X-ray image preprocessing method and X-ray image preprocessing system

Abstract

The invention provides an X-ray image preprocessing method and an X-ray image preprocessing system. The X-ray image preprocessing method comprises the following steps: acquiring an X-ray image; acquiring a plurality of regions from the X-ray image; acquiring an X-ray image representative value of a pixel value of each of the plurality of regions; and performing contrast adjustment on the X-ray image so that each of the X-ray image representative values becomes a predetermined value acquired on the basis of the input teacher data when the learning completion model is created, thereby acquiring a contrast-adjusted X-ray image.

Description

Technical Field

[0001] This invention relates to an X-ray image preprocessing method and system, and particularly to an X-ray image preprocessing method and system for analyzing X-ray images using a learned model. Background Technology

[0002] Previously, devices for analyzing X-ray images using learned models were known. For example, such a device for analyzing X-ray images was disclosed in Japanese Patent Application Publication No. 2024-029975.

[0003] Japanese Patent Application Publication No. 2024-029975 discloses an X-ray imaging system that uses a learned model to identify at least one of a region of an object to be inspected and a region of abnormality contained within the object. The X-ray imaging system disclosed in Japanese Patent Application Publication No. 2024-029975 includes a fluoroscopic device and an analysis device. In the structure disclosed in Japanese Patent Application Publication No. 2024-029975, the analysis device analyzes the X-ray image generated by the fluoroscopic device capturing the object to be inspected. Specifically, in the structure disclosed in Japanese Patent Application Publication No. 2024-029975, the analysis device is configured to identify a region of the object to be inspected by inputting an X-ray image into a learned model.

[0004] Here, although not described in Japanese Patent Application Publication No. 2024-029975, in the process of identifying (obtaining) a region of an object from an X-ray image using a learned model, the user sets the shooting conditions, including at least one of tube voltage and tube current, to be equal when taking X-ray images for both generating teacher data (teacher X-ray images) and taking X-ray images for analysis using the learned model. However, with the evolution of X-ray imaging devices over the years, even when the user sets the tube voltage and tube current to irradiate with X-rays of a specified energy and dose, the actual energy and dose of X-rays irradiated from the X-ray irradiation unit may sometimes be lower than the user's intended energy and dose. In this case, since the shooting conditions for creating the learned model and the shooting conditions for taking X-ray images are different, the analytical accuracy of the processing (analysis) of obtaining the region of the object using the learned model is reduced. Furthermore, when the X-ray irradiation unit is changed after capturing the teacher data used to generate the learned model, even if the tube voltage and current values ​​are set to the values ​​used when capturing the teacher data, the energy and dose of X-rays actually irradiated from the X-ray irradiation unit during X-ray image capture may sometimes be higher than those used when capturing the teacher data. In this case, because the capturing conditions when creating the learned model are different from the capturing conditions when capturing the X-ray image, the analysis accuracy of processing (analyzing) the area of ​​the object to be examined using the learned model decreases. In addition, due to changes in X-ray imaging equipment over the years, the capturing conditions set by the user may differ from the actual capturing conditions. Therefore, even if multiple learned models are generated based on teacher data captured using multiple capturing conditions, it is difficult to select the learned model generated based on teacher data captured under capturing conditions consistent with the actual capturing conditions. Therefore, a technique is desired that can suppress the decrease in processing accuracy of the area of ​​the object to be examined even when the capturing conditions when capturing the teacher data used to generate the learned model are different from the capturing conditions when capturing the X-ray image of the object to be analyzed in the learned model. Summary of the Invention

[0005] The present invention was made to solve the aforementioned problems. One object of the present invention is to provide an X-ray image preprocessing method and an X-ray image preprocessing system that can suppress the reduction in processing accuracy of the area of ​​the object to be examined by using the learned model, even when the shooting conditions when shooting teacher data used to generate the learned model are different from the shooting conditions when shooting the X-ray image of the object to be analyzed by the learned model.

[0006] To achieve the aforementioned objective, the X-ray image preprocessing method of the first aspect of the present invention is performed before processing the region of the object to be examined obtained from the X-ray image of the object to be examined using a learned model. The X-ray image preprocessing method includes the following steps: obtaining an X-ray image; obtaining multiple regions from the X-ray image based on the pixel values ​​of the X-ray image; obtaining X-ray image representative values ​​of the pixel values ​​of each of the multiple regions; and obtaining a contrast-adjusted X-ray image by adjusting the contrast of the X-ray image in such a way that the X-ray image representative values ​​of each of the multiple regions of the X-ray image become predetermined values ​​obtained from teacher data when the learned model was created.

[0007] The X-ray image preprocessing system of the second aspect of the present invention is performed before processing the acquisition of regions of an object to be examined from an X-ray image of an object to be examined using a learned model. The X-ray image preprocessing system includes an X-ray imaging device having an X-ray irradiation unit for irradiating X-rays and an X-ray detector for detecting X-rays irradiated from the X-ray irradiation unit; and an image processing device for generating an X-ray image. The image processing device performs the following control: generating an X-ray image; and acquiring multiple regions from the X-ray image based on the pixel values ​​of the X-ray image, and acquiring X-ray image representative values ​​of the pixel values ​​of each of the multiple regions, and adjusting the contrast of the X-ray image in such a way that the X-ray image representative values ​​of each of the multiple regions of the X-ray image become predetermined values ​​obtained based on teacher data when the learned model was created, thereby obtaining a contrast-adjusted X-ray image.

[0008] Here, due to changes in the X-ray imaging equipment over the years, even if the user sets the tube voltage and tube current to irradiate with the specified energy and dose of X-rays, at least one of the energy and dose of the X-rays actually irradiated from the X-ray irradiation unit may sometimes decrease. Therefore, due to these changes, the contrast between multiple regions in the X-ray image is reduced because the X-ray image is captured under conditions where at least one of the X-ray energy and dose is lower than the user-set imaging conditions. Furthermore, if the X-ray irradiation unit is replaced after capturing the teacher's data used to generate the learned model, even if the user sets the tube voltage and tube current values ​​to the same values ​​as when capturing the teacher's data, at least one of the energy and dose of the X-rays actually irradiated from the X-ray irradiation unit may sometimes be higher than when capturing the teacher's data. In this case, the contrast between multiple regions in the X-ray image is greater than the contrast between multiple regions in the teacher's data because the X-ray image is captured under conditions where at least one of the X-ray energy and dose is higher than the contrast between multiple regions in the teacher's data.

[0009] Therefore, in the X-ray image preprocessing method of the first aspect and the X-ray image preprocessing system of the second aspect, the contrast of the X-ray image is adjusted by using the representative X-ray values ​​of each region of the X-ray image as predetermined values ​​obtained from the teacher data when creating the learned model, thereby obtaining a contrast-adjusted X-ray image. Thus, regardless of whether at least one of the energy or dose of the X-ray irradiated from the X-ray irradiation unit decreases, resulting in a lower contrast between regions in the X-ray image than the contrast between regions in the teacher data, or a higher contrast between regions in the X-ray image than the contrast between regions in the teacher data, a contrast-adjusted X-ray image with the representative values ​​of each region as predetermined values ​​can be obtained. Therefore, the contrast between regions in the X-ray image can be made close to the contrast between regions in the teacher data. As a result, even when the shooting conditions when capturing the teacher data used to generate the learned model are different from the shooting conditions when capturing the X-ray image of the object to be analyzed using the learned model, the reduction in processing accuracy of the region of the object to be examined using the learned model can be suppressed. Furthermore, the ability to suppress the reduction in processing accuracy in the area where the object to be inspected is obtained has been confirmed in experiments conducted by the inventors of this application, as described later. Attached Figure Description

[0010] Figure 1 This is a block diagram illustrating the overall structure of an X-ray imaging system based on an embodiment of the present invention.

[0011] Figure 2 This is a diagram showing an example of an object being inspected.

[0012] Figure 3 It is a block diagram used to illustrate the functional structure of the control unit.

[0013] Figure 4 This is a diagram showing an example of an X-ray image.

[0014] Figure 5 This diagram illustrates the structure of generating a learned model and the structure of using the learned model to analyze X-ray images.

[0015] Figure 6 This is a diagram used to illustrate the structure of the control unit performing logarithmic transformation processing.

[0016] Figure 7 This is a diagram used to illustrate the structure of the solder ball area and the background area obtained by the control unit.

[0017] Figure 8 This diagram illustrates the structure by which the control unit obtains the first average pixel value and the second average pixel value.

[0018] Figure 9 This diagram illustrates the structure of the control unit that acquires X-ray images after contrast adjustment.

[0019] Figure 10 This is a flowchart illustrating the structure of the control unit acquiring the contrast-adjusted X-ray image.

[0020] Figure 11 This is a diagram used to illustrate the results obtained by analyzing X-ray images based on a first comparative example using the learned model.

[0021] Figure 12 This is a figure used to illustrate the results obtained by analyzing the contrast-adjusted X-ray image based on the first embodiment using the learned model.

[0022] Figure 13 This is a figure used to illustrate the results obtained by analyzing X-ray images based on a second comparative example using the learned model.

[0023] Figure 14 This is a figure used to illustrate the results obtained by analyzing the contrast-adjusted X-ray image based on the second embodiment using the learned model. Detailed Implementation

[0024] Hereinafter, embodiments embodying the present invention will be described with reference to the accompanying drawings.

[0025] (Overall structure of the X-ray imaging system)

[0026] Reference Figures 1-10 An X-ray image preprocessing system 100 based on an embodiment of the present invention will be described.

[0027] like Figure 1 As shown, the X-ray image preprocessing system 100 based on this embodiment is a system for imaging the interior of the object to be inspected 90 by detecting X-rays passing through it. The X-ray image preprocessing system 100 is used, for example, in non-destructive inspection applications for imaging the interior of the object to be inspected 90.

[0028] like Figure 2As shown, the object 90 under inspection is an electronic device including a substrate 91. Electronic components 92 are mounted on the substrate 91. The electronic components 92 are electrically connected to the substrate 91 via multiple solder balls 93 (bumps). The multiple solder balls 93 are arranged in a regular pattern. Specifically, the multiple solder balls 93 are arranged in a grid pattern on the substrate 91 in a regular manner. That is, the electronic components 92 are connected to the substrate 91 via a ball grid array (BGA). Furthermore, multiple solder balls 93 are arranged on one side of the substrate 91. The electronic components 92 include, for example, integrated circuits (ICs). In the X-ray image preprocessing system 100, the X-ray image 40 (refer to...) is... Figure 4 The X-ray image 40 is used for preprocessing and is used for non-destructive inspection of anomalies such as voids (holes) and bridging of multiple solder balls 93. In addition, in addition to electronic components 92, electronic components 94 such as surface-mount resistors or capacitors are also mounted on the substrate 91.

[0029] like Figure 1 As shown, the X-ray image preprocessing system 100 includes an X-ray imaging device 1 and an image processing device 2. The X-ray imaging device 1 performs X-ray imaging on the object 90 to be inspected. Additionally, the image processing device 2 generates an X-ray image 40 (see reference). Figure 4 In addition, the image processing device 2 performs analysis processing on the generated X-ray image 40 using the learned model 31, as well as preprocessing of the analysis processing. That is, the X-ray image preprocessing system 100 is an X-ray image preprocessing system performed before processing the area of ​​the object to be inspected 90 obtained from the X-ray image 40 of the object to be inspected using the learned model. The X-ray imaging device 1 and the image processing device 2 each have a communication module and exchange information via a network or the like.

[0030] The X-ray imaging apparatus 1 includes an X-ray irradiation unit 10 and an X-ray detector 11. The X-ray irradiation unit 10 is configured to irradiate X-rays. In this embodiment, the X-ray irradiation unit 10 irradiates an object 90, which includes a plurality of solder balls 93, with X-rays. The X-ray irradiation unit 10 includes an X-ray tube that is powered by a power supply device (not shown) to irradiate X-rays.

[0031] The X-ray detector 11 detects the X-rays irradiated from the X-ray irradiation unit 10. The X-ray detector 11 outputs an electrical signal corresponding to the detected X-rays. The X-ray detector 11 includes, for example, a flat panel detector (FPD) that serves as an X-ray detector. The X-ray irradiation unit 10 and the X-ray detector 11 are disposed inside a frame (not shown) of the X-ray imaging apparatus 1.

[0032] like Figure 1 As shown, the image processing apparatus 2 includes a control unit 20 and a storage unit 21. The image processing apparatus 2 is, for example, a personal computer communicatively connected to the X-ray imaging apparatus 1. The control unit 20 controls the operation of various parts of the X-ray imaging apparatus 1. For example, the control unit 20 controls the X-ray irradiation performed by the X-ray irradiation unit 10 by controlling a power supply device (not shown). The control unit 20 includes a central processing unit (CPU), read-only memory (ROM), and random access memory (RAM), etc. Alternatively, the control unit 20 may also include a graphics processing unit (GPU) or a processor configured for image processing purposes, such as a field-programmable gate array (FPGA).

[0033] The storage unit 21 is configured to store various programs 30 and parameters executed by the storage control unit 20. Furthermore, the storage unit 21 is configured to store the learned completion model 31 and the specified value 70. Details regarding the learned completion model 31 and the specified value 70 will be described later. The storage unit 21 may include, for example, non-volatile memory such as a hard disk drive (HDD) or a solid-state drive (SSD).

[0034] Additionally, the image processing apparatus 2 is connected to a display unit 22 and an operation unit 23. The display unit 22 includes, for example, a liquid crystal monitor. The display unit 22 displays images and text information under control by the control unit 20. The operation unit 23 receives input operations performed by the operator. The operation unit 23 includes, for example, a pointing device such as a keyboard and mouse. The operation unit 23 outputs operation signals to the control unit 20 based on the received input operations.

[0035] like Figure 3 As shown, the control unit 20 includes an image generation unit 20a, a logarithmic processing unit 20b, a region acquisition unit 20c, a representative value acquisition unit 20d, and a contrast adjustment unit 20e, which are function blocks. The image generation unit 20a, logarithmic processing unit 20b, region acquisition unit 20c, representative value acquisition unit 20d, and contrast adjustment unit 20e are executed by the control unit 20 and stored in the storage unit 21 (see reference 20). Figure 1 Program 30 in ) (refer to) Figure 1The program 30 is configured in software form to implement the functional blocks. In other words, the program 30 is configured to enable the computer (control unit 20) to perform the controls performed by the image generation unit 20a, the logarithmic processing unit 20b, the region acquisition unit 20c, the representative value acquisition unit 20d, and the contrast adjustment unit 20e.

[0036] The image generation unit 20a, logarithmic processing unit 20b, region acquisition unit 20c, representative value acquisition unit 20d, and contrast adjustment unit 20e may also be equipped with dedicated processors (processing circuits), containing independent hardware.

[0037] Image generation unit 20a is based on X-ray detector 11 (reference) Figure 1 X-ray images generated from detected X-rays 40 (reference) Figure 4 ).

[0038] The details of the functions of the logarithmic processing unit 20b, the region acquisition unit 20c, the representative value acquisition unit 20d, and the contrast adjustment unit 20e will be described later.

[0039] (X-ray image)

[0040] like Figure 4 As shown, X-ray image 40 captures a plurality of solder balls 93 arranged in a grid pattern in a regular manner on substrate 91.

[0041] (Model completed)

[0042] like Figure 5 As shown, the learned model 31 is used to analyze X-ray image 40. Specifically, the learned model 31 is used to identify solder balls 93 (see reference). Figure 4 The area is the solder ball area 80a (refer to...) Figure 7 After learning, model 31 is processed by image processing device 2 (refer to...). Figure 1 (or generated by a computer different from the image processing device 2 and pre-stored in the storage unit 21 (see reference)) Figure 1 )middle.

[0043] In addition, such as Figure 5 As shown, a learned model 31 is generated through machine learning using input teacher data 32 and output teacher data 33 as datasets. Input teacher data 32 is generated based on teacher X-ray images (not shown). The teacher X-ray images and the analyzed X-ray image 40 (see reference) Figure 4 Similarly, it is generated by the X-ray imaging device 1. The analyzed X-ray image 40 and the teacher's X-ray image used to generate the completed learning model 31 are images containing the inspection object 90 (solder ball 93) having a common structure. Furthermore, the input teacher data 32 is an example of the "teacher's data" of the claim.

[0044] Additionally, the output teacher data 33 is generated based on the teacher's X-ray image. The output teacher data 33 is generated by the user labeling the area of ​​the solder ball 93 captured in the teacher's X-ray image.

[0045] Model 31, after learning, was generated using deep learning machine learning. Deep learning includes, for example, machine learning based on U-Net, which is a Fully Convolutional Network (FCN). Model 31, after learning, can identify the region of solder ball 93 (solder ball region 80a) and the background region 81a (see reference) outside the solder ball 93 by performing pixel-by-pixel analysis on the input X-ray image 40. Figure 7 It is generated by learning the image transformation (image reconstruction) method.

[0046] like Figure 5 As shown, after learning, model 31 is configured to output a recognition result image 60 when an X-ray image 40 is input. The recognition result image 60 is a label image of the region identified as the solder ball 93 in the X-ray image 40, as well as the other regions.

[0047] Here, when analyzing X-ray image 40 using the learned model 31, the imaging conditions (tube voltage and tube current) for capturing the input teacher data 32 used to generate the learned model 31 are made consistent with the imaging conditions for capturing X-ray image 40. However, with the X-ray imaging apparatus 1 (refer to...) Figure 1 During the period of use, the X-ray irradiation unit 10 (refer to) Figure 1Over time, the X-ray irradiation unit 10 may change. Even if the user sets the tube voltage and current to provide the required energy and dose of X-rays, the X-ray irradiation unit 10 may still irradiate with lower energy and dose than intended. Furthermore, if the X-ray irradiation unit 10 is replaced after the input teacher data 32 is captured, even if the user sets the same tube voltage and current as when the input teacher data 32 was captured, the actual energy and dose of the X-rays irradiated from the X-ray irradiation unit 10 may be higher than when the input teacher data 32 was captured. In other words, even if the user sets the same tube voltage and current, the conditions under which the teacher data is captured when the learning model 31 is generated may differ from the actual conditions under which the X-ray image 40 is captured. When the actual conditions under which the X-ray image 40 is captured differ from the conditions under which the teacher data is captured when the learning model 31 is generated, the properties of the obtained image change non-linearly. Therefore, if the shooting conditions when capturing the teacher's data differ from those when capturing the X-ray image 40, the analysis accuracy based on the learned model 31 will decrease.

[0048] (Obtaining the X-ray image after contrast adjustment)

[0049] Therefore, in this embodiment, by preprocessing the X-ray image 40, a contrast-adjusted X-ray image 44 (see reference) is obtained. Figure 9 Then, the contrast-adjusted X-ray image 44 is used for analysis using the learned model 31.

[0050] Next, refer to Figures 6-9 For control unit 20 (refer to) Figure 3 Obtain the contrast-adjusted X-ray image 44 (reference). Figure 9 The structure of the control unit 20 acquiring the contrast-adjusted X-ray image 44 generally includes: a structure for performing logarithmic transformation on the X-ray image 40; a structure for acquiring multiple regions from the X-ray image 40; and a structure for acquiring representative X-ray image values ​​50 from each of the multiple regions (see reference). Figure 8 The structure of the X-ray image 40 and the structure for adjusting the contrast of the X-ray image 40 based on the representative value 50 of the X-ray image.

[0051] First, refer to Figure 6 The structure of the logarithmic processing unit 20b performing logarithmic processing on the X-ray image 40 to obtain the logarithmically transformed X-ray image 41 will be explained below.

[0052] In this embodiment, the logarithmic processing unit 20b performs logarithmic transformation on the pixels of each of the 40 X-ray images using the natural logarithm (logarithm to the base Napier's constant e). As a result, the change in pixel values ​​of the X-ray image changes from exponential to linear. Consequently, the degree of change in pixel values ​​in the low-dose region increases, and consequently, the resolution of the low-dose region is improved. Furthermore, in Figure 6 In the example shown, by making the shadows marked on the background of the X-ray image 40 before logarithmic transformation different from the shadows marked on the background of the X-ray image 41 after logarithmic transformation, the resolution of the low-dose portion of the X-ray image 41 after logarithmic transformation is shown to be improved.

[0053] Next, refer to Figure 7 This explains the structure of the region acquisition unit 20c, which acquires multiple regions.

[0054] like Figure 7 As shown, the area acquisition unit 20c is configured to acquire an object 90 including the object to be inspected (see reference). Figure 1 The regions are defined as follows: the first region 80 and the second region 81, which is the region outside the object being inspected 90. In this embodiment, the region acquisition unit 20c is configured to acquire the first region 80 and the second region 81 by performing binarization processing on the logarithmically transformed X-ray image 41. The region acquisition unit 20c acquires the first region 80 and the second region 81, for example, by using Otsu binarization processing (discriminant analysis method) that can automatically calculate the threshold for binarization.

[0055] In this embodiment, the first region 80 is the solder ball region 80a containing multiple solder balls 93. The second region 81 is the background region 81a excluding the solder ball region 80a. That is, in this embodiment, the region acquisition unit 20c acquires the solder ball region 80a and the background region 81a from the logarithmically transformed X-ray image 41 by performing binarization processing. Furthermore, in this embodiment, the solder ball region 80a and the background region 81a are roughly divided by Otsu binarization, therefore the solder ball region 80a may contain electronic components 94 such as chip capacitors (see reference). Figure 2 ).

[0056] Figure 7 The first region image 42 shown is an image representing the solder ball region 80a. In the first region image 42, the solder ball region 80a is illustrated in white. In other words, the first region image 42 is a mask image of the solder ball region 80a.

[0057] Additionally, the second region image 43 is an image representing the background region 81a. In the second region image 43, the background region 81a is illustrated in white. In other words, the second region image 43 is a mask image of the background region 81a.

[0058] Next, refer to Figure 8 This section explains the structure of the representative value acquisition unit 20d, which acquires the representative value 50 of the X-ray image.

[0059] The representative value acquisition unit 20d is configured to acquire the X-ray image representative value 50 of the solder ball region 80a and the X-ray image representative value 50 of the background region 81a. In this embodiment, the representative value acquisition unit 20d is configured to acquire the average pixel value of the first region 80, i.e., the first average pixel value 50a, and the average pixel value of the second region 81, i.e., the second average pixel value 50b, as the X-ray image representative value 50.

[0060] Specifically, the representative value acquisition unit 20d acquires the solder ball region 80a in the X-ray image 41 based on the logarithmically transformed X-ray image 41 and the first region image 42. Furthermore, the representative value acquisition unit 20d acquires the average pixel value of each pixel contained in the solder ball region 80a in the X-ray image 41 as a first average pixel value 50a.

[0061] Furthermore, the representative value acquisition unit 20d acquires the background region 81a in the X-ray image 41 based on the logarithmically transformed X-ray image 41 and the second region image 43. Moreover, the representative value acquisition unit 20d acquires the average pixel value of each pixel contained in the background region 81a in the X-ray image 41 as the second average pixel value 50b.

[0062] Next, refer to Figure 9 To illustrate the contrast adjustment unit 20e's effect on the X-ray image 40 (refer to...) Figure 4 The structure adjusts the contrast.

[0063] The contrast adjustment unit 20e is configured to be based on the first region 80 (refer to...) Figure 7 The X-ray image representative value of 50 (reference) Figure 8 ) and the second area 81 (refer to) Figure 7The contrast adjustment unit 20e adjusts the contrast of the logarithmically transformed X-ray image 41 by taking the representative X-ray image value 50 of each region of the logarithmically transformed X-ray image 41 as a predetermined value 70 obtained from the input teacher data 32 when the learning model 31 was created, thereby obtaining a contrast-adjusted X-ray image 44. Specifically, the contrast adjustment unit 20e is configured to adjust the contrast of the logarithmically transformed X-ray image 41 by taking the representative X-ray image value 50 of each region of the logarithmically transformed X-ray image 41 as a predetermined value 70 based on the first average pixel value 50a and the second average pixel value 50b, thereby obtaining a contrast-adjusted X-ray image 44.

[0064] The specified value of 70 is based on the input teacher data 32 (refer to). Figure 5 The representative values ​​of the pixels in each of the multiple regions of the input teacher data 32 are obtained, namely, the teacher data representative values. The teacher data representative values ​​are, for example, the average value (average pixel value) of the pixel values ​​in each of the multiple regions of the input teacher data 32. The specified value 70 is obtained when the learning model 31 is completed and is stored in the storage unit 21.

[0065] In this embodiment, the contrast adjustment unit 20e is configured to adjust the pixel values ​​of the logarithmically transformed X-ray image 41 by setting a predetermined value 70 for both the first average pixel value 50a and the second average pixel value 50b, thereby obtaining a contrast-adjusted X-ray image 44. The contrast adjustment unit 20e performs contrast adjustment on the logarithmically transformed X-ray image 41 such that the value of the first average pixel value 50a is within 2 / 3 of the lower grayscale range of the logarithmically transformed X-ray image 41, and the second average pixel value 50b is within 1 / 3 of the higher grayscale range of the logarithmically transformed X-ray image 41. Specifically, the contrast adjustment unit 20e performs contrast adjustment on the logarithmically transformed X-ray image 41 based on the following equation (1).

[0066] [Formula 1]

[0067]

[0068] Here, I i，j J is the pixel value (brightness value) of the pixel in the i-th row and j-th column of the input image I (the logarithmically transformed X-ray image 41). Additionally, J... i，j It is the pixel value (luminance value) of the pixel in the i-th row and j-th column of the contrast-adjusted image J (contrast-adjusted X-ray image 44). Additionally, μ t I and μ b IThese are the first average pixel value 50a and the second average pixel value 50b, respectively. Additionally, c... t and c b These are the specified values ​​70 (average pixel value) for the first region 80 (solder ball region 80a) and the second region 81 (background region 81a), respectively. Furthermore, the specified value 70 for the first region 80 (solder ball region 80a) is a representative value (average pixel value) of the pixel values ​​of the solder ball region in the input teacher data 32. Similarly, the specified value 70 for the second region 81 (background region 81a) is a representative value (average pixel value) of the pixel values ​​of the background region in the input teacher data 32.

[0069] Next, refer to Figure 10 For control unit 20 (refer to) Figure 1 Obtain the contrast-adjusted X-ray image 44 (reference). Figure 9 The X-ray image preprocessing method is explained.

[0070] In step 101, the image generation unit 20a (refer to...) Figure 3 ) Obtain X-ray images 40 (refer to) Figure 4 In this embodiment, in step 101, the image generation unit 20a uses an X-ray detector 11 (see reference 11) to generate the image. Figure 1 The detected X-rays generate an X-ray image 40, thereby obtaining an X-ray image 40.

[0071] Next, in step 102, as Figure 6 As shown, the logarithmic processing unit 20b performs logarithmic transformation on each pixel of the X-ray image 40. In this embodiment, the logarithmic processing unit 20b obtains the logarithmically transformed X-ray image 41 by performing logarithmic transformation on the X-ray image 40.

[0072] Next, in step 103, as Figure 7 As shown, the region acquisition unit 20c acquires multiple regions from the X-ray image 40 based on the pixel values ​​of the X-ray image 40. In this embodiment, the region acquisition unit 20c acquires two regions, the solder ball region 80a and the background region 81a, by binarizing the logarithmically transformed X-ray image 41.

[0073] Next, in step 104, as Figure 8 As shown, the representative value acquisition unit 20d acquires X-ray image representative values ​​50 for the pixel values ​​of multiple regions. In this embodiment, the representative value acquisition unit 20d acquires a first average pixel value 50a from the solder ball region 80a and a second average pixel value 50b from the background region 81a, as the X-ray image representative value 50.

[0074] Next, in step 105, the contrast adjustment unit 20e obtains a predetermined value 70 (refer to...). Figure 1 Specifically, the contrast adjustment unit 20e acquires a predetermined value 70 stored in the storage unit.

[0075] Next, in step 106, as Figure 9 As shown, the contrast adjustment unit 20e adjusts the contrast of the X-ray image 40 by making the representative X-ray image values ​​50 of each of the multiple regions of the X-ray image 40 a predetermined value 70 obtained based on the input teacher data 32 when the learning model 31 was created, thereby obtaining a contrast-adjusted X-ray image 44. In this embodiment, the contrast adjustment unit 20e obtains the contrast-adjusted X-ray image 44 based on the logarithmically transformed X-ray image 41, the first average pixel value 50a, and the second average pixel value 50b. Then, the processing ends.

[0076] (Effects of this implementation method)

[0077] In this embodiment, the following effects can be achieved.

[0078] In this embodiment, as described above, the X-ray image preprocessing method is performed before processing the region of the object 90 to be examined from the X-ray image 40 of the object 90 using the learned model 31. The X-ray image preprocessing method includes the following steps: obtaining the X-ray image 40; obtaining multiple regions from the X-ray image 40 based on the pixel values ​​of the X-ray image 40; obtaining X-ray image representative values ​​50 for the pixel values ​​of each of the multiple regions; and obtaining a contrast-adjusted X-ray image 44 by adjusting the contrast of the X-ray image 40 in such a way that the X-ray image representative values ​​50 for each of the multiple regions of the X-ray image 40 become predetermined values ​​70 obtained based on the input teacher data 32 when the learned model 31 was created.

[0079] Here, given the changes in the X-ray imaging device 1 over the years, even if the user sets the tube voltage and tube current to irradiate with the specified energy and dose of X-rays, at least one of the energy and dose of the X-rays actually irradiated from the X-ray irradiation unit may sometimes be lower. Therefore, given the changes in the X-ray imaging device 1 over the years, since the X-ray image 40 is captured under imaging conditions where at least one of the X-ray energy and dose is lower than the imaging conditions set by the user, the contrast between multiple regions in the X-ray image 40 is reduced. Furthermore, if the X-ray irradiation unit 10 is replaced after capturing the input teacher data 32 used when generating the learned model 31, even if the user sets the tube voltage and tube current values ​​to be equal to those set when capturing the input teacher data 32, at least one of the energy and dose of the X-rays actually irradiated from the X-ray irradiation unit 10 may sometimes be higher than at least one of the energy and dose of the X-rays irradiated from the X-ray irradiation unit 10 when capturing the input teacher data 32. In this case, since the X-ray image 40 is captured under shooting conditions where at least one of the X-ray energy and dose is higher than the shooting conditions when the input teacher data 32 is captured, the contrast between multiple regions in the X-ray image 40 is greater than the contrast between multiple regions in the input teacher data 32.

[0080] Therefore, as described above, by adjusting the contrast of the X-ray image 40 so that the representative X-ray image values ​​50 of each region of the X-ray image 40 are predetermined values ​​70 obtained from the input teacher data 32 when the learning model 31 was created, a contrast-adjusted X-ray image 44 is obtained. Thus, regardless of whether at least one of the energy or dose of the X-ray irradiated from the X-ray irradiation unit decreases, resulting in a lower contrast between the regions of the X-ray image 40 than the contrast between the regions of the input teacher data 32, or a higher contrast between the regions of the X-ray image 40 than the contrast between the regions of the input teacher data 32, a contrast-adjusted X-ray image 44 can be obtained where the representative X-ray image values ​​50 of each region are predetermined values ​​70 obtained from the input teacher data 32 when the learning model 31 was created. Therefore, the contrast between the regions of the X-ray image 40 can be made close to the contrast between the regions of the input teacher data 32. As a result, even when the shooting conditions for capturing the teacher data used to generate the learned model 31 are different from the shooting conditions for capturing the X-ray image 40, the object of analysis of the learned model 31, the reduction in processing accuracy of acquiring (identifying) the area of ​​the inspection object 90 using the learned model 31 can be suppressed. Furthermore, the ability to suppress the reduction in processing accuracy of acquiring the area of ​​the inspection object 90 has been confirmed in experiments conducted by the inventors of this application, as described later.

[0081] Furthermore, in this embodiment, as described above, program 30 is an X-ray image preprocessing program performed before the processing of obtaining regions of the object 90 from the X-ray image 40 of the object 90 using the learned model 31, and the computer performs the following control: obtaining the X-ray image 40; obtaining multiple regions from the X-ray image 40 based on the pixel values ​​of the X-ray image 40; obtaining X-ray image representative values ​​50 for the pixel values ​​of each of the multiple regions; and obtaining a contrast-adjusted X-ray image 44 by adjusting the contrast of the X-ray image 40 in such a way that the X-ray image representative values ​​50 for each of the multiple regions of the X-ray image 40 become predetermined values ​​70 obtained based on the input teacher data 32 when the learned model 31 was created.

[0082] Therefore, a procedure 30 can be provided that, similar to the X-ray image preprocessing method, can suppress the reduction in processing accuracy of the area of ​​the object to be examined 90 obtained by using the learned model 31, even when the shooting conditions for capturing teacher data used to generate the learned model 31 are different from the shooting conditions for capturing the X-ray image 40, the analysis object of the learned model 31.

[0083] In addition, in this embodiment, the X-ray image preprocessing system 100 is an X-ray image preprocessing system performed before processing the acquisition of the region of the object 90 from the X-ray image 40 of the object 90 using the learned model 31, and includes: an X-ray imaging device 1 having an X-ray irradiation unit 10 for irradiating X-rays and an X-ray detector 11 for detecting X-rays irradiated from the X-ray irradiation unit 10; and an image processing device 2 for generating the X-ray image 40. The image processing device 2 performs the following control: generating the X-ray image 40; and obtaining multiple regions from the X-ray image 40 based on the pixel values ​​of the X-ray image 40, and obtaining X-ray image representative values ​​50 for the pixel values ​​of each of the multiple regions, and adjusting the contrast of the X-ray image 40 in such a way that the X-ray image representative values ​​50 for each of the multiple regions of the X-ray image 40 become a predetermined value 70, thereby obtaining a contrast-adjusted X-ray image 44.

[0084] Therefore, an X-ray image preprocessing system 100 can be provided, which, similar to the X-ray image preprocessing method, can suppress the reduction in processing accuracy of the area of ​​the object to be examined 90 obtained by using the learned model 31, even when the shooting conditions for shooting teacher data used to generate the learned model 31 are different from the shooting conditions for shooting the analysis object of the learned model 31, i.e., the X-ray image 40.

[0085] Furthermore, in this embodiment, the following further effects can be obtained through the following configuration.

[0086] That is, in this embodiment, as described above, the predetermined value 70 is obtained based on the representative values ​​of the pixels of each region in the input teacher data 32 (teacher data) when the learning completed model 31 is created. Therefore, since the predetermined value 70 is the representative value of the pixels of each region in the input teacher data 32 (teacher data), the contrast of the X-ray image can be easily adjusted so that the contrast between the multiple regions of the X-ray image 40 is close to the contrast between the multiple regions in the input teacher data 32 (teacher data) by adjusting the contrast of the X-ray image in such a way that the representative values ​​50 of the X-ray image of each region of the X-ray image become the predetermined value 70. As a result, even when the shooting conditions when shooting the input teacher data 32 (teacher data) are different from the shooting conditions when shooting the X-ray image 40, the object of analysis of the learning completed model 31, the reduction in processing accuracy of obtaining (identifying) the area of ​​the inspection object 90 using the learning completed model 31 can be easily suppressed.

[0087] In the step of acquiring multiple regions from X-ray image 40, two regions are acquired: the region containing the object to be inspected 90, namely the first region 80, and the region excluding the object to be inspected 90, namely the second region 81. In the step of acquiring contrast-adjusted X-ray image 44, the contrast of X-ray image 40 is adjusted by making the X-ray image representative value 50 of the first region 80 and the X-ray image representative value 50 of the second region 81 the predetermined values ​​70 of the first region 80 and the second region 81, respectively, thereby acquiring contrast-adjusted X-ray image 44. As a result, in contrast-adjusted X-ray image 44, compared with X-ray image 40 without contrast adjustment, the contrast between the regions of the first region 80 and the second region 81 is close to the contrast between the regions of the first region and the second region in the input teacher data 32 (teacher data). As a result, by using contrast-adjusted X-ray image 44 in the processing of acquiring the region of the object to be inspected 90 using the learned model 31, the decrease in processing accuracy (analysis accuracy of the region of the object to be inspected 90) of acquiring (identifying) the region of the object to be inspected 90 can be suppressed.

[0088] Furthermore, in this embodiment, as described above, before the step of acquiring multiple regions from the X-ray image 40, a step of performing logarithmic transformation on each pixel of the X-ray image 40 is included. In the step of acquiring multiple regions from the X-ray image 40, the first region 80 and the second region 81 are obtained by binarizing the logarithmically transformed X-ray image 41. Here, the X-rays detected by the X-ray detector 11 after the X-ray irradiation unit 10 irradiates and transmits through the object 90 being inspected are exponentially attenuated according to the distance between the X-ray irradiation unit 10 and the X-ray detector 11. Therefore, the resolution of the low-dose portion in the X-ray image 40 is lower than the resolution of the high-dose portion in the X-ray image 40. Therefore, by performing logarithmic transformation as described above, the change in pixel values ​​in the X-ray image 40 is linear, thus eliminating the blurring of the low-dose portion in the X-ray image 40 and improving the resolution of the low-dose portion in the X-ray image 40. In addition, by performing logarithmic transformation on the X-ray image 40, the change in pixel values ​​in the logarithmically transformed X-ray image 41 can be converted into a linear change. As a result, the resolution of the low-dose portion can be improved, and the first region 80 and the second region 81 are obtained by binarizing the X-ray image 41 after logarithmic transformation which converts the changes in pixel values ​​into linear changes. Therefore, compared with the structure of obtaining the first region 80 and the second region 81 by binarizing the X-ray image 40 without logarithmic transformation, the first region 80 and the second region 81 can be obtained with good accuracy.

[0089] Furthermore, in this embodiment, as described above, the X-ray image 40 is an image of a substrate 91 with a plurality of solder balls 93. The first region 80 is a solder ball region 80a with a plurality of solder balls 93, and the second region 81 is a background region 81a other than the solder ball region 80a. In the step of obtaining multiple regions from the X-ray image 40, the solder ball region 80a and the background region 81a are obtained from the logarithmically transformed X-ray image 41 by performing binarization processing. In the step of obtaining the X-ray image representative value 50, the X-ray image representative value 50 of the solder ball region 80a and the X-ray image representative value 50 of the background region 81a are obtained. As a result, the contrast between the solder ball region 80a and the background region 81a can be made close to the contrast between the solder ball region and the background region in the input teacher data 32 (teacher data). As a result, the processing accuracy reduction of obtaining (recognizing) the solder ball region 80a in the contrast-adjusted X-ray image 44 using the learned model 31 can be suppressed. Furthermore, the ability to suppress the decrease in processing accuracy of obtaining (identifying) solder ball region 80a using the learned model 31 has been confirmed in experiments conducted by the inventors of this application, as described later.

[0090] Furthermore, in this embodiment, as described above, in the step of obtaining the representative value 50 of the X-ray image, the average pixel value of the first region 80, i.e., the first average pixel value 50a, and the average pixel value of the second region 81, i.e., the second average pixel value 50b, are obtained as the representative value 50 of the X-ray image. In the step of obtaining the contrast-adjusted X-ray image 44, the contrast of the X-ray image 40 is adjusted by setting the first average pixel value 50a and the second average pixel value 50b to predetermined values ​​70, thereby obtaining the contrast-adjusted X-ray image 44. Therefore, when adjusting the contrast between the first region 80 and the second region 81, the first average pixel value 50a and the second average pixel value 50b, which are the representative values ​​50 of the X-ray image, are used to adjust the contrast of the X-ray image 40. Thus, excessive reduction of pixel values ​​in the first region 80 or excessive increase of pixel values ​​in the second region 81 can be suppressed. Additionally, excessive increase of pixel values ​​in the first region 80 or excessive reduction of pixel values ​​in the second region 81 can be suppressed. As a result, the contrast of the X-ray image 40 can be easily and accurately adjusted.

[0091] [First Example (Effect Confirmation Experiment)]

[0092] To confirm the effectiveness of the described implementation, an experiment was conducted as follows. Specifically, based on the recognition result image 61 (refer to the first comparison example)... Figure 11 ) and the recognition result image 62 based on the first embodiment (refer to) Figure 12 In the solder ball area 80a (refer to) Figure 11 The recognition accuracy of ) was compared.

[0093] Figure 11 The recognition result image 61 shown based on the first comparison example is the following recognition result, namely, using the generated fully learned model 31 (refer to...). Figure 1 The comparison example uses X-ray images taken under different tube voltage conditions during the acquisition of teacher data, along with the recognition results output by the learned model 31. Specifically, the learned model 31 was generated using teacher data acquired under acquisition conditions of 120 kV and 100 μA. The X-ray images were also acquired under acquisition conditions of 100 kV and 100 μA. Furthermore, in the first comparative example, by intentionally setting the tube voltage to a low value, the state of reduced X-ray energy due to changes in the X-ray irradiation section over 10 years was reproduced.

[0094] In the recognition result image 61, the recognition results based on the learned model 31 are superimposed as labels 82. However, in the recognition result image 61, labels 82 are not superimposed on most of the solder ball region 80a. That is, it can be confirmed that when the tube voltage value when capturing teacher data is different from the tube voltage value when capturing X-ray images, the analysis accuracy based on the learned model 31 decreases.

[0095] exist Figure 12 In the first embodiment shown, the contrast-adjusted X-ray image 44—that is, the X-ray image obtained by adjusting the contrast of an X-ray image captured under the shooting conditions of 100kV tube voltage and 100μA tube current using the X-ray image preprocessing method described in the embodiment—is input into the learned model 31 to obtain the recognition result image 62. Furthermore, in the first embodiment, by intentionally setting the tube voltage value to a low level, the state of reduced X-ray energy irradiated by the X-ray irradiation unit 10 due to years of change is reproduced.

[0096] In the recognition result image 62, the recognition results based on the learned model 31 are also superimposed in the form of labels 82. In the recognition result image 62, labels 82 are superimposed in the same manner as the solder ball region 80a. That is, it can be confirmed that even when the tube voltage value when capturing teacher data is different from the tube voltage value when capturing X-ray images, the reduction in analysis accuracy based on the learned model 31 can be suppressed.

[0097] [Second Example (Effect Confirmation Experiment)]

[0098] Furthermore, to confirm the effectiveness of the described implementation, the following experiment was conducted. Specifically, based on the recognition result image 63 (refer to the second comparison example)... Figure 13 ) and the recognition result image 64 based on the second embodiment (refer to) Figure 14 In the solder ball area 80a (refer to) Figure 13 The recognition accuracy of ) was compared.

[0099] Figure 13The recognition result image 63 shown based on the second comparative example is the recognition result of an X-ray image taken under different tube current conditions when the teacher data used to generate the learned model 31 was captured, and the recognition result output by the learned model 31. Specifically, the learned model 31 generated using teacher data captured under the conditions of a tube voltage of 120 kV and a tube current of 100 μA was used. In addition, the X-ray image was captured under the conditions of a tube voltage of 120 kV and a tube current of 50 μA. Furthermore, in the second comparative example, by deliberately setting the tube current value low, the state of reduced X-ray dose caused by changes in the X-ray irradiation section 10 over the years was reproduced.

[0100] In the recognition result image 63, the recognition results based on the learned model 31 are superimposed as labels 82. In the recognition result image 63, labels 82 are superimposed on most of the solder ball region 80a, but not on a portion of the solder ball region 80a. That is, it can be confirmed that when the tube current value when capturing teacher data is different from the tube current value when capturing X-ray images, the analysis accuracy based on the learned model 31 decreases.

[0101] exist Figure 14 In the second embodiment shown, the contrast-adjusted X-ray image 44—that is, the X-ray image obtained by adjusting the contrast of an X-ray image captured under the shooting conditions of tube voltage 120kV and tube current 50μA using the X-ray image preprocessing method described in the embodiment—is input into the learned model 31 to obtain the recognition result image 64. Furthermore, in the second embodiment, by intentionally setting the tube current value low, the state of reduced X-ray dose caused by years of changes in the X-ray irradiation section 10 is reproduced.

[0102] In the recognition result image 64, the recognition results based on the learned model 31 are also superimposed in the form of labels 82. In the recognition result image 64, labels 82 are superimposed in the same manner as the solder ball region 80a. That is, it can be confirmed that even when the tube current value when capturing teacher data is different from the tube current value when capturing X-ray images, the reduction in analysis accuracy based on the learned model 31 can be suppressed.

[0103] Based on the experimental results, it can be confirmed that when analyzing the contrast-adjusted X-ray image 44 obtained using the X-ray image preprocessing method based on the described embodiment using the learned model 31, even if the energy and dose of the X-rays irradiated from the X-ray irradiation unit are different from the energy and dose of the X-rays when the teacher's data was taken, the reduction in analysis accuracy based on the learned model 31 can be suppressed. That is, it can be confirmed that the reduction in processing accuracy of obtaining (identifying) the first region 80 (solder ball region 80a) using the learned model 31 can be suppressed.

[0104] [Variation Example]

[0105] Furthermore, the embodiments and examples disclosed herein should be considered exemplary in all respects and not restrictive. The scope of the invention is defined by the claims rather than the description of the embodiments and examples, and also includes all modifications (variations) within the meaning and scope equivalent to the claims.

[0106] For example, in the described embodiment, an example of a structure is shown where the control unit 20 (region acquisition unit 20c) acquires a first region 80 and a second region 81 from the X-ray image 40 as multiple regions, but the present invention is not limited thereto. For example, the control unit (region acquisition unit) may also be configured to acquire three or more regions from the X-ray image as multiple regions. The number of regions acquired by the control unit (region acquisition unit) can be varied depending on the number of objects analyzed after the model has been learned.

[0107] Furthermore, in the described embodiment, an example of a structure is shown where the control unit 20 (logarithmic processing unit 20b) performs logarithmic transformation on the X-ray image 40 before processing to obtain multiple regions from the X-ray image 40; however, the present invention is not limited to this. For example, the control unit (logarithmic transformation processing unit) may not perform logarithmic transformation on the X-ray image. However, when the control unit (logarithmic processing unit) does not perform logarithmic transformation on the X-ray image, binarization processing is performed when the resolution of the low-dose portion of the X-ray image is low and the pixel value changes exponentially. In this case, the accuracy of obtaining the first and second regions based on the binarization processing is reduced. Therefore, it is preferable that the control unit (logarithmic processing unit) is configured to perform logarithmic transformation on the X-ray image.

[0108] Furthermore, in the described embodiment, an example of a structure is shown where the control unit 20 (region acquisition unit 20c) performs binarization processing on the logarithmically transformed X-ray image 41 to obtain the first region 80 and the second region 81, but the present invention is not limited thereto. As long as the first region and the second region can be obtained from the X-ray image, the control unit (region acquisition unit) may also obtain the first region and the second region using methods other than binarization processing. For example, the control unit (region acquisition unit) may also be configured to obtain the first region and the second region from the X-ray image using a split-merge method that divides the image into regions with equal features.

[0109] Furthermore, in the described embodiment, an example is shown where the control unit 20 (region acquisition unit 20c) acquires the first region 80 and the second region 81 through Otsu binarization processing; however, the present invention is not limited to this. For example, the control unit (region acquisition unit) may also be configured to acquire the first region and the second region from the X-ray image through binarization processing other than Otsu binarization processing. For example, the control unit (region acquisition unit) may also be configured to acquire the first region and the second region through binarization processing based on a threshold set (input) by the user.

[0110] Furthermore, in the described embodiment, an example is shown where the X-ray image 40 is an image of a substrate 91 with solder balls 93 disposed thereon, but the present invention is not limited thereto. For example, the X-ray image may also be an image of an object to be inspected other than solder balls. In this case, the control unit (area acquisition unit) only needs to be configured to acquire the area where the object to be inspected is captured and the area other than the object to be inspected from the X-ray image.

[0111] Furthermore, in the described embodiment, an example of a structure is shown where the control unit 20 (representative value acquisition unit 20d) acquires a first average pixel value 50a and a second average pixel value 50b as a representative value 50 of the X-ray image, but the present invention is not limited thereto. For example, the control unit (representative value acquisition unit) may also be configured to acquire the median, mode, etc., of the pixel values ​​of the first region and the second region respectively as representative values ​​of the X-ray image.

[0112] Furthermore, in the described embodiment, an example of a structure is shown where the control unit 20 (contrast adjustment unit 20e) adjusts the pixel values ​​of the X-ray image 40 (logarithmically transformed X-ray image 41) such that the first average pixel value 50a and the second average pixel value 50b are both predetermined values ​​70. However, the present invention is not limited to this. For example, the control unit (contrast adjustment unit) may also be configured to adjust the pixel values ​​of the X-ray image (logarithmically transformed X-ray image) such that either the median or the mode of the pixel values ​​in the first region and the second region are predetermined values ​​70.

[0113] Furthermore, in the described embodiment, an example is shown where the control unit 20 (area acquisition unit 20c) acquires the area of ​​the solder balls 93 of the ball grid array (BGA), namely the solder ball area 80a, but the present invention is not limited thereto. For example, the control unit (area acquisition unit) may also be configured to acquire areas of multiple solder materials in the connection portions of multiple terminals of a planar grid array (LGA) with terminals arranged in a grid pattern. Alternatively, the control unit (area acquisition unit) may also acquire areas of multiple terminals rather than solder materials.

[0114] Furthermore, in the described embodiment, an example is shown where the specified value 70 is the average pixel value of the first region and the average pixel value of the second region of the input teacher data 32 (teacher data), but the present invention is not limited thereto. For example, the specified value may be either the median or the mode of the pixel values ​​of the first region of the input teacher data (teacher data), or either the median or the mode of the pixel values ​​of the second region.

[0115] Furthermore, in the described embodiment, an example is shown where the X-ray imaging device 1 and the image processing device 2 are provided separately, but the present invention is not limited thereto. In the present invention, the X-ray imaging device and the image processing device may also be integrally formed.

[0116] Furthermore, in the described embodiment, an example of the structure of the image processing apparatus 2 storing the learned model 31 is shown, but the present invention is not limited thereto. When preprocessing of the X-ray image and analysis of the preprocessed X-ray image (contrast-adjusted X-ray image) are performed by a separately provided control device, the image processing apparatus may not need to store the learned model.

[0117] Furthermore, in the described embodiment, an example is shown where a plurality of solder balls 93 are arranged on one side of the substrate 91, but the present invention is not limited thereto. For example, the present invention can also be applied to a structure in which a plurality of solder balls 93 are arranged on both the surface and the back surface of the substrate 91.

[0118] Furthermore, in the described embodiment, the processing of the control unit 20 to obtain the contrast-adjusted X-ray image 44 was explained using a process-driven flowchart that processes sequentially according to the processing flow; however, the present invention is not limited thereto. In the present invention, the processing performed by the control unit can also be performed using an event-driven type of processing that executes processing on an event-by-event basis. In this case, it can be performed in a completely event-driven manner, or a combination of event-driven and process-driven processing can be used.

[0119] [Example]

[0120] Those skilled in the art will understand that the exemplary embodiments described are specific examples of the following approaches.

[0121] (Project 1)

[0122] An X-ray image preprocessing method is performed before processing the region of an object to be examined from an X-ray image using a learned model. The X-ray image preprocessing method includes the following steps:

[0123] Obtain the X-ray image;

[0124] Based on the pixel values ​​of the X-ray image, multiple regions are obtained from the X-ray image;

[0125] Obtain the X-ray image representative values ​​of the pixel values ​​of each of the multiple regions; and

[0126] A contrast-adjusted X-ray image is obtained by adjusting the contrast of the X-ray image in such a way that the representative X-ray image values ​​of each of the plurality of regions of the X-ray image become predetermined values ​​obtained from teacher data when the learning completed model was created.

[0127] (Project 2)

[0128] According to the X-ray image preprocessing method described in Project 1, the specified value is obtained based on the representative values ​​of pixels in multiple regions of the teacher's data, i.e., the representative values ​​of the teacher's data.

[0129] (Project 3)

[0130] According to the X-ray image preprocessing method described in Project 1 or Project 2, in the step of obtaining the plurality of regions from the X-ray image, two regions are obtained: the region containing the object to be examined, i.e., the first region, and the region other than the object to be examined, i.e., the second region.

[0131] In the step of obtaining the contrast-adjusted X-ray image, the contrast of the X-ray image is adjusted by making the representative X-ray image value of the first region and the representative X-ray image value of the second region the predetermined values ​​of the first region and the second region, respectively, thereby obtaining the contrast-adjusted X-ray image.

[0132] (Project 4)

[0133] According to the X-ray image preprocessing method described in Project 3, before the step of obtaining the plurality of regions from the X-ray image, a step of performing a logarithmic transformation on each pixel of the X-ray image is further included.

[0134] In the step of obtaining the plurality of regions from the X-ray image, the first region and the second region are obtained by binarizing the X-ray image after logarithmic transformation.

[0135] (Project 5)

[0136] According to the X-ray image preprocessing method described in Project 4, the X-ray image is an image of a substrate with multiple solder balls disposed thereon.

[0137] The first area is the area where the plurality of solder balls have been implanted.

[0138] The second region is the background region outside the solder ball region.

[0139] In the step of obtaining the plurality of regions from the X-ray image, the solder ball region and the background region are obtained from the logarithmically transformed X-ray image by performing binarization processing.

[0140] In the step of obtaining the representative X-ray image value, the representative X-ray image value of the solder ball region is obtained, and the representative X-ray image value of the background region is obtained.

[0141] (Project 6)

[0142] According to any one of Projects 3 to 5, in the X-ray image preprocessing method, in the step of obtaining the representative value of the X-ray image, the average pixel value of the first region (i.e., the first average pixel value) and the average pixel value of the second region (i.e., the second average pixel value) are obtained as the representative value of the X-ray image.

[0143] In the step of obtaining the contrast-adjusted X-ray image, the contrast of the X-ray image is adjusted by setting the first average pixel value and the second average pixel value to the predetermined values, thereby obtaining the contrast-adjusted X-ray image.

[0144] (Project 7)

[0145] An X-ray image preprocessing procedure is performed before processing to obtain a region of the object being examined from an X-ray image using a learned model, the X-ray image preprocessing procedure causing a computer to perform the following controls:

[0146] Obtain the X-ray image;

[0147] Based on the pixel values ​​of the X-ray image, multiple regions are obtained from the X-ray image;

[0148] Obtain the X-ray image representative values ​​of the pixel values ​​of each of the multiple regions; and

[0149] A contrast-adjusted X-ray image is obtained by adjusting the contrast of the X-ray image in such a way that the representative X-ray image values ​​of each of the plurality of regions of the X-ray image become predetermined values ​​obtained from teacher data when the learning completed model was created.

[0150] (Project 8)

[0151] An X-ray image preprocessing system is performed before processing to obtain a region of an object from an X-ray image of the object being examined using a learned model. The X-ray image preprocessing system includes:

[0152] An X-ray imaging apparatus includes an X-ray irradiation unit for irradiating X-rays and an X-ray detector for detecting X-rays irradiated from the X-ray irradiation unit; and

[0153] An image processing device generates the X-ray image.

[0154] The image processing device performs the following control:

[0155] Generate the X-ray image; and

[0156] Based on the pixel values ​​of the X-ray image, multiple regions are obtained from the X-ray image, and representative X-ray image values ​​of the pixel values ​​of each of the multiple regions are obtained. The contrast of the X-ray image is adjusted by making the representative X-ray image values ​​of each of the multiple regions of the obtained X-ray image the predetermined values ​​obtained from the teacher's data when the learning completed model was made, thereby obtaining a contrast-adjusted X-ray image.

Claims

1. An X-ray image preprocessing method, performed before processing to obtain a region of the object to be examined from an X-ray image using a learned model, the X-ray image preprocessing method comprising the following steps: Obtain the X-ray image; Based on the pixel values ​​of the X-ray image, multiple regions are obtained from the X-ray image; Obtain the representative values ​​of the pixel values ​​of each of the multiple regions, i.e., the representative values ​​of the X-ray image; and A contrast-adjusted X-ray image is obtained by adjusting the contrast of the X-ray image in such a way that the representative X-ray image values ​​of each of the plurality of regions of the X-ray image become predetermined values ​​obtained from teacher data when the learning completed model was created.

2. The X-ray image preprocessing method according to claim 1, wherein, The specified values ​​are obtained based on the representative values ​​of pixels in multiple regions of the teacher's data, i.e., the representative values ​​of the teacher's data.

3. The X-ray image preprocessing method according to claim 2, wherein, In the step of obtaining the plurality of regions from the X-ray image, two regions are obtained: the region containing the object under examination, i.e., the first region, and the region other than the object under examination, i.e., the second region. In the step of obtaining the contrast-adjusted X-ray image, the contrast of the X-ray image is adjusted by making the representative X-ray image value of the first region and the representative X-ray image value of the second region the predetermined values ​​of the first region and the second region, respectively, thereby obtaining the contrast-adjusted X-ray image.

4. The X-ray image preprocessing method according to claim 3, wherein, Prior to the step of obtaining the plurality of regions from the X-ray image, the method further includes a step of performing a logarithmic transformation on each pixel of the X-ray image. In the step of obtaining the plurality of regions from the X-ray image, the first region and the second region are obtained by binarizing the X-ray image after logarithmic transformation.

5. The X-ray image preprocessing method according to claim 4, wherein, The X-ray image is an image of a substrate with multiple solder balls arranged on it. The first area is the area where the plurality of solder balls have been implanted. The second region is the background region outside the solder ball region. In the step of obtaining the plurality of regions from the X-ray image, the solder ball region and the background region are obtained from the logarithmically transformed X-ray image by performing binarization processing. In the step of obtaining the representative X-ray image value, the representative X-ray image value of the solder ball region is obtained, and the representative X-ray image value of the background region is obtained.

6. The X-ray image preprocessing method according to claim 3, wherein, In the step of obtaining the representative value of the X-ray image, the average pixel value of the first region (i.e., the first average pixel value) and the average pixel value of the second region (i.e., the second average pixel value) are obtained as the representative value of the X-ray image. In the step of obtaining the contrast-adjusted X-ray image, the contrast of the X-ray image is adjusted by setting the first average pixel value and the second average pixel value to the predetermined values, thereby obtaining the contrast-adjusted X-ray image.

7. An X-ray image preprocessing system, performed before processing to obtain a region of an object from an X-ray image of an object to be examined using a learned model, the X-ray image preprocessing system comprising: An X-ray imaging apparatus includes an X-ray irradiation unit that irradiates X-rays and an X-ray detector that detects the X-rays irradiated from the X-ray irradiation unit. as well as An image processing device generates the X-ray image. The image processing device performs the following control: Generate the X-ray image; and Based on the pixel values ​​of the X-ray image, multiple regions are obtained from the X-ray image, and representative X-ray image values ​​of the pixel values ​​of each of the multiple regions are obtained. The contrast of the X-ray image is adjusted by making the representative X-ray image values ​​of each of the multiple regions of the X-ray image the predetermined values ​​obtained from the teacher's data when the learning model was created, thereby obtaining a contrast-adjusted X-ray image.

Images