WO2020209312A1

WO2020209312A1 - Inspecting device and inspecting method

Info

Publication number: WO2020209312A1
Application number: PCT/JP2020/015888
Authority: WO
Inventors: 聖正岡
Original assignee: 朝日レントゲン工業株式会社
Priority date: 2019-04-09
Filing date: 2020-04-08
Publication date: 2020-10-15
Also published as: JP7178620B2; JPWO2020209312A1

Abstract

This inspecting device is provided with a first image generating unit, a reconstructing unit, an identifying unit, and a removing unit. The first image generating unit generates a plurality of two-dimensional X-ray images in which a positional relationship between an object being inspected and an X-ray radiating unit and an X-ray detecting unit employed for X-ray imaging differ. The reconstructing unit generates images in which the plurality of two-dimensional X-ray images are projected respectively onto a plurality of slices at different depths, and a projected image is then reconstructed from each of the generated images. The identifying unit converts each of the plurality of projected images into a binarized projected image in which the position of a foreign body has been identified. The removing unit compares a plurality of the binarized projected images obtained from images captured in slices having the same depth, and removes erroneously detected parts of the position identification of the foreign body on the basis of the comparison results.

Description

Inspection equipment and inspection method

The present invention relates to a technique for detecting a foreign substance that may be contained in an object to be inspected.

Conventionally, various inspection devices have been developed to detect foreign substances that may be contained in the object to be inspected. For example, the inspection apparatus disclosed in Patent Document 1 detects an highlighted line from an X-ray transmission image of an object to be inspected, and detects a foreign substance based on the highlighted line.

Japanese Unexamined Patent Publication No. 2016-156647

However, if the difference in X-ray attenuation between the object to be inspected and the foreign matter is small, it is difficult to detect the emphasized line, and as a result, there arises a problem that the detection accuracy of the foreign matter is lowered. Examples of cases where the difference in X-ray attenuation between the object to be inspected and the foreign substance is small include a case where the object to be inspected is a fish body and the foreign substance is a small bone.

An object of the present invention is to provide an inspection device and an inspection method having high foreign matter detection accuracy in view of the above situation.

In order to achieve the above object, the inspection apparatus according to the present invention targets an object to be inspected for X-ray photography and generates a two-dimensional X-ray image based on the detection result of the X-ray detection unit used for the X-ray photography. A first image generation unit and an image obtained by projecting the two-dimensional X-ray image onto a fault set at a predetermined position other than the position of the X-ray detection unit are generated, and then the generated image is two-dimensional. A reconstruction unit that reconstructs a projected image, which is an image, a specific unit that identifies the position of a foreign object based on the projected image, and a removal unit are provided. A plurality of the two-dimensional X-ray images having different positional relationships between the X-ray irradiation unit used for the X-ray photography and the X-ray detection unit are generated, and the reconstruction unit generates the plurality of the two-dimensional X-ray images. An image obtained by projecting each of the captured images onto a plurality of faults having different depths is generated, and then each of the generated images is reconstructed into the projected image, and the specific unit uses each of the plurality of the projected images as a foreign substance. Converted to a position-specified binarized projection image, the removal unit compares a plurality of the binarized projection images obtained from images projected on a tomography of the same depth with each other, and based on the comparison result. The configuration is such that a false detection portion for specifying the position of a foreign object is removed (first configuration).

In the inspection device having the first configuration, the specific unit identifies the position of a foreign substance by using artificial intelligence learned from the projected image of the object to be inspected in the past (second configuration). It may be.

Even if the inspection device having the first or second configuration includes a second image generation unit that generates an image that two-dimensionally displays the position of the foreign matter specified by the specific unit (third configuration). Good.

Even if the inspection device having the first or second configuration includes a second image generation unit that generates an image that three-dimensionally displays the position of the foreign matter specified by the specific unit (fourth configuration). Good.

In the inspection device having any of the first to fourth configurations, the specific unit includes a division unit that divides the projected image into predetermined regions and a calculation unit that calculates an average luminance value in each of the predetermined regions. When the luminance value of the pixel of interest is smaller than the average luminance value of the predetermined region to which the pixel of interest belongs by a predetermined value or more, the number of the first set number arranged on one side of the first direction of the pixel of interest. The first luminance value that becomes the maximum in the pixel is calculated, the second luminance value that becomes the maximum in the second set number of pixels arranged on the other side in the first direction of the pixel of interest is calculated, and the first luminance value of the pixel of interest is calculated. The maximum third luminance value is calculated within the third set number of pixels arranged on one side of the second direction, which is a direction different from the direction, and within the fourth set number of pixels arranged on the other side of the second direction of the pixel of interest. When the predetermined condition is satisfied, the pixel of interest is specified as the position of the foreign matter, and the predetermined condition is (a) the pixel of interest with respect to the first luminance value. The first condition that the ratio of the brightness value and the ratio of the brightness value of the pixel of interest to the second brightness value are equal to or less than the first threshold value, (b) the ratio of the brightness value of the pixel of interest to the third brightness value and The configuration is one of the second condition that the ratio of the brightness value of the pixel of interest to the fourth brightness value is equal to or less than the second threshold value, and (c) the first condition and the second condition. (Fifth configuration) may be used.

In the inspection device having any of the first to fourth configurations, the specific portion includes a division portion that divides an image based on X-ray photography of an object to be inspected into predetermined regions and an average brightness in each of the predetermined regions. A calculation unit for calculating a value is provided, and when the luminance value of the pixel of interest is smaller than the average luminance value in the predetermined region to which the pixel of interest belongs by a predetermined value or more, one side of the pixel of interest in the first direction. Within the first set number of pixels lined up in, the first brightness value, which is the brightness value of a pixel having a larger brightness value than the focus pixel and as far as possible from the focus pixel, or the brightness value of a pixel as close as possible to the focus pixel, is calculated. Then, among the second set number of pixels arranged on the other side of the first direction of the focus pixel, the brightness value of the pixel having a larger brightness value than the focus pixel and as far as possible from the focus pixel, or a pixel as close as possible to the focus pixel. The second luminance value, which is the luminance value of the above, is calculated, and the luminance value is larger than that of the pixel of interest and the luminance value is larger than that of the pixel of interest within the third set number of pixels arranged on one side of the second direction, which is a direction different from the first direction of the pixel of interest. The third brightness value, which is the brightness value of the pixel as far as possible from the focus pixel or the brightness value of the pixel as close as possible to the focus pixel, is calculated, and within the fourth set number of pixels arranged on the other side of the second direction of the focus pixel. If the fourth brightness value, which is the brightness value of a pixel having a brightness value larger than that of the pixel of interest and as far away as possible from the pixel of interest or the brightness value of a pixel as close as possible to the pixel of interest, is satisfied. The pixel of interest is specified as the position of a foreign object, and the predetermined conditions are (a) the ratio of the brightness value of the pixel of interest to the first luminance value and the ratio of the luminance value of the pixel of interest to the second luminance value. The first condition that each of the above is equal to or less than the first threshold value, (b) the ratio of the brightness value of the pixel of interest to the third brightness value and the ratio of the brightness value of the pixel of interest to the fourth brightness value are the second threshold values, respectively. It may be a configuration (sixth configuration) that is any one of the second condition (c) the first condition and the second condition.

In the inspection device having any of the first to fourth configurations, the specific portion includes a division portion that divides an image based on X-ray photography of an object to be inspected into predetermined regions and an average brightness in each of the predetermined regions. A calculation unit for calculating a value is provided, and when the luminance value of the pixel of interest is smaller than the average luminance value in the predetermined region to which the pixel of interest belongs by a predetermined value or more, one side of the pixel of interest in the first direction. The first is the average luminance value of the first set number of pixels lined up at a first predetermined position away from the focus pixel and the second set number of pixels lined up at a second predetermined position on the other side of the first direction of the focus pixel. The brightness value is calculated, and the third set number of pixels lined up on one side of the second direction, which is a direction different from the first direction of the pixel of interest, at a third predetermined position away from the pixel of interest, and the other of the second direction of the pixel of interest. The second luminance value, which is the average luminance value of the fourth set number of pixels arranged on the side at a fourth predetermined position away from the pixel of interest, is calculated, and if the predetermined condition is satisfied, the pixel of interest is specified as the position of the foreign matter. The predetermined conditions are (a) the first condition that the ratio of the brightness value of the pixel of interest to the first luminance value is equal to or less than the first threshold value, and (b) the pixel of interest with respect to the second luminance value. It may be a configuration (seventh configuration) in which one of the second condition (c) the first condition and the second condition that the ratio of the luminance values is equal to or less than the second threshold value. ..

In the inspection device having any of the first to fourth configurations, the specific portion includes a division portion that divides an image based on X-ray photography of an object to be inspected into predetermined regions and an average brightness in each of the predetermined regions. A calculation unit for calculating a value is provided, and when the luminance value of the pixel of interest is greater than a predetermined value or more than the average luminance value in the predetermined region to which the pixel of interest belongs, one side of the pixel of interest in the first direction. Calculate the minimum first luminance value within the first set number of pixels lined up in, and calculate the smallest second luminance value within the second set number of pixels lined up on the other side of the first direction of the pixel of interest. , Calculates the minimum third luminance value among the third set number of pixels arranged on one side of the second direction, which is a direction different from the first direction of the pixel of interest, and arranges them on the other side of the second direction of the pixel of interest. The fourth luminance value, which is the minimum among the fourth set number of pixels, is calculated, and if the predetermined condition is satisfied, the pixel of interest is specified as the position of the foreign matter, and the predetermined condition is (a) the first. The first condition that the ratio of the brightness value of the pixel of interest to the luminance value and the ratio of the luminance value of the pixel of interest to the second luminance value are equal to or higher than the first threshold value, (b) the focus on the third luminance value. Of the second condition that the ratio of the brightness value of the pixel and the ratio of the brightness value of the pixel of interest to the fourth brightness value are equal to or higher than the second threshold value, (c) the first condition and the second condition. It may be a configuration (eighth configuration) which is any one.

In the inspection device having any of the first to fourth configurations, the specific portion includes a division portion that divides an image based on X-ray photography of an object to be inspected into predetermined regions and an average brightness in each of the predetermined regions. A calculation unit for calculating a value is provided, and when the luminance value of the pixel of interest is greater than a predetermined value or more than the average luminance value in the predetermined region to which the pixel of interest belongs, one side of the pixel of interest in the first direction. Within the first set number of pixels lined up in, the first brightness value, which is the brightness value of a pixel whose brightness value is smaller than that of the attention pixel and is as far as possible from the attention pixel or the brightness value of a pixel as close as possible to the attention pixel is calculated. Then, among the second set number of pixels arranged on the other side of the first direction of the focus pixel, the brightness value of the pixel having a brightness value smaller than that of the focus pixel and as far as possible from the focus pixel, or a pixel as close as possible to the focus pixel. The second luminance value, which is the luminance value of the above, is calculated, and the luminance value is smaller than that of the pixel of interest and the luminance value is smaller than that of the pixel of interest within the third set number of pixels arranged on one side of the second direction, which is a direction different from the first direction of the pixel of interest. The third brightness value, which is the brightness value of the pixel as far as possible from the focus pixel or the brightness value of the pixel as close as possible to the focus pixel, is calculated, and within the fourth set number of pixels arranged on the other side of the second direction of the focus pixel. If the fourth brightness value, which is the brightness value of a pixel having a brightness value smaller than that of the pixel of interest and as far as possible from the pixel of interest or the brightness value of a pixel as close as possible to the pixel of interest, is satisfied. The pixel of interest is specified as the position of a foreign object, and the predetermined conditions are (a) the ratio of the brightness value of the pixel of interest to the first luminance value and the ratio of the luminance value of the pixel of interest to the second luminance value. The first condition that each of the above is equal to or higher than the first threshold value, (b) the ratio of the brightness value of the pixel of interest to the third brightness value and the ratio of the brightness value of the pixel of interest to the fourth brightness value are the second threshold values, respectively. The configuration may be one of the second condition (c) the first condition and the second condition (the ninth configuration).

In the inspection device having any of the first to fourth configurations, the specific portion includes a division portion that divides an image based on X-ray photography of an object to be inspected into predetermined regions and an average brightness in each of the predetermined regions. A calculation unit for calculating a value is provided, and when the luminance value of the pixel of interest is greater than a predetermined value or more than the average luminance value in the predetermined region to which the pixel of interest belongs, one side of the pixel of interest in the first direction. The first is the average luminance value of the first set number of pixels lined up at a first predetermined position away from the focus pixel and the second set number of pixels lined up at a second predetermined position on the other side of the first direction of the focus pixel. The brightness value is calculated, and the third set number of pixels lined up on one side of the second direction, which is a direction different from the first direction of the pixel of interest, at a third predetermined position away from the pixel of interest, and the other of the second direction of the pixel of interest. The second luminance value, which is the average luminance value of the fourth set number of pixels arranged on the side at a fourth predetermined position away from the pixel of interest, is calculated, and if the predetermined condition is satisfied, the pixel of interest is specified as the position of the foreign matter. The predetermined conditions are (a) the first condition that the ratio of the brightness value of the pixel of interest to the first luminance value is equal to or higher than the first threshold value, and (b) the pixel of interest with respect to the second luminance value. It may be a configuration (10th configuration) in which one of the second condition (c) the first condition and the second condition that the ratio of the luminance values becomes equal to or higher than the second threshold value. ..

In order to achieve the above object, the inspection method according to the present invention is an inspection method using an X-ray imaging device including an X-ray irradiation unit and an X-ray detection unit, and X-rays an object to be inspected by the X-ray imaging apparatus. A storage step for storing a two-dimensional X-ray image obtained by radiography and an image obtained by projecting the two-dimensional X-ray image onto a fault set at a predetermined position other than the position of the X-ray detection unit are generated. Then, the storage step includes a reconstruction step of reconstructing the generated image into a projected image which is a two-dimensional image, a specific step of identifying the position of a foreign object based on the projected image, and a removal step. In the reconstruction step, a plurality of the two-dimensional X-ray images having different positional relationships between the object to be inspected, the X-ray irradiation unit used for the X-ray photography, and the X-ray detection unit are stored. Images obtained by projecting a plurality of the two-dimensional X-ray images onto a plurality of faults having different depths are generated, and then each of the generated images is reconstructed into the projected images, and in the specific step, a plurality of images are generated. Each of the projected images is converted into a binarized projected image in which the position of the foreign substance is specified, and in the removal step, the plurality of the binarized projected images obtained from the images projected on the tomography of the same depth are used. Are compared, and the erroneous detection portion for specifying the position of the foreign matter is removed based on the comparison result (11th configuration).

According to the present invention, the detection accuracy of foreign matter can be improved.

The figure which shows the schematic structure of the inspection apparatus which concerns on one Embodiment of this invention. A flowchart showing the schematic operation of the inspection device according to the embodiment of the present invention. Flowchart showing an example of processing to reconstruct a projected image Diagram showing the coordinate transformation of the system Perspective view showing an example of the positional relationship between a group of pixels which is a part of the detection surface of the X-ray detector and four voxels. Top view showing the positional relationship shown in FIG. 5A Perspective view showing an example of the positional relationship between a pixel group which is a part of the detection surface of the X-ray detector and one voxel. Top view showing the positional relationship shown in FIG. 6A The figure which shows the thickness of a voxel with respect to the X-ray incident direction Diagram showing an example of an oblique quadrangular prism that approximates a voxel Perspective view showing a rectangle formed in a voxel Figure showing another example of an oblique prism that approximates a voxel Figure showing another example of an oblique prism that approximates a voxel Figure showing another example of an oblique prism that approximates a voxel Figure showing another example of an oblique prism that approximates a voxel The figure which shows an example of the case where the X-ray transmitted through a voxel is obliquely incident on the detection surface of the X-ray detector in the Z-axis direction. Flow chart showing an example of processing to identify the position of foreign matter for each projected image A diagram in which voxels with predetermined positions on multiple faults are arranged in a straight line for each fault for convenience. Flowchart showing an example of position correction The figure which shows the 1st modification of the inspection apparatus The figure which shows the 2nd modification of the inspection apparatus The figure which shows the 2nd modification of the inspection apparatus The figure which shows the 3rd modification of the inspection apparatus The figure which shows the 3rd modification of the inspection apparatus

An embodiment of the present invention will be described below with reference to the drawings. In the present specification, as a definition of a word, in a two-dimensional image generated by a tom set at a predetermined position excluding the position of an X-ray detection unit, at least one of the X-axis direction and the Y-axis direction is used. For convenience, the 2D image produces an image that is a "projection" of the 2D X-ray image onto the fault, even if the 2D X-ray image is simply moved in parallel and the length does not change. , It shall be obtained by the process of reconstructing the generated image, and the two-dimensional image shall be referred to as a "projected image".

<1. Outline configuration of inspection equipment>
FIG. 1 is a diagram showing a schematic configuration of an inspection device according to an embodiment of the present invention. The inspection device 100 shown in FIG. 1 includes a first X-ray irradiation unit 1A, a second X-ray irradiation unit 1B, a first X-ray detection unit 2A, a second X-ray detection unit 2B, a belt conveyor 3, a CPU 4, and a ROM 5. , RAM 6, VRAM 7, display unit 8, HDD 9, and input unit 10. In the present embodiment, the image processing device for processing the image is composed of the CPU 4, the ROM 5, the RAM 6, and the HDD 9.

The first X-ray irradiation unit 1A and the second X-ray irradiation unit 1B each irradiate the object T1 to be inspected with X-rays. The X-rays emitted from the first X-ray irradiation unit 1A and the X-rays emitted from the second X-ray irradiation unit 1B each have a fan beam shape extending along the Y axis, and more specifically, a narrow fan beam shape. The first X-ray irradiation unit 1A and the second X-ray irradiation unit 1B may be shared to form a single X-ray irradiation unit. The X-rays emitted from the single X-ray irradiation unit may have a wide fan beam shape or a cone beam shape.

The X-ray irradiation direction from the first X-ray irradiation unit 1A to the first X-ray detection unit 2A and the X-ray irradiation direction from the second X-ray irradiation unit 1B to the second X-ray detection unit 2B are different from each other. .. In the present embodiment, the irradiation direction of the X rays emitted from the first X-ray irradiation unit 1A to the first X-ray detection unit 2A is the direction orthogonal to the X-axis and the Y-axis, and the second X-ray irradiation unit 1B to the second X-ray. The irradiation direction of the X-rays emitted to the detection unit 2B is a direction inclined from the direction orthogonal to the X-axis and the Y-axis.

The first X-ray detection unit 2A and the second X-ray detection unit 2B each output a digital amount of electric signals corresponding to the incident X-rays at a constant frame rate. The first X-ray detection unit 2A and the second X-ray detection unit 2B are line sensors extending along the Y-axis, respectively. Since the inspection device 100 employs the tomosynthesis method, the first X-ray detection unit 2A and the second X-ray detection unit 2B each have a plurality of X-ray detection elements in the X-axis direction as well. The first X-ray detection unit 2A and the second X-ray detection unit 2B can each collect incident X-rays at a predetermined frame rate as image data of a digital electric amount corresponding to the amount of the X-rays. Hereinafter, this collected data is referred to as "frame data" (an example of a two-dimensional X-ray photographed image). The first X-ray detection unit 2A and the second X-ray detection unit 2B may be shared to form a single X-ray detection unit. However, when the first X-ray irradiation unit 1A and the second X-ray irradiation unit 1B are shared, the first X-ray detection unit 2A and the second X-ray detection unit 2B are not shared.

The belt conveyor 3 is an object to be inspected placed on the belt with respect to the pair of the first X-ray irradiation unit 1A and the first X-ray detection unit 2A and the pair of the second X-ray irradiation unit 1B and the second X-ray detection unit 2B. Move T1 toward the negative side of the X-axis. That is, the longitudinal direction of the belt, which is a component of the belt conveyor 3, is along the X axis, and the width direction of the belt, which is a component of the belt conveyor 3, is along the Y axis. In the present embodiment, the object to be inspected T1 is placed on the X-axis with respect to the pair of the first X-ray irradiation unit 1A and the first X-ray detection unit 2A and the pair of the second X-ray irradiation unit 1B and the second X-ray detection unit 2B. The first moving mechanism (belt conveyor 3) that moves toward the negative side was used, but instead of the first moving mechanism, the pair of the first X-ray irradiation unit 1A and the first X-ray detection unit 2A and the second X-ray irradiation unit 1B A second movement mechanism that moves the pair of the second X-ray detection unit 2B and the second X-ray detection unit 2B toward the positive side of the X-axis with respect to the object T1 to be inspected may be used.

The CPU 4 controls the entire inspection device 100 according to the programs and data stored in the ROM 5 and the HDD 9. ROM 5 records fixed programs and data. The RAM 6 provides working memory. The CPU operates so as to perform a function of generating an image according to a program stored in the HDD 9. That is, the CPU 41 also serves as an image generation unit that generates an image.

VRAM 7 temporarily stores image data. The display unit 8 displays an image based on the image data stored in the VRAM 7.

The HDD 9 has various types of radiography control programs for controlling X-ray photography operations, an image reconstruction processing program for generating a reconstructed image, a foreign matter position identification processing program for identifying the position of a foreign matter, a position correction program, and the like. Stores various data such as setting values of various parameters and image data used when executing a program and various programs.

The input unit 10 is, for example, a keyboard, a pointing device, or the like, and inputs the content of the user operation.

<2. Outline operation of inspection equipment>
The schematic operation of the inspection device 100 will be described with reference to the flowchart of FIG. First, the inspection device 100 performs X-ray imaging (step S1). Specifically, while the belt conveyor 3 is moving the object to be inspected T1, X-rays are exposed from the first X-ray irradiation unit 1A and the second X-ray irradiation unit 1B. The X-rays emitted from the first X-ray irradiation unit 1A pass through the imaging region of the object to be inspected T1 and enter the first X-ray detection unit 2A, and the X-rays emitted from the second X-ray irradiation unit 1B are the objects to be inspected. It passes through the imaging region of T1 and is incident on the second X-ray detection unit 2B. As described above, the first X-ray detection unit 2A and the second X-ray detection unit 2B detect incident X-rays at a predetermined frame rate and sequentially output two-dimensional digital data of the corresponding digital electric amount in frame units. .. This frame data is stored in the HDD 9.

Next, the inspection device 100 generates a projected image obtained by projecting onto a tomography set at a predetermined position excluding the positions of the

X-ray detection units

2A and 2B (step S2). Specifically, the inspection device 100 generates an image in which the frame data is projected onto each of the 60 layers of tomography set at predetermined positions other than the positions of the

X-ray detectors

2A and 2B, and then the generated image is generated. Is reconstructed into a two-dimensional image (projected image). As a method of projecting frame data onto a certain fault to obtain one projected image, for example, the frame data is projected onto the central plane in the depth direction (one fault plane) of the certain fault. A method of obtaining the one projected image, a method of projecting the frame data on the uppermost surface (one tomographic plane) in the depth direction of the one fault, and obtaining the one projected image, the above frame data A method of obtaining the one projected image by projecting onto the lowest surface (one fault plane) in the depth direction of one fault, and projecting the frame data on each of a plurality of fault planes included in the one fault. Examples thereof include a method of obtaining the above-mentioned one projected image by synthesizing a plurality of projected images (for example, simple averaging process, weighted averaging process, etc.). In the present embodiment, the projected image obtained by projecting onto each tomographic image set at a predetermined position other than the positions of the

X-ray detectors

2A and 2B is a projected image on each tomographic image obtained based on the principle of tomosynthesis. .. In the present embodiment, the direction perpendicular to the X-axis and the Y-axis is the depth direction of the fault, and the first X-ray detection unit 2A and the second X-ray detection unit 2B are from the first X-ray irradiation unit 1A and the second X-ray irradiation unit 1B side. A 60-layer fault is set at a predetermined position excluding the positions of the

X-ray detectors

2A and 2B at a pitch of 0.5 mm on the side. In the present embodiment, the predetermined positions other than the positions of the

X-ray detection units

2A and 2B include the position of the object to be inspected T1, but the present invention is not limited thereto.

The 60-layer projected image derived from the first X-ray irradiation unit 1A and the first X-ray detection unit 2A is stored in the HDD 9. Similarly, the projected images of the 60 layers derived from the second X-ray irradiation unit 1B and the second X-ray detection unit 2B are also stored in the HDD 9.

Next, the inspection device 100 identifies the position of the foreign matter for each projected image (step S3). The details of the method for identifying the position of the foreign matter will be described later. By specifying the position of the foreign matter for each projected image, the detection accuracy of the foreign matter can be improved as compared with the case where the position of the foreign matter is specified for an X-ray image such as a simple X-ray transmission image. The result of specifying the foreign matter position can be, for example, a binarized image showing the position of the foreign matter and the position other than the foreign matter with different luminance values.

Next, the inspection device 100 removes the erroneous detection portion for specifying the position of the foreign matter (step S4). Specifically, the inspection device 100 includes a projection image derived from the first X-ray irradiation unit 1A and the first X-ray detection unit 2A and a projection image derived from the second X-ray irradiation unit 1B and the second X-ray detection unit 2B at the same depth. The image is specified based on the position of the foreign matter specified based on the projected images derived from the first X-ray irradiation unit 1A and the first X-ray detection unit 2A and the projected images derived from the second X-ray irradiation unit 1B and the second X-ray detection unit 2B. The position of the foreign matter is compared with that of the foreign matter, and the erroneous detection portion of the foreign matter is removed based on the comparison result. More specifically, the inspection device 100 adopts only the pixel identified as the position of the foreign matter in both projected images as the position of the foreign matter by the above comparison, and at the position of the foreign matter in only one projected image. Do not use the identified pixel as the position of the foreign object. In the process of step S4, the foreign matter existing in the fault to be processed is detected at the same coordinate position of both projected images even if the irradiation angle of X-rays is different, whereas it is present in the fault to be processed. This is an erroneous detection partial removal process utilizing the fact that when foreign matter or the like that is not projected is projected in the X-ray irradiation direction, it is detected at different coordinate positions of both projected images. The inspection device 100 executes this false detection partial removal process on all faults.

Next, the inspection device 100 corrects the position (step S5). The details of the position correction will be described later.

Finally, the inspection device 100 generates an output image and displays the output image on the display unit 8 (step S6). As the output image, for example, an image obtained by adding all the projected images showing the positions of the foreign objects reflecting the false detection portion removal process and the position correction, that is, an image that displays the positions of the foreign objects in two dimensions can be mentioned. it can. As another example of the output image, there is an image obtained by stacking each projection image showing the position of the foreign matter reflecting the false detection portion removal process and the position correction, that is, an image that displays the position of the foreign matter in three dimensions. Can be done.

<3. Projected image>
An example of the process of step S2 described above, that is, a process of generating an image obtained by projecting frame data onto a tomographic image and then reconstructing the generated image into a projected image will be described with reference to the flowchart of FIG.

First, the CPU 4 reads the defect registration data and creates a defect table (step S11).

Next, the CPU 4 reads the density correction image and creates the density correction data (step S12). Note that, unlike the present embodiment, steps S11 and S12 may be executed before step S1.

The CPU 4 reads the projection data (frame data) (step S13), and performs defect correction and density correction on the projection data (step S14).

In each projection data read in step S13, regarding the projection data derived from the first X-ray irradiation unit 1A and the first X-ray detection unit 2A, only the voxels through which the X-rays exposed from the first X-ray irradiation unit 1A are transmitted. The calculation may be performed, and the projection data derived from the second X-ray irradiation unit 1B and the second X-ray detection unit 2B need to be calculated only for the voxels through which the X-rays exposed from the second X-ray irradiation unit 1B are transmitted. .. Since the range of voxels through which X-rays are transmitted is narrow in each projection data, it is possible to shorten the calculation time by performing the calculation only for the voxels through which X-rays are transmitted. Therefore, the CPU 4 may set the range of voxels to be calculated at each fault for each frame data in advance according to the size of the object T1 to be inspected.

Next, the CPU 4 calculates the actual position coordinates of each voxel vertex occupying the reconstructed area (step S15).

Subsequently, the CPU 4 sequentially reads the projection data obtained in the data collection process in step S13 for each frame (step S16). In the process of one step S16, one frame of projection data is read.

Next, the CPU 4 convolves and integrates the projection data and the filter function (step S17).

After that, in order to simplify the calculation, the CPU 4 performs coordinate conversion of the system so as to be the coordinate system shown in FIG. 4 for each calculation result of the convolution integral (step S18). Specifically, regarding the calculation result of the convolution integral derived from the second X-ray irradiation unit 1B and the second X-ray detection unit 2B, the second X-ray irradiation unit 1B is the origin and the center position of the second X-ray detection unit 2B is on the Z axis. The system including the second X-ray irradiation unit 1B, the reconstruction region R1, and the second X-ray detection unit 2B is rotationally moved and translated so as to be in the positive direction of. Here, the Z-axis is an axis orthogonal to the X-axis and the Y-axis, and the direction from the first X-ray irradiation unit 1A to the first X-ray detection unit 2A is the positive direction of the Z-axis. Regarding the calculation result of the convolution integral derived from the first X-ray irradiation unit 1A and the first X-ray detection unit 2A, since the coordinate system is originally shown in FIG. 4, the coordinate conversion of the system is not performed.

Next, the reconstruction calculation using the FBP (Filtered Back Projection) method executed by the CPU 4 in step S19 will be described with reference to FIGS. 5A to 13. In the following description, when it is not necessary to distinguish between the first X-ray irradiation unit 1A and the second X-ray irradiation unit 1B, they are referred to as the X-ray irradiation unit 1, and the first X-ray detection unit 2A and the second X-ray detection unit 2B are referred to. When it is not necessary to distinguish between them, they are referred to as an X-ray detector 2.

First, pay attention to a certain pixel on the detection surface of the X-ray detection unit 2, and consider a case where X-rays from the X-ray irradiation unit 1 are incident on the pixel of interest.

The X-rays incident on the pixel of interest are attenuated when passing through the subject, and the X-rays captured by the X-ray detection unit 2 are reduced. That is, when the reconstruction region R1 shown in FIG. 4 composed of a plurality of voxels is set, X-rays transmitted through the voxels are incident on the pixel of interest, and the X-rays are attenuated by the amount of the voxels transmitted by the X-rays. It will be reflected in the brightness value of the pixel of interest.

When the X-rays incident on the pixel of interest pass through a part of multiple voxels in the fault, the X-rays are attenuated according to the volume ratio of the part where the X-rays of each voxel are transmitted, and the degree of attenuation is the degree of attenuation of the pixel of interest. It is reflected in the brightness value.

That is, each voxel transmitted with X-rays incident on the pixel of interest contributes to the brightness value of the pixel of interest at the ratio of the volume ratio of the portion of each voxel transmitted with X-rays. Conversely, if the brightness value of the pixel of interest is divided by the volume ratio of the part where the X-rays of each voxel transmitted to the pixel of interest are transmitted, the X-rays of each voxel are transmitted at each divided value. Corresponds to the X-ray attenuation of the affected part.

FIG. 5A is a perspective view showing an example of the positional relationship between the pixel group PXG, which is a part of the detection surface of the X-ray detection unit 2, and the voxels VX1 to VX4. FIG. 5B is a top view showing the positional relationship shown in FIG. 5A. The pixel group PXG is composed of pixels PX1 to PX16. In the positional relationship shown in FIGS. 5A and 5B, assuming that the pixel of interest is pixel PX11, voxels VX1 to VX4 become voxels through which X-rays incident on the pixel of interest have passed.

On the other hand, the X-ray attenuation reflected in the pixel of interest, that is, the brightness value of the pixel of interest, corresponds to the brightness value of the pixel of interest and the total volume of the portion of all voxels in which the X-rays incident on the pixel of interest are transmitted. The product of the product of the volume ratio of the X-ray-transmitted portion of one voxel in all voxels is summed for each voxel in the whole voxel. Therefore, when focusing on a voxel, the X-rays that pass through the voxel of interest are incident on a plurality of pixels, so that the effect of X-ray attenuation on the voxel of interest is applied to each pixel according to the volume ratio corresponding to each pixel. Will be given. That is, for each pixel, the ratio (volume ratio) of the effect of the voxel of interest on that pixel among the affected X-ray attenuation is taken as the multiplication value with the brightness value, and the multiplication value is integrated for each pixel. By doing so, it is possible to obtain the X-ray attenuation of the entire voxel of interest. That is, the back projection to the voxel of interest can be obtained.

FIG. 6A is a perspective view showing an example of the positional relationship between the pixel group PXG, which is a part of the detection surface of the X-ray detection unit 2, and the voxel VX1. FIG. 6B is a top view showing the positional relationship shown in FIG. 6A. In the positional relationship shown in FIGS. 6A and 6B, assuming that the voxel of interest is voxel VX1, the X-ray attenuation of the entire voxel VX1 of interest is (1) the brightness value of the pixel PX5 and the pixel of the voxel VX1 of interest with respect to the volume of the voxel VX1 of interest. Multiplying value by the ratio of the volume of the part where the X-ray incident on the PX5 is transmitted, (2) the brightness value of the pixel PX6 and the part where the X-ray incident on the pixel PX6 of the voxel VX1 of interest is transmitted to the volume of the voxel VX1 of interest. Multiplying value with the volume ratio of, (3) Multiplying value of the brightness value of the pixel PX7 with the volume ratio of the portion through which the X-ray transmitted to the pixel PX7 of the voxel VX1 of interest is transmitted to the volume of the voxel VX1 of interest, ( 4) Multiplying the brightness value of pixel PX9 and the ratio of the volume of the portion through which X-rays incident on pixel PX9 of voxel VX1 of interest to the volume of voxel VX1 of interest, (5) the brightness value of pixel PX10 and attention Multiplying the volume of the voxel VX1 with respect to the volume of the portion through which the X-ray incident on the pixel PX10 of the voxel VX1 is transmitted, and (6) the brightness value of the pixel PX11 and the volume of the voxel VX1 of interest with respect to the volume of the voxel VX1 It is the sum of the multiplication value with the ratio of the volume of the portion through which the X-ray incident on the pixel PX11 is transmitted.

Specifically, since the X-rays that pass through the voxels travel straight, the detection surface of the X-ray detection unit 2 is projected to the voxel position in the X-ray incident direction for each voxel. On the contrary, for each voxel, the voxel may be projected to the position of the detection surface of the X-ray detection unit 2 in the X-ray incident direction.

If none of the constituent planes of a rectangular parallelepiped voxel is perpendicular to the X-ray incident direction, for example, as shown in FIG. 7, the thickness of the voxel with respect to the X-ray incident direction is not uniform within the voxel VX1, depending on the pixel. X-rays that have passed through the thin part of the voxel may be incident, and if this is calculated exactly, the calculation time for the back projection will be enormous.

Therefore, in the present embodiment, the rectangular parallelepiped voxel is approximated to an oblique quadrangular prism having the same volume as the rectangular parallelepiped voxel and having a uniform thickness with respect to the X-ray incident direction, and back projection is performed. The oblique quadrangular prism has two bottom surfaces by translating a pair of facing constituent surfaces having overlapping regions when viewed from the X-ray incident direction on a plane including both or one of the opposing constituent surfaces. It has the same volume as a rectangular parallelepiped voxel and has a uniform thickness with respect to the X-ray incident direction. Even if such an approximation is performed, the image quality of the projected image is hardly affected.

FIG. 8A is a diagram showing an example of an oblique quadrangular prism that approximates voxel VX1. The oblique quadrangular prism OP1 shown in FIG. 8A has rectangles RT2 and RT3 as bottom surfaces, and has a uniform thickness with respect to the X-ray incident direction. The rectangle RT1 is a side of each side of the voxel VX1 of interest other than the side included in the pair of facing constituent surfaces having overlapping regions when viewed from the X-ray incident direction among the constituent surfaces of the voxel VX1 shown in FIGS. 6A and 6B. It is a rectangle with the midpoint as the apex. A perspective view of the rectangle RT1 formed in the voxel VX1 of interest is as shown in FIG. 8B. The rectangular RT2 detects the X-rays on the constituent planes of the voxel VX1 shown in FIGS. 6A and 6B, which are closer to the X-ray detector 2 on the pair of opposed constituent planes having overlapping regions when viewed from the X-ray incident direction. It is translated in parallel on a plane including the facing constituent surfaces closer to the portion 2. The rectangle RT3 detects the X-ray detection unit 2 of the pair of facing constituent planes having overlapping regions when viewed from the X-ray incident direction among the constituent planes of the voxel VX1 shown in FIGS. 6A and 6B. The portion 2 is translated in parallel on a plane including the distant facing configuration surface. The outer circumferences of the rectangles RT1 to RT3 coincide with each other when viewed from the X-ray incident direction.

Assuming that the boxel of interest is the boxel VX1 shown in FIGS. 6A and 6B, the X-ray attenuation of the entire boxel VX1 of interest is (1) the brightness value of the pixel PX5 and the X-ray incident on the pixel PX5 of the rectangle RT1 with respect to the area of the rectangle RT1. Multiplying value with the ratio of the area of the transmitted part, (2) Multiplying the brightness value of the pixel PX6 with the ratio of the area of the part where the X-ray incident on the pixel PX6 of the rectangle RT1 is transmitted to the area of the rectangle RT1 , (3) Multiplying the brightness value of the pixel PX7 by the ratio of the area of the portion where the X-ray incident on the pixel PX7 of the rectangle RT1 is transmitted to the area of the rectangle RT1, (4) The brightness value of the pixel PX9 and the rectangle. Multiplying the area of RT1 by the ratio of the area of the portion where the X-rays incident on the pixel PX9 of the rectangle RT1 are transmitted, (5) the brightness value of the pixel PX10 and the pixel PX10 of the rectangle RT1 with respect to the area of the rectangle RT1. Multiplying value with the ratio of the area of the part where the X-ray is transmitted, (6) The ratio of the brightness value of the pixel PX11 and the area of the part where the X-ray incident on the pixel PX11 of the rectangle RT1 is transmitted to the area of the rectangle RT1. It is the sum of the multiplication values. The coordinates of each vertex of the rectangle RT1 can be calculated from the coordinates of each vertex of the voxel VX1 of interest. Further, each of the above areas can be calculated from the X-coordinate and Z-coordinate of the rectangle RT1 and the X-coordinate and Y-coordinate of each lattice point of the pixel formed on the detection surface of the X-ray detection unit 2.

As described above, the rectangle RT1 is the midpoint of each side of the voxel VX1 of interest other than the side included in the pair of facing constituent planes having overlapping regions when viewed from the X-ray incident direction among the constituent surfaces of the voxel VX1 of interest. It is a rectangle whose apex is. As a result, it is possible to prevent the oblique quadrangular prism OP1 from being biased with respect to the voxel VX1 of interest in the direction parallel to the detection surface of the X-ray detection unit 2, and the effect of approximation on the image quality of the projected image is minimized. can do. However, the position setting of the oblique quadrangular prism that approximates the voxel VX1 of interest is not limited to the setting of the present embodiment. For example, if it is not a problem that the oblique quadrangular prism OP1 is biased with respect to the voxel VX1 of interest with respect to the direction parallel to the detection surface of the X-ray detection unit 2, an oblique quadrangular prism that approximates the voxel VX1 of interest is used. The oblique quadrangular prism OP2 shown in FIG. 9, the oblique quadrangular prism OP3 shown in FIG. 10, the oblique quadrangular prism OP4 shown in FIG. 11, the oblique quadrangular prism OP5 shown in FIG.

When performing the approximation shown in FIG. 9, assuming that the boxel of interest is the boxel VX1 shown in FIGS. 6A and 6B, the X-ray attenuation of the entire boxel VX1 of interest is (1) a rectangle with respect to the brightness value of the pixel PX5 and the area of the rectangle RT4. Multiplying the ratio of the area of the part where the X-ray incident on the pixel PX5 of the RT4 is transmitted, (2) the brightness value of the pixel PX6 and the X-ray incident on the pixel PX6 of the rectangle RT4 with respect to the area of the rectangle RT4 are transmitted. Multiplying value with the ratio of the area of the part, (3) Multiplying the brightness value of the pixel PX7 with the ratio of the area of the part through which the X-ray incident on the pixel PX7 of the rectangle RT4 is transmitted to the area of the rectangle RT4, (4) ) Multiplying the brightness value of the pixel PX9 and the ratio of the area of the portion where the X-ray incident on the pixel PX9 of the rectangle RT4 to the area of the rectangle RT4, (5) the brightness value of the pixel PX10 and the area of the rectangle RT4 The multiplication value of the ratio of the area of the portion where the X-ray incident on the pixel PX10 of the rectangle RT4 is transmitted, (6) the brightness value of the pixel PX11 and the X-ray incident on the pixel PX11 of the rectangle RT4 with respect to the area of the rectangle RT4. It is the sum of the multiplication value with the ratio of the area of the transparent part. The rectangle RT4 is farther from the X-ray detection unit 2 of the pair of facing constituent surfaces having overlapping regions when viewed from the X-ray incident direction among the constituent surfaces of the voxel VX1 shown in FIGS. 6A and 6B.

In the case of performing the approximation shown in FIG. 10, assuming that the boxel of interest is the boxel VX1 shown in FIGS. 6A and 6B, the X-ray attenuation of the entire boxel VX1 of interest is (1) a rectangle with respect to the brightness value of the pixel PX5 and the area of the rectangle RT5. Multiplying the area ratio of the area where the X-ray incident on the pixel PX5 of the RT5 is transmitted, (2) the brightness value of the pixel PX6 and the X-ray incident on the pixel PX6 of the rectangle RT5 with respect to the area of the rectangle RT5 are transmitted. Multiplying value with the ratio of the area of the part, (3) Multiplying the brightness value of the pixel PX7 with the ratio of the area of the part through which the X-ray incident on the pixel PX7 of the rectangle RT5 is transmitted to the area of the rectangle RT5, (4) ) Multiplying the brightness value of pixel PX9 and the ratio of the area of the portion where X-rays incident on pixel PX9 of rectangle RT5 to the area of rectangle RT5, (5) the brightness value of pixel PX10 and the area of rectangle RT5 The multiplication value of the ratio of the area of the portion where the X-ray incident on the pixel PX10 of the rectangle RT5 is transmitted, (6) the brightness value of the pixel PX11, and the X-ray incident on the pixel PX11 of the rectangle RT5 with respect to the area of the rectangle RT5. It is the sum of the multiplication value with the ratio of the area of the transparent part. The rectangle RT5 is closer to the X-ray detection unit 2 of the pair of facing constituent surfaces having overlapping regions when viewed from the X-ray incident direction among the constituent surfaces of the voxel VX1 shown in FIGS. 6A and 6B.

In the case of performing the approximation shown in FIG. 11, assuming that the boxel of interest is the boxel VX1 shown in FIGS. 6A and 6B, the X-ray attenuation of the entire boxel VX1 of interest is (1) the brightness value of the pixel PX5 and the rectangle with respect to the area of the rectangle RT6. Multiplying the ratio of the area of the part where the X-ray incident on the pixel PX5 of the RT6 is transmitted, (2) the brightness value of the pixel PX6 and the X-ray incident on the pixel PX6 of the rectangle RT6 with respect to the area of the rectangle RT6 are transmitted. Multiplying value with the ratio of the area of the part, (3) Multiplying the brightness value of the pixel PX7 with the ratio of the area of the part through which the X-ray incident on the pixel PX7 of the rectangle RT6 is transmitted to the area of the rectangle RT6, (4) ) Multiplying the brightness value of the pixel PX9 and the ratio of the area of the portion where the X-ray incident on the pixel PX9 of the rectangle RT6 to the area of the rectangle RT6, (5) the brightness value of the pixel PX10 and the area of the rectangle RT6. The multiplication value of the ratio of the area of the portion where the X-ray incident on the pixel PX10 of the rectangle RT6 is transmitted, (6) the brightness value of the pixel PX11, and the X-ray incident on the pixel PX11 of the rectangle RT6 with respect to the area of the rectangle RT6. It is the sum of the multiplication value with the ratio of the area of the transparent part. It should be noted that the rectangle RT6 is each side of the voxel VX1 of interest other than the side included in the pair of facing constituent surfaces in which the overlapping regions exist when viewed from the X-ray incident direction among the constituent surfaces of the voxel VX1 shown in FIGS. 6A and 6B. Is a rectangle whose apex is a division point that divides the X-ray detector 2 from the far side at a ratio of 1: 2.

When performing the approximation shown in FIG. 12, assuming that the boxel of interest is the boxel VX1 shown in FIGS. 6A and 6B, the X-ray attenuation of the entire boxel VX1 of interest is (1) a rectangle with respect to the brightness value of the pixel PX5 and the area of the rectangle RT7. Multiplying the ratio of the area of the part where the X-ray incident on the pixel PX5 of the RT7 is transmitted, (2) the brightness value of the pixel PX6 and the X-ray incident on the pixel PX6 of the rectangle RT7 with respect to the area of the rectangle RT7 are transmitted. Multiplying value with the ratio of the area of the part, (3) Multiplying the brightness value of the pixel PX7 with the ratio of the area of the part through which the X-ray incident on the pixel PX7 of the rectangle RT7 is transmitted to the area of the rectangle RT7, (4) ) Multiplying the brightness value of pixel PX9 and the ratio of the area of the portion where X-rays incident on pixel PX9 of rectangle RT7 to the area of rectangle RT7, (5) the brightness value of pixel PX10 and the area of rectangle RT7 The multiplication value of the ratio of the area of the portion where the X-ray incident on the pixel PX10 of the rectangle RT7 is transmitted, (6) the brightness value of the pixel PX11, and the X-ray incident on the pixel PX11 of the rectangle RT7 with respect to the area of the rectangle RT7. It is the sum of the multiplication value with the ratio of the area of the transparent part. It should be noted that the rectangle RT7 is each side of the voxel VX1 of interest other than the side included in the pair of facing constituent surfaces having overlapping regions when viewed from the X-ray incident direction among the constituent surfaces of the voxel VX1 shown in FIGS. 6A and 6B. Is a rectangle whose apex is a division point that divides the X-ray detector 2 from the far side at a ratio of 2: 1.

Since the vertically long X-ray gap beam is emitted from the X-ray irradiation unit 1 in step S1, the X-rays emitted from the X-ray irradiation unit 1 spread in the X-ray direction (= lateral direction of the X-ray detection unit 2). Is small, and the spread of X-rays emitted from the X-ray irradiation unit 1 in the Y-axis direction (= vertical direction of the X-ray detection unit) is large. Therefore, with respect to the X-axis direction (= lateral direction of the X-ray detection unit 2), the X-ray incident direction is the X-ray detection unit 2 as shown in FIGS. 7 to 12 described above regardless of the position of the boxel of interest. Can be treated as perpendicular to the detection surface of, but in the Y-axis direction (= vertical direction of the X-ray detection unit 2), the X-ray incident direction is uniformly perpendicular to the detection surface of the X-ray detection unit 2. The farther the position of the focus box cell is from the origin with respect to the Y axis direction, the more X-rays transmitted through the focus box cell are obliquely incident on the detection surface of the X-ray detection unit 2 in the Y axis direction. Here, the back projection when the X-rays passing through the voxel of interest are obliquely incident on the detection surface of the X-ray detection unit 2 in the Y-axis direction will be described with reference to FIG. FIG. 13 is a diagram showing an example of a case where X-rays transmitted through the voxel VX1 of interest are obliquely incident on the detection surface of the X-ray detection unit 2 in the Y-axis direction. Also in FIG. 13, similarly to FIGS. 7 to 12, a rectangular parallelepiped voxel (for example, the voxel of interest VX1 in FIG. 13) has the same volume as the rectangular parallelepiped voxel and has a uniform thickness with respect to the X-ray incident direction. Back projection is performed by approximating a prism (for example, the oblique prism OP6 in FIG. 13). The oblique quadrangular prism has two bottom surfaces by translating a pair of facing constituent surfaces having overlapping regions when viewed from the X-ray incident direction on a plane including both or one of the opposing constituent surfaces. It has the same volume as a rectangular parallelepiped voxel and has a uniform thickness with respect to the X-ray incident direction.

In the case of performing the approximation shown in FIG. 13, assuming that the boxel of interest is the boxel VX1 shown in FIGS. 6A and 6B, the X-ray attenuation of the entire boxel VX1 of interest is (1) a rectangle with respect to the brightness value of the pixel PX5 and the area of the rectangle RT1. Multiplying the ratio of the area of the part where the X-ray incident on the pixel PX5 of RT1 is transmitted, (2) the brightness value of the pixel PX6 and the X-ray incident on the pixel PX6 of the rectangle RT1 with respect to the area of the rectangle RT1 are transmitted. Multiplying value with the ratio of the area of the part, (3) Multiplying the brightness value of the pixel PX7 with the ratio of the area of the part through which the X-ray incident on the pixel PX7 of the rectangle RT1 is transmitted to the area of the rectangle RT1, (4) ) Multiplying the brightness value of the pixel PX9 and the ratio of the area of the portion where the X-ray incident on the pixel PX9 of the rectangle RT1 is transmitted to the area of the rectangle RT1, (5) the brightness value of the pixel PX10 and the area of the rectangle RT1. The multiplication value of the ratio of the area of the portion through which the X-ray incident on the pixel PX10 of the rectangle RT1 is transmitted, (6) the brightness value of the pixel PX11, and the X-ray incident on the pixel PX11 of the rectangle RT1 with respect to the area of the rectangle RT1. It is the sum of the multiplication value with the ratio of the area of the transparent part. As described above, the rectangle RT1 is formed on each side of the voxel VX1 of interest other than the side included in the pair of facing constituent surfaces having overlapping regions when viewed from the X-ray incident direction among the constituent surfaces of the voxel VX1 of interest. It is a rectangle with the midpoint as the apex.

In the above description, there are cases where none of the constituent surfaces of the boxel of interest is parallel to the detection surface of the X-ray detector 2 and the incident direction of the X-rays passing through the boxel of interest is perpendicular to the detection surface of the X-ray detector 2. It was shown that the same approximation can be performed for both cases, but none of the constituent planes of the focus box cell is parallel to the detection plane of the X-ray detector 2 and the X-rays that pass through the focus box cell. The same approximation can be performed when the incident direction is not perpendicular to the detection surface of the X-ray detection unit 2.

Although only voxels VX1 are shown in FIGS. 6A and 6B, the same back projection is performed on all voxels in the reconstruction region R1. In the calculation of the reconstruction, a Cartesian coordinate system defined by the X-axis and the Z-axis shown in FIG. 4 and the Y-axis orthogonal to them may be used, and the moving diameter r, the first declination θ, and the first deviation angle θ are used. A polar coordinate system defined by an argument φ of 2 may be used. When a polar coordinate system is used, the distance from the center (X-ray source) of the X-ray irradiation unit 1 to one voxel is defined as the moving diameter r. Further, since the vertical field angle theta ₁ necessary camera from the center of the X-ray irradiation unit 1 (X-ray source) from the end of one voxel to the end is very small, approximating the sin [theta ₁ to theta ₁ Can be done. Similarly, the lateral angle phi ₁ necessary camera from the center of the X-ray irradiation unit 1 (X-ray source) from the end of one voxel to the end is very small, approximating the sin [phi ₁ to phi ₁ be able to.

The shape of the voxel is a rectangular parallelepiped, but the rectangular parallelepiped also includes a cube, which is a special example in which the length, width, and height are all the same. Similarly, the constituent planes of the voxel and the cross-sections of the voxels parallel to the constituent planes are rectangular in shape, but the rectangle also includes a square, which is a special example of equal length and width. In addition, when there are two pairs of opposing configuration surfaces in which overlapping regions exist on the sides of the constituent surfaces of the voxel of interest when viewed from the X-ray incident direction, only the pair of opposing configuration surfaces are among the constituent surfaces of the voxel of interest. It may be selected as a pair of facing configuration surfaces having overlapping regions when viewed from the X-ray incident direction. Further, when there are three pairs of facing constituent planes in which overlapping regions exist at points among the constituent planes of the voxel of interest, only the pair of facing constituent planes are among the constituent planes of the voxel of interest. It may be selected as a pair of facing configuration surfaces having overlapping regions when viewed from the X-ray incident direction.

When the back projection calculation is completed for all the voxels in the reconstruction area R1, the reconstruction calculation using the FBP method in step S19 is completed. After that, the CPU 4 determines whether or not the frame data has ended (step S20), and if not, returns to step S16 and repeats the above-described operation.

On the other hand, the number of times (n) that X-rays have passed through each voxel differs depending on the relative position between the object to be inspected T1 and the irradiation area of X-rays, so the CPU 4 calculates the number of times in the calculation process (step S21). ), The final result is divided by n (step S22).

<4. Positioning of foreign matter>
An example of the process of step S3 described above, that is, the process of specifying the position of the foreign matter for each projected image will be described with reference to the flowchart of FIG.

First, the CPU 4 divides the projected image into predetermined areas (for example, 16 pixels × 16 pixels areas) (step S31). If there is a part of the projected image that is not filled with a predetermined area, the edge part of the projected image is excluded from the inspection target, and the position of the predetermined area group is set so that the predetermined area group is located in the center of the projected image. It may be set.

Next, the CPU 4 calculates the average luminance value L1 in each of the predetermined regions (step S32).

After that, the CPU 4 determines whether or not the brightness value L2 of the pixel of interest is smaller than the average brightness value L1 in the predetermined region to which the pixel of interest belongs by a predetermined value V1 or more (step S33). As the predetermined value V1, for example, the standard deviation of the brightness value of each pixel in the predetermined region to which the pixel of interest belongs can be mentioned.

When it is not determined that the luminance value L2 of the pixel of interest is smaller than the average luminance value L1 in the predetermined region to which the pixel of interest belongs by a predetermined value V1 or more (NO in step S33), the CPU 4 positions the pixel of interest as a foreign object. Not specified as.

On the other hand, when it is determined that the luminance value L2 of the pixel of interest is smaller than the average luminance value L1 in the predetermined region to which the pixel of interest belongs by a predetermined value V1 or more (YES in step S33), the process proceeds to step S34.

In step S34, the CPU 4 calculates the maximum first brightness value among the first set number of pixels arranged on the negative side in the horizontal direction of the pixel of interest, and the second set number of pixels arranged on the positive side in the horizontal direction of the pixel of interest. Calculate the maximum second brightness value within the pixel, calculate the maximum third brightness value within the third set number of pixels arranged on the negative side in the vertical direction of the pixel of interest, and line up on the positive side in the vertical direction of the pixel of interest. The fourth brightness value, which is the maximum within the fourth set number of pixels, is calculated. The first set number to the fourth set number may all have the same value, or may have two or more and four or less different values. If a predetermined area is located at the edge of the projected image and at least one of the first set number to the fourth set number cannot be secured, the number of pixels that can be secured is used, and if it cannot be secured at all. The pixel is excluded from inspection.

Next, the CPU 4 determines whether or not the ratio of the brightness value L2 of the pixel of interest to the first brightness value and the ratio of the brightness value L2 of the pixel of interest to the second brightness value are equal to or less than the threshold value TH1 (step S35). ..

When it is determined that the ratio of the brightness value L2 of the pixel of interest to the first brightness value and the ratio of the brightness value L2 of the pixel of interest to the second brightness value are each equal to or less than the threshold value TH1 (YES in step S35), step S37 described later. Move to.

On the other hand, when it is not determined that the ratio of the brightness value L2 of the pixel of interest to the first brightness value and the brightness value L2 of the pixel of interest to the second brightness value are each equal to or less than the threshold value TH1 (NO in step S35), step. Move to S36.

In step S36, the CPU 4 determines whether or not the ratio of the brightness value L2 of the pixel of interest to the third brightness value and the ratio of the brightness value L2 of the pixel of interest to the fourth brightness value are each equal to or less than the threshold value TH1 (step S36). ). In the present embodiment, the threshold value used in step S35 and the threshold value used in step S36 are set to the same value, but they may be different values from each other. Further, unlike the present embodiment, in step S35, it may be determined only whether or not the ratio of the brightness value L2 of the pixel of interest to the first brightness value is equal to or less than the threshold value TH1. On the contrary, in step S35, it may be determined only whether or not the ratio of the brightness value L2 of the pixel of interest to the second brightness value is equal to or less than the threshold value TH1. The same modification may be performed for step S36. In step S36, it may be determined only whether or not the ratio of the brightness value L2 of the pixel of interest to the third brightness value is equal to or less than the threshold value TH1. On the contrary, in step S36, it may be determined only whether or not the ratio of the brightness value L2 of the pixel of interest to the fourth brightness value is equal to or less than the threshold value TH1.

When it is determined that the ratio of the brightness value L2 of the pixel of interest to the third luminance value and the luminance value L2 of the pixel of interest to the fourth luminance value are each equal to or less than the threshold value TH1 (YES in step S36), step S37 described later. Move to.

On the other hand, when it is not determined that the ratio of the brightness value L2 of the pixel of interest to the third brightness value and the brightness value L2 of the pixel of interest to the fourth brightness value are equal to or less than the threshold value TH1 (NO in step S36), the CPU 4 Does not specify the pixel of interest as the position of the foreign object. It should be noted that, instead of immediately confirming that the pixel of interest is not specified as the position of the foreign matter, the size of the predetermined region and the value of the threshold value TH1 are changed to return to step S35, and the process proceeds from step S35 or step S36 to step S37. You may try to do it. The size of the predetermined region and the value of the threshold value TH1 may be changed not only once but also twice or more.

In step S37, the CPU 4 specifies the pixel of interest as the position of the foreign object.

Then, all the pixels belonging to the predetermined area are sequentially set as "pixels of interest", and the processing after step S33 is repeated. Further, all the predetermined areas are sequentially set as "predetermined areas to which the pixel of interest belongs" one by one, and the processes after step S32 are repeated.

According to the flowchart of FIG. 14, since the position of the foreign matter is specified based on the luminance characteristics of the entire minute region in a predetermined region, that is, the minute region, the detection accuracy of the foreign matter can be improved.

As the object to be inspected T1, for example, "fish" can be mentioned. When the object to be inspected T1 is a "fish", the foreign matter is a "small bone". The processing of the flowchart of FIG. 14 can be applied when the attenuation coefficient of the foreign matter is larger than the attenuation coefficient of the object T1 to be inspected (excluding the foreign matter). However, when the projected image is a black-and-white inverted image, in the processing of the flowchart of FIG. 14, "small" is replaced with "large", "maximum" is replaced with "minimum", and "threshold value TH1 or less" is replaced. It may be replaced with "threshold TH1 or higher".

When the attenuation coefficient of the foreign matter is smaller than the attenuation coefficient of the object T1 to be inspected (excluding the foreign matter), the processing of the flowchart of FIG. 14 is not applied as it is, but is "small" in the processing of the flowchart of FIG. Is replaced with "large", "maximum" is replaced with "minimum", and "threshold TH1 or less" is replaced with "threshold TH1 or more". However, when the projected image is a black-and-white inverted image, the processing of the flowchart of FIG. 14 may be applied as it is.

<5. Position correction>
Since each projected image after the reconstruction calculation is a projection onto the tomography when the inspected object T1 is viewed from the X-ray source, the position of the projected inspected object T1 shifts if the irradiation angle is different. In order to correct this deviation, the CPU 4 performs position correction as described above (step S5).

Specifically, as shown in FIG. 15, an arbitrary fault used as a correction reference when the brightness of a voxel on a fault other than the arbitrary fault used as a correction reference is projected onto an arbitrary fault used as a correction reference. Based on the brightness of the voxels above, position correction is performed to match the horizontal and vertical lengths of the projected images after the reconstruction calculation.

Projections to any voxel used as a correction criterion for voxel brightness on a fault other than any fault used as a correction criterion should normally be calculated by examining the X-ray transmission path for all pixels in each frame. However, since the width of the X-ray detector 2 (= the lateral length of the X-ray detector 2) is very narrow (for example, 64 pixels) and the change between adjacent frames is small, a certain fault can be identified. It is assumed that the X-ray path passing through the voxel does not change between adjacent frames, and the calculation is performed in consideration of only the X-ray transmission path incident on the center of the X-ray detector 2 for each frame.

Hereinafter, a voxel and two certain faults j = j1 in which the height direction (= vertical direction of the X-ray detection unit 2) on an arbitrary fault (hereinafter referred to as a reference fault lay) used as a correction reference is a predetermined position. , J2 The position correction will be described in detail with reference to FIG. 15 in which voxels whose height direction is a predetermined position are arranged in a straight line for each fault for convenience.

If the brightness of a boxel i = b on a fault j = a other than the reference fault lay is projected onto the reference fault lay, and the projected brightness is on the voxel i = c on the reference fault lay, then Basically, the projected brightness is treated as the brightness of voxel i = c on a certain fault j = a after position correction. a, b, and c are arbitrary natural numbers. When the X-rays of one frame pass through a plurality of voxels having the same fault and height direction, the X-rays of one frame overlap with each other when viewed from the intermediate surface (= X-ray incident direction). A rectangle whose apex is the midpoint of each side of the voxel other than the side included in the pair of facing constituent planes in which the region is present. See the dotted line in FIG. 15) Only the voxel that has passed through is one frame. A voxel that allows X-rays to pass through.

That is, in a certain frame, one voxel i = b through which X-rays pass on a certain fault j = a other than the reference fault lay has a one-to-one correspondence with the voxel i = c on the reference fault lay. If so (Pattern I shown in FIG. 15), the brightness projected on the reference fault lay is treated as the brightness of voxels i = c on a certain fault j = a after position correction. Similarly, even if there is no one-to-one correspondence, when there is one X-ray frame passing through one voxel on the reference fault lay (pattern I'shown in FIG. 15), in a certain frame as well. If one voxel i = b through which X-rays pass on a certain fault j = a other than the reference fault lay comes to the voxel i = c on the reference fault lay, the projected brightness is corrected. It is treated as the brightness of voxels i = c on a certain fault j = a in.

However, when the X-rays of a plurality of frames pass through one voxel on the reference fault lay (Pattern II shown in FIG. 15), the X-rays on the reference fault lay do not pass through the X-rays of any frame (Fig. 15). In pattern III) shown in 15, it is necessary to adjust the brightness value.

The idea is the same for the height direction.

An example of the position correction in step S5 according to the above-mentioned concept will be described with reference to the flowchart shown in FIG.

In the example of the position correction shown in FIG. 16, first, the CPU 4 reads the number of voxels of each fault in the reconstruction area R1 from the HDD 9 (step S41).

Next, the CPU 4 reads the data of the brightness value cal (i, j, k) of each voxel after the reconstruction calculation calculated in step S22 from the HDD 9 (step S42). i is a variable for specifying the coordinates (position) of the target voxel in the X-axis direction shown in FIG. 1, and j is a variable for specifying the coordinates (position) of the target voxel in the Y-axis direction shown in FIG. Yes, k is a variable to identify the fault to which the target voxel belongs.

Next, the CPU 4 reads from the HDD 9 the coordinates of the voxel through which the X-rays incident on the center of the X-ray detection unit 2 pass for each frame and each fault, and passes through the voxel in which the Y-axis direction of each fault is maximum. The ratio of the distances in the Y-axis direction of each line connecting the radiation source and the X-ray detection unit 2 is read from the HDD 9 (step S43).

Next, the CPU 4 uses the reading result of step S43 to transmit the n (i) th X-ray among the N (i) frames that pass through the voxels at the X-axis coordinate i on the reference fault lay. The X-ray coordinates ii (i, k, n (i)) of the voxel of a certain fault k are calculated for each frame and each fault (step S44).

Next, the CPU 4 uses the reading result of step S43 to transmit the h (j) th X-ray of the H (j) frames that pass through the voxel at the Y-axis coordinate k on the reference fault lay. The Y-axis coordinates kk (k, k, h (j)) of the voxels of a certain fault k are calculated for each frame and each fault (step S45).

Next, the CPU 4 projects the luminance value of the boxel of the fault k onto the reference fault lay in a loop of the X-axis coordinates i, Y-axis coordinates j, n (i), h (j) of the box cell, and the luminance value datawa ( i, j, k) is calculated from the luminance value cal (ii, jj, k) of the point (ii, jj, k) on the fault k (step S46).

When N (i) ≠ 0 and H (j) ≠ 0, that is, when both i and j correspond to pattern I or pattern II, the luminance value datawa (i, j, k) is expressed by the following equation (1). Will be done.

When N (i) ≠ 0 and H (j) = 0, that is, when i corresponds to pattern I or pattern II and j corresponds to pattern III, the luminance value datawa (i, j, k) is as follows. It is expressed by equation (2).

When N (i) = 0 and H (j) ≠ 0, that is, i corresponds to pattern III and j corresponds to pattern I or pattern II, the luminance value datawa (i, j, k) is as follows. It is represented by equation (3).

When N (i) = 0 and H (j) = 0, that is, when both i and j correspond to pattern III, the luminance value datawa (i, j, k) is expressed by the following equation (4).

However, i1 and i2 in the above equations (3) and (4) are before and after (n (i) = 0) when there is no frame corresponding to the X-axis coordinate i of the voxel on the reference fault lay. It is the X-axis coordinate ii on the fault k corresponding to i satisfying n (i)> 0 of the frame of, and satisfies i1 ≤ i2. J1 and j2 in the above equations (2) and (4) are also considered in the same manner in the Y-axis direction. If at least one of the coordinate positions indicated by i1, i2, j1, and j2 is out of the region forming the projected image, the luminance value datawa (i, j, k) is set to 0.

Considering the case of i1 + 1 <i2, in the case of pattern III, voxels in which the X-rays of the frame are not transmitted on the fault having more voxels than the reference fault lay (for example, the fault k = k2 shown in FIG. 15). (For example, the voxel shown in FIG. 15) will be present, but it is not desirable not to use the brightness value of this voxel. Therefore, in this case, it is desirable to perform the following processing.

Let i1 = ii (i−Δ1, k, N (i−Δ1)), i2 = ii (i ＋ Δ2, k, N (i ＋ Δ1)), and X on the fault k corresponding to the X-axis coordinate i on the reference fault lay. Let the axis coordinates be the real number i _k expressed by the following equation (5). Here, Δ1 + Δ2-1 is the number of continuous voxels in which the X-rays of the frame on the reference fault lay are not transmitted.

Consider j1 and j2 in the same way in the Y-axis direction. Then, in the case of i1 + 1 <i2 and / or j1 + 1 <j2, the following equation (6) is used instead of the above equation (2), and the following equation (7) is used instead of the above equation (3), and the above equation (4) is used. ) Instead of the following equation (8). However, flr (x) in the following equations (6) to (8) represents the largest integer among the integers equal to or less than the real number x.

When the calculation of the luminance value datawa (i, j, k) described above is completed, the position correction process shown in FIG. 16 (an example of the process in step S5) is completed.

As described above, in the present embodiment, in the reconstruction calculation using the FBP method, instead of back-projecting the rectangular parallelepiped voxels constituting the reconstruction region onto the voxels, the constituent surfaces of the rectangular parallelepiped voxels constituting the reconstruction region. Of these, back projection is performed on one of the pair of facing constituent planes having overlapping regions when viewed from the X-ray incident direction, or on the cross section of the voxel parallel to the facing constituent planes, so that the calculation time for back projection can be reduced. It can be shortened significantly. Therefore, the projected image can be reconstructed with a small amount of reconstruction calculation using the FBP method.

Further, in the present embodiment, each projected image after the reconstruction calculation is not uniformly enlarged or reduced in each of the horizontal and vertical directions to perform position correction, but a certain fault other than an arbitrary fault used as a correction reference. Position correction is performed based on the brightness of the voxel on the arbitrary fault used as the correction reference when projected onto an arbitrary fault using the brightness of the above voxel as the correction reference. As a result, it is possible to correct the position of each projected image after the reconstruction calculation using the FBP method while suppressing the occurrence of distortion.

Further, in the above-described embodiment, the horizontal and vertical lengths of the projected images after the reconstruction calculation are matched by the position correction, but the horizontal lengths of the projected images after the reconstruction calculation are different. If this is not a problem, the vertical length of each projected image after reconstruction calculation may be matched by position correction, and the vertical length of each projected image after reconstruction calculation may be matched. If the difference does not matter much, the lengths of each projected image after the reconstruction calculation may be matched only in the lateral direction by the position correction.

Further, in the above-described embodiment, the horizontal and vertical lengths of the projected images after the reconstruction calculation are matched by the position correction, but the horizontal and vertical lengths of the projected images after the reconstruction calculation are matched by the position correction. The lengths in the directions may be substantially the same. That is, for example, the side of each projected image after the position correction and the reconstruction calculation is not inconvenient when comparing the projected images of a plurality of layers with each other and comparing the state of the object T1 to be inspected according to the tomographic depth. There may be differences in directional and vertical lengths.

<6. Others>
In the above-described embodiment, the false detection partial removal process of step S4 is executed, but if the specification that requires the foreign matter detection accuracy is satisfied without executing the false detection partial removal process of step S4, step S4 The false detection partial removal process may be omitted. When omitting the erroneous detection portion removal process in step S4, either the pair of the first X-ray irradiation unit 1A and the first X-ray detection unit 2A or the pair of the second X-ray irradiation unit 1B and the second X-ray detection unit 2B is used. It is recommended to remove it from the inspection device 100.

In the above-described embodiment, in the erroneous detection portion removal process in step S4, the position of the foreign matter specified based on the projected images derived from the first X-ray irradiation unit 1A and the first X-ray detection unit 2A, and the second X-ray irradiation unit 1B and the second X-ray irradiation unit 1B. The position of the foreign matter specified based on the projected image derived from the 2X ray detection unit 2B was compared, and the false detection part of the foreign matter was removed based on the comparison result, but this false detection part removal process is just an example. Is.

Therefore, the inspection device 100 is provided with the first to m (m is a natural number of 3 or more) X-ray irradiation units and the first to mX-ray detection units, which have different X-ray irradiation directions, and the k (k is). Any natural number of m or less) X-rays irradiated from the X-ray irradiation unit and transmitted through the object T1 to be inspected may be incident on the kX-ray detection unit. In this case, the position of the foreign matter is specified based on the k-th projected image, which is a projected image obtained by the X-rays emitted from the kX-ray irradiation unit, and the first to mth projected images of the same depth are respectively. The positions of the foreign substances specified based on the above are compared with each other, and the erroneous detection portion of the position of the foreign matter may be removed based on the comparison result.

In the above-described embodiment, in step S34, the CPU 4 calculates the maximum first brightness value among the first set number of pixels arranged on the negative side in the horizontal direction of the pixel of interest, and arranges them on the positive side in the horizontal direction of the pixel of interest. The maximum second brightness value within the second set number of pixels is calculated, the maximum third brightness value within the third set number of pixels arranged on the negative side in the vertical direction of the pixel of interest is calculated, and the pixel of interest is calculated. The maximum fourth brightness value within the fourth set number of pixels arranged on the positive side in the vertical direction was calculated, but this calculation process is only an example. That is, the above-mentioned "horizontal negative side", "horizontal positive side", "vertical negative side", and "vertical positive side" are "first direction one side" and "first direction other side", respectively. , "One side in the second direction", and "the other side in the second direction" can be generalized. The first direction and the second direction are different directions. The first direction and the second direction do not have to be orthogonal to each other.

Further, the first brightness value is not the maximum brightness value among the first set number of pixels arranged on one side of the first direction of the pixel of interest, but the first set number of pixels arranged on one side of the first direction of the pixel of interest. The brightness value of the pixel having a brightness value larger than that of the pixel of interest and being as far away as possible from the pixel of interest may be used, and the same substitution may be performed for the second to fourth brightness values. That is, the second brightness value is not the maximum brightness value among the second set number of pixels arranged on the other side of the first direction of the pixel of interest, but the second set number of pixels arranged on the other side of the first direction of the pixel of interest. The brightness value of a pixel having a brightness value larger than that of the pixel of interest and as far away as possible from the pixel of interest is used. Further, the third brightness value is not the maximum brightness value among the pixels of the third set number arranged on one side of the second direction of the pixel of interest, but the pixels of the third set number arranged on one side of the second direction of the pixel of interest. The brightness value of a pixel having a brightness value larger than that of the pixel of interest and as far away as possible from the pixel of interest is used. Further, the fourth brightness value is not the maximum brightness value among the fourth set number of pixels arranged on the other side of the second direction of the pixel of interest, but the fourth set number of pixels arranged on the other side of the second direction of the pixel of interest. The brightness value of a pixel having a brightness value larger than that of the pixel of interest and as far away as possible from the pixel of interest is used. Even when the above replacement is performed, since the position of the foreign matter is specified based on the luminance characteristics of the entire minute region as in the above-described embodiment, the detection accuracy of the foreign matter can be improved. However, in the case of performing the above substitution, when the attenuation coefficient of the foreign matter is smaller than the attenuation coefficient of the object T1 to be inspected (excluding the foreign matter), "small" is changed to "large" in the process of step S33. In the process of replacing step S34, "large" is replaced with "small", and "threshold TH1 or less" is replaced with "threshold TH1 or more". Further, instead of the above-mentioned "luminance value of a pixel as far as possible from the pixel of interest", "luminance value of a pixel as close as possible to the pixel of interest" may be used.

Further, the first brightness value is not the maximum brightness value among the first set number of pixels arranged on one side of the focus pixel in the first direction, but the first predetermined position from the focus pixel on one side of the focus pixel in the first direction. The first set number of pixels lined up apart (for example, in the case of four pixels lined up 3 pixels away from the pixel of interest, the third to sixth pixels lined up on one side in the first direction from the pixel of interest) and the focus It is the average brightness value of the second set number of pixels arranged on the other side of the first direction of the pixels at a second predetermined position away from the pixel of interest. Then, in steps S34 to S36, the processing related to the second luminance value is not performed. Further, the third brightness value is not the maximum brightness value among the third set number of pixels arranged on one side in the second direction of the pixel of interest, but the third predetermined position from the pixel of interest on one side of the second direction of the pixel of interest. The average brightness value of the pixels of the third set number arranged apart and the pixels of the fourth set number arranged at the other side in the second direction of the pixel of interest at a fourth predetermined position away from the pixel of interest (“second brightness value” in claim 8”. Equivalent to). Then, in steps S34 to S36, the processing related to the fourth luminance value is not performed. Even when the above replacement is performed, since the position of the foreign matter is specified based on the luminance characteristics of the entire minute region as in the above-described embodiment, the detection accuracy of the foreign matter can be improved. However, in the case of performing the above substitution, when the attenuation coefficient of the foreign matter is smaller than the attenuation coefficient of the object T1 to be inspected (excluding the foreign matter), "small" is changed to "large" in the process of step S33. In the process of replacing step S34, "threshold TH1 or less" may be replaced with "threshold TH1 or more".

The CPU 4 may specify the position of the foreign matter by using the artificial intelligence learned from the projected image of the object to be inspected T1 inspected in the past. By using artificial intelligence, the accuracy of detecting foreign matter can be further improved. The place where artificial intelligence is provided is not particularly limited. For example, the CPU 4 may be provided with artificial intelligence. Further, for example, artificial intelligence may be provided on a cloud that can be accessed by the inspection device 100 via a communication network.

Unlike the above-described embodiment, the execution order of step S4 and step S5 is exchanged, and in step S6, all the projected images showing the positions of the foreign matter reflecting the position correction and the false detection partial removal process are added together. An image to be generated, that is, an image displaying the position of a foreign object in two dimensions may be generated as an output image. In this case, in step S3, it is not always necessary to specify the position of the foreign matter for each projected image, and the position of the foreign matter may be specified for each of a plurality of projected images.

Unlike the above-described embodiment, step S5 is executed immediately after step S3, the positions of foreign substances on all faults are added for each irradiation angle after the position correction is completed, and the irradiation after adding the positions of foreign substances on all faults. Step S4 is executed for the data for each angle, and in step S6, the image obtained by adding all the projected images showing the positions of the foreign matter reflecting the position correction and the false detection portion removal processing, that is, the position of the foreign matter is 2. An image to be displayed in two dimensions may be generated as an output image. In this case, in step S3, it is not always necessary to specify the position of the foreign matter for each projected image, and the position of the foreign matter may be specified for each of a plurality of projected images.

In the above-described embodiment, the projected image obtained by projecting onto each tomographic image set at a predetermined position other than the positions of the

X-ray detectors

2A and 2B is a projected image on each tomographic image obtained based on the principle of tomosynthesis. There is, but this is just an example. That is, the projected image obtained by projecting on each tomographic image set at a predetermined position other than the positions of the

X-ray detectors

2A and 2B does not have to be the projected image on each tomographic image obtained based on the principle of tomosynthesis. .. When the tomosynthesis method is not adopted, the 1X-ray detection unit 2A and the second X-ray detection unit 2B may each be a single-line line sensor. When the 1X-ray detection unit 2A and the 2nd X-ray detection unit 2B are each a single line, the pixel size of the line sensor and the movement amount of the object to be inspected per frame are matched.

Note that, due to structural restrictions of the inspection device 100, the first X-ray detection unit 2A and the second X-ray detection unit 2B may not be arranged at the same height. In this case, the Z-axis direction position of the first X-ray detection unit 2A and the Z-axis direction position of the second X-ray detection unit 2B are different, and the pixel in which the X-ray passing through the same position of the inspected object T1 is incident is the first X-ray. The detection unit 2A and the second X-ray detection unit 2B are misaligned. Therefore, in consideration of this deviation, the second X-ray irradiation unit 1B and the second X-ray detection unit corresponding to the pixels of the projected image derived from the first X-ray irradiation unit 1A and the first X-ray detection unit 2A in the Y-axis direction. The pixels of the projected image derived from 2B may be specified.

Instead of generating a captured image (two-dimensional digital data) by simply arranging pixel data (data obtained by a line sensor) or simply integrating as in the above-described embodiment, it is possible to perform original image processing. The captured image may be generated based on the above. When the captured image is generated based on the original image processing, it is not always necessary to match the moving speed of the object to be inspected T1 with the pixel size of the line sensor.

Further, instead of the first X-ray detection unit 2A and the second X-ray detection unit 2B which are line sensors, a two-dimensional detector 2C may be used as shown in FIG. By using the detection result of the shaded area (see FIG. 17) of the two-dimensional detector 2C, the two-dimensional detector 2C can be used as a pseudo two-line sensor. In the configuration shown in FIG. 17, similarly to the configuration shown in FIG. 1, the first X-ray irradiation unit 1A and the second X-ray irradiation unit 1B may be shared to form a single X-ray irradiation unit.

Further, by using a plurality of two-dimensional detectors instead of the single two-dimensional detector 2C and using the detection results of a part of each of the plurality of two-dimensional detectors, a plurality of two-dimensional detectors can be simulated. Can be a plurality of line sensors.

In addition, by using a single line sensor instead of the single two-dimensional detector 2C and using the detection results of multiple regions of the single line sensor as in the case of the single two-dimensional detector 2C, A single line sensor can be a plurality of pseudo line sensors. The single line sensor is a line sensor having a plurality of lines extending along the Y-axis direction.

Further, instead of the first X-ray detection unit 2A and the second X-ray detection unit 2B which are line sensors, two-

dimensional detectors

2D and 2E may be used as shown in FIGS. 18 and 19. The inspection device 100 shown in FIGS. 18 and 19 may perform the following operations, for example. The object to be inspected T1 is moved by the drive of the belt conveyor 3 to a position where the image can be taken by the second X-ray irradiation unit 1B and the two-dimensional detector 2E, and then the belt conveyor 3 is stopped, and the second X-ray irradiation unit 1B and the two-dimensional The object to be inspected T1 is stationary at a position where it can be photographed by the detector 2E (see FIG. 18). In the stationary state shown in FIG. 18, one shot of the object to be inspected T1 is photographed by the second X-ray irradiation unit 1B and the two-dimensional detector 2E. Then, the object to be inspected T1 is moved by the drive of the belt conveyor 3 to a position where the first X-ray irradiation unit 1A and the two-dimensional detector 2D can take a picture, and then the belt conveyor 3 is stopped, and the first X-ray irradiation unit 1A and the first X-ray irradiation unit 1A and The object to be inspected T1 is stationary at a position where it can be photographed by the two-dimensional detector 2D (see FIG. 19). In the stationary state shown in FIG. 19, one shot of the object to be inspected T1 is photographed by the first X-ray irradiation unit 1A and the two-dimensional detector 2D.

Further, instead of the first X-ray irradiation unit 1A, the second X-ray irradiation unit 1B, and the line sensors the first X-ray detection unit 2A and the second X-ray detection unit 2B, X-ray irradiation is performed as shown in FIGS. 20 and 21. The part 1C and the two-dimensional detector 2F may be used. The inspection device 100 shown in FIGS. 20 and 21 may perform the following operations, for example. The object to be inspected T1 is moved to the first predetermined position by driving the belt conveyor 3. At this time, the two-dimensional detector 2F is also moved by a moving mechanism (not shown) in conjunction with the movement of the object T1 to be inspected. After that, the belt conveyor 3 and the moving mechanism (not shown) are stopped, the object T1 to be inspected is stopped at the first predetermined position, and the two-dimensional detector 2F is stopped at the position corresponding to the first predetermined position (see FIG. 20). ). In the stationary state shown in FIG. 20, one shot of the object to be inspected T1 is taken by the X-ray irradiation unit 1C and the two-dimensional detector 2F. Then, the object to be inspected T1 is moved to the second predetermined position by driving the belt conveyor 3. At this time, the two-dimensional detector 2F is also moved by a moving mechanism (not shown) in conjunction with the movement of the object T1 to be inspected. After that, the belt conveyor 3 and the moving mechanism (not shown) are stopped, the object T1 to be inspected is stopped at the second predetermined position, and the two-dimensional detector 2F is stopped at the position corresponding to the second predetermined position (see FIG. 21). ). In the stationary state shown in FIG. 21, one shot of the object to be inspected T1 is photographed by the X-ray irradiation unit 1C and the two-dimensional detector 2F.

In the above description, a plurality of two-dimensional X-ray images in which the positional relationship between the X-ray object T1 and the X-ray irradiation unit used for X-ray photography are different only in the X-axis direction are generated. A plurality of two-dimensional X-ray images may be generated in which the positional relationship between the X-ray and the X-ray irradiation unit used for the X-ray photography is different only in the Y-axis direction. A plurality of two-dimensional X-ray images having different positional relationships with the line irradiation unit in both the X-axis direction and the Y-axis direction may be generated.

When a line sensor is used as an X-ray detection unit, frame data is projected onto a certain fault for each X-ray irradiation region, and each two-dimensional image generated by reconstruction is converted into "two-dimensional X-rays". The present invention also includes an inspection device having a configuration that is regarded as a "captured image" and is subjected to subsequent processing.

1 X-ray irradiation unit 1A 1st X-ray irradiation unit 1B 2nd X-ray irradiation unit 2 X-ray detection unit 2A 1st X-ray detection unit 2B 2nd X-ray detection unit 3 Belt conveyor 4 CPU
5 ROM
6 RAM
7 VRAM
8 Display 9 HDD
10 Input unit 100 Inspection device

Claims

A first image generation unit that targets an object to be inspected for X-ray photography and generates a two-dimensional X-ray image based on the detection result of the X-ray detection unit used for the X-ray photography.
An image obtained by projecting the two-dimensional X-ray photographed image onto a fault set at a predetermined position other than the position of the X-ray detection unit is generated, and then the generated image is reconstructed into a projected image which is a two-dimensional image. Reconstruction part and
A specific part that identifies the position of a foreign object based on the projected image,
With the removal part
With
The first image generation unit generates a plurality of two-dimensional X-ray images having different positional relationships between the object to be inspected, the X-ray irradiation unit used for the X-ray imaging, and the X-ray detection unit.
The reconstructing unit generates images obtained by projecting a plurality of the two-dimensional X-ray images onto a plurality of tomographic images having different depths, and then reconstructs each of the generated images into the projected images.
The specific unit converts each of the plurality of the projected images into a binarized projected image in which the position of the foreign matter is specified.
The removing unit compares a plurality of the binarized projected images obtained from the images projected on the tomographic image of the same depth, and removes the erroneously detected portion of the position of the foreign matter based on the comparison result. A featured inspection device.
The inspection device according to claim 1, wherein the specific unit uses artificial intelligence learned from the projected image of the object to be inspected in the past to identify the position of a foreign substance.
The inspection device according to claim 1 or 2, further comprising a second image generation unit that generates an image that two-dimensionally displays the position of the foreign matter specified by the specific unit.
The inspection device according to claim 1 or 2, further comprising a second image generation unit that generates an image that three-dimensionally displays the position of the foreign matter specified by the specific unit.
The specific part is
A division portion that divides the projected image into predetermined areas, and
A calculation unit that calculates the average luminance value in each of the predetermined regions,
With
When the brightness value of the pixel of interest is smaller than the average brightness value in the predetermined region to which the pixel of interest belongs by a predetermined value or more.
The first brightness value that is the maximum within the first set number of pixels arranged on one side of the first direction of the pixel of interest is calculated, and the maximum among the pixels of the second set number arranged on the other side of the first direction of the pixel of interest. The second brightness value is calculated, and the maximum third brightness value among the third set number of pixels arranged on one side of the second direction, which is a direction different from the first direction of the pixel of interest, is calculated, and the focus is calculated. Calculate the maximum fourth brightness value within the fourth set number of pixels lined up on the other side of the second direction of the pixels.
If a predetermined condition is satisfied, the pixel of interest is specified as the position of a foreign object, and the pixel of interest is specified.
The predetermined conditions are
(A) The first condition that the ratio of the brightness value of the pixel of interest to the first brightness value and the ratio of the brightness value of the pixel of interest to the second brightness value are each equal to or less than the first threshold value.
(B) The second condition that the ratio of the brightness value of the pixel of interest to the third brightness value and the ratio of the brightness value of the pixel of interest to the fourth brightness value are each equal to or less than the second threshold value.
(C) The first condition and the second condition,
The inspection device according to any one of claims 1 to 4, which is any one of the above.
The specific part is
A division unit that divides an image based on X-ray photography of the object to be inspected into predetermined areas, and
A calculation unit that calculates the average luminance value in each of the predetermined regions,
With
When the brightness value of the pixel of interest is smaller than the average brightness value in the predetermined region to which the pixel of interest belongs by a predetermined value or more.
Within the first set number of pixels arranged on one side in the first direction of the pixel of interest, the brightness value of the pixel having a higher brightness value than the pixel of interest and as far as possible from the pixel of interest, or the brightness of a pixel as close as possible to the pixel of interest. The first brightness value, which is a value, is calculated, and the brightness of the pixels having a brightness value larger than that of the pixel of interest and as far as possible from the pixel of interest within the second set number of pixels arranged on the other side of the first direction of the pixel of interest. The second brightness value, which is the value or the brightness value of the pixel as close as possible to the pixel of interest, is calculated, and the pixel is within the third set number of pixels arranged on one side of the second direction, which is a direction different from the first direction of the pixel of interest. A third brightness value, which is a brightness value of a pixel having a brightness value larger than that of the pixel of interest and as far as possible from the pixel of interest or a brightness value of a pixel as close as possible to the pixel of interest, is calculated, and the other side of the pixel of interest in the second direction. Among the fourth set number of pixels lined up in, the fourth brightness value, which is the brightness value of a pixel having a brightness value larger than that of the pixel of interest and as far as possible from the pixel of interest, or the brightness value of a pixel as close as possible to the pixel of interest, is calculated. And
If a predetermined condition is satisfied, the pixel of interest is specified as the position of a foreign object, and the pixel of interest is specified.
The predetermined conditions are
(A) The first condition that the ratio of the brightness value of the pixel of interest to the first brightness value and the ratio of the brightness value of the pixel of interest to the second brightness value are each equal to or less than the first threshold value.
(B) The second condition that the ratio of the brightness value of the pixel of interest to the third brightness value and the ratio of the brightness value of the pixel of interest to the fourth brightness value are each equal to or less than the second threshold value.
(C) The first condition and the second condition,
The inspection device according to any one of claims 1 to 4, which is any one of the above.
The specific part is
A division unit that divides an image based on X-ray photography of the object to be inspected into predetermined areas, and
A calculation unit that calculates the average luminance value in each of the predetermined regions,
With
When the brightness value of the pixel of interest is smaller than the average brightness value in the predetermined region to which the pixel of interest belongs by a predetermined value or more.
A first set number of pixels arranged on one side of the pixel of interest in the first direction separated from the pixel of interest by a first predetermined position, and a second set number of pixels arranged on the other side of the pixel of interest in the first direction separated by a second predetermined position. The first brightness value, which is the average brightness value of the pixels, is calculated, and a third set number of pixels are arranged on one side of the second direction, which is a direction different from the first direction of the pixel of interest, at a third predetermined position from the pixel of interest. And the second brightness value, which is the average brightness value of the fourth set number of pixels arranged at the other side in the second direction of the pixel of interest at a fourth predetermined position from the pixel of interest, is calculated.
If a predetermined condition is satisfied, the pixel of interest is specified as the position of a foreign object, and the pixel of interest is specified.
The predetermined conditions are
(A) The first condition that the ratio of the brightness value of the pixel of interest to the first brightness value is equal to or less than the first threshold value.
(B) The second condition that the ratio of the brightness value of the pixel of interest to the second brightness value is equal to or less than the second threshold value.
(C) The first condition and the second condition,
The inspection device according to any one of claims 1 to 4, which is any one of the above.
The specific part is
A division unit that divides an image based on X-ray photography of the object to be inspected into predetermined areas, and
A calculation unit that calculates the average luminance value in each of the predetermined regions,
With
When the brightness value of the pixel of interest is greater than or equal to the average brightness value in the predetermined region to which the pixel of interest belongs.
The first brightness value that is the smallest among the first set number of pixels arranged on one side of the first direction of the pixel of interest is calculated, and the smallest among the second set number of pixels arranged on the other side of the first direction of the pixel of interest. The second brightness value is calculated, and the minimum third brightness value among the third set number of pixels arranged on one side of the second direction, which is a direction different from the first direction of the pixel of interest, is calculated, and the focus is calculated. The fourth brightness value, which is the smallest among the fourth set number of pixels arranged on the other side in the second direction of the pixels, is calculated.
If a predetermined condition is satisfied, the pixel of interest is specified as the position of a foreign object, and the pixel of interest is specified.
The predetermined conditions are
(A) The first condition that the ratio of the brightness value of the pixel of interest to the first brightness value and the ratio of the brightness value of the pixel of interest to the second brightness value are each equal to or higher than the first threshold value.
(B) The second condition that the ratio of the brightness value of the pixel of interest to the third brightness value and the ratio of the brightness value of the pixel of interest to the fourth brightness value are each equal to or higher than the second threshold value.
(C) The first condition and the second condition,
The inspection device according to any one of claims 1 to 4, which is any one of the above.
The specific part is
A division unit that divides an image based on X-ray photography of the object to be inspected into predetermined areas, and
A calculation unit that calculates the average luminance value in each of the predetermined regions,
With
When the brightness value of the pixel of interest is greater than or equal to the average brightness value in the predetermined region to which the pixel of interest belongs.
Within the first set number of pixels arranged on one side in the first direction of the pixel of interest, the brightness value of a pixel having a smaller brightness value than the pixel of interest and as far as possible from the pixel of interest, or the brightness of a pixel as close as possible to the pixel of interest. The first brightness value, which is a value, is calculated, and the brightness of the pixels having a brightness value smaller than that of the pixel of interest and as far as possible from the pixel of interest within the second set number of pixels arranged on the other side of the first direction of the pixel of interest. The second brightness value, which is the value or the brightness value of the pixel as close as possible to the pixel of interest, is calculated, and the pixel is within the third set number of pixels arranged on one side of the second direction, which is a direction different from the first direction of the pixel of interest. A third brightness value, which is a brightness value of a pixel having a brightness value smaller than that of the pixel of interest and as far as possible from the pixel of interest or a brightness value of a pixel as close as possible to the pixel of interest, is calculated, and the other side of the pixel of interest in the second direction. Among the fourth set number of pixels lined up in, the fourth brightness value, which is the brightness value of a pixel whose brightness value is smaller than that of the pixel of interest and is as far as possible from the pixel of interest, or the brightness value of a pixel as close as possible to the pixel of interest, is calculated. And
If a predetermined condition is satisfied, the pixel of interest is specified as the position of a foreign object, and the pixel of interest is specified.
The predetermined conditions are
(A) The first condition that the ratio of the brightness value of the pixel of interest to the first brightness value and the ratio of the brightness value of the pixel of interest to the second brightness value are each equal to or higher than the first threshold value.
(B) The second condition that the ratio of the brightness value of the pixel of interest to the third brightness value and the ratio of the brightness value of the pixel of interest to the fourth brightness value are each equal to or higher than the second threshold value.
(C) The first condition and the second condition,
The inspection device according to any one of claims 1 to 4, which is any one of the above.
The specific part is
A division unit that divides an image based on X-ray photography of the object to be inspected into predetermined areas, and
A calculation unit that calculates the average luminance value in each of the predetermined regions,
With
When the brightness value of the pixel of interest is greater than or equal to the average brightness value in the predetermined region to which the pixel of interest belongs.
A first set number of pixels arranged on one side of the pixel of interest in the first direction separated from the pixel of interest by a first predetermined position, and a second set number of pixels arranged on the other side of the pixel of interest in the first direction separated by a second predetermined position. The first brightness value, which is the average brightness value of the pixels, is calculated, and a third set number of pixels are arranged on one side of the second direction, which is a direction different from the first direction of the pixel of interest, at a third predetermined position from the pixel of interest. And the second brightness value, which is the average brightness value of the fourth set number of pixels arranged at the other side in the second direction of the pixel of interest at a fourth predetermined position from the pixel of interest, is calculated.
If a predetermined condition is satisfied, the pixel of interest is specified as the position of a foreign object, and the pixel of interest is specified.
The predetermined conditions are
(A) The first condition that the ratio of the brightness value of the pixel of interest to the first brightness value is equal to or higher than the first threshold value.
(B) The second condition that the ratio of the brightness value of the pixel of interest to the second brightness value is equal to or higher than the second threshold value.
(C) The first condition and the second condition,
The inspection device according to any one of claims 1 to 4, which is any one of the above.
It is an inspection method using an X-ray imaging apparatus including an X-ray irradiation unit and an X-ray detection unit.
A storage step of storing a two-dimensional X-ray image obtained by X-raying an object to be inspected by the X-ray apparatus, and a storage step.
An image obtained by projecting the two-dimensional X-ray photographed image onto a fault set at a predetermined position other than the position of the X-ray detection unit is generated, and then the generated image is reconstructed into a projected image which is a two-dimensional image. Reconstruction steps and
A specific step of identifying the position of a foreign object based on the projected image, and
With the removal step,
With
In the storage step, a plurality of two-dimensional X-ray images having different positional relationships between the object to be inspected, the X-ray irradiation unit used for the X-ray imaging, and the X-ray detection unit are stored.
In the reconstruction step, images obtained by projecting each of the plurality of two-dimensional X-ray images onto a plurality of tomographic images having different depths are generated, and then each of the generated images is reconstructed into the projected images.
In the specific step, each of the plurality of projected images is converted into a binarized projected image in which the position of the foreign matter is specified.
In the removal step, a plurality of the binarized projected images obtained from the images projected on the tomography of the same depth are compared with each other, and the erroneously detected portion of the foreign matter is removed based on the comparison result. An inspection method characterized by.