JPH0829145A - Inspection of surface defect - Google Patents

Abstract

PURPOSE:To provide a method for inspecting a surface, which can inspect defect of the surface effectively and highly accurately in a short time and can discriminates the kind of the defect. CONSTITUTION:In a scanning step, laser light is made to scan on a substance under inspection 6 with a polygon scanner 2. In the step for detecting the amount of received light, the reflected light of the laser light from the substance under inspection 6 is received with photodetectors 9 and 10, and the electric signal of the received light is obtained. The ordinary defective part of the substance under inspection is detected in a detection storage step based on the electric signal of the received light. In the detection storage step, the size and the position of the detected ordinary defective part are stored in a defective size memory in the main scanning direction, a defective position memory in the main scanning direction and a defective position memory in the secondary scanning direction. In an image pickup step, the image of the ordinary defective part, which is detected and stored, is picked up.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a surface defect inspection method for inspecting a surface defect of an object.

■

2. Description of the Related Art As a method of inspecting surface defects of various kinds of test objects such as steel plates, plastic films and papers, a laser is used which irradiates the test object with laser light and detects a change in intensity of reflected light or transmitted light. Known are a scanning inspection method and an image processing method in which image processing is performed to determine a defect based on image data obtained by capturing an image of a test object. Of these, in the inspection of surface defects by the laser scanning method, minute defects can be detected with high detection sensitivity. In addition, according to the image processing method, the feature amount of a defect is calculated from image data by obtaining visual information, and a spatially low frequency defect such as surface unevenness and a spatial defect such as a scratch are spatially calculated. It is easy to identify a high-frequency defect and set a criterion for each defect type.

[0003]

By the way, the above-described laser scanning inspection method has a problem that it is difficult to obtain visual information and the type of defect cannot be accurately determined. However, there is a problem in that the number of data becomes huge, a large memory capacity is required, and the inspection time becomes long.

The present invention has been made in view of the current state of surface defect inspection as described above, and an object of the present invention is to perform surface defect inspection efficiently in a short time and with high precision, and It is to provide a surface inspection method that can also determine

[0005]

In order to achieve the above object, the invention according to claim 1 comprises: a scanning step of scanning a laser beam on a test object by a polarizer; and the test object of the laser beam. The reflected light or the transmitted light from the received light is received, and a received light amount detection step for obtaining a received light electric signal, and based on the received light electric signal obtained in the received light amount detection step, a normal defective portion of the test object is detected and detected. The present invention is characterized by including a detection storage step of storing the size and position of the generated normal defect portion and an imaging step of capturing an image of the normal defect portion detected and stored in the detection storage step.

[0006] Similarly, in order to achieve the above object, the invention according to claim 2 comprises a scanning step of scanning a laser beam on a test object by a polarizer, and a reflected light of the laser beam from the test object. Alternatively, it is obtained in a received light amount detection step of receiving transmitted light to obtain a received light electric signal, a high frequency region blocking step of passing the received light electrical signal obtained in the received light amount detection step through a low pass filter, and a high frequency region blocking step. A detection storage step of detecting a low frequency defect portion of the test object based on the output signal and storing the size and position of the detected low frequency defect portion, and a low frequency defect detected and stored in the detection storage step. And an imaging step of imaging the part.

[0007] Similarly, in order to achieve the above object, the invention according to claim 3 comprises a scanning step of scanning a laser beam on an object to be inspected by a polarizer, and a reflected light of the laser beam from the object to be inspected. Alternatively, a received light amount detection step of receiving transmitted light to obtain a received light electrical signal, a low frequency region blocking step of passing the received light electrical signal obtained in the received light amount detection step through a differentiating circuit, and a low frequency region blocking step. A detection storage step of detecting a high-frequency defect portion of the test object based on the obtained output signal and storing the size and position of the detected high-frequency defect portion, and a high-frequency defect portion detected and stored in the detection storage step. And an image capturing step for capturing the image.

[0008] Similarly, in order to achieve the above object, the invention according to claim 4 comprises a scanning step of scanning a laser beam on an object to be inspected by a polarizer, and a reflected light of the laser beam from the object to be inspected. Alternatively, a received light amount detection step of receiving transmitted light to obtain a received light electrical signal, a high frequency region blocking step of dividing the received light electrical signal obtained in the received light amount detection step into one and passing one through a low pass filter, and the received light amount detection Based on the one output signal of the one half of the received electric signal obtained in step, the low frequency defect portion of the test object,
Detecting and storing a normal defective portion of the test object based on the other output signal, and storing the size and position of the detected defective portion, and imaging the defective portion detected and stored in the detecting and storing step. And an imaging step for performing.

[0009] Similarly, in order to achieve the above object, the invention according to claim 5 is a scanning step of scanning a laser beam on an object to be inspected by a polarizer, and a reflected light of the laser beam from the object to be inspected. Alternatively, the transmitted light is received and a received light amount detection step for obtaining a received light electric signal and the received electric signal obtained in the received light amount detection step are divided into two, one is passed through a low-pass filter to block the high frequency region, and the other is A frequency domain blocking step of blocking a low frequency domain by passing through a differentiating circuit, and a low frequency defective section and a high frequency defective section are respectively detected based on the output signal obtained in the frequency domain blocking step, and the detected defect is detected. It is characterized by including a detection storage step of storing the size and position of a copy, and an imaging step of imaging the defective part detected and stored in the detection storage step.

[0010] Similarly, in order to achieve the above object, the invention according to claim 6 is the same as the surface defect inspection method according to any one of claims 1 to 5, wherein the defect obtained in the imaging step is used. The present invention is characterized by having a neural network identification step in which a feature amount extracted from image data is used as an input layer and a defect type is used as an output layer.

[0011]

According to the first aspect of the invention, in the scanning step, the laser light is scanned by the polarizer by the polarizer, and in the light receiving amount detecting step, the reflected light or the transmitted light of the laser light from the object is received. The received light electrical signal is obtained, and the normal defective portion of the test object is detected in the detection and storage step based on the received light electrical signal, and the size and position of the detected normal defective portion is stored in the detection and storage step. Then, the normal defective portion detected and stored in the imaging step is imaged.

According to the second aspect of the invention, in the scanning step, the laser light is scanned on the object by the polarizer, and in the step of detecting the amount of received light, the reflected light or the transmitted light of the laser light from the object is received. Thus, the received light electric signal is obtained, and the low frequency component is extracted from the received light electric signal by passing the received light electric signal through the low-pass filter in the high frequency region blocking step. Then, based on the output signal of the low-pass filter, in the detection storage step, the low-frequency defect portion of the test object is detected, and in the detection storage step, the size and position of the detected low-frequency defect portion are stored, and the imaging step An image of the low-frequency defect portion detected and stored at is picked up.

According to the third aspect of the invention, in the scanning step, the laser light is scanned by the polarizer by the polarizer, and in the light receiving amount detecting step, the reflected light or the transmitted light of the laser light from the object is received. Thus, the received light electric signal is obtained, and the high frequency component is extracted from the received light electric signal by passing the received light electric signal through the differentiating circuit in the low frequency region blocking step. Then, based on the output signal of the differentiating circuit, in the detection storage step, the high frequency defect portion of the test object is detected,
In the detection and storage step, the size and position of the detected high frequency defect portion are stored, and the high frequency defect portion detected and stored in the imaging step is imaged.

According to the fourth aspect of the invention, in the scanning step, the laser light is scanned by the polarizer on the test object, and in the light receiving amount detecting step, the reflected light or the transmitted light of the laser light from the test object is received. Thus, the received light electric signal is obtained, the received light electric signal is divided into two, and one of the received light electric signal is passed through the low-pass filter in the high frequency region blocking step, whereby the low frequency component is extracted from the received light electric signal.
Then, in the detection storage step, based on one of the two divided output signals of the received light electrical signal obtained in the received light amount detection step, the low-frequency defect portion of the test object is detected based on the other output signal of the test object. The normal defect portion is detected, the size and position of the detected defect portion are stored, and the low frequency defect portion and the normal defect portion detected and stored in the imaging step are imaged.

According to the fifth aspect of the present invention, in the scanning step, the laser light is scanned by the polarizer on the test object, and in the received light amount detecting step, the reflected light or the transmitted light of the laser light from the test object is received. The received light electric signal is obtained, the received light electric signal is divided into two, and one of the received light electric signals is passed through the low-pass filter in the high frequency region blocking step, whereby the low frequency component is extracted from the received light electric signal, and the other is received. The high-frequency component is extracted from the received light electric signal by passing the received light electric signal of (1) through the differentiating circuit in the low frequency region blocking step. Then, in the detection storage step, based on one of the two divided output signals of the received light electrical signal obtained in the received light amount detection step, the low-frequency defect portion of the test object is detected based on the other output signal of the test object. Each high-frequency defect portion is detected, the size and position of the detected defect portion are stored, and the low-frequency defect portion and the high-frequency defect portion detected and stored in the imaging step are imaged.

According to a sixth aspect of the invention, in the invention according to any one of the first to fifth aspects, the feature amount extracted from the defect image data obtained in the imaging step is used as an input layer, and the type of the defect. The type of defect such as the shape and grade of the defect is identified by the neural network having the output layer as the output layer.

[0017]

【Example】

[First Embodiment] A first embodiment of the present invention will be described with reference to FIGS. FIG. 1 is an explanatory diagram of a laser scanning inspection method of this embodiment, FIG. 2 is an explanatory diagram of an image inspection method of this embodiment,
FIG. 3 is an explanatory diagram showing the relationship between the beam diameter and the sampling pitch of this embodiment, and FIG. 4 is an explanatory diagram of the signal processing of this embodiment.
FIG. 5 is a block diagram showing the configuration of the signal processing apparatus used in this embodiment, and FIG. 6 is an explanatory diagram of the labeling processing of this embodiment.

In the present embodiment, the inspection of the surface defects of the test object such as steel plate, plastic film and paper is carried out in two steps,
First, a laser scanning inspection method detects a predetermined number of defects or more, and then, with respect to a defective portion obtained by the laser scanning inspection method, based on inspection information of the laser inspection method, a CCD
An imaging inspection with a camera or the like is performed. First, the laser scanning inspection method will be described. As shown in FIG. 3, it is assumed that the inspection object 6 is a square sheet having a width WT of 300 mm and a length L of 300 mm, and a defect having a size of about 20 μm is detected. The laser light source used in this case has a wavelength of 633.
As a laser light source for irradiating a He-Ne laser beam of nm,
When the diameter of the laser beam is narrowed down to about 40 μm and used, it is possible to detect a defect portion of about half the diameter of the laser beam.

In the laser scanning inspection, as shown in FIG. 1, the laser light from the laser light source 1 has a focal length f of 390 m.
The polygon scanner 2 having m, a rotation speed R of 6000 rpm, and a surface number N of 8 is deflected to scan the inspection object 6. As shown in FIG. 3, the sampling pitch in this case is set such that the main scanning direction sampling pitch P1 is 20 μm and the sub scanning direction sampling pitch P2 is 10 μm. Then, as the inspection parameter, one main scanning time t (s
ec), inspection time T (sec), main scanning linear velocity V (mm
/ Sec), main scanning direction sampling frequency S (MH
z) is set as shown in, for example, equations (1) to (4).

T = 60 / (RN) (1) T = (L / P2) t = (300000/10) (60 / 6000.8) = 3.75 sec (2) V = 2f (Dθ / dt) = 2f (2πR / 60) = (πRf / 15) (3) S = V / P1 = (πRf / 15P1) = (π6000.3900000) / (15.20) = 25 MHz・ ・ (4)

Returning to FIG. 1, the reflected light 7 from the test object 6
Is incident on the light guide rod 8 and is guided to the light receivers 9 and 10 arranged at both ends of the light guide rod 8 to undergo photoelectric conversion. When a defect is present on the object 6 to be inspected, scattering of laser light by the defect causes
Since the amount of light reaching the light receivers 9 and 10 decreases, it is possible to detect a defect.

In this embodiment, the signal processing device 14 as shown in FIG. 5 is used. The output signals of the photo detectors 9 and 10 are input to the OR circuit 21, and the sum signal output from the OR circuit 21 is A level signal having a predetermined threshold value Th1 is input to one input terminal of the comparator 22 and the other input terminal of the comparator 22. Therefore, FIG.
As shown in, the sum signal output from the OR circuit 21 is
When the defect signal F is detected below the threshold Th1, it is determined that there is a defect on the test object 6, and the comparator 22
A defect signal is output from. The signal processing device 14 includes a main scanning direction address counter 23 that counts the number of samplings in the main scanning direction as address data, and a sub scanning direction address counter 24 that counts the detection signal of the trigger PD 4 shown in FIG. 1 as address data. Is provided,
Further, a main scanning direction defect position memory 25 and a sub scanning direction defect position memory 26 are provided. When the defect signal is detected, the address data N of the main scanning direction address counter 23 and the sub scanning direction address counter 24
1 and N2 are written in the main scanning direction defect position memory 25 and the sub scanning direction defect position memory 26, respectively.

Further, the signal processing device 14 is provided with a defect size counter 27 and a defect size memory 28 in the main scanning direction, and when a defect signal is output from the comparator 22, the defect size counter 27 will show the defect shown in FIG. Size N
3 is counted, and this count value is written in the main scanning direction defect size memory 28.

In this way, when a defect is detected,
The lateral position of the defect is determined at N1, the vertical position of the defect at N2, and the lateral size of the defect at N3. When a defect signal for a defect is detected over a plurality of main scans, address data N1, N2, defect size N
Based on 3, the labeling process is performed as shown in FIG. 6, and the label value is given to each defect. After the labeling process, the quadrangular endpoints that circumscribe each defect are calculated,
Data T1 to T4 shown in FIG. 6 are written in the memory for each defect.

In the stable manufacturing process, it is known that the number of defects generated in the test object 6 having the size shown in FIG. 3 is about several. Therefore, in this embodiment, the labeling process is executed at high speed. Since it is possible and data is written only when a defect occurs, it is not necessary to increase the memory capacity.

Although the position and the optical size of the defect are detected by the laser scanning inspection method described above, in the present embodiment, the laser scanning inspection is further performed on the defect detected by the laser scanning inspection method. Based on the defect information obtained by the method,
The type of defect is identified by an imaging inspection method in which the defect is imaged with a CCD or the like.

Next, the imaging inspection method of this embodiment will be described.
For the inspection object 6 whose position and optical size have been detected by the laser scanning inspection method, the defective portion is provided with two CCDs.
Images are picked up by 11 and 12. In this case, the sampling pitch is set to 20 μm in the main scanning direction (the number of pixels of the CCDs 11 and 12 is 7500), the test object 6 is moved so that the sampling pitch in the sub scanning direction is 10 μm, and the same sampling as in the laser scanning inspection method is performed. The pitch is set. Then, the coordinate data of the defect detected by the laser scanning inspection method is read from the defect position memory 25 in the main scanning direction and the defect position memory 26 in the sub-scanning direction, and the CCD 11 and 12 perform an image inspection on the read coordinate position. Be done.
Based on the visual information obtained by this imaging inspection, the visual inspector determines the feature amount and type of the defect, and determines the quality of the inspection object based on the empirical reference information.

For example, if a defect having a diameter of about 20 μm on a test object 6 of 300 mm × 300 mm is to be detected only by the imaging inspection method, the sampling pitch in the main scanning direction is 20 μm and the sampling pitch in the sub scanning direction is 10
.mu.m is required, and if one pixel is represented by 8 bits, the number of data becomes 4.5 million, the required memory capacity becomes enormous, and the time for data processing becomes long. However, according to the present invention, a laser scanning inspection method is used in advance to confirm spatially a predetermined number or more of normal defects ranging from a high frequency to a low frequency and their positions, and perform an image inspection on the obtained normal defect portion. The memory capacity can be reduced, and the data processing time can be shortened.

As described above, according to the present embodiment, the position and the optical size of the normal defect ranging from high frequency to low frequency are detected by the laser scanning inspection method, and the image inspection is performed on the detected normal defect portion. It is possible to obtain the imaging data of the defective portion in a short time, to accurately grasp the surface defect of the inspection object 6, and to judge the quality of the inspection object 6 accurately and efficiently. .

[Second Embodiment] A second embodiment of the present invention will be described with reference to FIGS. 7 and 8. FIG. 7 is a characteristic diagram of a low frequency defect which is a detection target of this embodiment, and FIG. 8 is a block diagram showing a configuration of a signal processing device used in this embodiment.
The same reference numerals are given to the same parts.

As shown in FIG. 8, the signal processing circuit 15 of the present embodiment differs from the signal processing circuit 14 of the first embodiment already described with reference to FIG. The low-pass filter 15 is connected to.
The configuration of the other parts of the signal processing circuit 15 of this embodiment is as follows.
This is the same as the signal processing circuit 14 of the first embodiment already described with reference to FIG. By the way, in the case where the defect on the surface of the object to be inspected is a low frequency defect having a spatially low frequency such as unevenness, the intensity of the sum signal output from the OR circuit 21 and the main scanning direction of the object to be inspected. Have a relationship as shown in FIG. In order to detect the defect signal F1 having such a characteristic, it is necessary to bring the threshold level Th1 close to a normal signal level, and noise may be detected. Therefore, in this embodiment, the low-pass filter 15 is provided to block the high frequency component of the sum signal.

As described above, in the present embodiment, in the laser scanning inspection, the sum signal output from the OR circuit 21 is passed through the low-pass filter 15 to block high-frequency components, so that the surface unevenness of the object to be inspected or the like. Spatial low-frequency low-frequency defects are detected with high accuracy without detecting noise, and the image inspection of the detected low-frequency defects is performed. The surface defect of the inspection object 6 can be accurately grasped, and the inspection object 6
It is possible to accurately and efficiently determine whether the quality of the item is good or bad.

[Third Embodiment] A third embodiment of the present invention will be described with reference to FIGS. 9 and 10. FIG. 9 is a characteristic diagram of a high frequency defect to be detected in this embodiment, and FIG. 10 is a block diagram showing a configuration of a signal processing device used in this embodiment. The same parts as those in FIG. ing.

As shown in FIG. 10, the signal processing circuit 16 of the present embodiment is different from the signal processing circuit 14 of the first embodiment already described with reference to FIG. 5 between the OR circuit 21 and the comparator 22. Differentiating circuit 20 is connected to. The configuration of the other parts of the signal processing circuit 16 of this embodiment is the same as that of the signal processing circuit 1 of the first embodiment already described with reference to FIG.
Same as 4. By the way, in the case where a defect on the surface of the object to be inspected is a high frequency defect having a high spatial frequency such as a flaw, between the intensity of the sum signal output from the OR circuit 21 and the main scanning direction of the object to be inspected, There is a relationship as shown in FIG. In order to detect the defect signal F2 having such a characteristic with high accuracy, the differentiating circuit 20 is provided in the present embodiment to cut off the low frequency component of the sum signal.

As described above, in the present embodiment, in the laser scanning inspection, the sum signal output from the OR circuit 21 is passed through the differentiating circuit 20 to cut off the low frequency components, so that the flaw of the object to be inspected or the like is detected. High-frequency defects of high frequency are detected spatially with high accuracy, and the image inspection of the detected high-frequency defects is performed. Therefore, the imaging data of the high-frequency defects can be obtained in a short time, and the surface defects of the inspection object 6 can be accurately determined. Therefore, the quality of the object 6 can be accurately and efficiently determined.

[Fourth Embodiment] A fourth embodiment of the present invention will be described with reference to FIG. FIG. 11 is a block diagram showing the configuration of the signal processing apparatus of this embodiment, and the same parts as those in FIGS. 5 and 8 are designated by the same reference numerals.

In the signal processing device 19 of this embodiment, as shown in FIG.
As shown in (1), the sum signal output from the OR circuit 21 is divided into two, and one of them is input to the low-pass filter 15 of the circuit 15A excluding the OR circuit 21 from the signal processing circuit 15 used in the second embodiment, The other is input to the comparator 22 of the circuit 14A in which the OR circuit 21 is removed from the signal processing circuit 14 used in the first embodiment.

In this embodiment, the normal defect inspection of the surface described in the first embodiment and the low frequency defect inspection such as the surface unevenness described in the second embodiment are simultaneously performed.
The operation is as described in the first and second embodiments.

As described above, according to the present embodiment, in the laser scanning inspection, the sum signal output from the OR circuit 21 is divided into two parts, and one of them is passed through the low-pass filter 15 to block the high frequency component, and thereby A low-frequency defect having a spatially low frequency such as surface unevenness of an inspection object is detected with high accuracy without detecting noise, and on the other hand, a normal defect ranging from high frequency to low frequency is detected. Then, since the image inspection is performed on the detected low-frequency defect portion and the normal defect portion, the image data of the normal defect portion and the low-frequency defect portion is obtained in a short time, and the surface defect of the inspection object 6 can be accurately determined. Therefore, the quality of the object 6 can be accurately and efficiently determined.

[Fifth Embodiment] A fifth embodiment of the present invention will be described with reference to FIG. FIG. 12 is a block diagram showing the configuration of the signal processing apparatus of this embodiment, and the same parts as those in FIGS. 8 and 10 are designated by the same reference numerals.

In the signal processing device 30 of this embodiment, FIG.
As shown in FIG. 7, the sum signal output from the OR circuit 21 is divided into two, one of which is input to the low-pass filter 15 of the circuit 15A described in the fourth embodiment, and the other is a signal used in the third embodiment. It is input to the differentiating circuit 20 of the circuit 16A except the OR circuit 21 from the processing circuit 16.

In this embodiment, the inspection of low frequency defects such as surface irregularities described in the second embodiment and the inspection of high frequency defects such as surface scratches described in the third embodiment are performed simultaneously. Be done. The operation is as described in the second and third embodiments.

As described above, according to the present embodiment, in the laser scanning inspection, the sum signal output from the OR circuit 21 is divided into two, and one of them is passed through the low-pass filter 15 to cut off a high frequency component, so that A low-frequency defect having a spatially low frequency such as surface unevenness of the inspection object is detected with high accuracy without detecting noise, and the other is cut off by passing through the differentiating circuit 20 to remove the low-frequency component. It is possible to detect spatially high frequency high frequency defects such as surface scratches of objects with high accuracy. Then, since the image inspection is performed on the detected low frequency defect portion and high frequency defect portion, the image data of the low frequency defect portion and the high frequency defect portion can be obtained in a short time, and the surface defect of the inspection object 6 can be accurately determined. Therefore, the quality of the object 6 can be accurately and efficiently determined.

[Sixth Embodiment] A sixth embodiment of the present invention will be described with reference to FIGS. 13 is a characteristic diagram of the shape category of the present embodiment, FIG. 14 is a structural diagram of a shape recognition neural network of the present embodiment, and FIG.
5 is a characteristic diagram of grade categories.

In the sixth embodiment, the laser scanning inspection method and the imaging inspection method of any of the first to fifth embodiments already described are used, and then the image obtained by the imaging inspection method is used. On the basis of the feature amount obtained from the data, a shape category is identified by a neural network, and a decision tree is used to perform fine classification for each defect shape. For example, the area and the width / length are obtained as the feature amount from the imaged data, and the obtained feature amount is input to the neural network shown in FIG. In this neural network, the number of units in the input layer is matched with the number of dimensions of the feature vector that has the feature quantity as the component, the number of units in the intermediate layer is matched with the number of identification planes, and the number of units in the output layer is matched with the number of shape categories. The input feature quantity is classified into points, planes, and lines according to the identification characteristics shown in FIG.
The categories of shapes identified in this way are points,
By a judgment tree (not shown) that judges the characteristics of lines and surfaces,
Each defect shape is subdivided.

Further, when the class recognition of defects is performed using the neutral network for class recognition, the area and density average value are obtained as the feature quantity, and the discrimination characteristics of FIG. 15 are obtained for these feature quantities. According to the above, a neural network (not shown) similar to the case of FIG. 14 is used to classify the defects into three grades of light, medium, and heavy.

As described above, according to this embodiment, in addition to the effects obtained in the first to fifth embodiments, the shapes and grades of surface defects are automatically and accurately classified. Therefore, empirical processing and correction operation by an inspector are not required, and the ability to determine the quality of the inspection object 6 can be significantly improved.

[0048]

According to the first aspect of the invention, the laser light is scanned on the object to be inspected, and the reflected light or the transmitted light of the laser light from the object to be inspected is received, and based on the received light electric signal, Since the normal defect portion of the test object is detected, the size and position of the normal defect portion are stored, and only the normal defect portion is imaged, the surface defect inspection of the test object can be performed efficiently in a short time. It becomes possible to do it accurately. According to the second aspect of the present invention, the laser light is scanned on the test object, the reflected light or the transmitted light of the laser light from the test object is received, and the low frequency component is extracted from the received electric signal. The low-frequency defect portion of the test object is detected based on the low-frequency component, the size and position of the low-frequency defect portion are stored, and the imaging inspection of only the low-frequency defect portion is performed. It becomes possible to perform a spatially low frequency surface defect inspection efficiently and accurately in a short time. According to the third aspect of the present invention, the laser light is scanned on the test object, the reflected light or the transmitted light of the laser light from the test object is received, the high frequency component is extracted from the received electric signal, and the high frequency component is extracted. A high-frequency defect portion of the test object is detected based on the component, the size and position of the high-frequency defect portion are stored, and imaging inspection of only the high-frequency defect portion is performed. It becomes possible to perform surface defect inspection efficiently and accurately in a short time. According to the invention of claim 4, the laser light is scanned on the object to be detected, the reflected light or the transmitted light of the laser light from the object is received, the received light electric signal is divided into two, and one received electric signal is received. , A low-frequency component is extracted from the received light electrical signal by passing through a low-pass filter, and based on the one output signal of the received light electrical signal halved, the low-frequency defective portion of the test object outputs the other output signal. The normal defect portion of the inspection object is detected based on the above, the size and position of the detected defect portion are stored, and the image inspection of only the low frequency defect portion and the normal defect portion is performed. It is possible to efficiently and appropriately inspect the normal defect portion and the spatially low frequency defect portion on the surface of the. According to the invention described in claim 5, the laser light is scanned on the object to be detected, the reflected light or the transmitted light of the laser light from the object is received, the received light electric signal is divided into two, and one of the received light electricity is received. By passing the signal through the low-pass filter, the low-frequency component is extracted from the received light electrical signal, and by passing the other received light electrical signal through the differentiating circuit, the high-frequency component is extracted from the received light electrical signal. Based on one of the two output signals, the low-frequency defect part of the test object is detected, and the high-frequency defect part of the test object is detected based on the other output signal, and the size and position of the detected defect part are detected. Is stored and the imaging inspection of only the low-frequency defect portion and the high-frequency defect portion is performed, so that the inspection of the spatially low-frequency defect portion and the high-frequency defect portion of the surface of the test object can be performed efficiently in a short time. And accurately Door is possible. According to the invention described in claim 6, in addition to the effect of the invention described in any one of claims 1 to 5, the feature amount extracted from the defect image data obtained by imaging is input to the input layer to detect the defect. The neural network with the type as the output layer makes it possible to automatically and accurately identify the type of defect such as the shape and grade of the defect.

[Brief description of drawings]

FIG. 1 is an explanatory diagram of a laser scanning inspection method according to a first embodiment of the present invention.

FIG. 2 is an explanatory diagram of an image inspection method of the same embodiment.

FIG. 3 is an explanatory diagram showing a relationship between a beam diameter and a sampling pitch in the example.

FIG. 4 is an explanatory diagram of signal processing according to the same embodiment.

FIG. 5 is a block diagram showing a configuration of a signal processing device used in the embodiment.

FIG. 6 is an explanatory diagram of a labeling process of the embodiment.

FIG. 7 is a characteristic diagram of a low frequency defect which is a detection target of the second embodiment of the present invention.

FIG. 8 is a block diagram showing a configuration of a signal processing device used in the embodiment.

FIG. 9 is a characteristic diagram of a high frequency defect which is a detection target in the third embodiment of the present invention.

FIG. 10 is a block diagram showing a configuration of a signal processing device used in the embodiment.

FIG. 11 is a block diagram showing a configuration of a signal processing device according to a fourth exemplary embodiment of the present invention.

FIG. 12 is a block diagram showing a configuration of a signal processing device according to a fifth exemplary embodiment of the present invention.

FIG. 13 is a characteristic diagram of a shape category according to the sixth embodiment of the present invention.

FIG. 14 is a structural diagram of a neural network for shape recognition of the embodiment.

FIG. 15 is a characteristic diagram of grade categories in the example.

[Explanation of reference numerals] 1 laser light source 2 polygon scanner 8 optical rod 9, 10 light receiver 15 low-pass filter 20 differentiating circuit 22 comparator 25 main operation direction defect position memory 26 sub operation direction defect position memory 28 main operation direction defect size memory

Claims (6)

[Claims] 1. A scanning step of scanning a laser beam on a test object by a polarizer, and a received light amount detecting step of receiving reflected light or transmitted light of the laser beam from the test object to obtain a received light electric signal. A detection storage step of detecting a normal defect portion of the test object based on the received light electric signal obtained in the received light amount detection step, and storing the size and position of the detected normal defect portion; An image pickup step of picking up an image of the normal defect portion detected and stored in the step, and a surface defect inspection method. 2. A scanning step of scanning a laser beam on an object to be inspected by a polarizer, and a received light amount detecting step of receiving a reflected light or a transmitted light of the laser beam from the object to obtain a received light electric signal. A high-frequency region blocking step of passing a received light electrical signal obtained in the received light amount detecting step through a low-pass filter, and a low-frequency defect portion of the test object based on the output signal obtained in the high-frequency region blocking step. A surface having a detection storage step of storing the size and position of the detected low-frequency defect portion, and an imaging step of imaging the low-frequency defect portion detected and stored in the detection storage step. Defect inspection method. 3. A scanning step of scanning a laser beam on an object to be inspected by a polarizer, and a received light amount detecting step of receiving a reflected light or a transmitted light of the laser beam from the object to obtain a received light electric signal. A low-frequency region blocking step of passing a received light electrical signal obtained in the received light amount detecting step through a differentiating circuit, and an output signal obtained in the low-frequency region blocking step. Surface defect, which includes a detection storage step of detecting the high frequency defect portion and storing the size and position of the detected high frequency defect portion, and an imaging step of capturing an image of the high frequency defect portion detected and stored in the detection storage step. Inspection methods. 4. A scanning step of scanning a laser beam on an object to be inspected by a polarizer, and a received light amount detecting step of receiving reflected light or transmitted light of the laser beam from the object to be inspected to obtain a received light electric signal. A high-frequency region blocking step of dividing the received light electric signal obtained in the received light amount detection step into one and passing one through a low-pass filter; and one of the two divided output signals of the received light electric signal obtained in the received light amount detection step. Based on the above, a low-frequency defect portion of the test object, a detection storage step for detecting the normal defect portion of the test object based on the other output signal, and storing the size and position of the detected defect portion. And an imaging step of imaging the defective portion detected and stored in the detection and storage step. 5. A scanning step of scanning a laser beam on an object to be inspected by a polarizer, and a received light amount detecting step of receiving a reflected light or a transmitted light of the laser beam from the object to obtain a received light electric signal. And a frequency domain cutoff step of dividing the received electric signal obtained in the received light amount detection step into two, passing one through a low pass filter to cut off a high frequency area and passing the other through a differentiating circuit to cut off a low frequency area. And detecting a low frequency defect portion and a high frequency defect portion, respectively, based on the output signal obtained in the frequency domain cutoff step,
A surface defect inspection method comprising: a detection storage step of storing the size and position of the detected defect portion; and an imaging step of capturing an image of the defect portion detected and stored in the detection storage step. 6. The method according to claim 1, further comprising a neural network identification step in which the feature amount extracted from the defect image data obtained in the imaging step is an input layer and the type of defect is an output layer. Item 6. The surface defect inspection method according to any one of Items 5.