JP2003067724A - Oblateness detection method for ringed pattern and oblateness detection system for ringed pattern - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an oblateness detection system that can precisely detect an oblate defect occurring in cut faces of a hollow fiber bundle set with a resin material at both end faces of a hollow fiber module to cause a degradation in dialysis performance. SOLUTION: An oblateness detection method for a ringed pattern binarizes an inspection surface image where many ringed patterns collect with respect to a ringed portion and a background portion, classifies a small area district present in the processed image into a circular pattern and a noncircular pattern consisting of patterns except the circular one and a pseudo pattern, removes the pseudo pattern present within a constant distance of a circular pattern circumference from oblateness detection targets, and determines whether the forms of all the remaining patterns are oblate beyond a threshold or not.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a device for inspecting a hollow fiber module used in an artificial dialyzer for removing waste products in blood by blood passing through the hollow fiber and dialysate passing outside the hollow fiber, In particular, a portion of a yarn bundle that is inserted in a cylindrical container with both ends open and is made of a large number of hollow fibers whose material is a semipermeable membrane and whose both ends are open. TECHNICAL FIELD The present invention relates to a hollow fiber module flatness detection method and flatness detection device for detecting the presence or absence of flatness of a hollow fiber shape on a surface.

[0002]

2. Description of the Related Art A hollow fiber module has a cylindrical container whose both ends are open, in which a yarn bundle composed of a large number of hollow fibers whose both ends are opened by inserting a semipermeable membrane is inserted into the container. It has a resin in the gap.

In the process of assembling a hollow fiber module, when the shape of the hollow fiber at both end faces of a normally circular fiber bundle is flattened due to a problem during transportation of the fiber bundle or cutting of the end face, during the artificial dialysis. In addition, the flow of blood in the hollow fiber may be hindered, the dialysis performance of the hollow module may be deteriorated, and sufficient artificial dialysis may not be performed.

In detecting the flatness of the hollow fiber, an image obtained by picking up the end face of the hollow fiber module with a high-resolution CCD camera is subjected to a differential filter by an image processing device to emphasize the hollow fiber portion, and the image is hollowed. The flatness of the hollow part of the hollow fiber labeled by the labeling process, which is divided into the thread part and the resin part / hollow part of the thread (aspect ratio: the maximum width of the region and the length orthogonal thereto) ratio)
The presence or absence of flatness of the hollow fiber was determined by whether or not the value exceeds the threshold value defined by.

[0005]

SUMMARY OF THE INVENTION An object of the present invention is to accurately produce flat defects that occur on the cut surfaces of hollow fiber bundles hardened with a resin material on both end surfaces of a hollow fiber module and cause deterioration of dialysis performance. It is to provide a flat detection device capable of detecting.

When the labeling process is performed on the binarized image of the hollow fiber module end face image, the region (simulated pattern) in which the resin portion is surrounded by a plurality of hollow fibers is also labeled. Since such a pseudo pattern is similar to the shape of the hollow part when the hollow fiber is flat, it is difficult to distinguish the pseudo pattern and the flat pattern based on the information from the pattern shape, and many pseudo patterns are flat when making the judgment. Is falsely detected.

Therefore, another object of the present invention is to provide a flatness detecting device for a hollow fiber module which can accurately inspect only the cut shape of the hollow fiber in order to prevent such erroneous detection.

[0008]

In order to achieve the above object, the present invention adopts the following constitutions. That is, (1) an inspection plane image in which a large number of ring-shaped patterns are aggregated is binarized by a ring portion and a background portion, and a small area group existing in the processed image is converted into a circular pattern and a shape other than a circle. Patterns and non-circular patterns consisting of pseudo patterns are removed, and the pseudo patterns existing within a certain distance from the outer circumference of the circular pattern are removed from the flatness detection target, and the shapes of all the remaining patterns are flattened beyond a certain threshold. A flat pattern detection method of a ring pattern for determining whether or not. (2) The method of removing the pseudo pattern from the inspection target is to perform an expansion process of expanding the pattern to a distance in which the ring width and the diameter of the ring inner circle are combined in the image in which only the circular pattern is extracted, The flat pattern detection method for a ring-shaped pattern according to (1), wherein the non-circular pattern existing in the mask area is removed by a distance mask image created by performing contraction processing for the diameter. (3) The inspection surface on which the large number of ring-shaped patterns are aggregated is the end surface of a hollow fiber module in which both end surfaces of a hollow fiber bundle composed of a large number of hollow fibers are fixed with a resin material (1) or The method for detecting flatness of a ring-shaped pattern according to (2). (4) The flattening detection method according to (1) or (2), which has an image input unit for capturing an image of an inspection surface on which a large number of ring-shaped patterns are gathered, and an image processing unit. A flat pattern flatness detecting device using a ring. (5) In the above (4), the inspection surface on which the large number of ring-shaped patterns are gathered is the end surface of a hollow fiber module in which both end surfaces of a hollow fiber bundle composed of a large number of hollow fibers are fixed with a resin material. The flat detection device of the described ring-shaped pattern. (6) After the end face of the hollow fiber module is cut, the flat pattern detecting method for a ring-shaped pattern according to any one of (1) to (3), or the inspection device according to (4) or (5). A method for manufacturing a hollow fiber module, in which the quality of the hollow fiber shape is determined by an inspection using, and only a good product is post-processed.

[0009]

BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of the present invention will be described below with reference to the drawings. The flattening detection device of FIG. 1 includes a support device 101 for a hollow fiber module M, a light irradiation device 102, and an imaging device 1.
03, an image processing device 104, an operation command device 105, a display device 106, and a module transfer device 107. This flatness detecting device detects a flat defect 501 of the hollow fiber generated on the end surface of the hollow fiber module M. The configuration and operation of each unit will be described below. <Module Support Device> The module support device 101 is
The hollow fiber module M is mounted, and the inclination of the hollow fiber module M can be adjusted so that the axial center of the end surface of the hollow fiber module M is parallel to the optical central axes of the lens 103A and the CCD camera 103B described later. By moving the installation position vertically and horizontally for each type of hollow fiber module M having different outer diameters and cylindrical lengths, it is possible to match the optical center axis. In this embodiment, since the end surface of the hollow fiber module is imaged by the line sensor camera described later, the module supporting device electric stage is installed and the module is moved horizontally at a predetermined speed. <Light Irradiation Device> The light irradiation device 102 includes a light source 102A and a light irradiation unit 102B. The light source 102A adjusts the light amount according to a command from the operation command device 105 described later.
The light irradiation unit 102B is configured such that the irradiation port 102B1 and the irradiation port 102B2 are line symmetrical with respect to the optical center axis, and is installed at the same light irradiation angle θ with respect to the optical center axis of the end face of the hollow fiber module M (see FIG. 2). In this embodiment, a halogen lamp light source is used as the light source 102A, and a line type optical fiber light guide with a bifurcated irradiation port is used for the light irradiation unit 102B. <Imaging Device> The imaging device 103 is for measuring the end surface of the hollow fiber module M mounted on the supporting device 101 to obtain an image, and is composed of a lens 103A and a camera 103B. An end face image of the hollow fiber module M is obtained by the image pickup device 103. In this embodiment, a 5000-pixel line sensor camera is used. <Image Processing Device> The image processing device 104 processes the captured image of the end surface of the hollow fiber module M obtained by the imaging device 103 to detect the flatness of the hollow fiber. This will be described based on the flowchart showing the flow of processing in the image processing apparatus of FIG.

First, the image input to the image processing apparatus 104 is subjected to a pre-processing by the high-pass filter 201 so that the wall thickness region 301 of the hollow fiber shown in FIG. 5 is emphasized. By the filter 201, the hollow fiber portion 301 is emphasized in the high brightness direction, and the hollow portion 302 and the resin portion 303 of the hollow fiber are emphasized in the low brightness direction. When the processed image is binarized 202 by a predetermined threshold value, the binarized image in which the hollow fiber portion 301 is a white pixel and the hollow fiber hollow portion 302 and the resin portion 303 are black pixels is obtained as illustrated in FIG. .

Next, a labeling process 203 is performed to label the black pixel areas within an arbitrary area range. The pattern subjected to the labeling process includes a circular pattern 401 which is a hollow portion of a normal hollow fiber, a pattern 402 other than a circular shape in which the hollow fiber shape is abnormally deformed, and a resin portion surrounded by a plurality of hollow fibers. Is recognized as one area and there is a pseudo pattern 403. In order to perform the classification process 205 for classifying these patterns into circular patterns and other non-circular patterns, the following formula (1)
The ratio α α = L / of the long side D and the short side L when the rectangle 404 circumscribing the pattern is calculated so that the long side D becomes the maximum value of the pattern length as shown in FIG. D ... (1) and the outer peripheral length ρ of the pattern, the calculated value A1 of the pattern area calculated assuming that the pattern is a perfect circle and the number of pattern pixels existing in the area The index value calculation 204 of either of the ratios β β = A1 / A2 = ρ / 4πA2 (2) of the area A2 is performed.

A pattern below a threshold value set for each of the two index values is a circular pattern 401, and a pattern exceeding at least one of the index values is a non-circular pattern (a pattern other than a circle 402, a pseudo pattern). When the pattern is a perfect circle, the index values α and β are both 1 as shown in the following equation.

Α: In the case of a perfect circle, L = D, so that α = 1. Β: In the case of a perfect circle, A1 = A2 = πR2, so β = 1. Next, the removal process 20 for removing the pseudo pattern 403.
Do 6. Since the pseudo pattern 403 is generated in the resin portion surrounded by the plurality of hollow fibers, the distance D1 from the outer periphery of the circular pattern 401 is always the distance D2 between the hollow portions of the hollow fibers as shown in the following formula (3). Is less than half of that (see FIG. 9). D1 ≦ D2 / 2 (3) Then, from the circular pattern 401 to the pseudo pattern 403
The pseudo pattern 403 is removed from the inspection target by utilizing the difference between the distance D1 up to and the distance D2 up to the non-circular pattern 402.

First, only the circular pattern 401 is taken out from the pattern distribution image as shown in FIG. 7, and the distribution image of only the circular pattern shown in FIG. 8 is created. Next, with respect to the image, an expansion process is performed in which the area of the circular pattern is enlarged from the outer circumference of the circular pattern 401 by a distance combining the wall thickness design value of the hollow fiber and the radius design value of the inner circle of the hollow fiber. Shrinkage processing is performed to reduce the area of the circular pattern by the inner diameter design value of the hollow fiber. As a result, the white pixel 4 in which the circular pattern 401 is expanded is an area in which the pseudo pattern 403, which is an area surrounded by a plurality of hollow fibers, can occur.
The non-circular pattern portion 4 which is filled with 05 but is separated from the circular pattern by at least twice the wall thickness of the hollow fiber 4
The area 02 remains as black pixels, and the distance mask image illustrated in FIG. 11 is created.

In the mask image, the white pixel portion is a pattern classified into a circular pattern 401 and a pseudo pattern 4.
The area where 03 is present is the area where the pattern 402 other than the circular shape can exist in the black pixel portion.

When the mask image and the pattern image shown in FIG. 7 are superposed and the pattern image pattern overlapping the white pixel portion of the mask image is removed, the circular pattern 401 and the pseudo pattern 403 of the pattern image are deleted. The image shown in FIG. 12 in which only the pattern 402 other than the circular shape remains is created. If the logical sum of the pattern distribution image other than the circular shape and the circular pattern image shown in FIG. 8 is obtained,
The hollow part pattern image of the hollow fiber shown in FIG. 13 in which only the hollow part of the hollow fiber including the circular shape and the non-circular shape has white pixels is created. For all patterns of this hollow fiber pattern image,
The index value α and β calculation processing 207 is performed, and a pattern that exceeds either or both of the threshold values is determined to be the flat pattern 501, and the flat pattern position of the binary image shown in FIG. 6 is marked. The flat detection result image shown in FIG. 14 is obtained.

Further, the number of the flat hollow fibers 501 existing on the end surface of the hollow fiber module M is counted, and when the number exceeds a predetermined threshold value, the hollow fiber module M to be inspected is regarded as a defective module. <Operation command device> In the operation command device 105, an inspection index value suitable for an inspection target by sending a command to the image processing unit 104 in response to differences in module outer diameter, outer diameter and inner diameter of hollow fiber, assembly method, and the like. And binarization threshold, high-pass filter coefficient, etc. are set. A command is sent to the light irradiation device 102 to adjust the light amount. -A command is sent to the module support device 101, and the stage is moved laterally at a speed suitable for imaging with a line sensor camera. <Display Device> In the display device 106, as the flatness detection result,
The result image shown in FIG. 14 in which the flat portion 501 is marked and the results such as the number of flattened hollow fibers of the hollow fiber module are output. <Module Transfer Device> In accordance with a command from the operation command device, the hollow fiber module M for which the flatness detection result has been determined is transferred to the subsequent step when the hollow fiber module M is normal. -If the hollow fiber module M is defective, remove it from the production line.

[0018]

Embodiments of the present invention will be described with reference to the drawings.

The image in FIG. 16 is an image of the end face of the hollow fiber module. A binary image as shown in FIG. 17 is obtained by applying a differential filter to the image by the image processing device 104 and then binarizing the image with a predetermined threshold value. In this image, there are 15323 small patterns composed of black pixels to be inspected, and when a rectangle circumscribing the pattern is calculated so that the long side D becomes the maximum value of the pattern length for each small pattern. There are 13408 non-circular patterns in which the ratio α (= L / D) between the long side D and the short side L is 0.7 or less. Of the patterns other than this circle,
There are 1903 non-circular patterns whose distance from the circular pattern is shorter than 6 pixels which is twice the wall thickness of the hollow fiber. Since these patterns are pseudo patterns that are not hollow portions of the hollow fiber, the circular pattern is expanded by 6 pixels which is the radius of the inner circle of the hollow fiber, and then the wall thickness of the hollow fiber is increased. Since it overlaps with the white pixel portion of the distance mask image shown in FIG. 18 which is obtained by contracting the circular pattern by 3 pixels, which is a minute portion, it is removed from the flat inspection object. Since the pattern other than the remaining circles is determined to be the hollow portion of the hollow fiber, an inspection defect image marking the portion determined to be flat as shown in FIG. 19 is output, and the hollow fiber flat defect in the end image of the hollow fiber module is output. The number is 12.

[0020]

According to the method of claims 1 to 3 and the device of claims 4 to 5, the pseudo pattern, which is the background portion surrounded by the hollow fibers and forming one region, is removed from the flat detection object, Accurate flatness detection of the hollow fiber can be performed without erroneous determination due to the pseudo pattern.

According to the method for producing a hollow fiber module of claim 6, the defective hollow fiber module, which is present in excess of the threshold value determined by the flat hollow fibers, is prevented from flowing to the subsequent step.
Even for defective hollow fiber modules, it is possible to eliminate the irrationality of performing the remaining processing operations.

[Brief description of drawings]

FIG. 1 is a configuration diagram of a flat detector in an embodiment of the present invention.

FIG. 2 is a configuration diagram of a light irradiation device according to an embodiment of the present invention.

FIG. 3 is a configuration diagram of an image pickup apparatus according to an embodiment of the present invention.

FIG. 4 is a flowchart showing a processing flow in the image processing apparatus in the embodiment.

FIG. 5 is a picked-up image showing the fiber shape of the hollow fiber in the example.

FIG. 6 is a binarization processing result diagram of a captured image showing the fiber shape of the hollow fiber in the example.

FIG. 7 is an image showing a fiber shape of only a pattern obtained by extracting a labeled pattern from a binarized image.

8 is an image showing the shape of fibers having only a circular pattern obtained by extracting only the circular pattern from the image of FIG. 7. FIG.

FIG. 9 is an image showing a fiber shape showing a difference in distances D1 and D2 between a circular pattern, a non-circular pattern, and a pseudo pattern.

FIG. 10 is an image showing a shape of a fiber showing a long side D and a short side L when a pattern is surrounded by a circumscribing rectangle.

11 is an image showing the shape of fibers of a distance mask created by performing expansion and contraction processing on the image of FIG.

FIG. 12 is an image showing the shape of fibers in which only a pattern 402 other than a circle, which did not fall within the mask range of FIG. 10, remains.

FIG. 13 is an image showing the fiber shape of only the hollow part pattern of the hollow fiber created by the logical sum of FIG. 7 and FIG. 11.

FIG. 14 is an inspection result output image showing the shape of a fiber marking a defective portion.

FIG. 15 is a flowchart showing an instruction procedure to the module transfer device in the operation instruction device.

FIG. 16 is a captured image showing the shape of fibers on the end surface of the hollow fiber module.

FIG. 17 shows a differential filter applied to the image of FIG.
It is a binarized image which shows the shape of the fiber which carried out the binarization process.

18 is an image showing the fiber shape of the distance mask created by performing expansion and contraction processing on an image in which only a circular pattern is extracted from the binarized image of FIG.

FIG. 19 is an image showing the shape of fibers in which only a pattern other than a circle that did not fall within the mask range of FIG. 18 remained.

[Explanation of symbols]

101: Module support device 102A: Light irradiation device Light source 102B: Light irradiation device Irradiation port 103A: Imaging device lens 103B: Imaging device CCD camera 104: Image processing device 105: operation command device 106: Display device 107: Module transfer device 301: Thick part of hollow fiber 302: hollow part of hollow fiber 303: Resin part 401: circular pattern 402: pattern other than circular 403: Pseudo pattern 404: A rectangle circumscribing the pattern 501: Hollow part of flat hollow fiber

─────────────────────────────────────────────────── ─── Continuation of front page (51) Int.Cl. ⁷ Identification code FI theme code (reference) G06T 7/60 150 G06T 7/60 150S // A61M 1/18 513 A61M 1/18 513 (72) Inventor Satoshi Kurashiki 1-1-1, Sonoyama, Otsu-shi, Shiga Toray Co., Ltd. Shiga Business Site F-term (reference) 4C077 AA05 BB01 CC04 HH10 HH20 KK30 LL05 4D006 GA13 HA02 LA10 MA01 PA01 PB09 PB42 PC47 5B057 AA07 BA29 CA08 CA12 CB06 CA12 CB06 CA12 CB01 CE12 DA03 DA16 DB02 DB08 DC09 5L096 BA06 EA02 EA43 FA04 FA66 GA51

Claims (6)

[Claims] 1. A method for detecting flatness of a ring-shaped pattern, wherein an inspection surface image in which a large number of ring-shaped patterns are aggregated is binarized by a ring portion and a background portion and is present in the processed image. The small area group is classified into a circular pattern, a non-circular pattern composed of a pattern other than a circular pattern and a pseudo pattern, and the pseudo pattern existing within a certain distance from the outer periphery of the circular pattern is removed from the flat detection target, and all the remaining patterns are A method for detecting flatness of a ring-shaped pattern, comprising determining whether or not a shape is flattening beyond a certain threshold. 2. The method for removing the pseudo pattern from the inspection object is to perform expansion processing to expand the pattern to a distance where the ring width and the diameter of the ring inner circle are combined in an image in which only the circular pattern is taken out, and then the inside of the ring is expanded. By the distance mask image created by performing contraction processing for the diameter of the circle,
The flat pattern detection method for a ring-shaped pattern according to claim 1, wherein the non-circular pattern existing in the mask area is removed. 3. An inspection surface on which a large number of ring-shaped patterns are gathered is an end surface of a hollow fiber module in which both end surfaces of a hollow fiber bundle composed of a large number of hollow fibers are fixed with a resin material. The method for detecting flatness of a ring-shaped pattern according to claim 1 or 2. 4. An image input means for picking up an inspection surface on which a large number of ring-shaped patterns are gathered, and an image processing means,
A flat pattern flatness detection device having a ring-shaped pattern, wherein the image processing means uses the flatness detection method according to claim 1. 5. The inspection surface on which the large number of ring-shaped patterns are aggregated is an end surface of a hollow fiber module in which both end surfaces of a hollow fiber bundle composed of a large number of hollow fibers are fixed with a resin material. The flattening device for a ring-shaped pattern according to claim 4. 6. The hollow fiber shape at the end face of the hollow fiber module by the method for detecting flatness of a ring-shaped pattern according to any one of claims 1 to 3 or an inspection using the flatness detection device according to claim 4 or 5. 2. A method for manufacturing a hollow fiber module, which comprises determining whether the product is good or bad and performing post-processing only on a good product.