Article inspection device

Abstract

To provide an article inspection device with which it is possible to easily and accurately select and set an algorithm that suits the article to be inspected from a plurality of image processing algorithms.SOLUTION: Provided is an article inspection device for inspecting an article W to be inspected by applying a prescribed image processing algorithm Pgm to an inspection image Dpx in which the article W to be inspected is captured. The article inspection device comprises: an algorithm storage unit 44 for preliminarily storing a plurality of image processing algorithms; a learning unit 41 for evaluating about the stored image processing algorithms on the basis of performance information indicating inspection performance when the suitability of inspection to an inspection image is applied to the data of a plurality of acquired images obtained by imaging articles other than the article to be inspected, and calculating a plurality of evaluation values representing respective evaluation results about the plurality of image processing algorithms; and an image processing algorithm setting unit 45 for setting a prescribed image processing algorithm used for determining the quality state of the article W to be inspected, on the basis of the plurality of evaluation values calculated by the learning unit 41.SELECTED DRAWING: Figure 1

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 to be inspected by applying a predetermined image processing algorithm to an inspection image obtained by imaging the article to be inspected of a predetermined variety.

Background Art

[0002] Conventionally, in an article inspection apparatus, a predetermined image processing algorithm having a predetermined image processing filter corresponding to an inspection item or a combination thereof is applied to imaging data of an article to be inspected, so that a predetermined quality state of the article to be inspected can be inspected with high accuracy. There are some such apparatuses.

[0003] In such an article inspection apparatus, it is necessary to select and set an image processing filter or the like corresponding to an inspection item of a specific article to be inspected from among a plurality of image processing filters and the like that are created in advance based on the characteristics of the article to be inspected and the characteristics of foreign matter to be detected and stored in a memory in advance. Therefore, there is an apparatus in which the function of selecting and setting an image processing filter or the like required for inspection is automated so that such a selection and setting operation can be easily performed without being influenced by the experience of an operator.

[0004] As this type of article inspection apparatus, for example, a plurality of types of filter processing corresponding to characteristics are performed on input inspection image data based on detection data of an X-ray dose transmitted through an article to be inspected in which no foreign matter is mixed, and a plurality of X-ray images are generated. A filter that generates data of an X-ray image such that the maximum pixel value is minimized among the data of the plurality of X-ray images in which the edge portion image of the article has an increased luminance is known to be selected and set as an optimal X-ray image processing filter (see, for example, Patent Document 1).

[0005] In addition, in order to extract a desired image processing algorithm that approximates the foreign object detection characteristics capable of detecting a foreign object to be detected from a plurality of image processing algorithms stored in the storage means, an image processing algorithm that can emphasize the foreign object while reducing the influence of the article to be inspected is ranked and displayed, thereby facilitating the selection and setting operation (see, for example, Patent Document 2).

Prior Art Documents

Patent Documents

[0006]

Patent Document 1

Patent Document 2

Summary of the Invention

Problems to be Solved by the Invention

[0007] However, in the conventional article inspection apparatus as described above, although high-precision foreign object detection is possible, the functions of detecting and displaying the characteristics of the article to be inspected are relatively suppressed in order to emphasize the foreign object. Therefore, considering the differences in the state such as the properties and forms of the article itself that affect the inspection results (the density, shape, luminance distribution, etc. of the inspection image corresponding to the differences), it is not easy to accurately select and set the necessary image processing algorithm, that is, the combination of the necessary plurality of image processing filters.

[0008] For example, it is not easy for an inexperienced user to accurately select and set an image processing algorithm corresponding to the article to be inspected, and it is not easy to accurately select and set the image processing algorithm also because a setting operation using a foreign object sample is required.

[0009] Therefore, for example, a selection operation input for roughly classifying based on typical forms, sizes, etc. that allow visual inspection of the inspected article is executed, and the optimal image processing algorithm is automatically selected based on the selection result, or when a sample of the foreign object to be detected is attached, it is necessary to confirm that the image processing algorithm is accurate enough for the detection waveform to clearly rise.

[0010] Furthermore, in the case of providing a learning stage for learning and memorizing X-ray transmission images of inspected articles with foreign objects in order to obtain high inspection accuracy, if overfitting occurs with learning data focused on defective parts, it becomes difficult to sufficiently learn the effects of the properties and forms of the article itself that affect the inspection results, and there was also concern that the optimal image processing algorithm might be omitted from the options.

[0011] The present invention has been made in view of the above-described conventional unsolved problems, and an object thereof is to provide an article inspection apparatus capable of easily and accurately selecting and setting an image processing algorithm suitable for an inspected article from a plurality of image processing algorithms.

Means for Solving the Problem

[0012] In order to achieve the above object, the article inspection apparatus according to the present invention is an article inspection apparatus that inspects the quality state of an inspected article by applying a predetermined image processing algorithm to an inspection image obtained by imaging the inspected article of a predetermined variety, comprising: an image processing algorithm storage unit that stores in advance a plurality of image processing algorithms including the predetermined image processing algorithm; an image processing algorithm evaluation unit that evaluates the suitability of the inspection for the inspection image for each of the plurality of image processing algorithms stored in the image processing algorithm storage unit based on performance information representing the inspection performance when the algorithms are applied to the data of a plurality of acquired images obtained by imaging articles other than the inspected article, and calculates a plurality of evaluation values representing the evaluation results for each of the plurality of image processing algorithms; and an image processing algorithm setting unit that sets the predetermined image processing algorithm to be used for determining the quality state of the inspected article based on the plurality of evaluation values calculated by the image processing algorithm evaluation unit.

[0013] Therefore, in the present invention, the image processing algorithm evaluation unit evaluates the suitability of the inspection for the data of the inspection image of the inspected article based on the performance information representing the inspection performance of the inspection when it is applied to the data of a plurality of acquired images obtained by imaging articles other than the inspected article, and calculates a plurality of evaluation values representing the evaluation results. Then, based on the calculated plurality of evaluation values, the image processing algorithm setting unit sets the predetermined image processing algorithm to be used for determining the quality state of the inspected article. As a result, it is possible to easily, quickly, and accurately select and set the optimal image processing algorithm from among the plurality of image processing algorithms, and the article inspection apparatus can quickly and accurately set the inspection algorithm for performing article inspection including image processing and determination processing.

[0014] Note that the data of the inspection image of the inspected article referred to here is, for example, the data of the inspection image whose label is unknown when using a model trained with labeled learning data for the selection setting of the image processing algorithm. In that case, the performance information representing the inspection performance is, for example, the performance information labeled on the acquired image data, and the evaluation value is, for example, an evaluation index equivalent to the correct rate with respect to the correct label, and is a value that can be displayed as a score.

[0015] In a preferred embodiment of the present invention, (2) the image processing algorithm evaluation unit calculates the plurality of evaluation values so that the larger the numerical value is as the inspection performance represented by the performance information is superior, and the image processing algorithm setting unit selects a part of the plurality of image processing algorithms that are superior based on the numerical values of the plurality of evaluation values, and can be configured to set the predetermined image processing algorithm.

[0016] With this configuration, it is possible to accurately and easily select and set an image processing algorithm with superior inspection performance from a plurality of image processing algorithms.

[0017] In a preferred embodiment of the present invention, (3) the plurality of evaluation values are calculated so as to be the maximum value when the performance information represents inspection performance that is superior to a predetermined level or more, and the image processing algorithm setting unit may display a predetermined number of the plurality of image processing algorithms on a display in order of superiority of the plurality of evaluation values so that they can be selected.

[0018] In this case, after extracting the image processing algorithms with inspection performance superior to a predetermined level or more among the plurality of image processing algorithms, it becomes possible to make judgments and confirmations for selecting the required inspection performance.

[0019] In a preferred embodiment of the present invention, (4) the image processing algorithm evaluation unit evaluates the suitability of each of the plurality of image processing algorithms for the inspection image based on a learning model created in advance by learning using inspection images of a plurality of types of product groups, which are the other articles, and image processing algorithms that match the inspection images of the product groups as teacher data.

[0020] In this case, learning is performed using teacher data in which inspection images of a plurality of types of product groups, which are inspection objects of a plurality of varieties, are associated with image processing algorithms that match the respective inspection images, and the learned learning model is created. Therefore, the learning model can exhibit high-precision classification performance for the inspection images of the inspection objects, and it becomes possible to select and set an appropriate image processing algorithm.

[0021] In a preferred embodiment of the present invention, (5) the image processing algorithm setting unit extracts at least one image processing algorithm candidate according to the evaluation values of the plurality of image processing algorithms based on the learning model in the image processing algorithm evaluation unit, and when there are a plurality of the extracted image processing algorithm candidates, the extracted image processing algorithm candidates can be displayed on a display so as to be selectable.

[0022] In this case, when the image processing algorithm with the best inspection performance among the plurality of image processing algorithms is clearly identified, the image processing algorithm can be automatically selected and set, or when there are a plurality of image processing algorithm candidates with a high probability of competing inspection performance, the candidates can be displayed and the selection judgment and confirmation operation input of the required inspection performance can be waited for and then the selection and setting can be performed.

[0023] In a preferred embodiment of the present invention, (6) it can be configured to include product group storage means for storing, as learning data, images of a product group associated with the optimal image processing algorithm among the plurality of image processing algorithms, and a learning unit for generating the learning model based on the learning data.

[0024] In this case, among a plurality of image processing algorithms, images of a product group associated with the optimal image processing algorithm can be used as teaching data, so that the optimal image processing algorithm can be accurately selected and set for the inspection image of the article to be inspected.

[0025] In a preferred embodiment of the present invention, (7) the learning model may be created using a deep learning classification method.

[0026] In this case, the learning model uses a deep learning classification method with inspection images of a plurality of types of product groups that image other articles than the article to be inspected and the image processing algorithms suitable for the respective inspection images as teaching data, so that accurate classification can be performed for the multi-dimensional features of the inspection images respectively, and based on the classification results, the necessary image processing algorithm that preferably corresponds to the inspection image can be more accurately selected and set.

[0027] In a preferred embodiment of the present invention, (8) the article to be inspected may be irradiated with X-rays to obtain an X-ray inspection image, and the quality state of the article to be inspected may be X-ray inspected by applying the predetermined image processing algorithm to the X-ray inspection image.

[0028] In this case, it becomes an X-ray inspection apparatus capable of easily, quickly, and accurately selecting and setting the optimal algorithm from a plurality of X-ray inspection image processing algorithms.

Effect of the Invention

[0029] According to the present invention, it is possible to provide an article inspection apparatus capable of easily and accurately selecting and setting an image processing algorithm suitable for an article to be inspected from a plurality of image processing algorithms.

Brief Description of the Drawings

[0030]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Embodiments for Carrying Out the Invention

[0031] Hereinafter, embodiments for carrying out the present invention will be described with reference to the drawings.

[0032] (One Embodiment) FIGS. 1 to 6 show an article inspection apparatus according to an embodiment of the present invention.

[0033] First, the configuration will be described.

[0034] As shown in Fig. 1, the article inspection apparatus 1 of this embodiment includes a conveyance unit 10, an inspection unit 20, and a control unit 30. While irradiating the article W to be inspected (article to be inspected) conveyed by the conveyor by the conveyance unit 10 with X-rays, the inspection unit 20 detects image data corresponding to the transmitted X-ray dose distribution, and inspects the quality state of the article W to be inspected based on the detected image data. Here, the quality state refers to the quality required for the article W as a product, the propriety of physical quantities, etc., for example, the presence or absence of foreign matter mixed in, the presence or absence of defective products, the propriety of the shape, size, storage state, etc. of the contents, the distribution of density, thickness, volume or mass, etc.

[0035] The conveyance unit 10 is a conveyor that winds a loop-shaped conveyance belt 11 around a plurality of conveyance rollers 12 and 13, and can sequentially convey the article W to be inspected in the right direction in Fig. 1 by the upper running section 11a of the conveyance belt 11, and is supported by a housing (not shown).

[0036] The inspection unit 20 is an X-ray inspection unit having an X-ray generator 21 (X-ray source) that generates X-rays in a predetermined energy band that passes through the article W to be inspected conveyed by the conveyance unit 10, and an X-ray detector 23 disposed directly below the upper running section 11a of the conveyance belt 11.

[0037] The X-ray generator 21 generates X-rays having a wavelength and intensity corresponding to the tube current and tube voltage with a known X-ray tube 22, and can irradiate the article W to be inspected within a predetermined inspection section Zx on the conveyance belt 11 with fan beam-shaped X-rays directed in the main observation direction orthogonal to the article conveyance direction of the conveyance unit 10 through the X-ray window portion of an outer casing (not shown in detail).

[0038] Although not shown in detail, this X-ray detector 23 is composed of, for example, a detection element composed of a scintillator that is a phosphor and a photodiode or a charge coupled device, which is arranged in an array at a predetermined pitch in the width direction of the conveyance path of the conveyance unit 10 to perform X-ray detection at a predetermined resolution, and is arranged at a predetermined position in the conveyance direction corresponding to the X-ray irradiation position from the X-ray generator 21.

[0039] That is, the X-ray detector 23 detects the X-rays irradiated from the X-ray generator 21 and transmitted through the inspection object W for each predetermined transmission region corresponding to the detection element, converts them into electrical signals according to the transmission amount of the X-rays, and outputs an X-ray detection signal for generating an X-ray transmission image with the direction in which the X-rays are transmitted as the observation direction. Here, it is assumed that the X-rays irradiated from the X-ray generator 21 or the X-rays detected by the X-ray detector 23 have a certain radiation quality (energy, wavelength) specified according to the quality of the inspection object W. However, by using a plurality of types of X-rays with different radiation qualities, it is also possible to generate so-called dual-energy or multi-energy X-ray images.

[0040] The control unit 30 includes a conveyance control means for controlling the conveyance speed, conveyance interval, etc. of the inspection object W by the conveyance belt 11 in the conveyance unit 10, and an inspection control means for controlling the X-ray irradiation intensity and irradiation period in the inspection unit 20, or controlling the X-ray detection cycle of the X-ray line sensor of the X-ray detector 23 according to the conveyance speed of the inspection object W and the detection period of each inspection object W, etc. However, detailed illustration is omitted.

[0041] The control unit 30 also includes a detection image data acquisition unit 31 that sequentially captures X-ray detection signals from the X-ray detector 23 at predetermined intervals to obtain and output X-ray transmission images of each inspected article W, an image processing unit 32 that captures the detection image data output from the detection image data acquisition unit 31 and performs image analysis processing such as predetermined filter processing (including preprocessing) capable of extracting image features and feature measurement for obtaining the feature amounts of the extracted image features, an inspection determination unit 33 that determines the presence or absence of a predetermined quality state of the inspected article W based on the data of the feature amounts extracted and measured by the image processing unit 32, for example, determination processing for the presence or absence of foreign matter contamination, the presence or absence of missing items, the shape, size, or storage state of the contents, etc., an image processing control unit 34 capable of updating and changing the inspection algorithm including the image processing algorithm used in the image processing unit 32 and the determination processing algorithm used in the inspection determination unit 33, and a display operation unit 35 (display) such as a touch panel capable of displaying and outputting the determination result in the inspection determination unit 33 and inputting requests for variety registration and other operations to the image processing control unit 34.

[0042] This control unit 30 is configured to include, for example, a microcomputer having a CPU, ROM, RAM, and I / O interface (not shown), an auxiliary storage device that stores a control program for exerting the functions of the image processing unit 32, the inspection determination unit 33, and the image processing control unit 34 in a manner readable in cooperation with the ROM, and a timer circuit, etc. According to the control program stored in the ROM, etc., the CPU exchanges data with the RAM, etc. while executing predetermined arithmetic processing and also executes the control program.

[0043] The detection image data acquisition unit 31 is an image input unit. For example, it performs A / D conversion on the X-ray detection signals from a plurality of detection elements of the X-ray detector 23, and for each predetermined unit transfer time corresponding to the detection element size in the X-ray detector 23, for all n detection element regions (n is an integer greater than 1, for example, 640), it writes data on the cumulative transmission amount within that unit time into the image memory as digital data of density levels representing gradations from 0 to 1023 (hereinafter referred to as line scanning).

[0044] Further, the detection image data acquisition unit 31 has a data processing program and a working memory (not shown) that generate data Dpx of an X-ray imaging image corresponding to the dose distribution of the X-rays transmitted through the inspected article W based on the detection data Lx of the line scanning image sequentially written into the image memory when the line scanning by the X-ray detector 23 is repeated a predetermined number of times according to the inspection period of the inspected article W, and output it to the image processing unit 32 as imaging data of the inspected article W.

[0045] In the image processing unit 32, a predetermined image processing algorithm combining an image processing filter and the like is preset and stored in order to perform a predetermined article inspection based on the imaging data Dpx (X-ray imaging image data) of the inspected article W taken in from the detection image data acquisition unit 31.

[0046] The image processing filter included in the image processing algorithm of the image processing unit 32 is a processing program for extracting image features (such as edges, lines, corners, regions, shades, textures) necessary for a predetermined article inspection based on the imaging data Dpx of the article under inspection W. When the image processing algorithm includes a filter for foreign object detection, for example, a feature extraction filter that performs edge detection processing to emphasize the contour of a foreign object in the article under inspection W, has a differential filter such as a Sobel filter, and performs differential processing based on a predetermined arithmetic formula in the vicinity of the pixel of interest to emphasize the edge of the foreign object. Note that the term "image processing filter, etc." means that it includes pre-processing such as shading correction and noise removal for the imaging data from the detected image data acquisition unit 31 in order to improve the detection processing accuracy of image features.

[0047] Also, the feature measurement of image features executed by the image processing unit 32 is for an image obtained by performing necessary pre-processing and image processing on the imaging data Dpx of the article under inspection W taken in from the detected image data acquisition unit 31, and calculates attributes related to shading features, color features, shape features, etc. (feature amounts that characterize edges, regions, distances, positions, shapes, etc.), calculates feature amounts representing the spatial relationship between such features, or calculates texture feature amounts related to the spatial frequency distribution and direction components, so as to calculate the feature amounts necessary for the determination processing in the inspection determination unit 33.

[0048] The inspection determination unit 33 detects the feature shapes and foreign objects, etc. detected in the article under inspection W based on the feature amounts extracted and feature-measured by the image processing unit 32, and compares feature amounts such as the area, contour length, and sum of densities of the detection target with predetermined determination reference values, so as to execute determination processing as to whether or not local feature shapes, etc. corresponding to foreign objects or defective parts that satisfy the determination conditions are included in the article under inspection W.

[0049] The article inspection device 1 thus inspects a predetermined quality state of the inspected article W by applying a predetermined image processing algorithm Pgm that combines a plurality of filter processes and the like to the detection image Dpx obtained by imaging the inspected article W of a predetermined variety and the image data obtained by adding necessary preprocessing and image processing thereto.

[0050] On the other hand, the image processing control unit 34 is capable of data communication with a learning unit 41 for machine learning configured by an external PC or the like. The learning unit 41 is provided with a learning model 42 and a learning data storage unit 43 that stores a data set of a large number of learning samples for learning the learning model 42, the data set including the image data of each sample and a label indicating the image processing algorithm that matches (is preferably used for) the image data. In the present embodiment, the label of each image processing algorithm stored in the learning data storage unit 43 includes at least the number (identification number) of the image processing algorithm, and may also include the name of the algorithm, the name of the inspection target variety indicating the content of the inspection by the algorithm, and the inspection items.

[0051] The image processing control unit 34 further includes an algorithm storage unit 44 (image processing algorithm storage unit) that stores in advance a plurality of image processing algorithms Pgm including a predetermined image processing algorithm, and an algorithm setting unit 45 (image processing algorithm setting unit) capable of executing an update setting process for updating the image processing algorithm set and stored in the image processing unit 32 to any one of the image processing algorithms stored in the algorithm storage unit 44.

[0052] The learning unit 41 has a function of creating a learning model 42 capable of associating the type of the image processing algorithm that matches with any one of the classifications of a plurality of image processing algorithms in which the image processing algorithms that match are associated with the input inspection image data of a variety corresponding to or approximated to any of the image data of a large number of learning samples by performing learning processing or machine learning processing based on the statistical pattern recognition method based on the data set of a large number of learning samples stored in the learning data storage unit 43.

[0053] The statistical pattern recognition method mentioned here represents the pattern to be classified with a specific high-dimensional feature vector according to the number of classifications, and classifies based on the position of the vector in the low-dimensional feature space. It consists of a learning stage and a recognition stage. Also, the data of the input inspection image mentioned here is the data of the inspection image with unknown labels when using the trained learning model 42 trained with labeled learning data regarding algorithm settings. In that case, the performance information representing the inspection performance is, for example, the performance information regarding the algorithm labeled for the acquired image data, and the evaluation value is, for example, an evaluation index equivalent to the correct rate for the correct label, which is a value that can be displayed as a score.

[0054] In the learning stage of the statistical pattern recognition executed by the learning unit 41, using a data set of a large number of learning samples where the type of the applicable image processing algorithm is known as label information for class classification, from the set of the data set, the type parameters of the image processing algorithm to be class classified and the parameters representing the distribution state in the feature space of the image features of each learning sample are calculated.

[0055] Also, in the recognition stage of the statistical pattern recognition executed by the learning unit 41, feature selection for extracting a low-dimensional feature vector effective for classification is executed based on the feature vector obtained from the input image pattern. The feature selection here is, for example, principal component analysis that linearly projects feature points in a multi-dimensional space onto a low-dimensional subspace with a large variance (an index of the variation of pixel values).

[0056] Next, it is classified and determined which class the extracted low-dimensional feature vector belongs to in light of the distribution in the feature space of each class learned in advance by a large number of learning samples, and the type of the image processing algorithm suitable for the inspected article W of the input image pattern is specified.

[0057] More specifically, the learning unit 41, for example, uses data of sample inspection images associated with an image processing algorithm, feature data obtained from the data of the inspection images, such as statistical feature quantities such as image density dispersion, average, and human histogram of imaging data of a product group that is of the same kind as or similar to the inspected article W but not the inspected article W itself, and for planar shapes, three-dimensional shapes, etc., obtains feature quantities with density, contour length, area, etc. as shape index values by a predetermined arithmetic expression, and based on these feature quantities (values with weights), creates a learning model 42 that can evaluate the suitability of the algorithm in the inspection image and output an evaluation value for each algorithm.

[0058] A plurality of types of image processing algorithms classifiable by the learning unit 41 are stored in the algorithm storage unit 44 as programs, setting parameters, etc. corresponding to various types of image processing algorithms. When the learning unit 41 identifies the type of image processing algorithm that suits the imaging data Dpx of the inspected article W taken in from the detection image data acquisition unit 31, the learning model 42 notifies and outputs the image processing algorithm of the suitable type to the algorithm setting unit 45 as the required image processing algorithm. Then, upon receiving the notification of this required image processing algorithm, the algorithm setting unit 45 executes an update setting process of updating the current image processing algorithm set and stored in the image processing unit 32 to the image processing algorithm of the suitable type stored in the algorithm storage unit 44. Note that the algorithm setting unit 45 may execute the update setting process when the image processing algorithm of the suitable type is notified and output not once but a predetermined number of times from the learning unit 41.

[0059] The plurality of image processing algorithms stored in the algorithm storage unit 44 include, for example, image filters and other image processing programs that are suitable for 30 varieties of articles that can be registered as varieties of the inspected article W. More specifically, for example, they are image processing algorithms suitable for any product group of products such as chicken breast meat, those with unevenness, powder, noodle products, and chocolate in bags.

[0060] In the learning unit 41, it is basically to perform supervised learning using a dataset of a large number of learning samples for which the type of the matching image processing algorithm is known as label information. However, semi-supervised learning may also be considered, in which the learning samples without labels are labeled based on the classification results obtained by the trained learning model 42, and the number of learning samples is increased. Alternatively, data augmentation may be considered, in which the image data of the learning samples are subjected to transformation processes such as left-right and up-down inversion, enlargement / reduction, or rotation, and then the re-classification by the trained learning model 42 is confirmed, and a dataset of sample images that do not change in classification is added to the learning data.

[0061] The learning unit 41 also has a function of an image processing algorithm evaluation unit that evaluates each of the plurality of image processing algorithms stored in the algorithm storage unit 44 based on the performance information indicating the inspection performance when the applicability of each image processing algorithm to the data of the input inspection image from the detection image data acquisition unit 31 is applied to the data of the plurality of acquired images obtained by imaging a large number of samples that are other articles than the article to be inspected by the learning model 42, and calculates a plurality of evaluation values representing the evaluation results of the respective image processing algorithms as values that can be score-displayed.

[0062] Then, based on the plurality of evaluation values calculated by the learning unit 41 for the plurality of image processing algorithms, the algorithm setting unit 45 determines a predetermined image processing algorithm necessary for determining the quality state of the article to be inspected W, and executes an update setting process for updating the current image processing algorithm set and stored in the image processing unit 32 to the required image processing algorithm of the applicable type.

[0063] More specifically, the learning unit 41 performs learning using inspection images of a plurality of types of product groups and image processing algorithms that match the inspection images of those product groups as teacher data. As a result, the learning unit 41 calculates the above-described score-displayable evaluation value to be equivalent to the probability (correct rate) of belonging to the class of the matching image processing algorithm, and to be a larger numerical value as the inspection performance represented by the performance information is more excellent, so that it can be score-displayed on the display operation unit 35.

[0064] As shown in FIGS. 2 and 3, the score display 51 from the learning unit 41 to the display operation unit 35 ranks and displays a dataset of a plurality of labeled image processing algorithms in the algorithm selection operation screen 50 (details will be described later) in descending order of score for a predetermined number of records (three in FIG. 2). From the ranking column 52 on the leftmost side, sequentially to the right, it is composed of an algorithm number column 53 that identifies the image processing algorithm, an algorithm detailed description column 54 that clarifies the inspection target of the image processing algorithm, and a score display column 55. Also, the score display in the score display column 55 in FIG. 2 is displayed as a percentage of a numerical value of 1 or less, which is the probability that the same matching image processing algorithm exists in the specific classification class of the sample image whose image characteristics are the same as the data of the input inspection image from the detection image data acquisition unit 31.

[0065] In this illustrated example, regarding the suitability of each image processing algorithm for the data of the input inspection image from the detection image data acquisition unit 31, the image processing algorithm with the algorithm number "1310", which has the most excellent inspection performance represented by the performance information, is the image processing algorithm candidate ranked first. It is described in the detailed description column 54 that it is an image processing algorithm for "chicken breast meat", and it is displayed that the score of that algorithm is 0.92 (92%). The image processing algorithm for "chicken breast meat" mentioned here is, for example, an image processing algorithm including an image processing filter suitable for detecting foreign substances such as bones remaining or adhering to chicken breast meat products.

[0066] In this case, it is shown that the data of the input inspection image conforms to the class in which the algorithm number "1310" is classified with a probability of 92%. Also, "for 2 kg pack of chicken breast" with algorithm number "1312" is ranked second, and the score of its image processing algorithm is displayed as 0.05. And "for chicken thigh meat" with algorithm number "1311" is ranked third, and the score of its image processing algorithm is displayed as 0.01. Therefore, it can be seen that for the image processing algorithms ranked second or lower, there is no compatibility.

[0067] Based on the compatibility evaluation result in the learning unit 41, the algorithm setting unit 45 can automatically select the optimal image processing algorithm and update and set the image processing algorithm stored in the image processing unit 32 without performing a selection operation input for roughly classifying based on typical forms, sizes, etc. that allow visual inspection of the state of the inspected article W as in the prior art.

[0068] Alternatively, based on the score display values (numerical values) of a plurality of evaluation values, the algorithm setting unit 45 can select and operate, through an operation screen input from the display operation unit 35, a superior image processing algorithm among a plurality of image processing algorithm candidates, for example, the image processing algorithm for "chicken breast meat" shown in FIGS. 2 and 3, and when the update button 56 in the figure is operated, update and set a predetermined image processing algorithm selected according to the operation input. Note that the cancel button 57 in both figures is for performing a cancel operation when a selection operation of any one of the plurality of image processing algorithm candidates is incorrect.

[0069] The algorithm selection operation screen 50 (see Fig. 6) for selecting the image processing algorithm candidates during variety registration shown in Fig. 3 is displayed on the touch panel screen (display) of the display operation unit 35. Below the common information display unit 61, an image display area 62 with a predetermined screen size and an inspection information display unit 63 are arranged. Below them, an operation input unit 64 having a plurality of operation buttons is arranged. And in the inspection information display unit 63, a variety number display unit 63a, an inspection result display unit 63b, a selection content display unit 63c for displaying the inspection content and image processing algorithm during inspection, or the operation target and operation buttons during setting operations, etc. are arranged. In the operation input unit 64 below the image display area 62, a menu button 64a, a display switching button 64b, an operation confirmation button 64c, a setting / adjustment button 64d, etc. are arranged. In the operation input unit 64 below the inspection information display unit 63, a stop button 64e and a start button 64f for the operation of the article inspection device 1 are arranged.

[0070] The learning unit 41 may calculate the score, which is the evaluation value of the inspection performance, so as to be the maximum value when the performance information regarding the inspection performance represents an inspection performance that is superior to a predetermined level or more. The algorithm setting unit 45 is configured to display, in the algorithm selection operation screen 50 (see Fig. 6) of the display operation unit 35, a predetermined plurality, for example, three image processing algorithms out of the plurality in the order of superiority of the evaluation values so that they can be selected.

[0071] That is, the algorithm setting unit 45 extracts at least one image processing algorithm candidate according to the evaluation values (scores) of a plurality of image processing algorithms based on the learning model 42 in the learning unit 41. When there are a plurality of the extracted image processing algorithm candidates, the plurality of the extracted image processing algorithm candidates can be list-displayed on the display 50 so that they can be selected.

[0072] In addition, as described above, the learning unit 41, which is an image processing algorithm evaluation unit, evaluates the suitability of inspections of a plurality of image processing algorithms (algorithm numbers 1310, 1311, 1312) for the input inspection image Dpx based on a learning model 42 created in advance by supervised learning using inspection images of a plurality of types of product groups, which are products other than the inspected article W, and the image processing algorithms that match the inspection images of those product groups as teacher data.

[0073] In this way, the image processing control unit 34 has a learning data storage unit 43 (product group storage means) that stores, as learning data, the image data of the product group associated with the optimal image processing algorithm among the plurality of image processing algorithms, and includes, outside or inside the apparatus main body of the article inspection apparatus 1, a learning unit 41 that generates the learning model 42 based on the learning data stored in the learning data storage unit 43 in a data communicable manner.

[0074] This learning model 42 is assumed to perform learning processing and machine learning processing using a statistical pattern recognition method, but it may also be one created using a deep learning classification method using a convolutional neural network, or a support vector machine (SVM) that performs two-class classification, for example, a non-linear SVM. Also, the input for learning may be numerical data such as the average value, variance value, maximum value, and product size of pixels, rather than the X-ray image itself. Further, when using an image for learning input, not only the transmission image by X-ray, but also a difference image using transmission images in different energy bands, an image after filtering the captured image, in addition to images obtained by imaging methods using different optical systems such as visible light NIR, can be used. Also, an image created by combining a plurality of types of images (each of the above images), such as an image obtained by assigning an arbitrary image to each channel of a color image such as an RGB image (see, for example, Japanese Unexamined Patent Application Publication Nos. 2023-114827 and 2023-114828), may be used.

[0075] In any case, the learning model 42 is typically created based on the learning data of individual articles in the product group other than the article under inspection W (a set of inspection images for which the image processing algorithm that already conforms is known), using machine learning (deep learning or other machine learning) of AI technology. The algorithm number or algorithm name for identifying the image processing algorithm can be obtained from the top of the evaluation values by the learning data of the sample articles stored in the learning data storage unit 43 and the learned learning model 42. Then, at the time of registering the variety of the article under inspection W, based on the new input inspection image that is the object of the inspection, using the learning model 42, the optimal image processing algorithm can be extracted so as to be ranked and displayed, and the numerical data (probability, etc.) of the evaluation value can also be displayed simultaneously.

[0076] As described above, the article inspection apparatus 1 of the present embodiment irradiates the article under inspection W being conveyed with X-rays in the inspection unit 20 to acquire the data of the input inspection image that is the X-ray inspection image, and applies a predetermined image processing algorithm to the data of the input inspection image in the image processing unit 32 to perform X-ray inspection on the quality state of the article under inspection W.

[0077] On the other hand, when the set variety of the article under inspection W is switched by the display operation unit 35 or newly set and registered in the control unit 30 of the article inspection apparatus 1, the data of the input inspection image from the detection image data acquisition unit 31 of the inspection unit 20 is taken into the learning unit 41 and the algorithm setting unit 45 of the image processing control unit 34, and at least one suitable, for example, a plurality of image processing algorithm candidates estimated to highly conform to the data of the input inspection image are extracted and displayed by the learning model 42. When the optimal image processing algorithm suitable for the article under inspection W of the set variety is automatically selected or selected by a selection operation, the control unit 30 causes the algorithm setting unit 45 to update the image processing algorithm used in the image processing unit 32 to the optimal image processing algorithm for the variety after the switching.

[0078] Next, the operation will be described.

[0079] In the present embodiment configured as described above, in the learning stage of the learning unit 41, as shown in FIG. 4, first, for a number of learning samples for learning the learning model 42, a data set of the image data of each learning sample and a label indicating the image processing algorithm conforming to the image data is acquired and stored in the learning data storage unit 43 (step S11).

[0080] Next, the learning unit 41 uses the data set of a number of learning samples stored in the learning data storage unit 43 to execute learning processing and machine learning processing by the statistical pattern recognition method as described above, thereby creating a learning model 42 that can be associated with any of a plurality of class classifications corresponding to the type of the image processing algorithm conforming to the data of the input inspection image of the inspected article W (step S12).

[0081] Next, as shown in FIG. 5, when a predetermined variety registration setting input is made to the control unit 30 by the display operation unit 35 and the inspected article W for variety registration is test-transported to the article inspection apparatus 1, the inspected article W being transported is irradiated with X-rays by the inspection unit 20, the X-rays transmitted through the inspected article W are detected by the X-ray detector 23, and the data of the input inspection image acquired by the detection image data acquisition unit 31 is respectively taken into the learning unit 41 and the algorithm setting unit 45 of the image processing control unit 34 (step S21).

[0082] Next, after at least one suitable image processing algorithm candidate estimated to highly conform to the data of the input inspection image is extracted by the learning model 42 (step S22), it is determined whether or not the number of the extracted image processing algorithm candidates is plural (step S23).

[0083] At this time, if there are a plurality of image processing algorithm candidates (YES in step S23), the plurality of image processing algorithm candidates are displayed in a ranking (step S24). At this time, the ranking may be displayed on the condition that there are a plurality of image processing algorithm candidates having a score equal to or higher than a predetermined value.

[0084] Next, wait for the user to refer to a plurality of candidate image processing algorithms ranked as shown in FIG. 2 and perform a selection operation on any of the candidate image processing algorithms (when NO in step S25). When a candidate image processing algorithm determined to be optimal for the inspection target article W to be inspected is selected (when YES in step S25), or when there is not a plurality of candidate image processing algorithms (when NO in step S23), then, one selected or one extracted image processing algorithm is parameter-set as part of an inspection algorithm suitable for the inspection target article W to be registered by variety (step S26), and the current process ends.

[0085] As described above, in the article inspection apparatus 1 of the present embodiment, the learning unit 41, which is an image processing algorithm evaluation unit, evaluates the suitability of inspection of the inspection image data of the inspection target article W based on performance information representing the inspection performance of inspections applied to the data of a plurality of acquired images obtained by imaging a product group that is other articles than the inspection target article W, respectively. A plurality of evaluation values (scores) representing the evaluation results for a plurality of image processing algorithms are calculated. Then, based on the plurality of calculated evaluation values, the algorithm setting unit 45 sets a predetermined image processing algorithm used for determining the quality state of the inspection target article W as a required image processing algorithm. Therefore, it becomes possible to easily, quickly, and accurately select and set the optimal image processing algorithm from a plurality of image processing algorithms.

[0086] Also, in the present embodiment, the learning unit 41 calculates the score, which is an evaluation value, so that the larger the numerical value is as the inspection performance is superior, and the algorithm setting unit 45 selects an image processing algorithm with superior inspection performance based on the numerical values of the plurality of scores and updates and sets the image processing algorithm. Therefore, it becomes possible to accurately and easily select and set an image processing algorithm with superior inspection performance from a plurality of image processing algorithms.

[0087] Furthermore, in the present embodiment, the evaluation value is calculated so as to reach the maximum value when the performance information represents inspection performance that is superior to a predetermined level or more. The algorithm setting unit 45 is configured to display, in the order of superiority of the evaluation values, a predetermined number, for example, three image processing algorithms out of the plurality of algorithms so that the user can select them. Therefore, after extracting the image processing algorithms that provide inspection performance superior to a predetermined level or more among the plurality of image processing algorithms, it becomes possible for the user to make a determination and confirmation to select the required inspection performance.

[0088] In addition, in the present embodiment, learning is performed using teacher data in which image processing algorithms suitable for respective inspection images are associated with inspection images of a plurality of types of product groups for a plurality of types of inspection objects W, and the learning model 42 is created. Therefore, the learning model 42 can exhibit high-precision classification performance for the inspection images of the inspection object W, and it becomes possible to perform accurate selection and setting of the image processing algorithm.

[0089] Also, in the present embodiment, when there are a plurality of extracted image processing algorithm candidates, the algorithm setting unit 45 displays them on the algorithm selection operation screen 50 of the display operation unit 35 (display) so that they can be selected. Therefore, in addition to being able to accurately select the image processing algorithm that provides superior inspection performance among the plurality of image processing algorithms, when a plurality of image processing algorithm candidates with a high probability of competing inspection performance are displayed, it becomes possible to wait for the selection determination and confirmation operation input of the required inspection performance and then perform selection and setting.

[0090] Furthermore, in the present embodiment, since the images of the product group associated with the optimal image processing algorithm among the plurality of image processing algorithms can be used as teacher data, it becomes possible to accurately select and set the optimal image processing algorithm for the inspection images of the inspection object W.

[0091] In addition, in this embodiment, when the learning model is created using the deep learning classification method, the learning model uses, as teacher data, inspection images of a plurality of types of product groups that image other items than the item under inspection W and the image processing algorithms that match the respective inspection images. By doing so, it becomes possible to accurately classify the multi-dimensional features of the inspection images. Therefore, based on these classification results, it is possible to more accurately select and set the image processing algorithm necessary for suitably corresponding to the inspection image.

[0092] In this embodiment, also, an X-ray is irradiated onto the item under inspection W to obtain an X-ray inspection image, and a predetermined image processing algorithm is applied to the X-ray inspection image to perform X-ray inspection on the quality state of the item under inspection W. Therefore, it becomes an item inspection apparatus capable of easily, quickly, and accurately selecting and setting the optimal image processing algorithm from a plurality of X-ray inspection image processing algorithms.

[0093] As described above, according to this embodiment, it is possible to provide the item inspection apparatus 1 that can easily select and set the accurate image processing algorithm corresponding to the item under inspection W from a plurality of image processing algorithms for item inspection.

[0094] (Other Embodiments) FIG. 7 shows an item inspection apparatus 1A according to another embodiment of the present invention.

[0095] As shown in the figure, the item inspection apparatus 1A of this embodiment is different from the item inspection apparatus 1 shown in FIG. 1 in that an algorithm evaluation determination unit 46 is further provided in the image processing control unit 34. However, other configurations are the same as those of the item inspection apparatus 1 of one embodiment. Therefore, in FIG. 7, the same components as those in the one embodiment shown in FIG. 1 are denoted by the same reference numerals as in FIG. 1. Hereinafter, the differences from the one embodiment will be described.

[0096] In the article inspection device 1A of the present embodiment, an algorithm evaluation determination unit 46 that evaluates the image processing algorithm selected by the user and determines the appropriateness of the selection is provided in the image processing control unit 34. When a predetermined type registration setting input is made to the control unit 30 by the display operation unit 35 and the test conveyance of the inspection object W for type registration is performed, any image processing algorithm candidate estimated to highly conform to the data of the input inspection image acquired by the detection image data acquisition unit 31 is extracted by the learning model 42 and operates when selected by the user.

[0097] During this operation, the algorithm evaluation determination unit 46 requests, for example, in the display screen to test-convey the inspection object W with a predetermined number of test pieces, and for each test-conveyed product, executes the image processing by the image processing unit 32 based on the selected image processing algorithm and the determination process by the inspection determination unit 33, and further has a function of evaluating and confirming the appropriateness of the selected image processing algorithm by, for example, displaying the result on the inspection result display unit 63b. Also, during the operation of the algorithm evaluation determination unit 46, the algorithm setting unit 45 can restore the image processing unit 32 to the state before the update without finalizing the update setting of the selected image processing algorithm.

[0098] The user can operate the update button 56 if the evaluation and determination result by the algorithm evaluation determination unit 46 is satisfactory after referring to the evaluation and determination result on the screen. However, not only that, when the evaluation values of the two image processing algorithms displayed in the ranking do not differ significantly, the user can operate the cancel button 57 to reselect another image processing algorithm whose evaluation value is next to the once-selected image processing algorithm.

[0099] Alternatively, by operating the setting / adjustment button 64d and adding an additional setting operation to finely adjust a part of the parameters when using the image processing algorithm selected by the algorithm setting unit 45, and then operating the update button 56, it becomes possible to finalize the update setting state.

[0100] Also in this embodiment, an algorithm evaluation determination unit 46 that can function as an image processing algorithm evaluation unit evaluates the suitability of the inspection of the inspection image data of the inspected article W based on performance information representing the inspection performance of the inspection applied to the data of a plurality of acquired images obtained by imaging a product group that is another article other than the inspected article W. For each, a plurality of evaluation values (scores) representing the respective evaluation results are calculated for a plurality of image processing algorithms. Then, based on the plurality of calculated evaluation values, the algorithm setting unit 45 sets a predetermined image processing algorithm used for determining the quality state of the inspected article W as the required image processing algorithm. Therefore, also in this embodiment, the same effects as those of the above-described one embodiment can be achieved.

[0101] Furthermore, in this embodiment, when an image processing algorithm that matches the data of the input inspection image of the new type of inspected article W is selected, prior to the update setting in the algorithm setting unit 45, the selected image processing algorithm can be evaluated by the algorithm evaluation determination unit 46 using a predetermined number of test articles, and its suitability can be enhanced.

[0102] In the above-described one embodiment, the article inspection apparatus 1 is an X-ray inspection method apparatus. However, the present invention is applicable to an article inspection apparatus of any other article inspection method that performs article inspection using imaging data. Also, as an example of an image processing algorithm, an image processing algorithm suitable for foreign object detection was exemplified. However, as described above, it may be one that inspects, for example, the presence or absence of defective products, the conformity of the shape, size, storage state, etc. of the contents, the distribution of density, thickness, volume, or mass, etc., such as the suitability of the quality and physical quantities required as a product for the inspected article W. Also, of course, the selection of the image processing algorithm may be accompanied by a selection operation made in units of inspection algorithms including the image processing algorithm used in the image processing unit 32 and the determination processing algorithm used in the inspection determination unit 33.

[0103] As described above, the present invention can provide an article inspection apparatus that can easily and accurately select and set an image processing algorithm suitable for an article to be inspected from a plurality of image processing algorithms, and is useful for all article inspection apparatuses that inspect the quality state of an article to be inspected by applying a predetermined image processing algorithm to an inspection image obtained by imaging an article of a predetermined type.

Explanation of Signs

[0104] 1, 1A Article inspection apparatus 10 Conveyor section 11 Conveyor belt 11a Upper running section 12, 13 Conveyor rollers 20 Inspection section (X-ray inspection section) 21 X-ray generator 22 X-ray tube 23 X-ray detector 30 Control section 31 Detection image data acquisition section (image input unit) 32 Image processing section 33 Inspection determination section 34 Image processing control section 35 Display operation section (display) 41 Learning section (image processing algorithm evaluation section) 42 Learning model 43 Learning data storage section 44 Algorithm storage section (image processing algorithm storage section) 45 Algorithm setting section (image processing algorithm setting section) 46 Algorithm evaluation determination section (image processing algorithm evaluation section) 50 Algorithm selection operation screen (touch panel, display) 51 Score display 52 Ranking column 53 Algorithm number column 54 Detailed explanation column 55 Score display column 56 Update button 57 Cancel button 61 Common information display section 62 Image display area 63 Inspection information display section 63a Variety number display section 63b Inspection result display section 63c Selection content display section 64 Operation input section 64a Menu button 64b Display switching button 64c Operation confirmation button 64d Setting / adjustment button 64e Stop button 64f Start button Dpx Imaging data (data of inspection images, photographed images, X-ray photographed images) Lx Detection data of line-scanned images W Object to be inspected Zx Inspection area

Claims

1. An article inspection apparatus that inspects the quality state of an inspection article (W) of a predetermined variety by applying a predetermined image processing algorithm (Pgm) to an inspection image (Dpx) obtained by imaging the inspection article, an image processing algorithm storage unit (44) that stores in advance a plurality of image processing algorithms including the predetermined image processing algorithm; for the plurality of image processing algorithms in the image processing algorithm storage unit, the suitability of the inspection for the inspection image is evaluated based on performance information representing the inspection performance when applied to the data of a plurality of acquired images obtained by imaging other articles other than the inspection article, and a plurality of evaluation values representing the respective evaluation results for the plurality of image processing algorithms are calculated by an image processing algorithm evaluation unit (41; 46); an image processing algorithm setting unit (45) that sets the predetermined image processing algorithm to be used for determining the quality state of the inspection article based on the plurality of evaluation values calculated by the image processing algorithm evaluation unit. The article inspection apparatus is characterized by comprising:

2. The image processing algorithm evaluation unit calculates the plurality of evaluation values so that the larger the numerical value, the more excellent the inspection performance represented by the performance information; The image processing algorithm setting unit selects a part of the plurality of image processing algorithms that are superior based on the numerical values of the plurality of evaluation values, and sets the predetermined image processing algorithm. The article inspection apparatus according to claim 1, characterized in that:

3. The plurality of evaluation values are calculated so as to have a maximum value when the performance information represents an inspection performance that is superior at a predetermined level or higher; The image processing algorithm setting unit causes a display to display a predetermined number of the plurality of image processing algorithms in descending order of superiority of the plurality of evaluation values in a selectable manner. The article inspection apparatus according to claim 1 or 2, characterized in that:

4. The image processing algorithm evaluation unit: evaluates the suitability of the plurality of image processing algorithms for the inspection image based on a learning model created in advance by learning using inspection images of a plurality of types of product groups that are the other articles and image processing algorithms that match the inspection images of the product groups as teacher data. The article inspection apparatus according to claim 1 or 2, characterized in that:

5. The image processing algorithm setting unit extracts at least one image processing algorithm candidate according to the evaluation values of the plurality of image processing algorithms based on the learning model in the image processing algorithm evaluation unit, and when there are a plurality of the extracted image processing algorithm candidates, the extracted image processing algorithm candidates are displayed on a display in a selectable manner. The article inspection apparatus according to claim 4, characterized in that.

6. The article inspection apparatus according to claim 4, further comprising product group storage means for storing images of a product group associated with the optimal image processing algorithm among the plurality of image processing algorithms as learning data, and a learning unit for generating the learning model based on the learning data.

7. The article inspection apparatus according to claim 4, characterized in that the learning model is created using a deep learning classification method.

8. The article inspection apparatus according to claim 1, characterized in that the article to be inspected is irradiated with X-rays to obtain an X-ray inspection image, and the quality state of the article to be inspected is inspected by applying the predetermined image processing algorithm to the X-ray inspection image.

Images