JP4369803B2 - Defect inspection method for patch - Google Patents

Description

  The present invention relates to a defect inspection method for a patch, and particularly suitable for inspecting defects such as foreign matter such as hair and dust, scratches, dirt, etc. in a transdermal drug that is a sheet-like patch. It relates to the inspection method.

  In the case of pharmaceuticals such as transdermally absorbable drugs and tablets, in general, inspections for foreign matter contamination, scratches, and dirt are performed. In particular, foreign substances derived from living organisms such as hair are mixed into the product, which is a fatal defect in the hygiene of the product. Therefore, in order to prevent the outflow of such a defective product, an automatic inspection method is used. It is being considered.

Conventionally, as a method for inspecting such a medicine, an inspection method for a PTP (Press Through Package) sheet or a tablet therein is known. Specifically, as an inspection method for the PTP sheet itself, methods disclosed in JP-A-11-39484 and JP-A-08-050101 are known, and the PTP sheet is packaged by the PTP sheet. As an inspection method for tablets, a method disclosed in Japanese Patent No. 0461136 is known.
However, although the inspection method disclosed here is a foreign substance inspection method for pharmaceutical products, it is based on a shape peculiar to a PTP sheet or the like, for example, an inspection method for a patch having a configuration like a transdermal drug. Is not applicable.

  On the other hand, there is a method disclosed in Japanese Patent No. 2597370 as a method for continuously inspecting a sheet-like inspection object on which periodic printing is applied. This method is an inspection method for detecting defects by image processing such as differentiation processing or binarization processing, and utilizes the fact that periodic printing is applied to a sheet when performing image processing. This is a method in which printing is not erroneously detected as a defect by masking the image of the printed part using the image of (1).

  By the way, in a patch material such as a transdermal drug, the separator is formed with a cut line called a back cut in order to facilitate the sticking operation and prevent the adhesive surfaces from adhering to each other and preventing wrinkles from entering. Yes. The shape of the back cut includes a straight shape, a curved shape, and the like, and the modes are various. In addition to such a back cut, the separator may be printed with precautions when pasting, or a product mark or the like may be printed on the separator or support. In addition, the adhesive material such as this type of transdermal drug is mainly formed as a piece by punching or cutting.

  When performing inspection for defects such as hair and dust, scratches, dirt, etc. with the patch of such a configuration as the object of inspection, if there is a back cut or printing, there is a possibility that these may be mistaken as defects depending on the optical system There is.

In particular, the back cut formed on the separator differs in appearance depending on the product configuration and the lighting conditions when inputting an image (that is, imaging). For example, the back cut is clearly visible and the entire back cut is recognized as a defect. In some cases, it looks thin and only a part of the back cut is recognized as a defect. Further, the back cut is not necessarily formed periodically as described above.
Therefore, when a patch having such a back cut is to be inspected, it is difficult to apply an inspection method using periodicity as in the prior art disclosed in Japanese Patent No. 2597370.

  In addition, it is difficult to apply the inspection method using the periodicity of printing as shown in the above-mentioned Japanese Patent No. 2597370 to the printed material that is configured as an individual piece.

  Therefore, the present invention is intended for inspection of a patch comprising a separator with a back cut, such as a transdermal absorption drug, and even when a back cut exists, foreign matter such as hair and dust, It is an object to provide a defect inspection method capable of accurately detecting defects such as scratches or dirt.

In order to solve the above problems, the defect inspection method for a patch of the present invention is a defect inspection method for a patch that includes an adhesive layer and a separator that covers the adhesive layer and has a back cut formed thereon. An imaging process for capturing an image of the adhesive material in a state in which the adhesive material is irradiated with light, and an image processing process for extracting a defect from the captured image and performing a pass / fail determination. In this case, an arbitrary point detected as a defect is set as a back-cut candidate point, and a defect detected within a predetermined range from the back-cut candidate point toward the back-cut forming direction is set as another back-cut candidate point. Extracting and masking the region where the back cut is formed by determining the back cut candidate point as a back cut based on the feature amount of a plurality of back cut candidate points detected by repeating. To do.

  According to the present invention, in the image processing step, by masking the region where the backcut is formed, it is possible to eliminate the detected defect due to the backcut, so that the backcut is mistaken as a defect. Can be prevented.

The defect detection method of the adhesive material of the present invention, when extracting the formed regions of the back cut and the arbitrary point detected as defects and back cutting candidate point, the back-cutting from the back cutting candidate points Based on feature quantities (for example, shape, brightness, etc.) of a plurality of back-traffic candidate points detected by repeating the operation with defects detected within a predetermined range in the forming direction as other back-traffic candidate points the said back cutting candidate point by determining the back cut, it is assumed to extract the formed regions of the spine cutting Te.

  This is focused on the fact that the back cut formed on the separator is formed with a directivity so that it can be easily peeled off, and the back cut is determined based on information on the formation direction of such a back cut. Is. That is, an arbitrary point detected as a defect is set as a back-cut candidate point, and a defect detected within a predetermined range from the back-cut candidate point toward the back-cut forming direction is set as another back-cut candidate point. The information regarding the direction of forming the back cut may be input in advance or may be obtained based on individual image data. Then, a new back crossing candidate point is tracked in the same manner from the other detected back crossing candidate points, and a plurality of back crossing candidate points are detected by repeating such operations. Since the original back cut has a predetermined range of feature values (for example, shape, brightness, etc.), is it a back cut or a defect based on the feature values of a plurality of detected back cut candidate points? Thus, it is possible to accurately extract the region where the back cut is formed.

Moreover, the defect detection method for a patch of the present invention preferably inspects the area masked as described above according to a determination criterion different from that of the unmasked area.
If the masked area is not inspected, it cannot be detected when there is a defect, and if the masked area is inspected according to the same standard as the unmasked area, backcutting, printing, etc. May cause many false detections. Therefore, as described above, by inspecting the masked area based on a determination criterion different from that of the unmasked area, it is possible to expand the inspection area to the mask area while preventing erroneous detection.

When inspecting the masked area, it is preferable that a defect is detected based on the shape or shading of the area extracted as the back cut.
Since the back cut is formed in a straight line or a curved line, it is extracted as a single line having a certain shading when image processing is performed. Therefore, when a defect exists so as to overlap on the back cut, the line width of the extracted back cut becomes thick or an intersection occurs at the back cut, which is different from one straight line. Or a shaded portion exceeding the shade of a certain range is generated.
Thus, by evaluating the shape or shading of the region extracted as a back cut in this way, it is possible to detect a defect that exists on the back cut.

  On the other hand, for the portion to which printing has been applied, a method of preventing printing from being taken in the imaging process by optical means is also conceivable, but the optical system in which the defect that is desired to be detected is most visible tends to show the printing portion. If there is sufficient contrast between the print part and the background part, the print part can be extracted from the feature value of the print part and masked. If they are different, it is difficult to create a mask by the above method.

  Therefore, in the defect inspection method for a patch according to the present invention, when the patch has periodic printing, the position of the print is preferably detected by pattern matching. It is assumed that an area on which printing has been performed is estimated based on position information and pre-input print pattern information, and a mask is automatically created from the estimated area and masked.

  As described above, if the printed area can be estimated, an optimal inspection method can be employed according to the appearance of the printing unit. That is, when the print looks thin, the printed portion is not simply masked and removed from the inspection area, but the masked area is separately inspected by changing the determination criterion. This makes it possible to detect a fatal defect that exists so as to overlap the print. Further, when printing is colored, a method of performing color image processing on the masked area, extracting a portion having the same color as the printing, and detecting a portion with another color as a defect is also conceivable. .

As described above, according to the defect detection method for a patch according to the present invention, it is possible to perform optimum processing according to the appearance of the back cut, and it is possible to improve the accuracy of defect detection.
That is, when the back cut can be erased in the imaging process, the defect can be detected without being affected by the back cut.
Also, if the backcut is clearly visible and the entire backcut is recognized as a defect, the effect of the backcut is removed by removing from the defect candidate points those that can be determined to be a backcut from the coordinates and features of the detected defect candidate points. Can be removed.
Furthermore, even if the backcut looks thin and only a part of the backcut is recognized as a defect, the backcut is correctly recognized by tracking the backcut by the above method, and this is masked. By doing so, it is possible to eliminate the influence of the back cut.

  In addition, according to the defect detection method for a patch according to the present invention, it is possible to accurately detect a defect even in a printed region.

Hereinafter, an example of an embodiment of the present invention will be described with reference to the drawings.
FIG. 1 is a cross-sectional view showing a configuration of a transdermally absorbable drug as an example of a patch to be examined according to the present invention. As shown in FIG. 1, the transdermally absorbable drug 10 comprises a support 11 made of paper or nonwoven fabric, an adhesive layer 12 laminated on the support 11 and containing a drug, and the adhesive layer 12 And a separator 13 made of PET or the like to be coated. The separator 13 is formed with a cut 14 (referred to as a back cut in the present invention) that allows the separator 13 to be partially peeled off. The back-cut 14 makes it easy to apply the transdermal absorbent 10 by partially peeling the separator 13 and partially exposing the adhesive layer 12, and the adhesive layers 12 adhere to each other at the time of application. This is to prevent the intrusion.

FIG. 2 is a plan view of the transdermally absorbable drug 10 as viewed from the separator 13 side. The shape of the back cut 14 is usually linear, curved, or various other shapes. In this embodiment, as shown in FIG. 2, a wavy back cut 14 is formed. ing.
In general, a product mark or the like may be printed on the surface of the separator or the support. In this embodiment, the case where the precautions at the time of affixing as the printing 15 is attached to the separator 13 is shown.
Further, the percutaneous absorption medicine 10 is manufactured by a manufacturing process in which an adhesive 12 is applied to a support 11 as a continuous sheet, a separator 13 is bonded, and then cut into pieces by punching or cutting. It is. Therefore, even if printing is periodically applied in the state of a continuous sheet, when it is divided into pieces, only about one cycle is attached. It is difficult to specify.

  FIG. 3 is a conceptual diagram showing the configuration of a defect inspection apparatus used in one embodiment of the defect inspection method of the present invention. The defect inspection apparatus 1 includes an imaging unit 2 that captures an image of a patch to be inspected, and an image processing unit 3 that determines a defect of the patch by image processing based on the obtained image data. Yes.

The imaging means 2 includes an illuminating device 21 that irradiates a patch (not shown) with light, and an image input device 22 that captures an image of a transdermal drug.
As the lighting device 21, a fluorescent lamp, a halogen lamp, an LED, or the like can be used in accordance with the patch to be inspected. Also, the illumination method can be appropriately changed depending on the configuration of the patch to be inspected and the type of defect to be detected. For example, if the patch to be inspected is light-transmitting, it is possible to detect a light-blocking defect such as hair by using transmitted illumination. Further, when the texture of the inspection object is rough, it is preferable to use reflected illumination together, and this can suppress an adverse effect on the image due to the roughness. Furthermore, when the patch to be inspected is not light transmissive, it is preferable to use reflected illumination. In this case, in order to suppress an adverse effect on the image due to the roughness of the texture, for example, light from all angles may be incident using dome-shaped illumination.

  The image input device 22 converts transmitted light or reflected light from a transdermal drug into image data. As the image input device 22, a line sensor, a two-dimensional camera, or the like can be used.

FIG. 4 shows an example of the arrangement of the illumination device 21 and the image input device 22.
When the patch to be inspected is a light-transmitting material, if reflected illumination is used, it is affected by the background, and if the background is dirty, it may be determined as a defect. Therefore, when a light-transmitting patch is to be inspected, as shown in FIG. 4, a transmission illumination 21 a disposed on the side opposite to the image input device 22 is provided on the patch 10 to be inspected. It is effective to use.

In addition, when the transmitted illumination 21a is applied, it is possible to irradiate with a light transmissive member such as glass disposed on the back surface, but in this case, there is a risk of erroneous detection due to contamination of the glass surface. is there. Therefore, as shown in FIG. 4, if the patch 10 is transported by the belt conveyor 23 and inspection is performed at the transfer portion between the conveyor 23 and the conveyor 23, the above background problem can be solved. Thus, when it is going to test | inspect at the delivery part of the conveyor 23, it is preferable that the image input device 22 uses a line sensor. In addition, if the distance between the conveyors 23 is wide, the product flutters. Therefore, it is preferable that the distance between the conveyors be as narrow as possible.
Further, when the patch to be inspected is not light transmissive, only reflected illumination can be employed. In such a case, it is also possible to input an image using a two-dimensional camera instead of a line sensor.

Further, when the patch 10 is provided with the separator 13 having the back cut 14 formed, the back cut 14 may be clearly visible or thin by simply using the transmitted illumination 21a. In such a case, it is preferable to arrange both the transmissive illumination 21a and the reflective illumination 21b, adjust the light quantity balance between them, and prevent a back-cut image from being taken.
Illumination has different conditions depending on the material of the adhesive layer and the separator, but here, as an example, the case where the adhesive layer is provided with a translucent non-woven fabric and the separator is made of transparent PET is described as an inspection target. To do.
First, as a light source for illumination, a fluorescent lamp is used for both transmission illumination and reflection illumination. In the case of only transmitted illumination, light diffuses at the edge of the back-cut portion, and that portion becomes darker than the surroundings. On the other hand, in the case of only the reflected illumination, light is scattered and looks bright at the edge of the spine. Therefore, the same brightness as that of the surroundings may be obtained by combining the reflected illumination and the transmitted illumination and combining the lights of both. As a method of actually setting such illumination conditions, first, only the transmitted illumination is turned on to take an image, and then the reflected illumination is turned on to gradually increase the light quantity of the reflected illumination. Find the condition that disappears.
In such a case, if the reflected illumination is irradiated from only one direction, the light is irradiated only on one side of the back-sliced edge, so the reflective illumination can be arranged so that the light is irradiated on both sides of the back-sliced edge. Preferably, for example, there may be mentioned a method of installing on the left and right sides in the upstream and downstream sides with respect to the flow direction, or in the direction perpendicular to the flow direction.

  On the other hand, the image processing means 3 includes a pre-processing unit 31 for performing noise removal processing on the captured image data, a defect detection unit 32 for detecting defects, and a back-cutting for creating a back-cut mask. A mask creating unit 33 and a print mask creating unit 34 for creating a printing mask are provided.

  The processing content in the preprocessing unit 31 can be appropriately selected depending on the noise to be removed. For example, when the noise is only a bright spot, an opening process is adopted to remove the bright spot. Conversely, when the noise is only a black spot, the closing process is adopted to remove the black spot. When both a bright spot and a black spot exist, a median filter or a smoothing process is adopted for smoothing.

The defect detection unit 32 detects a defect by a defect enhancement process 32a and a binarization process 32b. As a specific example of the defect enhancement processing 32a, differential processing can be cited. In the case of differential processing, only the edge of the defect is detected. Therefore, if there is a large defect whose luminance changes gently, the defect may not be detected. Therefore, in order to detect such a large defect, a method of binarization processing as it is without performing defect enhancement processing may be used together.
Also, the differential process as the defect enhancement process and the binarization process can be performed in parallel, and the defect detected by the differential process and the defect detected by the binarization process are combined to detect the defect. Also good.

The back-cut mask creating unit 33 and the print mask creating unit 34 detect back-cut and print in parallel with the defect detection unit 32, and create a mask for the area. By masking the back-cut and printed area, that is, by not inspecting the area, false detection due to the presence of back-cut and print can be prevented.
However, in this case, since defects in the mask area cannot be detected, preferably, the masked area is inspected by changing the determination criterion. This makes it possible to detect a fatal defect even in the mask area.

Here, a method for creating a mask for a region in which a back cut is formed will be described more specifically.
The appearance of the back cut varies depending on the configuration of the patch and the lighting conditions, and can be classified into a case where the back cut is not visible, a case where it is clearly visible, and a case where it appears thin.
If the patch is highly light transmissive, as described above, it is possible to obtain an image in which the back is not visible by adjusting the light quantity balance between the transmitted illumination and the reflected illumination. In such a case, it becomes possible to detect a defect without performing mask processing especially on the back cut.

  However, when the patch is not light-transmitting or is adjusted to an optical system in which a defect to be detected is most visible, the back cut cannot always be erased. Therefore, in such a case, the following processing is performed according to the appearance of the back cut.

First, as shown in FIG. 5, a case where the back-cut 14 can be clearly seen will be described.
When the back cut 14 is clearly visible, the entire back cut 14 is recognized as a defect, so that particles detected as a defect are connected as a line. Therefore, when the particles detected as defects in this way are connected as lines, these particles are determined to be a back cut and excluded from the defects. However, if there is a defect on the backcut, it will be difficult to determine whether it is a backcut or a defect. Therefore, in such a case, a backcut and a defect are distinguished as follows.

First, when the point-like defect 41 exists on the back cut 14, the shape of the back cut 14 is recognized as if the line width is widened. Therefore, when determining whether the line is a backcut or a defect, the line width of the detected particle is measured, and when the line width does not fall within a predetermined reference value, this is determined as a defect. Thereby, when there is a dot-like defect 41 on the back cut 14, the defect 41 can be distinguished from the back cut 14.
Further, when a linear defect 42 exists on the back cut 14, it is recognized that an intersection has occurred on the back cut 14. Therefore, it is determined whether or not the detected particles are connected by a single line. If they are connected, it is determined that they are back-crossed, and if they are not connected by a single line, an intersection occurs. If the detected particle is branched, it is determined as a defect. Thereby, it is possible to determine the presence or absence of the linear defect 42 existing on the back cut.
Further, it is possible to detect a defect from the gray value of the particles forming the back cut 14. That is, when the detected gray value is clearly different from the back-cut 14 (for example, when it is extremely dark or bright), it is possible to prevent the defect from being overlooked by determining the pixel as a defect.

Next, a case where the back cut 14 looks thin will be described.
When the backcut 14 appears thin, for example, as shown in FIG. 6, particles detected as defects are not connected as a single line, and therefore these particles cannot be determined as a backcut.

Therefore, first, a method of performing binarization processing with a threshold smaller than the threshold used for binarization processing for defect detection is adopted. When defects detected by this method are connected as a line, the connected defect can be determined as a back cut.
However, even if the binarization process is performed with a small threshold value, if the detected defect is not connected as a line, such a defect cannot be determined as a backcut, and this method cannot be used. .

Therefore, a processing method when the back cut is interrupted will be described with reference to FIGS. 6 and 7. In this case, it is assumed that the back-cut particles have a luminance value smaller than that of other portions.
First, a point that is the starting point for back-tracking is searched. The method for searching for the start point is to set a start point search line 52 on an arbitrary coordinate in the direction in which the back cut is formed (indicated by X in FIG. 6 and hereinafter also referred to as “X-axis direction”). A luminance value is measured along the start point search line 52, a point 54 where the luminance value is minimum is obtained from the measured luminance value distribution 53, and a difference between the minimum value and an average value of other portions is a predetermined value. If it is larger than this point, this point 54 is set as the starting point of back-tracking.
If the difference between the minimum luminance value and the average value of other parts is smaller than a predetermined value, this is not used as the starting point for back-traffic tracking, and the same search is performed again in different coordinates in the X-axis direction. This is repeated until the start point of backtracking is found.

Next, the back-crossing is traced along both the positive and negative directions of the X axis (left and right direction in FIG. 6) from the tracking start point. FIG. 7 illustrates the case of tracking in the positive direction (right direction) of the X axis. Here, FIG. 7 shows the luminance values for each pixel in the obtained image data arranged side by side on the coordinates of the pixel.
A pixel 54 marked with a circle at the left end is a tracking start point 54 discovered by the above-described start point search. First, an adjacent region (for example, three pixels as shown in FIG. 7) 55 on the right side of the start point 54 is selected. Search and find the minimum point. If the difference between the minimum value and the average value of the other part is larger than a predetermined value, that point is determined as a back-cut candidate point. Further, the adjacent region on the right side of the back-cutting candidate point is searched for, and similarly, a new back-cutting candidate point is found.

However, there may be almost no difference between the minimum value and the average value of the luminance values detected on the way. In such a case, it is determined that the back cut is interrupted, and the back cut candidate points are not recognized. When the backcut candidate point is not recognized, the search range is gradually expanded toward the X-axis direction in order to find the next candidate point. That is, when a back-off candidate point is searched for a second adjacent area (for example, 5 pixels as shown in FIG. 7) 56 adjacent to the adjacent area of the back-cut candidate point, and no back-off candidate point is found, The candidate for the back cut is searched for a third adjacent region 57 (for example, 7 pixels as shown in FIG. 7) 57 adjacent to the second adjacent region 56.
Then, when a back-cut candidate point is found again, the back-cut is tracked in the same manner as the method performed from the start point of the back-cut tracking.
In this way, after detecting candidate back-crossing points one after another, for example, when these multiple back-off candidate points are connected with a predetermined length, these back-off candidate points are finally set as back cuts. judge.
In this way, as shown in FIG. 7, the coordinates indicated by the circles are recognized as backcuts.

By repeating such a process, even when the back cut is interrupted, the back cut portion can be identified relatively accurately and a back cut mask can be created.
In addition, when the back cut is formed in a thin line shape and there is a possibility of erroneous detection as it is, it is preferable to determine a region to be masked by expanding a point to be determined as a back cut by so-called expansion processing. .

Next, a method for creating a mask for a printed area will be described.
When the printing portion is clearly visible, it is possible to extract a region printed by the binarization process and use it as a mask region. However, if the optical system is set so that the defect to be detected is most visible, the print may not be clearly visible. Even when there is shading, it is difficult to detect printing depending on the printing position.

In the present invention, when the printing position has periodicity, a method of masking printing using this can be adopted. Hereinafter, this method will be specifically described with reference to FIGS. First, a part of printing is detected by pattern matching, and an arbitrary printing position (coordinates) attached to the patch 10 is specified. Coordinates that are separated from the arbitrary printing by a predetermined printing pitch in the vertical and horizontal directions are calculated. Next, as shown in FIG. 9, the mask image 16 obtained by expanding the reference print image 15 is applied to the calculated individual coordinates, thereby completing the print mask as shown in FIG.
Note that the reference print image 15 may be created in advance from a non-defective image. However, when the image is captured by a line sensor, the image may not match the image created in advance because of the large expansion and contraction. Therefore, by performing binarization processing on the print portion detected by pattern matching and automatically creating a mask from the print of the inspection target product itself, a mask with higher accuracy can be made.
If the captured image has a tilt, the image may be rotated and leveled in advance, or the tilt may be calculated, and the coordinate calculation of the printing position may be corrected according to the tilt. . As a method of calculating the inclination angle, the inclination angle can be obtained by obtaining two or more boundary points of the edge of the product and calculating an approximate straight line connecting these points.

  As another method for creating a mask in a region where printing has been performed, when the printing is colored, it is also possible to employ a method in which only the printing color is extracted and used as a mask region. However, if there is a defect of the same color as the print color, the defect may be missed. Therefore, it is possible to prevent oversight by calculating feature quantities such as line width and area of printing, and using a method of judging this as a defect when the feature quantity is clearly different from the actual possible printing. it can.

  Furthermore, a method for performing defect inspection by changing the determination criterion only for the mask region will be described. First, after performing a defect inspection on the whole according to a predetermined standard (first standard), the area where the back-cut is formed and the area where printing is applied are masked as described above. Then, the entire defect is inspected according to the mask area standard (second standard), and areas other than the mask area are masked with an image obtained by inverting the mask area. In this way, by performing defect inspection with two criteria and synthesizing each other's inspection results, it is possible to perform the entire inspection with the determination criteria changed only for the mask region.

Sectional drawing which showed an example of the structure of the patch (transdermal drug) used as the test object of this invention. The top view at the time of seeing the patch (transdermal drug) used as the test object of this invention from the separator side. The conceptual diagram which showed one form of the inspection apparatus used for the inspection method which concerns on this invention The figure which showed one form of the imaging process among the inspection apparatuses used for the inspection method which concerns on this invention. The figure which showed an example of the defect which exists on a back cut and a back cut. The figure which showed an example of the case where a back cut appears to be interrupted, and distribution of the brightness | luminance measured. The figure explaining the back cut search method. The figure explaining the masking method of printing. The figure which showed the creation example of the mask image of printing.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Defect inspection apparatus 2 Imaging means 3 Image processing means 10 Adhesive material 11 Support body 12 Adhesive layer 13 Separator 14 Backcut 15 Printing 16 Print mask image 21 Illumination device 22 Image input device 23 Conveyor 31 Preprocessing section 32 Defect detection section 33 Backcut mask creation unit 34 Print mask creation unit 41 Point defect 42 on back cut Line defect 31 on back cut 31 Back cut extraction area 52 Back cut start point search line 53 Luminance distribution 54 of back cut start point search line 54 Start point (minimum part of brightness)

Claims (4)

A method for inspecting defects in a patch comprising an adhesive layer, and a separator that covers the adhesive layer and has a back cut formed thereon,
An imaging step of taking an image of the patch in a state in which the patch is irradiated with light;
And an image processing step of extracting a defect from the photographed image and determining pass / fail,
In the image processing step, an arbitrary point detected as a defect is set as a back-cut candidate point, and a defect detected within a predetermined range from the back-cut candidate point toward the back-cut forming direction is set as another back-cut candidate point. , Extracting and masking the region where the back cut is formed by determining the back cut candidate point as a back cut based on the feature amount of the plurality of back cut candidate points detected by repeating the operation A method for inspecting a defect of a patch material characterized by the above. The defect inspection method for a patch according to claim 1, wherein the masked area is inspected according to a determination criterion different from that of the unmasked area . The defect inspection method for a patch according to claim 1 or 2, wherein a defect is detected based on a shape or shading of an area extracted as the back cut . When the patch is one having periodic printing,
The position of the printing is detected by pattern matching, and an area where printing has been performed is estimated based on the detected position information and pre-input print pattern information, and a mask is automatically created from the estimated area. The defect inspection method for a patch according to any one of claims 1 to 3, wherein the defect is created and masked .

Images