Management System
The management system addresses the challenge of varying X-ray inspection device characteristics by associating images with device or factory information for tailored re-inspection using learning models, ensuring accurate detection across multiple devices.
Patent Information
- Authority / Receiving Office
- JP · JP
- Patent Type
- Patents
- Current Assignee / Owner
- ISHIDA CO LTD
- Filing Date
- 2026-03-19
- Publication Date
- 2026-05-15
AI Technical Summary
Conventional X-ray inspection devices struggle to distinguish between article characteristics and foreign substances with similar features, leading to missed detections, and variations in image quality across devices complicate re-inspection accuracy when multiple devices coexist.
A management system that associates X-ray transmission images with device characteristics or factory information and stores them for re-inspection, using learning models tailored to each device or factory configuration to enhance detection accuracy.
Enables highly accurate re-inspection by automatically selecting optimal learning models based on device or factory-specific characteristics, reducing missed and false detections across multiple X-ray inspection devices.
Smart Images
Figure 0007859728000001_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to a management system.
Background Art
[0002] X-ray inspection devices are widely used to detect foreign substances such as metal, stone, and bone fragments mixed in articles such as food and clothing.
[0003] In conventional X-ray inspection devices, a method of determining the presence or absence of foreign substances by threshold processing after performing image processing such as Laplacian filter or Sobel filter on the captured X-ray transmission image has been common.
[0004] In recent years, devices equipped with a function of saving all images during inspection have also become widespread and are used to ensure traceability after inspection.
Prior Art Documents
Patent Documents
[0005]
Patent Document 1
Summary of the Invention
Problems to be Solved by the Invention
[0006] However, when the characteristics of an article and those of a foreign substance are similar, it has been difficult to distinguish between the two with conventional image processing methods, and there has been a problem that foreign substances are overlooked.
[0007] Also, even when re-inspection is performed using the saved images, the X-ray inspection device has characteristics such as image quality and contrast of the generated images that differ for each individual depending on the hardware configuration such as the type of X-ray tube and sensor mounted. Therefore, it has been difficult to obtain sufficient detection accuracy even when applying a uniform determination criterion or learning model to the saved images in an environment where multiple X-ray inspection devices coexist.
[0008] Therefore, the present invention has been made in view of the above problems, and aims to provide a management system for an X-ray inspection device that can perform appropriate re-inspections. [Means for solving the problem]
[0009] One embodiment of the management system is a management system for managing a plurality of X-ray inspection devices, each comprising: a transport unit for transporting articles in a transport direction; an irradiation unit for irradiating the transported articles with X-rays; a detection unit for detecting the irradiated X-rays; a generation unit for generating an X-ray transmission image based on the X-rays detected by the detection unit; and an inspection unit for detecting defects in the articles based on the X-ray transmission image generated by the generation unit. The system comprises a first storage unit for associating the X-ray transmission image with device characteristics indicating the characteristics of each X-ray inspection device that generated the X-ray transmission image and storing them as a re-inspection image; and a re-inspection unit for a learning model for detecting defects in the articles, which re-inspects the re-inspection image stored in the first storage unit using the learning model learned for each device characteristic.
[0010] One embodiment of the management system is a management system for managing a plurality of X-ray inspection devices, each comprising: a transport unit for transporting articles in a transport direction; an irradiation unit for irradiating the transported articles with X-rays; a detection unit for detecting the irradiated X-rays; a generation unit for generating an X-ray transmission image based on the X-rays detected by the detection unit; and an inspection unit for detecting defects in the articles based on the X-ray transmission image generated by the generation unit. The system comprises a first storage unit for associating the X-ray transmission image with factory information indicating the factory where the X-ray inspection device that generated the X-ray transmission image is installed, and storing it as a re-inspection image; and a re-inspection unit for a learning model for detecting defects in the articles, which re-inspects the re-inspection image stored in the first storage unit using the learning model learned for each piece of factory information. [Effects of the Invention]
[0011] According to the present invention, it is possible to provide a management system for an X-ray inspection device that can perform appropriate re-examinations. [Brief explanation of the drawing]
[0012] [Figure 1] Figure 1 is a diagram showing an example of the overall configuration of the management system 1 according to the first embodiment. [Figure 2] Figure 2 shows an example of a table stored in the first storage unit 110 of the management system 1 according to the first embodiment. [Figure 3] Figure 3 shows an example of a table stored in the second storage unit 130 of the management system 1 according to the first embodiment. [Figure 4] Figure 4 shows an example of a table stored in the first storage unit 110 of the management system 1 according to the second embodiment. [Figure 5] Figure 5 shows an example of a table stored in the second storage unit 130 of the management system 1 according to the second embodiment. [Modes for carrying out the invention]
[0013] Preferred embodiments of the present invention will be described in detail below with reference to the attached drawings. In the following drawings, identical or similar parts are denoted by the same or similar reference numerals. However, it should be noted that the drawings are schematic, and the ratios of dimensions, etc., may differ from those of reality. Therefore, specific dimensions, etc., should be determined by referring to the following description. Furthermore, there may be parts in the drawings where the relationships and ratios of dimensions differ from those of other parts. In this specification and drawings, elements having substantially the same function and configuration are denoted by the same reference numerals to avoid redundant explanations, and elements not directly related to the present invention are omitted from the illustration.
[0014] (First Embodiment) Hereinafter, referring to FIGS. 1 to 5, the management system 1 according to the first embodiment will be described. FIG. 1 is a diagram showing an example of the overall configuration of the management system 1 according to the present embodiment, FIG. 2 is a diagram showing an example of a table stored by the first storage unit 110 of the management system 1 according to the present embodiment, and FIG. 3 is a diagram showing an example of a table stored by the second storage unit 130 of the management system 1 according to the present embodiment.
[0015] As shown in FIG. 1, the management system 1 according to the present embodiment is configured to manage a plurality of X-ray inspection devices 10.
[0016] The connection configuration between the management system 1 and the X-ray inspection device 10 is not particularly limited. For example, the management system 1 and the X-ray inspection device 10 may be connected via a LAN (wired or wireless) within the same factory, or may be configured to centrally manage the X-ray inspection devices 10 of a plurality of factories by means of a cloud service via the Internet.
[0017] Also, the management system 1 may be implemented as an on-premises server within the factory, may be constructed on a cloud server, or may be implemented as a general-purpose computer such as a notebook PC installed with re-inspection software. For example, an X-ray transmission image inspected by the X-ray inspection device 10 may be stored in a cloud storage, and a notebook PC installed with re-inspection software may refer to or download an image on the cloud storage to perform re-inspection, and the re-inspection result may be stored in the cloud storage.
[0018] Also, when the network connection environment is not ready, a portable storage medium such as an HDD may be connected to the X-ray inspection device 10 to store the X-ray transmission image, and the portable storage medium may be physically connected to the management system 1 to transfer data.
[0019] Each of the plurality of X-ray inspection devices 10 includes a conveyance unit 11, an irradiation unit 12, a detection unit 13, a generation unit 14, and an inspection unit 15.
[0020] The conveying unit 11 conveys articles in the conveying direction. For example, the conveying unit 11 is constituted by a conveying conveyor or the like. In the present embodiment, although the case where the article is food will be described, the present invention is not limited to such a case. For example, the article may be, in addition to food, pharmaceuticals, clothing, industrial parts, or the like.
[0021] The irradiation unit 12 irradiates the conveyed article with X-rays. Specifically, the irradiation unit 12 irradiates the article conveyed by the conveying unit 11 with X-rays.
[0022] The detection unit 13 detects the irradiated X-rays. Specifically, the detection unit 13 detects the X-rays that have been irradiated from the irradiation unit 12 and have passed through the article and the conveying unit 11. The detection unit 13 is constituted by a sensor such as a dual energy sensor, for example. The dual energy sensor may be a direct conversion type sensor or an indirect conversion type sensor.
[0023] The generation unit 14 generates an X-ray transmission image based on the X-rays detected by the detection unit 13. The inspection unit 15 detects defects of the article based on the X-ray transmission image generated by the generation unit 14. For example, examples of such defects of the article include foreign matter inclusion, shortage, shape defect, packaging defect, density abnormality, and the like.
[0024] In the present embodiment, it is assumed that the X-ray transmission images generated by the generation unit 14 differ in image quality and characteristics depending on the device characteristics described later.
[0025] Further, the management system 1 includes a first storage unit 110, a reinspection unit 120, a second storage unit 130, and a setting unit 140.
[0026] The first storage unit 110 stores the X-ray transmission image and the device characteristics indicating the characteristics of each X-ray inspection device 10 that generated the X-ray transmission image in association with each other as a reinspection image.
[0027] Here, the device characteristics include the name (identification information) and model of the X-ray inspection device that generated the X-ray transmission image, the type of radiation source or sensor used to generate the X-ray transmission image (e.g., dual-energy sensor, single-energy sensor, line sensor, etc.), sensor pitch, resolution of the X-ray transmission image, image adjustment settings applied to the X-ray transmission image (e.g., contrast enhancement intensity setting, noise reduction filter type and intensity setting, brightness / gamma correction setting value, etc.), image compression settings applied to the X-ray transmission image, inspection conditions applied to the X-ray transmission image, and identification information of the line where the X-ray inspection device that generated the X-ray transmission image is installed.
[0028] For example, the device characteristics may include at least one of the following: the model of the X-ray inspection device, the type of sensor mounted on the X-ray inspection device, the type of radiation source mounted on the X-ray inspection device, the resolution of the X-ray transmission image, the image adjustment settings applied to the X-ray transmission image, and the image compression settings applied to the X-ray transmission image.
[0029] For example, the first storage unit 110 may manage the re-examination images using a table as shown in Figure 2. For example, as shown in Figure 2, such a table may be a table that associates identification information of X-ray transmission images with identification information of device characteristics.
[0030] Furthermore, the first storage unit 110 may preferentially store X-ray transmission images of items that were determined to be good products in the initial inspection by the X-ray inspection device 10.
[0031] The re-inspection unit 120 re-inspects the re-inspection images stored in the first storage unit 110 using a learning model for detecting defects in the items. This learning model is a learning model that has been trained for each device characteristic.
[0032] The learning model may be, for example, an image classification model using a CNN (Convolutional Neural Network), an object detection model (e.g., YOLO, SSD, etc.), anomaly detection model (e.g., an autoencoder, etc.), a transfer learning-based model (e.g., a model utilizing pre-trained models such as ResNet, VGG, etc.), or a conventional machine learning model (e.g., SVN, Random Forest, etc.).
[0033] Furthermore, the learning model may be trained using a dataset obtained by collecting and classifying X-ray transmission images (including both normal and defective products) actually acquired by each X-ray inspection device 10 according to the characteristics of the device.
[0034] Furthermore, the learning model may be trained using X-ray transmission images that have been manually annotated (labeled as normal or defective), or it may be an existing learning model that has been fine-tuned (additional training) with data specific to the characteristics of each device. For example, since the way minute foreign objects are captured differs between an X-ray inspection device equipped with a sensor with a sensor pitch of 0.1 mm and an X-ray inspection device equipped with a sensor with a sensor pitch of 0.4 mm, it is preferable to use different learning models for each.
[0035] For example, the operator can manually set the timing of the re-inspection by the re-inspection unit 120 to any desired timing.
[0036] The second storage unit 130 stores multiple learning models (learning models for detecting defects in goods). For example, the second storage unit 130 may manage the above-mentioned learning models using a table as shown in Figure 3.
[0037] For example, as shown in Figure 3, such a table may be a table that associates identification information of device characteristics with identification information of the learning model.
[0038] Furthermore, such a table may be one that associates identification information of the type of item (product name), identification information of the device characteristics, and identification information of the learning model. For example, a separate learning model may be managed for each combination of item type and device characteristics, such as a learning model corresponding to the combination of product A and device a, a learning model corresponding to the combination of product A and device b, and a learning model corresponding to the combination of product B and device a. Alternatively, one general-purpose model may be maintained for each device characteristic, and differences in item types may be handled by inspection conditions (e.g., thresholds).
[0039] The re-examination unit 120 may re-examine the re-examination image stored in the first storage unit 110 using a learning model from among the multiple learning models stored in the second storage unit 130 that corresponds to the device characteristics of the X-ray inspection device 10 that generated the re-examination image stored in the first storage unit 110.
[0040] The setting unit 140 can set a learning model corresponding to the device characteristics of the X-ray inspection device 10 that generated the re-examination image stored in the first storage unit 110 as the learning model to be used for re-examination. Here, the setting unit 140 may make such a setting in response to user operation, or it may make such a setting based on machine learning or the like. For example, the setting method of the setting unit 140 can be manual selection, automatic matching, or switching of automatic settings using a correspondence table.
[0041] Furthermore, the setting unit 140 may be switchable between a manual setting mode, in which the learning model is set according to the operator's actions, and an automatic setting mode, in which the device characteristics of the X-ray inspection device 10 that generated the re-examination image are automatically read and the corresponding learning model is automatically set.
[0042] In such a case, the re-examination unit 120 may re-examine the re-examination image stored in the first storage unit 110 using a learning model set by the setting unit 140 from among the multiple learning models stored in the second storage unit 130.
[0043] According to the management system 1 of this embodiment, X-ray transmission images and device characteristics are associated and stored, and a learning model optimized for the device characteristics can be automatically selected during re-inspection. This enables highly accurate judgment that absorbs the individual "characteristics of X-ray transmission images" of the X-ray inspection device 10. Even in a site where multiple X-ray inspection devices 10 are present, the optimal learning model according to the hardware configuration of each X-ray inspection device 10 can be automatically applied, thereby suppressing variations in detection accuracy and reducing missed detections and false detections. As a result, it becomes possible to perform appropriate re-inspections.
[0044] A typical use case of the management system 1 according to this embodiment is when a market claim occurs. For example, if a defective product (product A) is returned in the market, the user creates or updates a learning model corresponding to the device characteristics (e.g., device a, device b, device c, etc.) of the X-ray inspection device 10 that inspected product A, and stores it in the second storage unit 130. Subsequently, the re-inspection unit 120 re-inspects the re-inspection image, which includes the X-ray transmission image generated during the inspection of product A, using the learning model corresponding to each device characteristic. This makes it possible to identify items that were judged as good products during the initial inspection but may actually be defective.
[0045] If, as a result of the re-inspection by the re-inspection unit 120, an X-ray transmission image that was judged as good in the initial inspection is judged as defective, the item can be disposed of as it poses a risk to shipment. However, since the re-inspection is performed using stricter standards for items that were judged as good, there is a possibility of false positives. Therefore, it may be possible to implement a procedure in which an operator visually checks items that have been judged as defective by the re-inspection unit 120, and only if the item is also determined to be defective after visual inspection is it disposed of.
[0046] Here, the management system 1 may further include an output unit that outputs the item corresponding to the re-inspection image as subject to disposal if, as a result of the re-inspection by the re-inspection unit 120, the re-inspection image that was judged as a good product in the initial inspection is judged as a defective product.
[0047] (Example of change 1) Hereinafter, with reference to Figures 4 and 5, the management system 1 according to Modified Example 1 of the present invention will be described, focusing on the differences from the management system 1 according to the First Embodiment described above.
[0048] In this modification example 1, the X-ray transmission images generated by the generation unit 14 will have different image quality and characteristics depending on the factory where the X-ray inspection device 10 is installed.
[0049] In this modification example 1, the first storage unit 110 associates the X-ray transmission image with factory information indicating the factory where the X-ray inspection device 10 that generated the X-ray transmission image is installed, and stores it as a re-inspection image.
[0050] For example, the first storage unit 110 may manage the re-examination images using a table as shown in Figure 4. For example, as shown in Figure 4, such a table may be a table that associates identification information of X-ray transmission images with factory information. For example, factory information may include the factory name, factory identification information (factory ID), location, line number, line name, etc.
[0051] The first storage unit 110 may store all X-ray transmission images of all items inspected by the X-ray inspection device 10. Alternatively, to shorten the calculation time, the first storage unit 110 may store only the X-ray transmission images of items that were determined to be good products in the initial inspection.
[0052] Furthermore, the first storage unit 110 may store only the X-ray transmission images that were determined to be defective in the re-inspection by the re-inspection unit 120, and may also have a function to sort and display the X-ray transmission images that were determined to be defective in the re-inspection.
[0053] Furthermore, the storage period for X-ray transmission images stored in the first storage unit 110 may be arbitrarily set by the user, and the first storage unit 110 may also have a function to automatically delete X-ray transmission images after the set storage period has elapsed.
[0054] The re-inspection unit 120 re-inspects the re-inspection images stored in the first storage unit 110 using a pre-learned model for each piece of factory information.
[0055] The second storage unit 130 stores multiple learning models (learning models for detecting defects in goods). For example, the second storage unit 130 may manage the above-mentioned learning models using a table as shown in Figure 5. For example, as shown in Figure 5, such a table may be a table that associates factory information with identification information of the learning models.
[0056] The re-inspection unit 120 may re-inspect the re-inspection image stored in the first storage unit 110 using a learning model from among the multiple learning models stored in the second storage unit 130 that corresponds to the factory information associated with the X-ray inspection device 10 that generated the re-inspection image stored in the first storage unit 110.
[0057] The setting unit 140 can set a learning model corresponding to the factory information associated with the X-ray inspection device 10 that generated the re-inspection image stored in the first storage unit 110 as the learning model to be used for re-inspection. Here, the setting unit 140 may make such a setting in response to user operation, or it may make such a setting based on machine learning or the like.
[0058] According to the management system 1 of this embodiment, X-ray transmission images and factory information are associated and stored, and a learning model optimized for the factory where the X-ray inspection device 10 is installed can be automatically selected during re-inspection. This enables highly accurate judgment by absorbing the individual "characteristics of X-ray transmission images" of the X-ray inspection device 10. Even in a site where multiple X-ray inspection devices 10 are present, the optimal learning model according to the hardware configuration of each X-ray inspection device 10 can be automatically applied, thereby suppressing variations in detection accuracy and reducing missed detections and false detections. As a result, it becomes possible to perform appropriate re-inspections.
[0059] Although the present invention has been described in detail using the embodiments described above, it will be clear to those skilled in the art that the present invention is not limited to the embodiments described herein. The present invention can be implemented in modified and altered forms without departing from the spirit and scope of the invention as defined by the claims. Therefore, the description herein is for illustrative purposes only and is not intended to be restrictive in any way to the present invention. [Explanation of Symbols]
[0060] 1…Management System 110...1st storage section 120... Re-examination Department 130…Second storage unit 140...Settings section 10...X-ray inspection equipment 11…Transportation section 12... Irradiation area 13...Detection unit 14...Generation section 15…Inspection Department
Claims
1. A conveying unit that conveys items in the conveying direction, An irradiation unit that irradiates the transported items with X-rays, A detection unit that detects the irradiated X-rays, A generation unit that generates an X-ray transmission image based on the X-rays detected by the detection unit, A management system for managing a plurality of X-ray inspection devices, each comprising: an inspection unit that detects defects in the article based on the X-ray transmission image generated by the generation unit; A first storage unit stores the aforementioned X-ray transmission image and the device characteristics indicating the characteristics of each X-ray inspection device that generated the aforementioned X-ray transmission image as a re-examination image, A management system comprising: a learning model for detecting defects in the aforementioned article, and a re-inspection unit that re-inspects the re-inspection images stored in a first storage unit using the learning model that has been learned for each of the device characteristics.
2. The system further comprises a second memory unit that stores multiple learning models, The management system according to claim 1, wherein the re-examination unit re-examines the re-examination image stored in the first storage unit using the learning model from among the plurality of learning models that corresponds to the device characteristics of the X-ray inspection apparatus that generated the re-examination image stored in the first storage unit.
3. A second memory unit that stores multiple learning models, The system further comprises a setting unit that sets the learning model corresponding to the device characteristics of the X-ray inspection apparatus that generated the re-examination image stored in the first storage unit as a learning model to be used for re-examination, The management system according to claim 1, wherein the re-examination unit re-examines the re-examination image stored in the first storage unit using the learning model set by the setting unit from among the plurality of learning models.
4. The management system according to claim 1, wherein the device characteristics include at least one of the following: the model of the X-ray inspection apparatus, the type of sensor mounted on the X-ray inspection apparatus, the type of radiation source mounted on the X-ray inspection apparatus, the resolution of the X-ray transmission image, the image adjustment settings applied to the X-ray transmission image, and the image compression settings applied to the X-ray transmission image.
5. The second storage unit stores the identification information of the type of article, the identification information of the device characteristics, and the identification information of the learning model in association with each other. The management system according to claim 2, wherein the re-inspection unit re-inspects the re-inspection image using the learning model among the plurality of learning models that corresponds to a combination of the type of article related to the re-inspection image and the device characteristics of the X-ray inspection device that generated the re-inspection image.
6. The management system according to claim 1, further comprising an output unit that outputs the item to be disposed of if, as a result of the re-inspection by the re-inspection unit, the re-inspection image that was determined to be a good product in the initial inspection is determined to be a defective product.
7. The management system according to claim 1, wherein the first storage unit preferentially stores the X-ray transmission images of articles that were determined to be good products in the initial inspection by the X-ray inspection device.
8. The management system according to claim 3, wherein the setting unit can switch between a manual setting mode in which the learning model is set according to an operation by an operator, and an automatic setting mode in which the device characteristics of the X-ray inspection device that generated the re-examination image are automatically read and the corresponding learning model is automatically set.
9. A conveying unit that conveys items in the conveying direction, An irradiation unit that irradiates the transported items with X-rays, A detection unit that detects the irradiated X-rays, A generation unit that generates an X-ray transmission image based on the X-rays detected by the detection unit, A management system for managing a plurality of X-ray inspection devices, each comprising: an inspection unit that detects defects in the article based on the X-ray transmission image generated by the generation unit; A first storage unit stores the aforementioned X-ray transmission image and factory information indicating the factory where the X-ray inspection device that generated the X-ray transmission image is installed, as a re-inspection image, A management system comprising a learning model for detecting defects in the aforementioned articles, and a re-inspection unit that re-inspects the re-inspection images stored in a first storage unit using the learning model that has been learned for each piece of factory information.