Foreign matter detection method and foreign matter detection program

Abstract

To precisely detect foreign matter by using a transmissive image obtained by applying electromagnetic waves.SOLUTION: A foreign matter detection method has the steps of: acquiring a transmissive tomographic image obtained by applying electromagnetic waves to a resin molding; fitting a normal distribution to a pixel value distribution of the acquired transmissive tomographic image, and setting a threshold for binarization processing on the basis of a statistical index of the fitted normal distribution; generating a binarization image from the transmissive tomographic image by using the set threshold; detecting an area corresponding to foreign matter contained in the resin molding from the generated binarization image; and determining the quality of the resin molding regarding whether an area corresponding to foreign matter is detected from a predetermined number of continuous, adjacent transmissive tomographic images.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present disclosure relates to a foreign object detection method and a foreign object detection program. [Background technology]

[0002] X-ray CT (Computed Tomography) devices can measure the internal state of an object, such as a resin molded product, non-destructively without cutting it, and are therefore used for a wide range of applications, including observing the internal structure of the object, detecting defects and foreign objects, measuring and comparing shapes, and evaluating density.

[0003] When the object being inspected is a resin molded product, X-ray CT scanning can be used to detect, as foreign objects, air bubbles caused by resin shrinkage, known as voids, or air bubbles trapped when measuring resin pellets in a molding machine. X-ray CT scanning can also be used to detect, as foreign objects, trace metals and inorganic fillers contained in the raw materials of the resin molded product or contaminated into the resin molded product due to damage or insufficient cleaning of cutters, screws, etc. during extrusion or molding. Because voids and other foreign objects contained in resin molded products can affect the strength and durability of the resin molded product depending on their location and size, X-ray CT scanning, which enables non-destructive testing, is effective for detecting foreign objects in resin molded products.

[0004] For example, Patent Document 1 describes a method for producing an aggregate of thermoplastic resin and discontinuous carbon fibers, in which the presence or absence of foreign matter is detected as a two-dimensional spatial distribution using an X-ray CT device. Also, Patent Document 2 describes identifying foreign matter based on images obtained by a Talbot-type X-ray imaging device. [Prior art documents] [Patent documents]

[0005] [Patent Document 1] Patent Publication No. 2021-000831 [Patent Document 2] International Publication No. 2021 / 002356 Summary of the Invention [Problem to be solved by the invention]

[0006] However, the problem with foreign object detection using X-ray CT scanners is that foreign objects that do not affect the quality of resin molded products are also detected, resulting in a decrease in the accuracy of quality assessment. Specifically, when product quality is checked using an X-ray CT scanner, the X-ray transmission image is a two-dimensional projection image, so even if the detected foreign object is thin enough to be shipped as a product, it may be judged as defective. Furthermore, noise or uneven brightness in the X-ray transmission image may be detected as a foreign object, resulting in the product being judged as lower quality than its actual quality.

[0007] It is also possible to synthesize three-dimensional images and measure the thickness of the foreign object, but the amount of processing required to synthesize three-dimensional images is large, and it is difficult to obtain a three-dimensional image that accurately represents the shape of the foreign object due to differences in color density between different tomographic images, etc.

[0008] The technology disclosed herein has been made in consideration of these points, and aims to provide a foreign object detection method and a foreign object detection program that can accurately detect foreign objects using a transmission image obtained by irradiating electromagnetic waves. [Means for solving the problem]

[0009] According to one aspect of the present disclosure, a foreign matter detection method includes the steps of: acquiring a transmission tomographic image obtained by irradiating a resin molded product with electromagnetic waves; fitting a normal distribution to the pixel value distribution of the acquired transmission tomographic image and setting a threshold value for binarization processing based on a statistical index of the fitted normal distribution; generating a binarized image from the transmission tomographic image using the set threshold value; detecting an area corresponding to a foreign matter contained in the resin molded product from the generated binarized image; and judging the quality of the resin molded product based on whether an area corresponding to the foreign matter is detected in a predetermined number of consecutive adjacent transmission tomographic images.

[0010] According to another aspect of the present disclosure, a foreign matter detection program causes a computer to execute the steps of: acquiring a transmission tomographic image obtained by irradiating an electromagnetic wave onto a resin molded product; fitting a normal distribution to the pixel value distribution of the acquired transmission tomographic image and setting a threshold value for binarization processing based on a statistical index of the fitted normal distribution; generating a binarized image from the transmission tomographic image using the set threshold value; detecting an area corresponding to a foreign matter contained in the resin molded product from the generated binarized image; and judging the quality of the resin molded product based on whether an area corresponding to the foreign matter is detected in a predetermined number of consecutive adjacent transmission tomographic images. [Brief explanation of the drawings]

[0011] [Figure 1] FIG. 1 is a flow diagram showing a foreign object detection method according to one embodiment. [Figure 2] FIG. 2 is a diagram showing an example of a transmission tomographic image. [Figure 3] FIG. 3 is a diagram illustrating an example of a pixel value distribution. [Figure 4] FIG. 4 is a diagram illustrating an example of threshold setting. [Figure 5] FIG. 5 is a diagram showing another example of threshold setting. [Figure 6] FIG. 6 is a diagram showing an example of a binarized image. [Figure 7] FIG. 7 is a diagram illustrating foreign object detection. [Figure 8] FIG. 8 is a diagram illustrating foreign matter determination. [Figure 9] FIG. 9 is a block diagram illustrating an example of the configuration of an information processing device. DETAILED DESCRIPTION OF THE INVENTION

[0012] Hereinafter, an embodiment according to the present disclosure will be described with reference to the accompanying drawings. The embodiment described below is an example and should not be construed as being limited by this description.

[0013] [Resin molded products] In one embodiment, a method for analyzing foreign matter such as voids inside a resin molded product is described. The resin molded product to be analyzed can be made using various conventionally known resin materials. The resin molded product according to this embodiment also includes a resin mixture obtained by blending multiple resins and a resin composition to which various fillers and colorants are added.

[0014] [Manufacturing method for resin molded products] The above-mentioned resin molded article can be obtained by a conventionally known molding method, such as compression molding, transfer molding, injection molding, extrusion molding, and blow molding.

[0015] [Foreign object detection method] 1 is a flow diagram showing a foreign object detection method according to one embodiment. This foreign object detection method is executed by an information processing device such as a personal computer or a server. The information processing device may be connected to an imaging device, such as an X-ray CT device, that captures transmission tomographic images of a resin molded product so that they can communicate with each other.

[0016] The resin molded product is imaged, for example, by an X-ray CT device, and a transmission tomographic image at a predetermined position is obtained (step S101). The range to be imaged may be a part of the resin molded product, or the entire resin molded product.

[0017] When capturing a transmission tomographic image, the resin molded product is irradiated with electromagnetic waves such as X-rays. The type of electromagnetic waves to be irradiated is not particularly limited, and a conventionally known irradiation device can be used, for example, an X-ray CT device that irradiates X-rays from an X-ray tube or the like. The irradiation conditions, such as tube current and irradiation time, are not particularly limited, and may be changed as appropriate depending on the shape of the target resin molded product, the type of resin used, etc.

[0018] Furthermore, when a resin molded product is irradiated with electromagnetic waves, a beam hardening phenomenon may occur in which pixel values ​​at the outer periphery of the resin molded product become high in the resulting transmission tomographic image. For this reason, a metal filter may be placed between the source of the electromagnetic waves and the resin molded product to adjust the current, or a dedicated application may be used to correct the transmission tomographic image.

[0019] The transmission tomographic image thus captured is an image of a cross section (tomogram) when a resin molded product is cut along a certain plane. An example of a transmission tomographic image is shown in Fig. 2. The transmission tomographic image 110 shown in Fig. 2 is an image of a tomogram of a resin molded product, and therefore is an almost uniform image in which the majority of pixels have pixel values ​​corresponding to resin. However, in an area 111 where a foreign substance such as a void exists, the pixel value is significantly different from that of the surrounding pixels.

[0020] When a transmission tomographic image is acquired, a pixel value distribution of the transmission tomographic image is obtained, and a threshold for binarizing the transmission tomographic image is set based on the pixel value distribution (step S102). Specifically, a normal distribution fitting (Gaussian fitting) is performed on the image value distribution of the transmission tomographic image, and a value obtained from a statistical index of the fitted normal distribution is set as the threshold. Examples of this statistical index include the median, standard deviation, two-sided probability, and one-sided probability.

[0021] That is, for example, as shown in the upper diagram of Fig. 3, pixel value distribution 120 of a transmission tomographic image is mostly the pixel value distribution of pixels corresponding to resin, and therefore roughly follows a normal distribution. Gaussian fitting is then performed on pixel value distribution 120 to derive normal distribution 130 representing pixel value distribution 120. However, because pixel value distribution 120 also includes pixel values ​​of pixels corresponding to foreign matter other than resin, pixel value distribution 120 and normal distribution 130 do not match, particularly in the pixel value range that is far from the average value, as shown in the enlarged view in the lower diagram of Fig. 3.

[0022] In other words, in a pixel value range that is far from the average value, the number of pixels with pixel values ​​corresponding to foreign matter increases, and the pixel value distribution 120 deviates from the normal distribution 130 by the amount of these pixels. For this reason, by using a pixel value with a predetermined probability, such as a lower probability or an upper probability, as the threshold for the binarization process, it is possible to obtain a binarized image that accurately distinguishes between resin and foreign matter.

[0023] When a one-sided probability is used as the statistical index, a threshold value based on a lower probability is set if the foreign object to be detected has a pixel value that is relatively low compared to the pixel value distribution 120, and a threshold value based on an upper probability is set if the foreign object to be detected has a pixel value that is relatively high compared to the pixel value distribution 120. In this case, if the distribution of pixel values ​​corresponding to the foreign object overlaps with the pixel value distribution 120 of the transmission tomographic image, a pixel value corresponding to a one-sided probability of 0.1 to 10%, for example, may be set as the threshold, and if the distribution of pixel values ​​corresponding to the foreign object does not overlap with the pixel value distribution 120 of the transmission tomographic image, a pixel value corresponding to a one-sided probability of 0.000001 to 0.1%, for example, may be set as the threshold.

[0024] Fig. 4 is a diagram showing an example in which a threshold value is set based on a one-sided probability. As shown in Fig. 4, when a pixel value for which the lower probability is a predetermined probability is set as a threshold Th, the pixels in pixel value distribution 120 can be separated into two groups: pixels whose pixel values ​​are less than threshold Th and pixels whose pixel values ​​are equal to or greater than threshold Th.

[0025] 5 is a diagram showing an example in which a threshold is set based on a two-sided probability. As shown in FIG. 5, when a pixel value for which the lower probability is a predetermined probability is set, the pixel value is set to a lower threshold T L , and the pixel value at which the upper probability is a predetermined probability is set to the upper threshold T U , each pixel in the pixel value distribution 120 is compared to a pixel value falling below a lower threshold T L Above upper threshold T U Pixels whose pixel values ​​are less than the lower threshold T L Below or above threshold T U The above pixels can be divided into two groups.

[0026] In this way, the threshold value for the binarization process is set from the pixel value distribution 120 of the transmission tomographic image, so that the threshold value is set according to the characteristics of the transmission tomographic image, and the effects of noise, brightness unevenness, etc. are reduced, making it possible to generate a binarized image for accurately detecting foreign objects.

[0027] In addition, when a resin molded product is formed from a resin composition containing, for example, glass fiber, the pixel value distribution of the transmission tomographic image will have two peaks: the pixel values ​​of pixels corresponding to the resin and the pixel values ​​of pixels corresponding to the glass fiber. In such a case, for example, peak detection for the pixel value distribution may be performed to detect the pixel values ​​of each peak, and fitting using a normal distribution may be performed within a predetermined pixel value range centered on the pixel value of each peak. Then, in setting the threshold, a value obtained from a statistical index of the normal distribution corresponding to each peak may be set as the threshold, and each pixel in the pixel value distribution may be separated into two groups: pixels in a pixel value range that includes one of the peaks, and pixels in a pixel value range that does not include either peak.

[0028] Once the threshold value for binarization is set, a binarized image is generated from the transmission tomographic image (step S103). That is, the pixel value of each pixel in the transmission tomographic image is compared with the threshold value, and a binarized image is generated in which the pixel value is binarized to 0 or 1 according to the comparison result.

[0029] Fig. 6 is a diagram showing an example of a binarized image 210. In the binarized image 210 shown in Fig. 6, most of the pixels have pixel values ​​corresponding to resin, but there are scattered pixels such as areas 211 and 212 that have pixel values ​​corresponding to foreign matter.

[0030] Therefore, foreign matter other than resin is detected from the binarized image 210 (step S104). Specifically, pixels having pixel values ​​corresponding to foreign matter in the binarized image 210 are detected, and the areas formed by these pixels are identified. That is, in the example shown in Fig. 6, for example, areas 211, 212, etc. are identified. The foreign matter detected here is different from resin, and therefore includes, for example, air bubbles such as voids.

[0031] The sizes of these regions 211, 212 are then calculated, and regions of a predetermined size or larger are determined to be regions corresponding to foreign objects. The sizes of the regions 211, 212 may be the number of pixels included in the regions 211, 212, or the diameter or radius of the smallest circumscribing circle circumscribing each of the regions 211, 212. For example, when detecting a foreign object having a substantially spherical shape, the size may be calculated by counting the number of pixels included in the regions 211, 212, and when detecting a foreign object having an elongated shape, the radius of the smallest circumscribing circle of the regions 211, 212 may be calculated as the size.

[0032] The number of pixels and the diameter or radius of the smallest circumscribing circle that are determined to be foreign matter may vary depending on the application or purpose of the resin molded product. For example, if the resin molded product is a housing for an actuator, sensor, or the like, the reduction in strength due to foreign matter is unlikely to be a problem, so the radius of the smallest circumscribing circle that is determined to be foreign matter may be less than 0.5 mm, preferably less than 0.2 mm, and more preferably less than 0.1 mm. On the other hand, if the resin molded product is a semiconductor component or the like, high quality is required, so the radius of the smallest circumscribing circle that is determined to be foreign matter may be less than 0.01 mm, for example.

[0033] 6 are compared with predetermined sizes, and if the size of region 211 is equal to or greater than the predetermined size while the size of region 212 is less than the predetermined size, region 211 is determined to be a region corresponding to a foreign object. That is, for example, as shown in FIG. 7, if the radius of the minimum circumscribing circle 221 of region 211 is equal to or greater than a predetermined value, region 211 is determined to be a region corresponding to a foreign object. On the other hand, if the radius of the minimum circumscribing circle 222 of region 212 is less than the predetermined value, region 212 is not determined to be a region corresponding to a foreign object.

[0034] In this way, when a foreign object is detected from one transmission tomographic image, it is determined whether or not all necessary transmission tomographic images have been acquired (step S105). That is, it is determined whether or not transmission tomographic images of multiple tomographic planes at different positions have been acquired for the resin molded product that is the target for foreign object detection, and if there are any transmission tomographic images that have not been acquired (step S105 No), the processes of steps S101 to S104 described above are repeated.

[0035] On the other hand, if all transmission tomographic images have been acquired (Yes in step S105), a region determined to be a foreign object in any of the transmission tomographic images is selected (step S106). Then, it is determined whether the selected foreign object has also been detected in transmission tomographic images captured on adjacent slices, and whether it has been detected in transmission tomographic images of a predetermined number of consecutive slices (step S107). In other words, it is determined whether a foreign object detected in one transmission tomographic image is detected in transmission tomographic images of multiple consecutive slices.

[0036] If a foreign object is detected from the transmission tomographic images of multiple consecutive slices (Yes in step S107), the foreign object is determined to be large in three-dimensional size, and therefore it is determined that there is a problem with the quality of the resin molded product (step S108).On the other hand, if a foreign object is not detected from the transmission tomographic images of multiple consecutive slices (No in step S107), it is determined that the foreign object is small in three-dimensional size, and therefore it is determined that there is no problem with the quality of the resin molded product (step S109).

[0037] Fig. 8 is a diagram illustrating the determination of a foreign substance. In Fig. 8, transmission tomographic images 210a, 210b, and 210c are each two-dimensional images on the xy plane, but are transmission tomographic images captured at different consecutive slices in the z-axis direction.

[0038] For example, when a foreign object 211a detected in the transmission tomographic image 210a is selected, the detection status of the foreign object in the transmission tomographic images 210b and 210c of consecutive tomographic layers is confirmed. Here, a foreign object 211b is detected in the transmission tomographic image 210b, and a foreign object 211c is detected in the transmission tomographic image 210c. Furthermore, the foreign objects 211a, 211b, and 211c in the transmission tomographic images 210a, 210b, and 210c contain points with the same x- and y-coordinates. That is, the foreign objects 211a, 211b, and 211c are detected in different tomographic layers in the z-axis direction, but have overlapping portions in the x- and y-axis directions. Therefore, the foreign objects 211a, 211b, and 211c are determined to be the same foreign object appearing in the transmission tomographic images of different tomographic layers. If such foreign objects are detected in the transmission tomographic images of a predetermined number of consecutive tomographic layers, it is determined that there is a problem with the quality of the resin molded product.

[0039] On the other hand, for example, when a foreign object 213a detected in the transmission tomographic image 210a is selected, the detection status of the foreign object in the transmission tomographic images 210b and 210c of successive tomographic layers is confirmed. Here, a foreign object having an overlapping portion with the foreign object 213a in the x-axis direction and the y-axis direction is not detected in the transmission tomographic images 210b and 210c. Therefore, the foreign object 213a is not detected in the transmission tomographic images of a predetermined number of successive tomographic layers, and is determined to be a thin foreign object detected only in the transmission tomographic image 210a, and it is determined that there is no problem with the quality of the resin molded product.

[0040] Once it has been confirmed whether or not the foreign matter detected from the transmission tomographic images will affect the quality of the resin molded product in this way, it is determined whether or not confirmation has been completed for all foreign matters detected from all transmission tomographic images (step S110). If there are any unconfirmed foreign matters (step S110 No), the processes of steps S106 to S109 described above are repeated. If confirmation has been completed for all foreign matters (step S110 Yes), the determination of the quality of the resin molded product is completed. In other words, it is determined whether or not the resin molded product contains any foreign matter that will affect the quality, and the quality of the resin molded product is evaluated.

[0041] As described above, according to this embodiment, transmission tomographic images are acquired for multiple sections of a resin molded product, a threshold value for binarization processing is set based on a statistical index of a normal distribution fitted to the pixel value distribution of each transmission tomographic image, foreign matter is detected from the obtained binarized images, and if foreign matter is detected in a predetermined number of consecutive transmission tomographic images, it is determined that there is a problem with the quality of the resin molded product. Therefore, the threshold value for binarization processing is set according to the characteristics of the transmission images, making it possible to generate binarized images that can accurately detect foreign matter, and to determine the quality of the resin molded product by taking into account the three-dimensional size of the foreign matter. In other words, foreign matter can be detected with high accuracy using transmission images obtained by irradiating electromagnetic waves.

[0042] Next, an example of detecting a foreign object in a resin molded product using the foreign object detection method according to the embodiment described above will be described, although the technology of the present disclosure is not limited to this example.

[0043] First, a resin molded product was produced using PEEK resin. The shape of the resin molded product and the molding conditions for injection molding are shown below. Size: 14mm x 12mm x 18mm Resin: PEEK resin manufactured by Polypla-Evonik Mold temperature: 200℃ Resin temperature: 410℃ Injection speed: 6mm / sec Holding pressure: 130MPa, 40sec

[0044] Next, transmission tomographic images of the resin molded product were taken using an X-ray CT device manufactured by Voxel Works. The X-ray irradiation conditions are as follows: The interval between transmission tomographic images can be changed depending on the imaging conditions such as resolution and magnification, but in this case, transmission tomographic images were taken every 14 μm. Voltage: 90kV Tube current: 89mA

[0045] Next, an image analysis program was used to calculate the pixel value distribution of the transmission tomographic image, and each peak obtained was fitted with a normal distribution. Furthermore, the lower probability point of the normal distribution, 0.5%, was set as the threshold, and binarization processing was performed. The number of pixels in the image was 1232 x 1106.

[0046] Next, the black regions in the binarized image were subjected to minimum circumscribing circle approximation, and regions where the diameter of the minimum circumscribing circle was 100 μm or more were determined to be regions corresponding to foreign matter.

[0047] The same processing was performed on all transmission tomographic images, and if there was an area that was judged to be a foreign object in six consecutive transmission tomographic images, it was determined to be a foreign object that could affect the quality of the resin molded product.

[0048] The foreign object detection method according to the present disclosure can be executed by an information processing device. Fig. 9 is a block diagram showing an example of the hardware configuration of an information processing device 10 that executes the foreign object detection method. As shown in Fig. 9, the information processing device 10 includes a processor 11, a memory 12, a network interface (hereinafter abbreviated as "NW interface") 13, and an I / O (Input / Output) interface 14.

[0049] The processor 11 has, for example, a central processing unit (CPU), a field programmable gate array (FPGA), or a digital signal processor (DSP), and controls the entire information processing device 100 and executes various types of arithmetic processing.

[0050] The memory 12 includes, for example, a random access memory (RAM) or a read only memory (ROM), and stores information used in the arithmetic processing executed by the processor 11.

[0051] The NW interface 13 is an interface for connecting to a network via wire or wirelessly.

[0052] The I / O interface 14 is an interface through which the user inputs information and outputs information to the user, and may include, for example, a keyboard, a display, a touch panel, a microphone, or a speaker.

[0053] The information processing device 10 acquires a transmission tomographic image of a foreign object detection target via the NW interface 13 and the I / O interface 14. The processor 11 then executes programs using the memory 12 to perform various processes such as Gaussian fitting, threshold setting, binarization, and foreign object detection.

[0054] The processes executed by the information processing device 10 can also be written as a computer-executable program. In this case, the program can be stored on a computer-readable, non-transitory recording medium and installed on the computer. Examples of such recording media include portable recording media such as CD-ROMs, DVD discs, and USB memory, as well as semiconductor memories such as flash memories. [Explanation of symbols]

[0055] 11 processors 12 Memory 13 Network Interface 14 I / O Interfaces 110 Transmission Tomography 111, 211, 212 areas 120 pixel value distribution 130 Normal distribution 210 Binarized Images 221, 222 Minimum circumscribed circle

Claims

1. a step of irradiating an electromagnetic wave onto a resin molded product and acquiring a transmission tomographic image; fitting a normal distribution to the pixel value distribution of the acquired transmission tomographic image, and setting a threshold value for binarization processing based on a statistical index of the fitted normal distribution; generating a binarized image from the transmission tomographic image using a set threshold value; detecting an area corresponding to a foreign substance contained in the resin molded product from the generated binary image; determining the quality of the resin molded product based on whether or not a region corresponding to a foreign substance is detected from a predetermined number of consecutive adjacent transmission tomographic images; A foreign object detection method comprising:

2. The obtaining step includes: A transmission tomographic image with corrected pixel values ​​is obtained for the outer peripheral portion of the resin molded product where the beam hardening phenomenon occurs. The foreign object detection method according to claim 1 .

3. The setting step includes: The pixel value at which the one-tailed probability of the normal distribution becomes a predetermined probability is set as the threshold value for the binarization process. The foreign object detection method according to claim 1 .

4. The setting step includes: The pixel value at which the two-sided probability of the normal distribution is a predetermined probability is set as the threshold value for binarization processing. The foreign object detection method according to claim 1 .

5. On the computer, a step of irradiating an electromagnetic wave onto a resin molded product and acquiring a transmission tomographic image; fitting a normal distribution to the pixel value distribution of the acquired transmission tomographic image, and setting a threshold value for binarization processing based on a statistical index of the fitted normal distribution; generating a binarized image from the transmission tomographic image using a set threshold value; detecting an area corresponding to a foreign substance contained in the resin molded product from the generated binary image; determining the quality of the resin molded product based on whether or not a region corresponding to a foreign substance is detected from a predetermined number of consecutive adjacent transmission tomographic images; A foreign object detection program that executes the above.

Images