Product Inspection Equipment

Abstract

Provided is an article inspection device that can reduce downtime for checking operation and can perform operation checks in parallel with an operation for inspecting an article. [Solution] This is an item inspection device that includes an image processing unit 33 that performs image processing on image data Dpx of an item P using a predetermined image processing algorithm to output data for judging the quality state of the item P, a judgment unit 34 that judges whether the quality state is good or bad based on that data, and an operation display unit 41 that displays the judgment results.The device further includes an operation confirmation processing unit 54 that, during operating operation, performs operation confirmation processing using a predetermined image processing algorithm in parallel with the operation of the image processing unit 33 to judge whether the operation of the image processing unit 33 and the judgment unit 34 is normal, and an operation confirmation image generation unit 53 that generates, based on the inspection image during operating operation, an image that includes at least an NG characteristic image Cn that indicates that the quality state of the item P is judged to be poor, as an operation confirmation image Dpc to be used by the operation confirmation processing unit 54.

Description

[Technical Field]

[0001] The present invention relates to an article inspection apparatus, and more particularly to an article inspection apparatus that inspects the quality state of an article by applying a predetermined image processing algorithm to an inspection image obtained by photographing an article of a predetermined type. [Background technology]

[0002] Conventionally, some item inspection devices apply an image processing algorithm having a predetermined image processing filter or a combination thereof corresponding to the inspection item to the image data of the inspected item, thereby enabling high-precision inspection of the predetermined quality state of the inspected item.

[0003] In such an article inspection device, it is necessary to select and set an image processing filter, etc., that corresponds to the inspection item of a specific article to be inspected from a plurality of image processing filters, etc., that are created based on the characteristics of the article to be inspected and the characteristics of the foreign matter to be detected and that are stored in advance in memory. Therefore, there are some devices that automate the function of selecting and setting the image processing filters, etc., required for inspection, so that such selection and setting work can be easily performed without being dependent on the experience of the operator.

[0004] One known example of this type of item inspection device includes a detection unit that outputs a signal in response to a transported item, a foreign matter determination unit that determines whether or not a foreign matter is present in the transported item based on the signal from the detection unit, an operation confirmation processing unit that uses a foreign matter sample to determine whether or not the operation of the detection unit is normal, and a display unit that displays the determination results of the foreign matter determination unit and the determination results of the operation confirmation processing unit, where the operation confirmation processing unit determines whether or not the operation of the detection unit is normal based on whether or not the amplitude and phase of the output signal from the detection unit fall within an acceptable range when a foreign matter sample specified for each type is test transported (see, for example, Patent Document 1).

[0005] Another known item inspection device includes an inspection unit that inspects items being transported, first and second passage detection sensors that detect the passage of an item being inspected that is brought into the inspection unit at different positions in the transport direction, and a control unit that can switch the inspection unit between a normal inspection mode and an operation confirmation mode.The control unit automatically switches from the normal inspection mode to the operation confirmation mode when the second passage detection sensor, which is relatively upstream, does not detect the passage of an item being inspected and the first passage detection sensor, which is relatively downstream, detects the passage of an item being inspected, and when the detection interval between items being detected as passing by the first passage detection means in the operation confirmation mode becomes equal to or longer than a predetermined time (see, for example, Patent Document 2). [Prior art documents] [Patent documents]

[0006] [Patent Document 1] Japanese Patent Application Laid-Open No. 2009-31149 [Patent Document 2] Japanese Patent Application Laid-Open No. 2013-113784 Summary of the Invention [Problem to be solved by the invention]

[0007] However, in all of the conventional article inspection devices described above, the inspector manually and physically checks whether the detection operation of the article inspection device and the judgment process based on the detection signal are normal by sending a sample for operation check down an inspection line having an inspection unit and a conveying unit. This results in a problem that the inspector's work time, which is not converted into data, is long and the operation check work is inefficient.

[0008] Furthermore, if a mistake occurs, such as forgetting to return a sample for operational check, downtime occurs in the production line, including the inspection line, resulting in a drop in production efficiency.

[0009] Furthermore, because the operation of the item inspection device was checked not while the device was in operation, but while the device was in an operation check mode or maintenance mode, or other state where inspection was stopped, it was not possible to accurately check the operation for variations or biases in product quality during the operation of the device to inspect the item.

[0010] Furthermore, if operation checks were to be performed not only when switching between product types but also periodically after the switch, there was a problem that the availability rate of the product inspection equipment would decrease not only due to the inspector's operation check work, but also due to increased downtime and the effort required to manage inspection result data.

[0011] The present invention has been made in consideration of the unresolved issues described above, and aims to provide an item inspection device that reduces the downtime required to transition to an inspection stop state for operational checks, and that can perform operational checks in parallel with the operating operation of item inspections. [Means for solving the problem]

[0012] (1) In order to achieve the above-mentioned object, the article inspection device of the present invention is an article inspection device comprising: an image processing unit that performs image processing using a predetermined image processing algorithm on an inspection image obtained by capturing an image of a transported article using an imaging means, and outputs data for judging the quality state of the article; a judgment unit that judges whether the quality state of the article is good or bad based on the judgment data; and a display unit that displays the results judged by the judgment unit, and is further characterized by comprising: an operation confirmation processing unit that, during operating operations in which the image processing unit, the judgment unit, and the display unit are operating, performs operation confirmation processing using the predetermined image processing algorithm in parallel with the operation of the image processing unit, and judges whether the operation of the image processing unit and the judgment unit is normal; and an operation confirmation image generation unit that generates, based on the inspection image during the operating operation, a defective product image that includes at least defective characteristic parts that will result in a judgment that the quality state of the article is bad, as an operation confirmation image to be used by the operation confirmation processing unit.

[0013] With this configuration, the operation check process is performed in parallel with the operation of the image processing unit during operation, and the operation check processing unit determines whether the image processing unit and the judgment unit are operating normally based on the inspection image during operation. At this time, the operation check image generation unit generates a defective product image containing at least a defective characteristic part that will result in a defective product quality status being judged as defective as the operation check image used by the operation check processing unit. Therefore, during operation, the operation check image of the defective product image, which is generated to include a defective characteristic part based on the inspection image of the inspected product that is a non-defective product, is used to check whether the operation check image will result in a defective product status being judged as defective, thereby making it possible to determine whether the operation of the image processing unit and the judgment unit is normal. As a result, an item inspection device is able to perform operation checks, such as whether the detection signal from the detection unit is appropriate, during operation without transitioning to an inspection stop state such as an operation check mode.

[0014] Since the operation confirmation processing unit uses a predetermined image processing algorithm in the same way as the judgment unit, it is possible to perform effective operation confirmation by performing operation confirmation processing using an inspection image of an item judged to be non-defective by the judgment unit as an inspection image of a non-defective item, and then adding an image of the defective characteristic part to that inspection image to create an operation confirmation image of a defective item. However, the operation confirmation processing unit may also use a predetermined image processing algorithm in the same way as the judgment unit, and perform, in parallel during operation, a determination as to whether the inspection image of an item judged to be non-defective by the judgment unit will be judged to be non-defective, and a determination as to whether the operation confirmation image of a defective item will be judged to be non-defective.

[0015] (2) In the item inspection device of the present invention, the operation confirmation image generation unit can be configured to generate the operation confirmation image by synthesizing multiple partial images that are the defective feature parts at multiple locations in the inspection image during the driving operation that have different image features.

[0016] With this configuration, the image for operation confirmation is a composite of multiple partial images that are defective feature areas at multiple locations in the inspection image with different image features, making it less susceptible to variations and biases in the quality state of the inspected item.

[0017] (3) The operation confirmation image generation unit may be configured to generate the defective product image by combining images of multiple defective feature parts of the product that result in a defective quality condition being determined as defective with respect to a good product inspection image in which the quality condition of the product was determined as good by the determination unit during the operating operation.

[0018] With this configuration, during normal operation, an image for operation confirmation of a defective product is generated by combining defective feature parts based on a good product inspection image, and the defective features of the image for operation confirmation are stably ensured, enabling stable operation confirmation.

[0019] (4) The item inspection device of the present invention may further include a control unit having a first control function unit that controls the operation of the image processing unit, the judgment unit, and the display unit to judge the quality condition of each item, and may also be configured to have a second control function unit that controls the operation of the operation confirmation processing unit and the operation confirmation image generation unit while the control unit is performing the first control function to judge whether the operation of the image processing unit and the judgment unit is normal or not.

[0020] With this configuration, the control unit controls the operation of the image processing unit and the judgment unit using the first control function unit to judge the quality of the article, and in parallel, controls the operation confirmation image generation unit and the operation confirmation processing unit using the second control function unit to judge whether the image processing unit and the judgment unit, which are responsible for inspection, are operating normally. Here, the operation confirmation processing can be performed in the background, hidden from the display screen during operation, and in that case, if the operation confirmation result is abnormal, it is possible to output a warning, such as a warning display or a warning sound.

[0021] (5) In the article inspection device of the present invention, the inspection image may be an X-ray transmission image obtained by imaging the article with X-rays.

[0022] With this configuration, similar to the X-ray inspection images based on the imaging data of each item, it is possible to easily create an operation confirmation image of a defective product image by combining an X-ray transmission image based on the imaging data of each item with at least the defective characteristic parts that will result in the quality condition of the item being judged to be poor, as an operation confirmation image to be used in the operation confirmation processing unit during operating operation. [Effects of the Invention]

[0023] According to the present invention, it is possible to provide an item inspection device that can perform operation checks, such as whether the detection signal from the detection unit is appropriate, during operation without transitioning to an inspection stop state such as an operation check mode. [Brief explanation of the drawings]

[0024] [Figure 1] 1 is a schematic configuration diagram of an article inspection device according to an embodiment of the present invention; [Figure 2] 1 is an explanatory diagram illustrating an example of an operation confirmation image in an article inspection device according to an embodiment of the present invention. [Figure 3] 10 is a flowchart showing an outline of the processing procedure of an operation check processing program executed during operation of an article inspection device according to an embodiment of the present invention. [Figure 4] FIG. 10 is a schematic configuration diagram of an article inspection device according to another embodiment of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0025] Hereinafter, embodiments of the present invention will be described with reference to the drawings.

[0026] (One embodiment) 1 to 3 show an article inspection device according to one embodiment of the present invention.

[0027] First, the configuration will be described.

[0028] 1, the article inspection device 1 of this embodiment includes a conveying unit 10, an inspection unit 20, and a control unit 30. The inspection unit 20 irradiates X-rays onto an article P to be inspected (hereinafter simply referred to as article P) conveyed by the conveying unit 10, while detecting image data corresponding to the distribution of the transmitted X-ray dose, and inspects the quality state of the article P based on the detected image data. Note that the quality state referred to here refers to whether the article P meets the quality and physical quantities required for a product, such as the presence or absence of contaminants, the presence or absence of missing parts, the pass / fail of the shape, size, storage state, etc. of the contents, the distribution of density, thickness, volume, or mass, etc.

[0029] The conveying section 10 is a conveyor that winds a loop-shaped conveying belt 11 around multiple conveying rollers 12 and 13, and can sequentially convey items P to the right in Figure 1 using the upper running section 11a of the conveying belt 11, and is supported by a housing not shown.

[0030] The inspection section 20 is an X-ray inspection section that has an X-ray generator 21 (X-ray source) that generates X-rays of a predetermined energy band that pass through the item P transported by the conveying section 10, and an X-ray detector 23 arranged directly below the upper running section 11a of the conveying belt 11.

[0031] The X-ray generator 21 generates X-rays of a wavelength and intensity according to the tube current and tube voltage using a known X-ray tube 22, and is able to irradiate fan-beam-shaped X-rays that spread in a direction perpendicular to the item transport direction of the transport section 10 onto items P within a predetermined inspection zone Zx on the conveyor belt 11 through an X-ray window section of an enclosure not shown in detail.

[0032] Although the details of this X-ray detector 23 are not shown, it is composed of an X-ray line sensor camera in which detection elements consisting of, for example, a scintillator, which is a phosphor, and a photodiode or a charge-coupled device are arranged in an array at a predetermined pitch in the width direction of the conveying path of the conveying section 10, and which detects X-rays at a predetermined resolution, and is placed at a predetermined position in the conveying direction corresponding to the X-ray irradiation position from the X-ray generator 21.

[0033] That is, the X-ray detector 23 detects the X-rays emitted from the X-ray generator 21 and transmitted through the object P for each predetermined transmission area corresponding to the detection element, converts them into an electrical signal according to the amount of transmission of the X-rays, and outputs an X-ray detection signal for generating an X-ray transmission image with the direction of transmission of the X-rays as the observation direction.

[0034] In addition, the X-ray detector 23 performs scanning in the width direction according to the conveying speed of the conveyor belt 11 and sequentially outputs X-ray detection signals, and when the item detection sensor 28 detects that an item P has been placed on the conveyor belt 11 so as to be conveyed toward the specified inspection section Zx, an X-ray detection signal of the detected item P is output after a specified time has elapsed.

[0035] The control unit 30 is an inspection control means that controls the X-ray irradiation intensity and irradiation period in the inspection unit 20, and controls the X-ray detection period of the X-ray line sensor of the X-ray detector 23 and the detection period for each item P according to the conveying speed of the item P, and also serves as a conveying control means that controls the conveying speed and conveying interval of the item P by the conveying belt 11 in the conveying unit 10, but detailed illustration of the conveying control means is omitted.

[0036] This control unit 30 is configured to include, for example, a microcomputer (processor) having a CPU, ROM, RAM, and I / O interface (not shown), a program device that readably stores control programs for performing each function of the multiple functional units described below in a ROM, auxiliary storage device, or other recording medium, or downloads them from another computer via data communication, and a timer circuit, etc., and in accordance with the control program stored in the ROM, etc., the CPU performs predetermined arithmetic processing while exchanging data with the RAM, etc., and also executes the control programs of the multiple functional units.

[0037] Specifically, the control unit 30 has, as the aforementioned multiple functional units, an inspection image acquisition unit 31 that sequentially takes in detection data Lx from the X-ray detector 23 at predetermined intervals and acquires and outputs data Dpx (hereinafter also referred to as image data Dpx) of X-ray images corresponding to the dose distribution of X-rays that have passed through the item P, an inspection processing unit 32 that performs a predetermined inspection based on the image data Dpx, and an image processing algorithm setting unit 35 that variably sets the image processing algorithm used in the inspection processing unit 32 depending on the selected type of item P.

[0038] The imaging data Dpx of each item P output from the inspection image acquisition unit 31 is described here as having a constant radiation quality (energy, wavelength) specified according to the quality of the item P, but it may also be output as a so-called dual-energy or multi-energy X-ray image in which the radiation quality of the X-rays is multiple types of X-rays that are different from each other.

[0039] The inspection processing unit 32 is composed of an image processing unit 33 that takes in the image data Dpx output from the inspection image acquisition unit 31 and performs image analysis processing such as one or more predetermined filter processing with parameters and limits set to enable extraction of image features, and feature measurement to determine the feature quantities of the extracted image features, and a judgment unit 34 that performs judgment processing to determine whether or not the item P is in a predetermined quality state based on the feature quantity data extracted and measured by the image processing unit 33, such as whether or not there is any foreign matter mixed in, whether or not there are any missing parts, and whether or not the shape, size, or storage condition of the contents is acceptable.

[0040] The image processing algorithm setting unit 35 is connected to the inspection processing unit 32 so as to be able to communicate data with it, and is capable of updating and changing at least the image processing algorithm used in the image processing unit 33 (which may include the judgment processing algorithm used in the judgment unit 34). Furthermore, the inspection processing unit 32 is configured to display and output the judgment results of the judgment unit 34 to the operation display unit 41 (display), and when an operation input is made from the operation display unit 41 to request selection or switching of the type, the image processing algorithm setting unit 35 can be requested to download and / or set a specific image processing algorithm Pgm corresponding to the type requested from the operation display unit 41.

[0041] More specifically, the inspection image acquisition unit 31 is an image input unit that, for example, A / D converts each of the X-ray detection signals from the multiple detection elements of the X-ray detector 23, and outputs (hereinafter referred to as line scanning) data on the cumulative transmission amount within that unit time for all detection element areas of the number n of detection elements (n is an integer greater than 1, for example 640) corresponding to the detection element size in the X-ray detector 23 as digital data of density levels representing gradations from 0 to 1023, for example.

[0042] In addition, the inspection image acquisition unit 31 has a data processing program and working memory (not shown) that perform the function of generating image data Dpx of the item P based on the detection data Lx of the line scan image that is sequentially written into the image memory when the line scan by the X-ray detector 23 is repeated a predetermined number of times according to the inspection period of the item P, and outputting the data to the image processing unit 33 and the inspection image storage unit 51.

[0043] The image processing unit 33 of the inspection processing unit 32 has a predetermined image processing algorithm that is combined with image processing filters, etc., set and stored in an updatable and switchable manner in order to perform a predetermined item inspection based on the image data Dpx of the item P captured from the inspection image acquisition unit 31.

[0044] The image processing filter included in the image processing algorithm of the image processing unit 33 is a processing program for extracting image features (e.g., edges, lines, corners, areas, shading, texture) required for a specified item inspection based on the image data Dpx of the item P. When the image processing algorithm includes a filter for foreign object detection, it may have, for example, a feature extraction filter that performs edge detection processing to emphasize the outline of a foreign object in the item P, a differential filter such as a Sobel filter, and perform differential processing based on a predetermined arithmetic expression on an area near a pixel of interest to emphasize the edge of the foreign object. Note that the image processing filter, etc., includes preprocessing such as shading correction and noise removal for the image data from the inspection image acquisition unit 31 in order to improve the accuracy of the image feature detection processing.

[0045] Furthermore, the feature measurement of image features performed by the image processing unit 33 is a process of calculating the features required for the judgment process in the judgment unit 34 by performing calculations of attributes (features that characterize edges, areas, distances, positions, shapes, etc.) related to shading features, color features, shape features, etc., calculations of features that represent the spatial relationships between such features, or calculations of texture features related to spatial frequency distribution and directional components, for images that have been subjected to the necessary pre-processing and image processing on the image data Dpx of the item P imported from the inspection image acquisition unit 31.

[0046] The judgment unit 34 detects characteristic shapes, foreign matter, etc. detected in the item P based on the feature amounts extracted and measured by the image processing unit 33, and compares the feature amounts such as the area, contour length, and concentration sum of the detected object with limits, which are predetermined judgment reference values, to perform judgment processing to determine whether the item P contains a local characteristic shape, etc. corresponding to a foreign matter or a defective part that meets the judgment conditions for defect judgment.

[0047] In this way, the inspection processing unit 32 applies a predetermined image processing algorithm that combines multiple filter processes, etc. to the image data Dpx of a predetermined type of item P and the X-ray inspection image data to which the necessary pre-processing and image processing have been applied, thereby determining and inspecting the predetermined quality state of the item P from the image.

[0048] The image processing algorithm setting unit 35 stores in advance a plurality of image processing algorithms for image judgment that can be used at least in the image processing unit 33 (which may further include a judgment processing algorithm that can be used in the judgment unit 34).The image processing algorithm setting unit 35 has a function of updating the image processing algorithm currently being used in the inspection processing unit 32 to a specific image processing algorithm Pgm that corresponds to the newly set variety when a request is made to switch or update to another variety from the inspection processing unit 32 based on the selection of a variety or a switching operation from the operation display unit 41.

[0049] The control unit 30 also has a first control function that operates as the aforementioned inspection control means and conveying control means, and also has a second control function that confirms whether the inspection processing in the inspection processing unit 32 (i.e., the image processing in the image processing unit 33 and the judgment processing in the judgment unit 34) is operating normally while the item inspection device 1 is in operation, and here it is configured in a form that incorporates an operation confirmation control unit 50 that performs the second control function.

[0050] The operation checking control unit 50 can execute, in real time, for example, a determination process for determining whether the inspection process in the inspection processing unit 32 is normal or not.

[0051] This operation confirmation control unit 50 is configured to include an inspection image memory unit 51 that sequentially stores and retains the image data Dpx of each item P acquired sequentially by the inspection image acquisition unit 31 as inspection images; an NG feature memory unit 52 that pre-stores image data of NG features that characterize the quality state of the item P; an operation confirmation image generation unit 53 that digitally synthesizes (hereinafter simply referred to as synthesis) the inspection image of each item P from the inspection image memory unit 51 with the NG features stored in the NG feature memory unit 52 to generate image data of an operation confirmation image Dpc containing the NG features; and an operation confirmation processing unit 54 that performs inspection processing similar to the image processing of the image processing unit 33 and the judgment processing of the judgment unit 34 on the operation confirmation image Dpc containing the above-mentioned NG features to confirm whether the inspection processing in the inspection processing unit 32 is normal or not.

[0052] Here, the NG features that are synthesized into the inspection image, which is the image data Dpx of each item P, to create an operation confirmation image Dpc containing NG features are image features that indicate that the quality or physical quantity required of the item P as a product is not appropriate, such as defects due to the presence of foreign matter, defects due to missing parts, defects in the shape, size, storage condition, etc. of the contents, defects in the density, thickness, volume or mass distribution, etc.

[0053] In the following description, the detection of an NG characteristic by the operation checking processor 54 is referred to as "NG detection", and the failure of the operation checking processor 54 to detect an NG characteristic is referred to as "NG non-detection".

[0054] When the inspection processing unit 32 is switched to a specific image processing algorithm Pgm corresponding to the set product type by the image processing algorithm setting unit 35, the operation confirmation processing unit 54 simultaneously retrieves the specific image processing algorithm Pgm from the image processing algorithm setting unit 35.

[0055] Then, during the operating operation of the item inspection device 1, in which the inspection processing unit 32 performs inspection processing based on the image data Dpx of each item P and the inspection results are output to the operation display unit 41, the operation confirmation control unit 50 causes the operation confirmation image generation unit 53 to create an operation confirmation image Dpc using the image data Dpx of each item P and image data of a specified NG feature image Cn, in parallel with the operating operation, and causes the operation confirmation processing unit 54 to perform image processing and judgment processing similar to the inspection processing in the inspection processing unit 32 on the operation confirmation image Dpc.

[0056] That is, the operation confirmation image generation unit 53 generates an operation confirmation image Dpc of a defective product image that includes at least defective characteristic parts (NG characteristics) that will result in the quality condition of the item P being judged to be defective, as an operation confirmation image to be used for operation confirmation processing in the operation confirmation processing unit 54, based on the image data Dpx of each item P, which is an inspection image during operation of the item inspection device 1.

[0057] In addition, during the operating operation of the item inspection device 1 in which the image processing unit 33, judgment unit 34 and operation display unit 41 are operating, the operation confirmation processing unit 54 performs image processing and judgment processing similar to the inspection processing in the inspection processing unit 32, using a predetermined image processing algorithm Pgm and image data of an operation confirmation image Dpc in which NG features are combined with the image data Dpx of each item P, in parallel with the operation of the image processing unit 33 and judgment unit 34 of the inspection processing unit 32, so that it can reliably determine whether the processing result will be NG detection when a foreign matter with NG features is mixed in or other abnormality occurs in each inspection image Dpx processed by the inspection processing unit 32.

[0058] In this embodiment, the imaging data Dpx of each item P, which is the inspection image, is an X-ray transmission image obtained by imaging the item P with X-rays. Therefore, as shown in Fig. 2, during operation of the item inspection device 1, the operation check image generation unit 53 synthesizes, for multiple locations with different image features in the imaging data Dpx of each item P, which is the inspection image, multiple partial images containing NG features, namely, NG feature portions C1, C2, C3, C4, C5, and C6 (partial images of defective feature portions), and performs pixel value synthesis conversion using a predetermined arithmetic formula to generate an operation check image Dpc, which is an image of a pseudo-defective product.

[0059] In this case, the operation confirmation image generation unit 53 may generate an operation confirmation image Dpc, which is an image of a pseudo-defective product, by combining multiple NG feature parts C1, C2, C3, C4, C5, and C6 that indicate a defective quality state of the product P only for good product inspection images in which the quality state of the product P was judged to be good by the judgment unit 34 during driving operation.

[0060] However, the operation check image generation unit 53 does not necessarily have to synthesize the NG feature image Cn with each item P in real time to create the operation check image Dpc, which is a simulated defective product image. For products determined to be defective, the image determined to be defective may be output as is as the operation check image Dpc without synthesizing the NG feature image Cn with the image data Dpx, and the operation check processing may be performed by the operation check processing unit 54. In this case, position information of the defective part of the product determined to be defective may be stored and used as the result of the operation check, and the operation check processing is preferably performed after the limit of the judgment unit 34 is changed, etc. Furthermore, if the inspection image DPx determined to be defective is an image of a product with a foreign substance, the operation check processing unit 54 may create foreign substance position information for the image of the product with a foreign substance using the data output from the judgment unit 34.

[0061] Furthermore, the operation check image generation unit 53 does not necessarily have to output an operation check image Dpc of a defective product image for all items P, but may sequentially generate operation check images Dpc that are simulated defective product images by combining NG feature images Cn with the predetermined number of imaging data Dpx of a predetermined number of consecutive items P that have been determined to be non-defective, and then output the next imaging data Dpx that has been determined to be non-defective as an operation check image Dpc without combining the NG feature image Cn, and have the operation check processing performed by the operation check processing unit 54. Of course, it is possible to associate each item P with its imaging data Dpx based on the detection time of the entry of each succeeding item P and the imaging time of the imaging data Dpx, or based on other identification information.

[0062] In this embodiment, the product that is the object of the item P is not limited to a product of uniform thickness or composition, but is also a product in which the brightness of the imaging data Dpx, which is an X-ray transmission image, varies from place to place due to unevenness in shape or variations in composition. Specifically, the item P is, for example, a food product containing a plurality of cooked ingredients as its contents, or a predetermined net amount (weight) of a plurality of vegetables or fruits of the same type as its contents, contained in a container or bag of a predetermined shape.

[0063] The multiple locations with different image features in the image data Dpx of the item P are, for example, between multiple contents, near the edges of the contents, areas where the brightness of the inspection image corresponding to the amount of X-ray penetration of the contents is relatively lower than the average brightness of the surroundings, and areas where the brightness of the inspection image corresponding to the amount of X-ray penetration of the contents is relatively higher than the average brightness of the surroundings. Conventionally, these multiple locations with different image features correspond to locations that are preferable for imaging as positions where a sample (test piece) of a defective characteristic part to be detected is placed when an operator attaches the sample to the product of the item P to check its operation.

[0064] In the image data Dpx obtained by X-ray imaging of the item P of this embodiment, which has variations in brightness as described above, for example, when a foreign object is mixed in, the brightness of the same foreign object changes depending on the location of the foreign object, and the detection sensitivity also changes due to the influence of the brightness of the surrounding products. In such cases, if an X-ray image is taken by manually attaching a foreign object or a test piece equivalent to the foreign object to the product that will be the item P, it may be difficult to determine the location of the contents from the appearance of the product, such as in the case of a packaged product or a product with a cavity inside, and it is often difficult to evenly change the location of the foreign object and then take images to check operation.

[0065] Therefore, as illustrated in Figure 2, in the operation confirmation image Dpc of this embodiment, the imaging data Dpx obtained by X-ray imaging of each item P is combined into an NG feature image Cn, which includes multiple NG feature parts C1 and C2 located between or near the edges of multiple contents of the item P, multiple NG feature parts C3 and C5 located in areas where the brightness of the inspection image corresponding to the amount of X-ray penetration of the contents is relatively low compared to the surrounding area, and multiple NG feature parts C4 and C6 located in areas where the brightness of the inspection image corresponding to the amount of X-ray penetration of the contents is relatively high compared to the surrounding area.

[0066] In other words, in this embodiment, when NG feature images Cn are synthesized from image data Dpx obtained by X-ray imaging of each item P, multiple NG feature parts C1 to C6 are evenly arranged in multiple locations with different image features, such as high-brightness and low-brightness parts of the product that will become item P, and between adjacent contents within the product or near the edges of the contents.This makes it possible to quickly, accurately, and in real time create operation confirmation images Dpc that are more effective for image judgment, without having to perform the conventional work of attaching a foreign matter sample or an alternative test piece to the product of item P to collect an image.

[0067] This operation confirmation image Dpc can be created relatively easily by the operation confirmation image generation unit 53, for example, as an image for AI learning or sensitivity verification, by setting a rule for the operation confirmation image generation unit 53 to synthesize and arrange multiple NG feature parts C1 to C6 evenly at multiple locations with different image features in the image data Dpx obtained by X-ray imaging of the product that will become the item P.

[0068] In this case, the synthesis positions of the multiple NG feature parts C1 to C6 for the product P are parts of multiple small image areas of a predetermined size (number of pixels) with different image features, such as high-brightness parts of the product, low-brightness parts and product edges, and between adjacent contents in the product, and are extracted as multiple product image areas having image features suitable for image synthesis of NG features using a search size with a preset number of pixels. On the other hand, the synthesis positions of the multiple NG feature parts C1 to C6 themselves can be, for example, the center of gravity of each NG feature image.

[0069] The number and composition positions of the NG feature parts C1 to C6 can be changed depending on the number and types of small image areas with different image features, but it is preferable to set at least one image composition position for multiple small image areas with common image features and at least one image composition position for multiple small image areas with different image features.

[0070] Specifically, the operation confirmation control unit 50 is configured to substantially match or effectively approximate the conditions for X-ray imaging and X-ray detection (hereinafter referred to as X-ray detection conditions) for the digital X-ray image data Dpx obtained by X-ray imaging of a product that is the currently set variety of item P, and to synthesize an NG feature image Cn equivalent to a foreign object that has been pre-stored in the NG feature memory unit 52 with the image data Dpx.

[0071] That is, the inspection image storage unit 51 stores the imaging data Dpx of the product that is the currently set variety of item P in association with its X-ray detection conditions, for example, the stepped set values ​​of the tube voltage and tube current of the X-ray tube 22 that correspond to the X-ray output (tube voltage × tube current) of the X-ray generator 21. Meanwhile, the NG feature storage unit 52 generates NG feature images Cn corresponding to a plurality of types of foreign matter that have different X-ray detection conditions for the same foreign matter sample, for example, different stepped set values ​​of the tube voltage and tube current of the X-ray tube of the X-ray generator 21, and stores the images in association with the respective X-ray detection conditions in the NG feature storage unit 52, so that the NG feature images Cn can be matched to or effectively approximated to the X-ray detection conditions of the imaging data Dpx of the item P when selected.

[0072] In addition, taking into consideration the difference between the X-ray detection conditions of the imaging data Dpx of the article P and the X-ray detection conditions of the NG feature image Cn of the test piece or the like stored in the NG feature memory unit 52, the NG feature image Cn may be density converted to be compatible with many X-ray detection conditions so that the image density of the reference NG feature image Cn stored in the NG feature memory unit 52 is substantially the same as when it is imaged under the same X-ray detection conditions as the imaging data Dpx of the article P. In either case, the difference in product type is directly reflected in the imaging data Dpx of the article P, while the NG feature image Cn is not affected by the product type.

[0073] Furthermore, NG feature images Cn equivalent to foreign substances can be stored in the NG feature storage unit 52 in advance (or even during production) as foreign substance images of test pieces, for example, prior to starting operation of the article inspection apparatus 1, or a user of the article inspection apparatus 1 can image and X-ray detect a known foreign substance sample using the inspection unit 20 to create a NG feature image Cn corresponding to that foreign substance sample. Of course, the NG feature storage unit 52 can import NG feature images Cn as image data from a removable recording medium, or obtain known NG feature images Cn via data communication from other article inspection apparatuses, an in-house management computer, or the like. It goes without saying that the NG feature images Cn are not limited to images equivalent to foreign substances.

[0074] When identifying the positions of multiple locations with different image features (composite positions of NG feature images Cn) in the image data Dpx obtained by X-ray imaging of the product that will become item P, the features of each part of the image area of ​​the product that will become item P, such as high and low brightness areas, areas near the edges of the product, and areas between the contents of the product, can be extracted as features by extracting the collection of brightness differences in the image, or by extracting features by focusing on the distribution of brightness gradient directions within the image.These features can also be extracted using a mechanical method in which a computer automatically determines the process of converting an image into features.

[0075] Here, the X-ray transmission image of the article P is an image corresponding to the distribution of X-ray absorption amounts in the article P, and can be obtained from the difference between the logarithmically transformed values ​​of the transmitted X-ray amount detected when the article P is not on the conveyor belt 11 and the transmitted X-ray amount detected when the article P is present on the conveyor belt 11. Similarly, the NG characteristic image Cn of a foreign matter sample or the like can be obtained from the difference between the logarithmically transformed values ​​of the transmitted X-ray amount detected when the article P or foreign matter sample is not on the conveyor belt 11 and the transmitted X-ray amount detected when the foreign matter sample is present on the conveyor belt 11. The procedure for obtaining such an image can be substantially similar to the procedure described in paragraphs 0036 to 0046 of JP 2009-168740 A, for example.

[0076] For example, since the wavelength of X-rays depends on the tube voltage (voltage applied to the X-ray tube) of the X-ray generator 21 and the X-ray exposure dose depends on the tube current of the X-ray generator 21, the shorter the X-ray wavelength (the higher the tube voltage), the smaller the X-ray absorption amount. Furthermore, an X-ray transmission image or X-ray absorption image of item P shows the two-dimensional distribution of the X-ray dose irradiated on the object or the two-dimensional distribution of the X-ray absorption amount by the object, and if the X-ray exposure dose Io (X-ray intensity), the X-ray penetration dose I (intensity after passing through the object), the X-ray absorption rate μ, and the thickness of the object through which the X-rays penetrated is d, the X-ray absorption amount T can be calculated from the Lambert-Beer attenuation law (I / Io = e to the power of -μd), where log is the natural logarithm, by using the following equation (1):

[0077] T=(log Io-log I)=μd ... Equation (1) Furthermore, the absorption coefficient μ of X-rays has the relationship of the following formula (2), where λ is the wavelength of X-rays, ρ is the density of an object, Z is the atomic number, and C is a constant.

[0078] μ=λ 3 ρZC ... Equation (2) Equation (1) shows that the X-ray absorption amount is the difference between the logarithmically transformed X-ray exposure amount Io and the X-ray penetration amount I, and the X-ray exposure amount Io is the X-ray penetration amount when the X-ray absorption amount is zero. In other words, the X-ray penetration amount detected when there is no inspection object on the conveyor belt is the X-ray exposure amount Io. Therefore, as can be seen from equations (1) and (2), changing the wavelength of the X-rays causes the degree of variation in the X-ray absorption amount to differ for items (item P and foreign objects) with different densities and atomic numbers. Utilizing this fact, the item inspection device 1 can set the X-ray output conditions (X-ray tube voltage and tube current) of the X-ray generator 21 so as to differentiate between item P and foreign objects.

[0079] Then, if the foreign substance sample image (X-ray absorption image of the foreign substance) detected under predetermined X-ray detection conditions (X-ray tube voltage and tube current) stored in the NG feature storage unit 52 is denoted as S1, and the X-ray absorption image of the item P acquired under the X-ray detection conditions (X-ray tube voltage and tube current) of the X-ray generator 21 set corresponding to the item P is denoted as S2, the composite image S is expressed by the following formula (3): S = αS1 + S2 (where α is a conversion coefficient)... Equation (3) This becomes:

[0080] The conversion coefficient α in the above equation (3) can be calculated in advance by calculating the ratio of the shading of the foreign substance sample image Cn of the foreign substance sample detected under predetermined X-ray detection conditions (tube voltage and tube current of the X-ray tube 22) stored in the reference NG feature memory unit 52 to the shading of the foreign substance sample image acquired for each X-ray detection condition (X-ray tube voltage and tube current), and storing the ratio in a conversion table for each X-ray detection condition (X-ray tube voltage and tube current).

[0081] The NG feature storage unit 52 stores X-ray images of a plurality of foreign substance samples of different sizes detected under preset X-ray detection conditions as a plurality of types of NG feature images Cn, for example, foreign substance images C1 to C6.

[0082] The NG feature storage unit 52 also has storage means for storing NG feature images Cn of foreign matter samples for each type of foreign matter, i.e., for each shape (including size) and material (e.g., metal, glass, etc.), when irradiated with X-rays under the above-mentioned X-ray detection conditions. These NG feature images Cn include X-ray absorption images obtained by converting the amount of X-ray transmission into the amount of X-ray absorption when a single foreign matter is irradiated with X-rays under predetermined X-ray detection conditions, and X-ray absorption images obtained by converting the amount of X-ray transmission into the amount of X-ray absorption when foreign matters of different sizes and shapes (e.g., angular, spherical, linear, etc.) but the same material (e.g., metal, glass, resin, bone, etc.) arranged at a predetermined interval are irradiated with X-rays under predetermined X-ray detection conditions.

[0083] The control unit 30 thus has an inspection processing unit 32 as a first control function unit that controls the operation of the image processing unit 33, the judgment unit 34, and the operation display unit 41 to judge the quality condition of each item P, and in addition, has an operation confirmation control unit 50 as a second control function unit that controls the operation of the operation confirmation image generation unit 53 and the operation confirmation processing unit 54 to judge whether the operation of the image processing unit 33 and the judgment unit 34 is normal or not while the inspection processing unit 32 is performing the first control function.

[0084] The label of each image processing algorithm stored in the image processing algorithm setting unit 35 includes at least the number (identification number) of the image processing algorithm, but may also include the name of the algorithm, the name of the variety to be inspected indicating the content of the inspection using that algorithm, and the inspection items.

[0085] In addition, the image processing algorithm setting unit 35 can have an algorithm storage unit in which a plurality of image processing algorithms are stored in advance, and an algorithm setting unit capable of executing an update setting process to update the image processing algorithm set and stored in the image processing unit 33 to one of the image processing algorithms Pgm stored and stored in the algorithm storage unit.

[0086] Next, the operation will be described.

[0087] In the article inspection device of this embodiment configured as described above, prior to starting operation of the article inspection device 1 or at an appropriate time during a period when article inspection is stopped, an NG feature image Cn is imported from a recording medium into the NG feature storage unit 52, or a known NG feature image Cn is acquired via data communication from another article inspection device or an in-house management computer, etc., so that an image of an existing test piece is stored in advance as an NG feature image Cn in the NG feature storage unit 52. Needless to say, it is possible to additionally store and save an NG feature image Cn in this NG feature storage unit 52.

[0088] Next, when the operating operation of the item inspection device 1 that performs item inspection of the item P is started, item inspection of the items P that are sequentially inserted is performed with the inspection screen displayed on the operation display unit 41, while the operation confirmation process as shown in Figure 3 is usually performed as a background process in parallel with the item inspection process, without any screen display.

[0089] As shown in FIG. 3, first, the image processing algorithm used in the operation of the article inspection device 1 is identified in the form of an algorithm number or the like corresponding to the currently set product type (step S11).

[0090] Next, after determining the X-ray detection conditions (X-ray output (tube voltage x tube current) of the X-ray generator 21) at the time of capturing the image data Dpx of the item P input into the inspection image memory unit 51, in order to synthesize one of the multiple NG feature images Cn pre-stored in the NG feature memory unit 52 for the image data Dpx of the item P from the group of foreign body images for multiple types of foreign body samples stored in the NG feature memory unit 52, one NG feature image Cn whose X-ray detection conditions substantially match or effectively approximate the X-ray detection conditions at the time of capturing the image data Dpx of the item P is extracted from the multiple types of NG feature images Cn whose X-ray detection conditions differ in multiple stages, and is acquired as the NG feature image to be used for synthesis (step S12).

[0091] Next, a detection signal is detected from the item detection sensor 28, and image data Dpx of the item P is output from the inspection image acquisition unit 31 (or further, when the inspection of a predetermined number of items has been completed or a predetermined time has passed, etc.), and it is determined whether the conditions for confirming whether the inspection processing in the inspection processing unit 32 is normal are met (step S13).

[0092] At this time, if the conditions for operation confirmation are met (YES in step S13), the operation confirmation image generation unit 53 synthesizes an NG feature image Cn, which has the same X-ray detection conditions as when the image data Dpx of the item P imported into the inspection image memory unit 51, as multiple foreign object images C1 to C6 at multiple locations in the image of the item P in Figure 2 where the image features are different from each other, to generate an operation confirmation image Dpc (step S14).

[0093] Next, during the operating operation of the item inspection device 1 in which the image processing unit 33, judgment unit 34 and operation display unit 41 are operating, the operation confirmation processing unit 54 performs image processing similar to that of the image processing unit 33 of the inspection processing unit 32, using a predetermined image processing algorithm Pgm and image data of an operation confirmation image Dpc in which an NG feature image Cn is combined with the imaging data Dpx of each item P, in parallel with the operation of the image processing unit 33 and judgment unit 34 of the inspection processing unit 32 (step S15).

[0094] Next, based on the image processing result, an operation check determination process is executed to determine whether or not a processing result of NG detection can be reliably obtained by a determination process similar to that of the determination unit 34 (step S16).

[0095] If the processing result is determined to be "NG detected" (if "○" in step S16), then it is checked whether the operation of the item inspection device 1 is stopped (step S17), and the processing from step S13 onwards is repeatedly executed until the operation of the item inspection device 1 is stopped.

[0096] On the other hand, if the processing result is determined to be "NG not detected" ("X" in step S16), then, while the inspection image of item P is displayed on the operation display unit 41, a warning is issued to warn that there is a possibility of some kind of malfunction occurring in the operation during the inspection of the item, or if "NG not detected" recurs within a certain inspection period (predetermined number of inspections), a warning is output to warn that manual operation confirmation by an operator or checking of the judgment criteria is required (step S18), and then the operation stop confirmation in step S17 and / or the processing from step S13 onwards is repeatedly executed.

[0097] In this embodiment, it is determined whether the inspection processing in the inspection processing unit 32 is operating normally at each predetermined operation confirmation period, and this determination processing is performed sequentially in real time or for each predetermined number of inspections.

[0098] Next, the operation will be described.

[0099] In this embodiment, during the operating operation of the item inspection device 1, the operation confirmation process is executed in the operation confirmation control unit 50 in parallel with the operation of the image processing unit 33 of the inspection processing unit 32, and based on the image data Dpx, which is the inspection image during the operating operation, the operation confirmation processing unit 54 of the operation confirmation control unit 50 determines whether the operation of the image processing unit 33 and the judgment unit 34 is normal or not.

[0100] At this time, the operation check image Dpc used by the operation check processing unit 54 is generated by the operation check image generation unit 53 as a defective product image in which the defective feature portions C1 to C6, which are the NG feature image Cn, are combined with multiple locations with different image features in the image data Dpx of the product P. Therefore, by using the operation check image Dpc of the defective product image generated to include the defective feature portions C1 to C6 based on the image data Dpx (inspection image) of the product P, which is primarily a non-defective product during operation, and checking whether the operation check image Dpc is judged to be defective, it becomes possible to determine whether the image processing unit 33 and the judgment unit 34 of the inspection processing unit 32 are operating normally. As a result, it becomes possible to accurately perform operation checks, such as whether the detection signal from the inspection unit 20 is appropriate, sequentially in real time or simultaneously for each predetermined number of inspections during operation of the product inspection device 1, without transitioning the product inspection device 1 to an inspection stop state such as an operation check mode.

[0101] Furthermore, in this embodiment, the operation check image generation unit 53 generates an operation check image Dpc by combining NG feature parts C1 to C6 (plurality of partial images that become defective feature parts) with multiple locations with different image features in the inspection image based on the image capture data Dpx captured during operation of the item inspection device 1. Therefore, the operation check image Dpc is less susceptible to the influence of variations in the quality state of the item P and bias in the amount of transmitted X-rays.

[0102] Furthermore, in this embodiment, the operation check image generation unit 53 generates an operation check image Dpc, which is a defective product image, by combining images of multiple NG characteristic parts C1 to C6 that indicate a defective quality state of the product P with the imaging data Dpx (good product inspection image) for which the quality state of the product P was determined to be good by the determination unit 34 during driving operation. Therefore, during normal driving operation, an operation check image Dpc is generated for a defective product image by combining defective characteristic parts with the good product inspection image, and the defective characteristics of the operation check image are stably ensured, enabling stable operation check. Furthermore, when a defective product is detected, it is also possible to perform operation check using the inspection image of the defective product.

[0103] Additionally, in this embodiment, the control unit 30 includes a first control function unit that controls the operation of the image processing unit 33, the determination unit 34, and the operation display unit 41 to determine the quality state of each item P, and the control unit 30 includes an operation confirmation control unit 50 as a second control function unit that controls the operation of the operation confirmation processing unit 54 and the operation confirmation image generation unit 53 to determine whether the image processing unit 33 and the determination unit 34 are operating normally while performing the first control function. Therefore, during an operating operation in which the control unit 30 controls the operation of the image processing unit 33, the determination unit 34, and the operation display unit 41 using the inspection processing unit 32, which is the first control function unit, to determine the quality state of the item P, the control unit 30 controls the operation confirmation image generation unit 53 and the operation confirmation processing unit 54 using the second control function unit, which determines whether the image processing unit 33 and the determination unit 34, which are responsible for inspection, are operating normally, and this can be performed in parallel, here in the background hidden from the display screen during the operating operation. If the operation check result indicates an abnormality, a warning display, a warning sound, or other notification output can be performed.

[0104] Furthermore, in this embodiment, the inspection image based on the imaging data Dpx is an X-ray transmission image obtained by imaging the item P with X-rays, so the operation confirmation image generation unit 53 can easily create an operation confirmation image Dpc of the defective product image by image-compositing multiple NG feature parts C1 to C6 as an NG feature image Cn for each inspection image, which is the imaging data Dpx of each item P during operation.

[0105] In this way, in this embodiment, an item inspection device 1 can be provided that can perform operation checks, such as whether the detection signal from the detection unit is appropriate, during operating operation without transitioning to an inspection stop state such as an operation check mode.

[0106] As described above, the operation confirmation processing unit 54 in this embodiment may use a predetermined image processing algorithm, similar to the determination unit 34, and may perform, in parallel, an operation confirmation determination as to whether or not an inspection image of an article P determined to be a pass-quality item by the determination unit 34 will be determined to be a pass-quality item, and an operation confirmation determination as to whether or not an operation confirmation image Dpc of a defective article image will be determined to be a fail-quality item, during operation of the article inspection device 1. Furthermore, the operation confirmation processing unit 54 may generate operation confirmation images Dpc of defective article images for a predetermined number of inspection images of a plurality of articles P, and perform operation confirmation processing for the remaining inspection images to determine whether or not NG is not detected using operation confirmation images Dpc corresponding to a pass-quality item, without generating operation confirmation images for defective article images.

[0107] 1 to 3, the operation check control unit 50 is built into the control unit 30, but the operation check control unit 50 may be provided as a control device or management device independent of the control unit 30. Such other embodiments will be described next.

[0108] (Other embodiments) FIG. 4 shows an article inspection device according to another embodiment of the present invention.

[0109] In FIG. 4, the same components as those in the embodiment shown in FIG. 1 are designated by the same reference numerals, and redundant detailed explanations will be omitted.

[0110] As shown in Figure 4, the item inspection device 2 of this embodiment includes a conveying unit 10 and an inspection unit 20 configured in the same manner as in the first embodiment, and an inspection control unit 130 and an operation confirmation control unit 150 configured independently so as to be able to communicate data with each other so as to perform functions similar to those of the control unit 30 of the first embodiment.

[0111] The inspection control unit 130 has a hardware configuration similar to that of the control unit 30 of the embodiment, and includes a plurality of functional units, including an inspection image acquisition unit 31 that sequentially receives detection data Lx of line scan images from the X-ray detector 23 at predetermined intervals, acquires and outputs imaging data Dpx corresponding to the dose distribution of X-rays transmitted through the item P, performs a predetermined inspection based on the imaging data Dpx, and outputs imaging data for each inspection result, in which the inspection results of each item P are added as attribute information to the imaging data Dpx, which is an inspection image of each item P, as imaging data Dpx1 of a non-defective item or imaging data Dpx2 of a defective item, to an operation display unit 41 and an operation confirmation control unit 1 50, an image processing algorithm setting unit 35 that variably sets the image processing algorithm used in the inspection processing unit 132 depending on the selected variety of the article P, an inspection image memory unit 51 that stores images of non-defective articles with a hardware configuration similar to that of the control unit of one embodiment, and an operation confirmation processing unit 54 that also has a hardware configuration similar to that of the control unit of one embodiment and that, when the inspection processing unit 32 is switched to a specific image processing algorithm Pgm depending on the set variety by the image processing algorithm setting unit 35, retrieves the specific image processing algorithm Pgm from the image processing algorithm setting unit 35 in parallel.

[0112] The operation confirmation control unit 150 has a non-defective product image database 151a that stores and stores image data Dpx1 of non-defective products among the image data classified by inspection result from the inspection processing unit 132, a non-compliant image database 151b that stores and stores image data Dpx2 of defective products among the image data classified by inspection result, and a foreign substance image database 151c that stores and stores foreign substance images Dsc1, Dsc2, and Dsci, which are non-compliant characteristic images, by downloading them from a recording medium or via a network. Note that in Fig. 4, the database is referred to as DB.

[0113] The operation confirmation control unit 150 further includes an NG feature learning unit 152 that learns NG features, such as foreign substance image features, that are suitable for importing image data from the good product image database 151a, the NG image database 151b, and the foreign substance image database 151c and combining them into a good product image, and creates an NG feature image generation model NM; and an operation confirmation image generation unit 153 that uses the good product image Dpx1 from the inspection image memory unit 51 of the inspection control unit 130 and the NG feature image generation model NM from the NG feature learning unit 152 to create and output an operation confirmation image Dpc similar to that created by the operation confirmation image generation unit 53 of one embodiment.

[0114] The NG feature image generation model NM generated by the NG feature learning unit 152 may be, for example, a model in which NG features are learned using two neural networks (generative and discriminative) that make up a generative adversarial network (GAN), or a model in which NG features are learned using a distributed model in which two nodes on the network (each equipped with a GPU) divide and process the learning task. Using the NG feature image generation model NM provides the operation check image generation unit 153 with the function of generating a new NG feature image Cn that does not exist in the foreign object image DB or NG image DB and is to be combined with the good product image Dpx1 from the inspection image storage unit 51. Note that even if performance is maximized during the learning phase, if the AI ​​that combines defects into good product images cannot output realistic images, operation check may be performed using actual captured images of defective products.

[0115] In this embodiment, too, operation confirmation processing is performed in the operation confirmation control unit 150 in parallel with the operation of the image processing unit 33 of the inspection processing unit 132 during operating operation of the item inspection device 2, and based on the image data Dpx, which is the inspection image during operating operation, the operation confirmation control unit 150 and the operation confirmation processing unit 54 of the inspection control unit 130 determine whether the operation of the image processing unit 33 of the inspection processing unit 132 and the judgment unit 34 is normal or not.

[0116] In this case, the operation check image Dpc is generated by the operation check image generation unit 153 as a defective product image by combining the NG feature image Cn, e.g., the defective feature portions C1-C6 shown in FIG. 2, with multiple locations with different image features in the image data Dpx of the product P. Therefore, by using the operation check image Dpc of the NG image generated based on the image data Dpx (inspection image) of a non-defective product P to confirm whether the operation check image Dpc is judged to be defective, it is possible to accurately determine whether the image processing unit 33 and the judgment unit 34 of the inspection processing unit 132 are operating normally. As a result, as in the first embodiment, operation checks, such as whether the detection signal from the inspection unit 20 is appropriate, can be accurately performed sequentially in real time or simultaneously for each predetermined number of inspections during operation of the product inspection device 1 without transitioning the product inspection device 1 to an inspection stop state such as an operation check mode. Furthermore, the operation check image Dpc is less susceptible to the effects of variations in the quality state of the product P and biases in the amount of X-ray transmission.

[0117] In the above-described embodiments, an operation check image is created by combining the latest good-product image with a non-compliant feature image to check whether the product inspection device 1's non-compliant detection function is functioning properly. However, when the inspection result is a good product, the good-product images may be stored, and at a predetermined timing, an operation check image may be created from any stored good-product image or the latest good-product image to perform the operation check. Furthermore, the defective portion of the defective product image may be added to the non-compliant feature memory and used for non-compliant compositing. In this case, if the defective product image is an image of a product with a foreign object, it is preferable to provide a foreign object position information creation unit and add the foreign object image and its position information to the non-compliant feature memory. Furthermore, although there is a concern about processing redundancy, it is also possible to combine the non-compliant feature image with the non-compliant feature image, perform the operation check process, and determine whether the product is defective by comparing the inspection results with the operation check results.

[0118] As described above, the article inspection device of the present invention can reduce downtime caused by transitioning to an inspection stop state for operational checks, and can provide an article inspection device that can perform operational checks in parallel with the operation of the article inspection. The present invention is useful for article inspection devices in general that inspect the quality state of an article by applying a predetermined image processing algorithm to an inspection image obtained by capturing an image of an article to be inspected. [Explanation of symbols]

[0119] 1, 2 Item inspection equipment 10 Conveying section 11 Conveyor belt 11a Upper section 12, 13 Conveyor roller 20 Inspection Department 21 X-ray generator 22 X-ray tube 23 X-ray detector 28 Item detection sensor 30 Control Unit 31 Inspection image acquisition unit 32 Inspection processing section (inspection control section) 33 Image processing section 34 Judgment section 35 Image processing algorithm setting section 41 Operation display section (display section) 50 Operation confirmation control unit (second control function unit) 51 Inspection image storage unit (good product image storage unit) 52 NG feature memory section (feature memory section for foreign objects and other defective parts) 53 Operation check image generation unit 54 Operation confirmation processing section 130 Inspection control unit 132 Inspection processing section 150 Operation check control section 151a Good Product Image Database (Good Product Image DB) 151b NG Image Database (NG Image DB) 151c Foreign body image database (foreign body image DB) 152 NG feature learning unit 153 Operation check image generation unit C1, C2, C3, C4, C5, C6 NG feature (multiple partial images that are defective feature) Cn NG characteristic image (foreign matter image) Dsc1, Dsc2, Dsci Foreign body images (NG feature images) Dpc operation confirmation image Dpx Imaging data (inspection images, X-ray images, X-ray image data) Dpx1 Imaging data (imaging data of good products, inspection images, good product images) Dpx2 imaging data (imaging data of defective products, inspection images, defective product images) Lx detection data NM NG feature image generation model P Article (article to be inspected) Pgm specific image processing algorithm

Claims

1. An article inspection device comprising: an image processing unit (33) that performs image processing using a predetermined image processing algorithm on an inspection image (Dpx) obtained by capturing an image of an article (P) being conveyed by an imaging means (20) and outputs data for determining the quality state of the article; a determination unit (34) that determines whether the quality state of the article is good or bad based on the determination data; and a display unit (41) that displays the results determined by the determination unit, an operation confirmation processing unit (54) that executes an operation confirmation process using the predetermined image processing algorithm in parallel with the operation of the image processing unit during a driving operation in which the image processing unit, the determination unit, and the display unit are operating, and determines whether the image processing unit and the determination unit are operating normally; An item inspection device further characterized by comprising an operation confirmation image generation unit (53) that generates, based on the inspection image during the operating operation, a defective product image that includes at least defective characteristic parts that will result in the quality condition of the item being judged to be defective, as an operation confirmation image to be used in the operation confirmation processing unit.

2. The object inspection device described in claim 1, characterized in that the operation confirmation image generation unit generates the operation confirmation image by combining multiple partial images that become the defective feature parts at multiple locations in the inspection image during the driving operation that have different image features.

3. The object inspection device described in claim 1 or 2, characterized in that the operation confirmation image generation unit generates the defective image by combining images of multiple defective features that result in the quality condition of the object being judged as defective with a good inspection image in which the quality condition of the object was judged as good by the judgment unit during the operating operation.

4. a control unit (30) having a first control function unit that controls the operations of the image processing unit, the determination unit, and the display unit to determine the quality state of each of the items, The object inspection device described in claim 1 or 2, characterized in that the control unit also has a second control function unit (50) that controls the operation of the operation confirmation processing unit and the operation confirmation image generation unit while performing the first control function, and determines whether the operation of the image processing unit and the judgment unit is normal or not.

5. 3. An article inspection apparatus according to claim 1, wherein the inspection image is an X-ray transmission image (Dpx) obtained by imaging the article with X-rays.

Images