to the methods described in Patent Documents 1 and 2, defect candidates corresponding to minute defects are extracted from a captured image, and after combining defect candidates having a distance of not more than the distance threshold as coupling defect candidates, It can be said that binding defect candidates exceeding the size lower limit are detected as defects of the object to be inspected.

JP-A-8-101130 Unexamined-Japanese-Patent No. 2004-12177

However, the methods described in Patent Documents 1 and 2 have the following problems because defects are detected in one processing condition for one captured image.
For example, in the case where the object to be inspected is a cover glass of a liquid crystal panel, when detecting an irremovable minute defect cluster as described above as a defect (hereinafter, "detect a minute defect cluster as a defect" The criterion of “detecting a minute defect crowded body” is whether or not it is visible to the user. When the density of the micro defect cluster is large (the micro defects are dense), it is necessary to detect because the cluster size is not very small and it is visible to the user. On the other hand, if the density of the defect mass is small (the micro defects are not dense), it is not necessary to detect because the density size is small, so it is not necessary to detect it. .

As shown in FIG. 1 (A), the above-mentioned matter belongs to the visible region VS when it is expressed in a graph format in which the horizontal axis represents the density of minute defects and the vertical axis represents the concentration of minute defects. Although it is necessary to detect a micro defect cluster, it is not necessary to detect a micro defect cluster belonging to the invisible region NV.
In addition, there may be a case where a removable attachment such as dust, dirt or water droplets adheres to the surface of the inspection object. Such deposits generally exist randomly and often have a long distance from one another as compared to the microdefects. In addition, the deposits do not have to be detected as defects because they have no problem in quality.

Here, from the viewpoint of defect detection processing using a captured image, the relationship shown in FIG. 1 (A) can be grasped as shown in FIG. 1 (B).
For example, a micro defect cluster belonging to the visible region VS and having a small density has a long distance between the micro defects as compared with the case of a high density. For this reason, if the distance threshold is increased to make it easy to combine defect candidates, the size of the combined defect candidate is unlikely to be equal to or larger than the size lower limit, and it is difficult to detect the combined defect candidate as a defect. On the other hand, in the case of a minute defect cluster which belongs to the visible region VS and has a high density, the distance between the minute defects is short. Therefore, even if the distance threshold is small, defect candidates are combined and the size of the combined defect candidate tends to be equal to or larger than the size lower limit, and the combined defect candidate is easily detected as a defect.

Here, if the distance threshold value is increased in order to detect a minute defect crowded body having a low density, defect candidates corresponding to an adherend in addition to the combined defect candidate in which the defect candidates corresponding to the minute defects are joined There is also a possibility that a binding defect candidate in which is bound is generated. In this case, the bonding defect candidate corresponding to the deposit is detected as a defect, and the number of detected defects may be larger than the number of defects which need to be actually detected.
As described above, it is considered that the size of the bonding defect candidate is unlikely to increase because the deposits generally exist randomly. Further, the detection target is a minute defect dense body having a large dense size even if the density is low. For this reason, by increasing the size lower limit value, it is possible to detect the binding defect candidate of the minute defect crowd as the defect without detecting the binding defect candidate of the deposit as the defect.
On the other hand, if the distance threshold is reduced to detect a minute defect crowded body having a high density, the size of the joint defect candidate is unlikely to be increased. Therefore, by decreasing the size lower limit, it is possible to detect a minute defect crowded body having a large density.

  If the above is shown in the graph format in which the horizontal axis is set to the size threshold and the vertical axis is set to the distance threshold as shown in FIG. The region TG corresponds to the visible region VS in FIG. 1A, and it can be considered that the detection unnecessary region NT defined by the coupling defect candidate which does not require detection corresponds to the invisible region NV. In addition, as the size lower limit decreases, or as the distance threshold increases, that is, as the lower left of FIG. 1B, it can be considered that binding defect candidates for the deposit that can be detected as defects increase.

Therefore, as described in Patent Documents 1 and 2, in order to detect all the minute defect crowds that need to be detected under one processing condition for one captured image, it is shown in FIG. 1 (C). Thus, the size lower limit value is set to the minimum value DS0 that can detect the coupled defect candidate in the necessary detection area TG as a defect, and the distance threshold is set to the maximum value DL0 that can detect the coupled defect candidate as a defect. It is conceivable.
However, when the size lower limit value and the distance threshold value are set in this manner, the coupled defect candidate included in the detection area AD0 as shown in FIG. 1C is detected as a defect. Since the detection area AD0 includes the detection unnecessary area NT having the same size as the detection necessary area TG, there is a possibility that the possibility of detection of a coupling defect candidate which does not require detection as a defect becomes high. In particular, when the inspection is performed in an environment in which the adherend easily adheres to the object to be inspected, the object to be inspected which is determined to be a defective product may frequently occur.

  An object of the present invention is to provide a defect detection apparatus, an inspection apparatus, a cleaning apparatus, a defect detection method, and a defect detection program capable of appropriately detecting a mass of minute defects as a defect.

The defect detection device according to the present invention is a defect detection device for detecting a dense body of minute defects in an inspection object as a defect, which is a captured image acquisition unit for acquiring a captured image of the inspection object; Defect candidates are extracted based on the binarized image, and a plurality of defect candidates whose distances are less than the distance threshold are combined as new defect candidates, and the size of the combined defect candidate exceeds the size lower limit. And a detection unit for detecting a defect in the inspection object, wherein the detection unit is configured to generate the binarized image and the distance threshold, the size lower limit value, and the binary image for one captured image. The processing is performed under a plurality of conditions different from the transformation threshold , the plurality of conditions being a first condition, the binarization threshold being greater than the first condition, and the distance threshold and the size lower limit being the above A second condition less than the first condition and Characterized in that it contains.

Here, a new defect candidate (combined defect candidate) in which a plurality of defect candidates are combined may include all of each defect candidate or may include only a part. Further, the bonding defect candidate may have any shape such as a circle, an ellipse, or a polygon. Furthermore, for example, when there are three defect candidates, after combining two defect candidates to generate a joint defect candidate, the joint defect candidate and the remaining defect candidates are combined to form a joint defect candidate. Good.
Further, the size lower limit may be at least one of the area of the bonding defect candidate, the entire outer circumference, the length of at least one side, the diameter, and the like.
According to the present invention, the detection process is performed on one captured image under a plurality of conditions, for example, in which the size lower limit value is larger than DS0 shown in FIG. 1C and the distance threshold is smaller than DL0. As compared with the configuration of the above, the ratio of the detection unnecessary area NT included in the detection area based on the plurality of conditions can be reduced. As a result, the likelihood of detecting unwanted coupling defect candidate is detected as a defect can be lower than conventional construction, Ru can properly detect minute defects dense body as a defect. Furthermore, processing based on the first condition detects a small defect crowded body having small density and dark but large size as a defect, and processing based on the second condition has small size but large density and is bright and bright. Masses can be detected as defects.

  In the defect detection device according to the present invention, preferably, when the detection unit detects a defect at the same position under the plurality of conditions, the detection unit counts the defect at the same position as one defect.

  According to the present invention, it is possible to suppress one defect from being counted as a plurality of defects.

  The defect detection apparatus according to the present invention further includes an irradiation condition setting unit that controls an irradiation unit to change an irradiation condition of light to the inspection object, and the captured image acquisition unit irradiates the light to the inspection object. It is preferable that a plurality of captured images having different conditions be acquired, and the detection unit perform processing on each of the plurality of captured images under the plurality of conditions.

Here, as the irradiation condition of light, the wavelength of light, brightness, incident angle and the like can be exemplified.
In addition, on the surface or inside of the inspection object, there are minute defect clusters of types that easily reflect light of different wavelengths or types that do not easily reflect each other, or types that differ in reflection angle of light incident from one direction There may be a micro defect cluster of
According to the present invention, by performing processing under a plurality of conditions for each of a plurality of captured images under different irradiation conditions of light, the density of minute defects that differ in the wavelength of light that is likely to be reflected, the reflection angle of light, etc. A cluster can be properly detected as a defect.

The inspection apparatus according to the present invention detects, as a defect, a mass of minute defects in the inspection object based on an imaging unit that images the inspection object to generate an imaged image, and the imaged image generated by the imaging unit. And the above-mentioned defect detection device.
The inspection apparatus according to the present invention is characterized in that the inspection unit according to the present invention comprises: an irradiation unit for irradiating light to an inspection object; an imaging unit for imaging the inspection object to generate a captured image; and a captured image generated by the imaging unit. The apparatus is characterized by including the above-described defect detection device for detecting a dense defect of minute defects in an object to be inspected as a defect.
The cleaning apparatus according to the present invention is characterized by including a cleaning unit for cleaning an object to be inspected and the inspection apparatus described above.
The defect detection method according to the present invention is a defect detection method for detecting, as a defect, a dense body of minute defects in a subject to be inspected, which comprises: a captured image acquisition step of acquiring a captured image of the test object; Defect candidates are extracted based on the binarized image, and a plurality of defect candidates whose distances are less than the distance threshold are combined as new defect candidates, and the size of the combined defect candidate exceeds the size lower limit. And a detection step of detecting a defect of the inspection object, wherein the detection step is a binary value for generating the binarized image and the distance threshold and the size lower limit value for one captured image. The processing is performed under a plurality of conditions different from the transformation threshold , the plurality of conditions being a first condition, the binarization threshold being greater than the first condition, and the distance threshold and the size lower limit being the above Second less than first condition Characterized in that it comprises the matter.
A defect detection program according to the present invention is characterized by causing a computer to execute the defect detection method described above.

  According to the present invention, it is possible to provide an inspection apparatus, a cleaning apparatus, a defect detection method, and a defect detection program capable of appropriately detecting a mass of minute defects as a defect.

In the inspection apparatus according to the present invention, the imaging unit generates a dark field image as the captured image, and the defect detection device determines a dense defect of minute defects of the inspection object as a defect based on the dark field image. It is preferable to detect.
According to the present invention, since a dark-field image with high contrast is used as compared to a bright-field image, defects can be detected with high sensitivity.

  In the defect detection method of the present invention, preferably, the object to be inspected is a transparent plate, and the minute defects are metal oxides.

  According to the present invention, the metal oxide present in the transparent plate can be appropriately detected.

BRIEF DESCRIPTION OF THE DRAWINGS It is explanatory drawing of the subject of this invention, (A) shows the relationship of the visible area of a defect, and an invisible area, (B) shows the relationship of the detection required area of a defect, and a detection unnecessary area, (C) The detection result of the defect at the time of using the conventional defect detection method is shown. The schematic diagram of the washing | cleaning apparatus which concerns on one Embodiment of this invention. The block diagram of the inspection apparatus which comprises the said washing | cleaning apparatus. The flowchart which shows the effect | action of the said test | inspection apparatus. Explanatory drawing of the defect candidate coupling process in the effect | action of the said inspection apparatus. Explanatory drawing of the defect detection process in the effect | action of the said inspection apparatus. Explanatory drawing of the defect detection result in the effect | action of the said inspection apparatus. Explanatory drawing of the defect detection result which concerns on the modification of this invention.

Hereinafter, an embodiment of the present invention will be described based on the drawings.
In the present embodiment, a case where a binarization threshold, a distance threshold, and a size lower limit are set as defect detection conditions will be described. Moreover, if there is even one defect in the glass plate GL, the case where the glass plate GL is treated as a defective product will be described.

[Configuration of cleaning device]
First, the configuration of a cleaning apparatus according to an embodiment of the present invention will be described.
As shown in FIG. 2, the cleaning device 1 cleans the transport unit 2 that transports the glass plate GL, which is a transparent plate as an inspection object, by a plurality of rollers 21 and the glass plate GL transported by the transport unit 2. The cleaning unit 3, the drainage unit 4 for draining the glass plate GL cleaned by the cleaning unit 3, and the inspection device 5 for inspecting the glass plate GL subjected to drainage by the drainage unit 4 are provided.

The inspection apparatus 5 detects as a defect a non-removable dense body such as a metal oxide of about 10 μm as a minute defect fixed to the glass plate GL or a bubble inside the glass plate GL. The inspection apparatus 5 is provided with the irradiation part 51, the imaging part 52, and the defect detection apparatus 53 as a computer, as shown in FIG.
The irradiation part 51 is controlled by the defect detection apparatus 53, and irradiates light to the glass plate GL. Moreover, the irradiation part 51 is comprised so that a change of the irradiation conditions of light is possible. As irradiation conditions, the wavelength of light, brightness, the incident angle of light to glass plate GL, etc. can be illustrated.
The imaging unit 52 is controlled by the defect detection device 53, and captures an image of an object to be inspected irradiated with light to generate a captured image. Here, the imaging unit 52 configures a dark field optical system together with the irradiation unit 51, and generates a dark field image as a captured image. Then, the imaging unit 52 transmits the generated dark field image to the defect detection device 53. Such a dark field image makes minute defects and deposits appear white, and the other parts appear black.
The imaging unit 52 may configure a bright field optical system together with the irradiation unit 51, generate a bright field image, and transmit the bright field image to the defect detection device 53. In addition, the imaging unit 52 may be provided on the opposite side of the irradiation unit 51 with respect to the glass plate GL, or may be provided on the same side as the irradiation unit 51.

The defect detection device 53 includes an operation unit 54 for a worker to input an instruction to the defect detection device 53, a display unit 55, a storage unit 56, and a control unit 57.
The data stored in the storage unit 56 includes a dark field image captured by the irradiation unit 51, a binarized image of the dark field image, an irradiation condition of light to the glass plate GL, a detection condition and detection of a defect in the control unit 57 It is a result etc. As a defect detection condition, a binarization threshold, a distance threshold, and a size lower limit can be exemplified. The operator may change the irradiation condition of the light and the detection condition of the defect by the operation of the operation unit 54.

  The control unit 57 is configured by the CPU processing a defect detection program and other programs stored in a ROM (not shown). The control unit 57 includes an irradiation condition setting unit 571, a captured image acquisition unit 572, and a detection unit 573.

The irradiation condition setting unit 571 controls the irradiation unit 51 to set the irradiation condition of light to the glass plate GL.
The captured image acquisition unit 572 controls the imaging unit 52 to acquire a dark field image of the glass plate GL. Then, the captured image acquisition unit 572 stores the dark field image in the storage unit 56.
The detection unit 573 acquires a dark field image of the glass plate GL from the storage unit 56, and generates a binarized image obtained by binarizing the dark field image based on the binarization threshold value. The detection unit 573 also extracts defect candidates representing minute defects from the binarized image. Furthermore, the detection unit 573 combines a plurality of defect candidates whose mutual distances are equal to or less than the distance threshold as joint defect candidates. Then, the detection unit 573 detects, as a defect, a bond defect candidate in which the size of the bond defect candidate exceeds the size lower limit.
In addition, the detection unit 573 performs the above-described series of detection processing on one dark field image under a plurality of conditions in which the distance threshold, the size lower limit, and the binarization threshold are different. Then, the detection unit 573 causes the storage unit 56 to store the detection result.

[Function of washing apparatus]
Next, the operation of the cleaning device 1 will be described.
In addition, since the process until the glass plate GL is conveyed to the test | inspection apparatus 5 below is the same as the conventional washing | cleaning apparatus, description is abbreviate | omitted.
In addition, the detection unit 573 generates a binarized image based on the first to third conditions shown in Table 1 and the conditions shown in Table 2, a process of extracting defect candidates, and a plurality The process of combining the defect candidates of and the case of performing the process of detecting the defect candidate after combination as a defect will be described by way of example.

  First, the irradiation condition setting unit 571 of the control unit 57 constituting the inspection apparatus 5 reads the irradiation condition from the storage unit 56 when detecting the start operation of cleaning the glass plate GL. Further, the detection unit 573 of the control unit 57 reads the detection condition from the storage unit 56. Then, when cleaning of the glass plate GL is started and a sensor (not shown) detects that the glass plate GL has reached the imaging range of the imaging unit 52, the irradiation unit 51 detects the irradiation condition setting unit 571 as shown in FIG. The light based on the predetermined irradiation condition is irradiated to the glass plate GL by control of (step S1). Then, the imaging unit 52 captures an image of the glass plate GL under the control of the captured image acquisition unit 572 (step S2). The dark field image generated by this imaging is stored in the storage unit 56 by the captured image acquisition unit 572.

Next, the irradiation condition setting unit 571 determines whether the irradiation under all the irradiation conditions has ended (step S3). In step S3, when the irradiation condition setting unit 571 determines that the irradiation under all the irradiation conditions is not completed, the irradiation condition is changed (step S4), and the process of step S1 is performed. When the processes in steps S1 to S4 are performed a plurality of times, a plurality of dark field images with different light irradiation conditions are generated for one glass plate GL and stored in the storage unit 56.
On the other hand, when the irradiation condition setting unit 571 determines that the irradiation under all the irradiation conditions is completed in step S3, the captured image acquisition unit 572 acquires one dark field image from the storage unit 56 (step S5: captured image) Acquisition process). Then, the detection unit 573 performs processing based on the above Table 1 and Table 2.

  For example, the detection unit 573 first binarizes the dark field image with the binarization threshold value of the first condition as shown in Table 1 to generate a binarized image (step S6). Thereafter, the detection unit 573 extracts a white portion in the binarized image as a defect candidate representing a minute defect (step S7: detection step), and performs a process of combining defect candidates based on Table 1. (Step S8: detection step).

For example, as illustrated in FIG. 5A, when the defect candidates CF1 to CF3 exist in a predetermined range of the binarized image as illustrated in FIG. 5A, as illustrated in FIG. It is determined whether or not another defect candidate exists in the coupling target area EA1 centering on the. Here, in the bonding target area, the length in the transport direction of the glass plate GL is represented by the following formula (1), and the length in the width direction orthogonal to the transport direction is represented by the following formula (2) Square area.
Transfer direction length of combined target area =
(Distance threshold of transport direction DLT × 2) + length of transport direction of defect candidate (1)
Length in the width direction of the connection target area =
(Distance threshold in the width direction DLW × 2) + length in the width direction of the defect candidate (2)

Then, the detecting unit 573 determines that the defect candidate CF2 exists in the coupling target area EA1, combines the defect candidate CF1 and the defect candidate CF2, and as shown in FIG. 5C, a new defect. Generate candidate CF4 (referred to as binding defect candidate CF4). The bonding defect candidate CF4 is a smallest rectangular area including the defect candidate CF1 and the defect candidate CF2.
Next, as shown in FIG. 5D, it is determined whether or not another defect candidate exists in the bonding target area EA2 centering on the bonding defect candidate CF4, and the defect candidate CF3 exists. Therefore, as shown in FIG. 5E, the bonding defect candidate CF4 and the defect candidate CF3 are bonded to generate a new defect candidate CF5 (sometimes referred to as a bonding defect candidate CF5). Thereafter, the above process is repeated until the combining process can not be performed. In the case shown in FIGS. 5A to 5E, since there is no other defect candidate in the binding object area EA3 centering on the binding defect candidate CF5, the binding defect candidate CF5 is a defect candidate CF1. It will be the final candidate for the joint defect as a reference.
Then, the detection unit 573 performs the combining process until there is no defect candidate not subjected to the combining process.

Thereafter, as illustrated in FIG. 4, the detection unit 573 performs defect detection processing based on Tables 1 and 2 on all the bonding defect candidates (step S9: detection step).
Specifically, when the detection unit 573 performs a defect detection process on the bonding defect candidate CF6 as shown in FIG. 6A, the area is equal to or less than the size lower limit, but the length in the transport direction and the width direction Since the length of each exceeds the size lower limit value, it is detected as a defect of the glass plate GL. Similarly, in the joint defect candidates CF7 and CF8 shown in FIGS. 6B and 6C, the area is equal to or less than the size lower limit value, and one of the lengths in the transport direction and the width direction is the size. Because it is below the lower limit, it is not detected as a defect. Moreover, although the length in the width direction is equal to or less than the size lower limit, the joint defect candidate CF9 shown in FIG. 6D is detected as a defect because the area exceeds the size lower limit.

Next, the detection unit 573 determines whether the processing under all the inspection conditions is completed (step S10). In step S10, when the detection unit 573 determines that the process based on all the inspection conditions is not completed, the detection condition is changed (step S11), and the process of step S6 is performed. For example, when the process according to the first condition is completed, the process condition is changed to the second condition and the process of step S6 is performed.
On the other hand, when the detection unit 573 determines that the processing under all the inspection conditions is completed in step S10, the captured image acquisition unit 572 determines whether the processing of the entire dark field image for one glass plate GL is completed. (Step S12). In this step S12, when it is determined that the processing of the entire dark field image is not completed, the captured image acquisition unit 572 acquires another dark field image from the storage unit 56 (step S13). Then, the detection unit 573 performs the process after step S6 on the other dark field image.
On the other hand, in step S12, when the captured image acquisition unit 572 determines that the processing of the full dark field image is completed, the control unit 57 ends the processing.

Here, when the processing of steps S6 to S11 based on the first, second, and third conditions of Table 1 and Table 2 is performed on one dark field image, the first, second, and third as shown in FIG. The combined defect candidate included in the second and third detection areas AD1, AD2, and AD3 is detected as a defect.
Under the first condition, the binarization threshold is set to a smaller value than the second and third conditions, and the distance threshold and the size lower limit are set to a large value. Large minute defect clusters are detected as defects. In addition, under the third condition, the binarization threshold is set to a large value and the distance threshold and the size lower limit are set to small values compared to the first and second conditions, so the size is bright but the density is small. A large number of minute defect clusters are detected as defects. Under the second condition, a minute defect crowded body in a mode between the first condition and the third condition is detected as a defect.

Then, as shown in FIG. 7, a first overlap area AL1 in which the first to third detection areas AD1 to AD3 overlap, a second overlap area AL2 in which the first and second detection areas AD1 and AD2 overlap, a second, and so on. In the third overlap area AL3 in which the third detection areas AD2 and AD3 overlap, there is a possibility that defects at the same position are detected in duplicate under each condition. For example, a minute defect mass belonging to the third overlapping region AL3 and having a large light density and a large size is detected as one defect in each of the first condition and the third condition.
However, in this embodiment, since the glass plate GL having at least one defect is treated as a defective product, defects in the same position are redundantly detected under each condition and counted as a plurality of defects. There is no influence on the discrimination of non-defective products, and there is no particular problem.

[Operation and effect of the embodiment]
As described above, in the above embodiment, the following effects can be achieved.
(1) The defect detection device 53 extracts defect candidates based on the binarized image of the captured image, combines a plurality of defect candidates whose mutual distance is less than the distance threshold as combined defect candidates, and the size after combination is the size The bonding defect candidate exceeding the lower limit value is detected as a defect of the glass plate GL. Then, the defect detection device 53 performs the above detection process on one of the captured images under the first to third conditions, and detects a defect at the same position under the first to third conditions. Detect as a defect.
For this reason, as shown in FIG. 7, it is possible to detect as a defect a binding defect candidate included only in the first to third detection areas AD1 to AD3 and it is unnecessary to detect as compared with the conventional configuration shown in FIG. The possibility that the binding defect candidate of the detection unnecessary area NT is detected as a defect can be reduced, and the minute defect cluster can be appropriately detected as a defect.

(2) The first condition and the second condition in which the binarization threshold is set to a larger value than the first condition and the distance threshold and the size lower limit are set smaller than the first condition It is set up as follows.
For this reason, under the first condition, a minute defect dense body having a small size but large density is detected as a defect under the second condition, and under the second condition, a minute defect dense that is brighter but smaller in size than the first condition. A cluster can be detected as a defect.

(3) The defect detection device 53 is configured such that the detection unit 573 can acquire a plurality of dark field images with different light irradiation conditions.
Therefore, for example, if the light wavelengths are set to different conditions, even if there are minute defect clusters of types that easily reflect light of different wavelengths or types that hardly reflect light of these types, almost all of these types of minute defects dense A cluster can be properly detected as a defect. In addition, if the light brightness and incident angle are set to different conditions, almost all types of minute defect clusters are defects even when minute defect clusters with different types of reflection angles of light incident from one direction are present. Can be properly detected.

(4) The defect detection device 53 detects a defect based on the dark field image generated by the imaging unit 52.
As described above, since a dark-field image with high contrast is used as compared with a bright-field image, defects can be detected with high sensitivity.

[Other embodiments]
The present invention is not limited to the above embodiment, and various improvements and design changes can be made without departing from the scope of the present invention.

That is, the defect detection process may be performed under a plurality of conditions in which the binarization threshold is the same and the distance threshold and the size lower limit are different.
In the first to third conditions, only the distance threshold in the width direction is set to a different value for the distance threshold, but only the distance threshold in the transport direction may be set to a different value, or both distance thresholds may be different. It may be set to Also for the size lower limit value, at least one of the length in the conveyance direction, the length in the width direction, and the area may be set to different values under the first to third conditions.
The distance threshold may be set only in the transport direction or the width direction, and at least one of the length in the transport direction, the length in the width direction, and the area may be set as the size lower limit.
Two detection conditions as shown in Table 1 may be set, or four or more may be set.
The defect detection processing may be performed based on one dark-field image captured under one irradiation condition without configuring the irradiation unit 51 so as to be able to change the irradiation condition of light.
Although the processing based on each inspection condition is performed on each pickup image after generating the pickup images based on a plurality of irradiation conditions, each inspection condition is generated each time a pickup image based on one irradiation condition is generated. The processing based on may be performed.
Defect detection processing based on different detection conditions may be performed on each captured image captured under a plurality of irradiation conditions. For example, the dark-field image obtained under the first illumination condition is subjected to defect detection processing based on the first to third conditions, and is obtained under the second illumination condition different from the first illumination condition. Alternatively, defect detection processing based on fourth to sixth conditions different from the first to third conditions may be performed.

Although the case where the glass plate GL having at least one defect is treated as a defective product is exemplified in the above embodiment, when the number of defects of the glass sheet GL exceeds a predetermined number (plural), it is treated as a defective product and the predetermined number You may treat it as a non-defective item in the following cases. In such a configuration, it is not preferable to count defects at the same position detected under each condition as a plurality of defects, and the following processing may be performed.
For example, in the case shown in FIG. 7, in the first, second and third overlapping regions AL1, AL2 and AL3, since there is a possibility that the same defect is detected redundantly under each condition, a defect at the same position is It may be counted as one defect.
In addition, for example, the detection unit 573 detects bonding defect candidates exceeding the size lower limit and not more than the size upper limit as defects of the glass plate GL, and each detection condition is defined by the size lower limit and the size upper limit. The detection ranges may be set so as not to overlap. With such a configuration, as shown in FIG. 8, since the first, second and third detection areas AD11, AD12 and AD13 corresponding to the first, second and third conditions do not overlap, under each condition The same defect can be suppressed from being detected redundantly, and one defect can be suppressed from being counted redundantly. For example, a minute defect crowded body belonging to the third overlap region AL3 shown in FIG. 7 and having a large light density and a large size is not detected as a defect under the first condition but is detected as one defect under the third condition alone. be able to.

The inspection apparatus 5 may be provided not in the cleaning apparatus but in any apparatus such as a circuit board or an apparatus for manufacturing an electronic device that needs to detect a dense defect of minute defects in the inspection object as a defect.
The object to be inspected is not limited to the glass plate GL, but may be a plate, block, hollow box or the like made of resin, metal, wood or the like, or may not be transparent. The inspection object may be a circuit board or a photosensitive drum as described in Patent Documents 1 and 2.

DESCRIPTION OF SYMBOLS 1 ... Cleaning apparatus 5 ... Inspection apparatus 51 ... Irradiation part 52 ... Imaging part 53 ... Defect detection apparatus 571 ... Irradiation condition setting part 572 ... Captured image acquisition part 573 ... Detection part GL ... Glass (Transparent plate, to-be-inspected object)

Claims (10)

A defect detection apparatus for detecting a dense defect of minute defects in an inspection object as a defect, comprising:
A captured image acquisition unit that acquires a captured image of the test object;
Defect candidates are extracted based on a binarized image obtained by binarizing the captured image, a plurality of defect candidates whose distances are less than the distance threshold are combined as new defect candidates, and the size after combination is the size lower limit value And a detection unit that detects the defect candidate exceeding the defect candidate as a defect of the inspection object,
The detection unit performs processing on a single captured image under a plurality of conditions in which the distance threshold, the size lower limit, and a binarization threshold for generating the binarized image are different .
The plurality of conditions are
The first condition,
The defect detection device characterized in that the binarization threshold value is larger than the first condition, and the distance threshold and the size lower limit value are smaller than the first condition . In the defect detection device according to claim 1 ,
The defect detection device, wherein the detection unit counts the defects at the same position as one defect when the defects at the same position are detected under the plurality of conditions. In the defect detection device according to claim 1 or 2 ,
The irradiation condition setting unit controls the irradiation unit to change the irradiation condition of the light to the inspection object,
The captured image acquisition unit acquires a plurality of captured images having different irradiation conditions of light to the subject.
The defect detection apparatus, wherein the detection unit performs processing on each of the plurality of captured images under the plurality of conditions. An imaging unit that images a subject and generates a captured image;
The defect detection apparatus according to claim 1 or 2 , wherein a dense body of minute defects in the inspection object is detected as a defect based on a captured image generated by the imaging unit. apparatus. An irradiation unit that irradiates light to the subject;
An imaging unit for imaging the subject to generate a captured image;
The defect detection apparatus according to claim 3 , wherein a dense object of minute defects in the inspection object is detected as a defect based on a captured image generated by the imaging unit. In the inspection apparatus according to claim 4 or 5 ,
The imaging unit generates a dark field image as the captured image,
The inspection apparatus according to claim 1, wherein the defect detection device detects, as a defect, a mass of minute defects in the inspection object based on the dark field image. A cleaning unit for cleaning the test subject;
A cleaning apparatus comprising the inspection apparatus according to any one of claims 4 to 6 . A defect detection method for detecting a mass of minute defects in an inspection object as a defect, comprising:
A captured image acquisition step of acquiring a captured image of the inspection object;
Defect candidates are extracted based on a binarized image obtained by binarizing the captured image, a plurality of defect candidates whose distances are less than the distance threshold are combined as new defect candidates, and the size after combination is the size lower limit value And detecting a defect candidate exceeding the defect candidate as a defect of the inspection object,
The detection process performs processing on a single captured image under a plurality of conditions in which the distance threshold, the size lower limit, and a binarization threshold for generating the binarized image are different .
The plurality of conditions are
The first condition,
A defect detection method comprising: a second condition in which the binarization threshold is larger than the first condition and the distance threshold and the size lower limit value are smaller than the first condition . In the defect detection method according to claim 8 ,
The inspection object is a transparent plate,
The said micro defect is a metal oxide, The defect detection method characterized by the above-mentioned. A defect detection program which causes a computer to execute the defect detection method according to claim 8 or 9 .

Images