TWI871595B - X-ray inspection device, x-ray inspection method and x-ray inspection program - Google Patents

Abstract

The present invention provides a technique for easily distinguishing a conductor portion that is an inspection target and an insulating portion other than the conductor portion in a substrate.

The X-ray inspection device of the present invention has an X-ray image obtaining part, a reconstruction calculation part, a sliced image obtaining part, a classification part, and a contrast enhancement part. The X-ray image obtaining part is to obtain a plurality of X-ray images captured by irradiating the substrate with X-rays at an angle oblique to a predetermined direction; the reconstruction calculation part is to perform reconstruction calculations according to the X-ray images to obtain reconstruction information; the sliced image obtaining part is to obtain a sliced image of the substrate in the predetermined first direction according to the reconstruction information; the classification part is to classify the sliced image into a conductor part and an insulation part; the contrast enhancement part is to enhance the contrast between the conductor part and the insulation part.

Description

X-ray inspection device, X-ray inspection method and X-ray inspection program

The present invention relates to an X-ray inspection device, an X-ray inspection method and an X-ray inspection program.

It is known that there is an X-ray inspection device that uses X-rays to photograph a substrate from different directions, generates reconstruction information based on the photographed images, and performs inspection based on three-dimensional structures such as electroplating filled in the through-holes (for example, refer to Patent Document 1).

[Prior technical literature]

[Patent Literature]

[Patent document 1] Japanese Patent No. 6946516

The technology that generates reconstructed information based on multiple images of the substrate taken from different directions using X-rays may generate artifacts for various reasons. If artifacts are generated, the image will be unclear, making it difficult to inspect the conductor part of the substrate.

This invention was completed in view of the above-mentioned problem, and its purpose is to provide a technology that can easily distinguish the conductive part of the substrate that is the inspection object from the insulating part other than the conductive part.

To achieve the above-mentioned purpose, the X-ray inspection device is provided with: an X-ray image acquisition unit, a reconstruction calculation unit, a slice image acquisition unit, a classification unit and a contrast and emphasis unit; the X-ray image acquisition unit acquires a plurality of X-ray images taken by irradiating a substrate with X-rays at an angle inclined to a predetermined direction; the reconstruction calculation unit performs reconstruction calculations based on the X-ray images to acquire reconstruction information; the slice image acquisition unit acquires a slice image of the substrate cut in a preset first direction based on the reconstruction information; the classification unit classifies the slice image into a conductive part and an insulating part; and the contrast and emphasis unit emphasizes the contrast between the conductive part and the insulating part.

The above-described structure obtains reconstruction information from the X-ray images, but the reconstruction information is affected by artifacts. Therefore, if the artifacts have a greater impact on the inspection object, it is more difficult to distinguish the conductor part and the insulating part that constitute the substrate, making it more difficult to perform the inspection. However, in the X-ray inspection device, a slice image is obtained based on the reconstruction information, and the slice image is analyzed to classify the image contained in the slice image into the conductor part and the insulating part. Then, the contrast between the conductor part and the insulating part is emphasized. As a result, the reconstruction information can easily distinguish the conductor part from the insulating part.

10: X-ray imaging device

11: X-ray generator

11a: X-ray output unit

12: X-ray detector

12a: Detection surface

20: Control Department

21: Generator control unit

22: Detector control unit

23: Positioning mechanism control unit

24: Input section

25: Output section

26: Memory

26a: Program data

26b: X-ray image data

27:CPU

27a: X-ray image acquisition unit

27b: Reconstruction calculation department

27c: Slice image acquisition unit

27d: Classification Department

27e: Contrast emphasis section

27f: Display control unit

Ar,A:arrow

Ax: Rotation axis

D1: First direction

D2: Second direction

Nv:Normal

Pt,Vi: Component symbol

Rg1,Rg2: Ring

S100~S150,S200~S255: Steps

Tr1, Tr2: two-point chain curve

W: Substrate

Z1, Z2: Partial

Zo: Statistical scope of the conductor department

Zu: Statistical range of the insulated area

α: Tilt angle

[Figure 1] is a schematic block diagram of an X-ray inspection device.

In [Figure 2], [Figure 2A] is a diagram illustrating the relationship between the substrate, X-ray detector, and X-ray generator; [Figure 2B] and [Figure 2C] are examples of slice images obtained from the reconstructed information before correction; [Figure 2D] and [Figure 2E] are examples of slice images obtained from the reconstructed information after correction.

In [Figure 3], [Figure 3A] is a flowchart of X-ray inspection processing; [Figure 3B] is a flowchart of classification processing of the conductor part and the insulation part.

In [Figure 4], [Figure 4A] ~ [Figure 4F] are process diagrams of image processing based on slice images.

In [Figure 5], [Figure 5A] is a graph showing the brightness statistics and classification results of each region number; [Figure 5B] is a graph showing the classification results corresponding to the slice image; [Figure 5C] is a graph showing the corrected slice image; and [Figure 5D] is a graph showing overshoot and undershoot.

[Figure 6] is an illustration of overshoot and undershoot.

Here, the embodiments of the present invention are described in the following order.

(1) Composition of X-ray inspection device:

(2) X-ray examination and processing:

(3) Other embodiments:

(1) Composition of X-ray inspection device:

FIG1 is a schematic block diagram of an X-ray inspection device of an embodiment of the present invention. The X-ray inspection device comprises: an X-ray imaging mechanism 10 and a control unit 20. The X-ray imaging mechanism 10 comprises: an X-ray generator 11 and an X-ray detector 12. The X-ray imaging mechanism 10 irradiates X-rays from the X-ray generator 11 toward the substrate W when the substrate W, the X-ray generator 11 and the X-ray detector 12 are in a predetermined relative position relationship.

The X-ray generator 11 has an X-ray output unit 11a for outputting X-rays, and can irradiate the substrate W with X-rays at a predetermined intensity from the X-ray output unit 11a. The X-ray detector 12 has a detection surface 12a, and can detect the X-ray intensity reaching the detection surface 12a. The X-ray detector 12 can take an X-ray image reflecting the X-ray penetration amount penetrating the substrate W based on the X-ray intensity detected on the detection surface 12a. That is, the X-ray detector 12 generates X-ray image data 26b representing the X-ray penetration amount image at each position of the detection surface 12a.

The substrate W of this embodiment is a stacked substrate with multiple layers. The substrate W is a cube with two sides larger than the other side (i.e., a plate-like shape). Each layer is stacked in the perpendicular direction of the widest two sides of the substrate W. In each layer, a wiring pattern is formed by a conductor in a parallel plane to the widest surface of the substrate W. The wiring patterns of each layer are electrically coupled by through holes that are filled with electroplating. In this embodiment, the wiring pattern, the conductors such as the filling electroplating formed in the substrate W are called "conductor parts", and the part other than the conductor parts are called "insulating parts". In addition, in this embodiment, the diameter of the through hole is about 30μm, but the size of the through hole, the distance between layers, and the size of the substrate W are not limited.

In this embodiment, the size of the through hole and the conductor part around it are the inspection objects, but of course the inspection objects are not limited, and solder etc. can also be the inspection objects. The substrate W is transported along a predetermined plane by a transport mechanism not shown in the figure. That is, the uninspected substrate W is transported along a predetermined plane and arranged in the irradiation range of the X-ray, and is transported out again by the transport mechanism after inspection. In this embodiment, there is an unillustrated positioning mechanism that changes the relative position relationship between the X-ray generator 11, the X-ray detector 12 and the substrate W. In this embodiment, the positioning mechanism can move the substrate W two-dimensionally along a predetermined plane (called "X-Y plane") within the irradiation range of the X-ray emitted by the X-ray generator 11. In addition, the positioning mechanism can rotate the X-ray detector 12 around a predetermined rotation axis.

FIG2A is a schematic diagram showing how X-rays are irradiated on a substrate W. In FIG2A , the horizontal direction is parallel to the X-Y plane, and the vertical direction is the Z direction perpendicular to the X-Y plane. FIG2A schematically shows the X-ray generator 11 and the X-ray detector 12 on and around the substrate W. The positioning mechanism is omitted.

In this embodiment, the X-ray generator 11 outputs X-rays from the X-ray output portion 11a toward a predetermined range. The detection surface 12a of the X-ray detector 12 rotates around a predetermined rotation axis Ax. In this embodiment, the rotation axis Ax is a straight line parallel to the Z direction and passing through the X-ray output portion 11a. In addition, in FIG. 2A , the rotation axis Ax is represented by a single-point chain. The normal Nv of the center of the detection surface 12a of the X-ray detector 12 is tilted by an angle α relative to the rotation axis Ax. The detection surface 12a of the X-ray detector 12 rotates around the rotation axis Ax while maintaining a state of facing the X-ray output unit 11a (i.e., maintaining a state where the normal line Nv is inclined at an angle α to the rotation axis Ax). In addition, in FIG. 2A , the normal line Nv is represented by a dotted line, and the trajectory of the X-ray detector 12 when rotating is represented by a two-point chain curve Tr1.

In this embodiment, the positioning mechanism moves the position of the substrate W in a circular orbit in the X-Y plane. That is, the X-ray detector 12 rotates to change the shooting range of the X-ray detector 12, so the positioning mechanism also moves the position of the substrate W synchronously with the rotation of the X-ray detector 12. For example, the image of a specific position of the substrate W is shot at a specific position of the detection surface 12a of the X-ray detector 12, so that the substrate W moves. In Figure 2A, the trajectory of the substrate W in the X-Y plane is represented by a two-point chain curve Tr2.

In addition, the positioning mechanism can change the relative position relationship between the substrate W, the X-ray output unit 11a, and the detection surface 12a in a manner that the substrate W can be photographed from different directions. Therefore, by rotating at least one of the X-ray detector 12, the X-ray generator 11, and the substrate W around the rotation axis Ax, the substrate W can be photographed from different directions. The structure to achieve this function is not limited to the structure shown in Figure 2A. For example, the position of the substrate W can be fixed, and the X-ray generator 11 and the X-ray detector 12 can be rotated in opposite directions on the same axis. In addition, the X-ray generator 11 and the X-ray detector 12 can be fixed, and the substrate W can be rotated within the field of view of the X-ray detector 12.

Next, the control unit 20 is explained. The control unit 20 includes: a generator control unit 21, a detector control unit 22, a positioning mechanism control unit 23, an input unit 24, an output unit 25, a memory 26, and a CPU 27. The memory 26 is a storage medium that can store data, and stores program data 26a and X-ray image data 26b. The CPU 27 reads and executes the program data 26a to perform calculations for various processing described later. In addition, the memory 26 can be any storage medium as long as it can store data, such as RAM, EEPROM, HDD, etc.

The positioning mechanism control unit 23 controls the positioning mechanism not shown in the figure to adjust the positions of the substrate W and the X-ray detector 12 so that the substrate W is located at the shooting position shown in FIG. 2A. The generator control unit 21 controls the X-ray generator 11 to irradiate X-rays from the X-ray generator 11 toward the substrate W. The detector control unit 22 obtains the X-ray intensity detected by the X-ray detector 12 (i.e., the X-ray image data 26b representing the penetration image). The X-ray image data 26b is image data composed of the color gradation values of a plurality of pixels, and the color gradation value of each pixel represents the intensity of the X-ray detected by the X-ray detector 12. The detector control unit 22 obtains the X-ray image data 26b from the X-ray detector 12 and stores it in the memory 26. The output unit 25 is a display that displays the inspection results of the substrate W, etc. The input unit 24 is an operation input device that accepts user input.

CPU 27 executes the program represented by program data 26a to perform inspection of substrate W. When executing the program, CPU 27 has the functions of X-ray image acquisition unit 27a, reconstruction calculation unit 27b, slice image acquisition unit 27c, classification unit 27d, contrast emphasis unit 27e and display control unit 27f.

The X-ray image acquisition unit 27a enables the CPU 27 to execute the function of acquiring multiple X-ray images taken by irradiating the substrate W with X-rays at an angle inclined to the Z-axis direction. That is, the CPU 27 uses the processing of the X-ray image acquisition unit 27a to output predetermined instructions to the generator control unit 21, the detector control unit 22, and the positioning mechanism control unit 23, and photographs the substrate W at multiple rotation positions with different rotation angles around the rotation axis Ax to acquire X-ray image data 26b representing the X-ray image.

The reconstruction calculation unit 27b is a function for causing the CPU 27 to perform reconstruction calculation processing based on a plurality of X-ray images. That is, the CPU 27 uses the processing of the reconstruction calculation unit 27b to perform reconstruction calculation based on the X-ray image data 26b. The reconstruction calculation can adopt various known algorithms. In this embodiment, a reconstruction calculation including a filtered back projection method is adopted. If the reconstruction calculation is executed, the state of reconstruction information of the substrate W is obtained in the three-dimensional space formed by the X, Y, and Z axes (that is, the state of obtaining the corresponding value of the X-ray penetration amount for each coordinate in the three-dimensional space for the substrate W is formed). The reconstruction information thus generated represents the color scale value corresponding to the X-ray penetration of the substrate W at each location according to each coordinate in the three-dimensional space, and the three-dimensional structure of the substrate W can be analyzed based on the color scale value. For example, if the penetration of the reconstruction information at each position along a specific plane is displayed in two dimensions, the structure of the substrate W at the section can be analyzed. Therefore, if the analysis is performed on an arbitrary section, the three-dimensional structure of the substrate W can be analyzed. In this embodiment, the penetration of the reconstruction information at each position is expressed in 8 bits. That is, the penetration at each position is expressed in 256 color scales. In addition, in this embodiment, the color scale value is defined in such a way that the smaller the penetration, the larger the color scale value. Therefore, in an image that regards color values as brightness, the greater the brightness, the smaller the X-ray penetration caused by the material at each pixel position, and the smaller the brightness, the greater the X-ray penetration caused by the material at each pixel position. Therefore, the image of the conductor becomes brighter, while the image of the insulating part becomes darker. In addition, the penetration information at each position of the cross section in any direction will be called "slice image" below.

As mentioned above, the slice images obtained based on the reconstruction information are mostly unclear due to the influence of artifacts, which is particularly difficult when inspecting smaller inspection objects such as through-holes. Figure 2B shows an example of a slice image obtained from the reconstruction information. The slice image is a cross-sectional image of the through-hole cut in a direction parallel to the Z direction. In the slice image, the dark gray level represents the insulating part, and the light gray level represents the electroplating and wiring pattern in the through-hole. However, the slice image is unclear due to the influence of artifacts, making it difficult to clearly distinguish between the conductive parts such as the electroplating and wiring patterns and the insulating parts. Therefore, in this embodiment, by using the functions of the classification unit 27d and the contrast and emphasis unit 27e of the CPU 27 to correct and reconstruct information, the conductive part and the insulating part can be easily distinguished.

The slice image acquisition unit 27c has the function of obtaining a slice image of a substrate cut in a preset first direction according to reconstruction information. In this embodiment, the first direction is a direction parallel to the X-Y plane. The first direction D1 is illustrated in FIG2A. In this embodiment, the substrate W is a stacked substrate formed with a plurality of parallel layers on the X-Y plane, and the Z direction is the stacking direction of the layers. Therefore, the first direction is a direction perpendicular to the stacking direction of the plurality of layers of the substrate W. In this embodiment, a wiring pattern is formed on the surface of each parallel layer in the X-Y plane using a conductor. Therefore, the first direction is a direction parallel to the surface on which the wiring pattern is formed. In addition, in this embodiment, the through hole extends in a direction parallel to the stacking direction of the layer. Therefore, the first direction is a direction perpendicular to the extension direction of the through hole.

Furthermore, in this embodiment, the X-Y plane is perpendicular to the rotation axis Ax and the Z direction, and the first direction is the perpendicular direction of the rotation axis Ax. In this embodiment, CPU27 obtains slice images when cutting according to multiple sections at the cutting position parallel to the first direction based on the reconstruction information, and the details are described later. FIG. 2C shows an example of a slice image, and particularly illustrates the periphery of the through hole.

Because the reconstructed information is affected by the artifacts, the slice image is not clear. For example, in Figure 2C, the through hole image is a light gray circle represented by the element symbol Vi, and there are a few dark gray rings Rg1 around it, and there are dark gray rings Rg2 around it. In the slice image shown in Figure 2C, the ring Rg1 around the through hole image is actually an insulating part, but it looks like a conductive part from the slice image. Therefore, it is difficult to visually judge which is the conductive part and which is the insulating part from the slice image affected by the artifacts.

Here, in this embodiment, CPU27 classifies the slice image into the conductive part and the insulating part and emphasizes the contrast. Classification unit 27d has the function of classifying the slice image into the conductive part and the insulating part. In this embodiment, CPU27 classifies the slice image into the conductive part and the insulating part by analyzing the slice image that is not cut in the second direction D2 but in the first direction. In this embodiment, the slice image obtained by cutting in the first direction based on the reconstruction information is easier to process than the slice image obtained by cutting in the second direction perpendicular to the first direction. For example, if Figure 2B is compared with Figure 2C, the cross-sectional view of Figure 2B cut in the second direction (Z direction) shows that the substrate W is a multi-layer substrate, reflecting the stacking of thin layers, and there are multiple layers of wiring patterns in a narrow range. In addition, in Figure 2B, one of the wiring pattern images is represented by the component symbol Pt.

On the other hand, in the cross-sectional view 2C cut in the first direction (X-Y plane direction), the surrounding structure of the through hole is simpler than that of FIG. 2B. Here, in this embodiment, in order to facilitate the processing of extracting the conductor part and the insulating part from the slice image, CPU27 performs classification according to the slice image cut in the first direction. In addition, in FIG. 2C, the image of the through hole is represented by the element symbol Vi.

As will be described in detail later, CPU 27 utilizes the function of the classification unit 27d to distinguish the conductive part and the insulating part contained in the slice image according to the brightness value of the slice image, and assigns a corresponding association to each pixel as to which is the conductive part and which is the insulating part.

The contrast emphasis section 27e has the function of emphasizing the contrast between the conductor part and the insulating part. In this embodiment, the CPU 27 uses the function of the contrast emphasis section 27e to emphasize the contrast by maintaining the brightness of the conductor part and reducing the brightness of the insulating part. That is, because each pixel of the slice image is specified as a conductor part and an insulating part by using the classification section 27d, the CPU 27 reduces the brightness of the insulating part according to the classification result. FIG. 2D shows the state in which contrast emphasis is performed on FIG. 2C. If contrast emphasis is performed, as shown in FIG. 2D, the brightness of the insulating part is reduced, so that the false shadow existing in the insulating part is almost not visible. On the other hand, because the conductor part maintains its brightness, the existence of the conductor part is emphasized. Based on the above structure, the conductor part and the insulating part can be clearly distinguished.

The above classification and contrast emphasis are implemented for multiple slice images. The result is that by emphasizing the contrast in the reconstruction information representing the three-dimensional structure of the substrate W, the conductor part and the insulating part can be clearly distinguished. Therefore, when an arbitrary slice image is generated based on the reconstruction information corrected by contrast emphasis, the conductor part and the insulating part can be easily distinguished.

In this embodiment, because the display is performed based on the corrected reconstruction information, the CPU 27 performs the function of the display control unit 27f. The display control unit 27f obtains the analysis target image of the substrate W cut in the second direction perpendicular to the first direction based on the reconstruction information after contrast emphasis, and displays the analysis target image on the display unit. That is, the CPU 27 obtains the color level value at each position on the cross section of the substrate W cut in the second direction based on the corrected reconstruction information and generates a slice image, and controls the output unit 25 to display the slice image.

FIG2E shows an example of contrast emphasis applied to the example shown in FIG2B. As shown in FIG2E, in the slice image after contrast emphasis, the brightness of the insulating portion is reduced. Therefore, the false image contained in the insulating portion is not easily visible. In addition, because contrast emphasis is applied after the conductor portion and the insulating portion are classified, the boundary between the conductor portion and the insulating portion is clear. Therefore, the user who views the slice image can easily distinguish between the conductor portion containing the through hole and the wiring pattern and the insulating portion.

(2) X-ray examination and processing:

Next, the X-ray inspection process performed by the X-ray inspection device of this embodiment is described according to the flowchart shown in FIG. 3A and FIG. 3B. Before performing the X-ray inspection process, the CPU 27 uses the functions of the X-ray image acquisition unit 27a and the reconstruction calculation unit 27b to obtain multiple X-ray images of the inspection target substrate W, and further performs reconstruction calculation to obtain reconstruction information of the substrate W.

If reconstruction information is obtained, CPU27 performs processing for correcting the reconstruction information. To perform the correction, CPU27 first uses the function of the slice image acquisition unit 27c to perform pre-processing (step S100). Pre-processing is pre-processing for edge capture described later. Therefore, various processing can be performed to easily perform edge capture. In this embodiment, pre-processing is noise reduction processing. Various processing can be used to reduce noise, such as the well-known smoothing processing.

Because the smoothing process is performed on a two-dimensional image, CPU27 generates a plurality of slice images based on the reconstruction information. Specifically, CPU27 sets a plurality of cutting positions at different positions in the Z direction with reference to the reconstruction information. Furthermore, CPU27 obtains a slice image when the first direction is set as the cutting direction at each cutting position. In addition, each slice image corresponds to identification information. The identification information is an integer, the minimum value is 1, and the maximum value is the total number of slice images. Here, it is assumed that each slice image is given corresponding identification information in a manner in which the value increases from the negative direction to the positive direction in the Z direction. The more slice images there are, the easier it is to perform correct classification, and they can also be appropriately selected within a range of numbers that does not exceed the resolution limit of the reconstruction information. FIG. 4A shows the same image as FIG. 2C, and specifically shows a portion of the slice image. The following description will be made with reference to the image processing results of the slice image shown in FIG. 4A. If a slice image is obtained, CPU 27 performs smoothing processing on each slice image.

Next, CPU27 uses the function of the classification unit 27d to classify the conductor part and the insulating part (step S110). FIG. 3B shows a processing flow chart of the classification shown in step S110. In the classification process, CPU27 first initializes the slice number k (step S200). The slice number k is a variable that specifies the slice image generated by step S100. In this embodiment, the minimum value of the slice image identification information is 1, and the maximum value is the total number of slice images. Therefore, in step S100, CPU27 initializes the slice number k by substituting 1 into it.

Next, CPU27 executes the processing of steps S205 to S255 for the slice image of slice number k. Specifically, CPU27 captures the area from the slice image of slice number k (step S205). Area capture is a process of setting the image of the same substance as the same area in the slice image. That is, because the substrate W is composed of a conductor part and a wiring part, any pixel of the slice image is an image of the conductor part or an image of the wiring part. Here, CPU27 captures the image of the same substance as the same area.

The capture can be implemented using various processing. In this embodiment, CPU27 uses edge capture to capture the boundary from the slice image, and captures the area by treating the parts separated by the boundary as different areas. The capture process of the boundary can utilize known processing, such as Sobel filter and binarization. That is, CPU27 uses Sobel filter to obtain the edge intensity of each pixel for the slice image of slice number k, and regards the pixels whose edge intensity is greater than the predetermined reference value as the boundary. In this embodiment, CPU27 performs binarization by setting the pixel value belonging to the boundary to 1 and the pixel value not belonging to the boundary to 0. In addition, in this embodiment, the boundary is set in such a way that the width of the boundary becomes 1 pixel. FIG. 4B shows an image obtained by extracting the boundary line from the slice image shown in FIG. 4A and binarizing it. In the binarized image, pixels with a value of 1 are represented by white, and pixels with a value of 0 are represented by black (the same applies below). In this embodiment, the image after binarization is referred to as a "boundary line image". In addition, various processing can also be performed along with the region extraction processing, such as processing to reduce noise and processing to make the boundary line a predetermined width.

If a boundary image is obtained, CPU 27 performs processing to regard a series of pixels separated by a boundary as the same region, and divides the slice image into a plurality of regions. In addition, CPU 27 assigns identification information representing the region to each divided region in a corresponding manner. In this embodiment, the identification information representing the region is an integer greater than 1. FIG. 4C shows a region map captured from the example shown in FIG. 4A. That is, the boundary image shown in FIG. 4B is generated based on FIG. 4A, and if a region is captured, identification information of the region is assigned to each region in a corresponding manner. FIG. 4C shows an example in which identification information 1, 2, 3, and 4 are assigned in a corresponding manner in order from the region located on the outside. In Figure 4C, each region is illustrated by coloring it in a manner such that the smaller the identification information of the region, the darker the grayscale.

If the region is captured, CPU27 initializes the region number L (step S210). Region number L is a variable representing the region captured by step S205. In this embodiment, the identification information of the region has a minimum value of 1 and a maximum value of the total number of regions in the slice image. Therefore, CPU27 initializes by substituting 1 into region number L.

Next, CPU27 obtains the statistical range for the area numbered L (step S215). The statistical range is the range for collecting the brightness statistical values of each area. That is, in this embodiment, the statistical value of brightness is obtained for each area captured by step S205, and is regarded as the brightness value representing the area. The statistical value only needs to represent the brightness value of the area, and various values such as the average, majority, and median can be used. When obtaining the statistical value, CPU27 does not set the entire area as the statistical acquisition object, but excludes a part of it and then obtains the statistics. Here, the remaining area that excludes a part is called the "statistical range".

The statistical range is set so that the statistical value can easily show whether each area is a conductor or an insulating part. Specifically, because the conductor is made of metal and the insulating part is made of resin, the X-ray penetration of the conductor is generally smaller than that of the insulating part. In the slice image shown in Figure 4A, there is a tendency that the conductor with a smaller X-ray penetration is brighter (larger brightness value) than the insulating part with a larger penetration. That is, the reconstruction information is generated in such a way that the conductor part of the analysis object can be clearly seen and the conductor part becomes brighter than the insulating part. Therefore, among the areas contained in the slice image, the area with a larger brightness statistical value can be considered to be the conductor part, and the area with a smaller brightness statistical value is the insulating part.

On the other hand, it is known that in the slice image, there are brightness overshoots and undershoots from the boundary to the inner side of the region on both sides of the boundary line. Figure 6 shows an illustration of overshoots and undershoots. The upper figure of Figure 6 shows a part of the slice image. There is a conductor part with higher brightness in the center of the slice image, and there is an insulating part with lower brightness around it.

The lower figure of Figure 6 shows the brightness of each position of the arrow Ar part in the slice image. As shown in the figure, the brightness at the boundary between the conductor part and the insulating part varies greatly. Specifically, there is a part Z1 with almost constant brightness in the center of the conductor part. If the insulating part is away from the boundary line to a certain extent, there is a part Z2 with almost constant brightness. Between parts Z1 and Z2, if the part Z1 approaches the boundary, the brightness rises sharply and then drops. If it further crosses the boundary and approaches part Z2, the brightness drops sharply and then rises again, and finally becomes almost constant.

In this way, the phenomenon that the brightness inside the conductor part near the boundary between the conductor part and the insulating part is higher than the brightness of part Z1 is called "overshoot". In addition, the phenomenon that the brightness inside the insulating part near the boundary between the conductor part and the insulating part is lower than the brightness of part Z2 is called "undershoot". Such overshoot and undershoot are generated by performing a reconstruction algorithm including the filtered back projection method. That is, if the filtered back projection method is performed, a filter that allows high-harmonic components of spatial frequency to pass is applied to emphasize the edge. Because the edge is emphasized, an overshoot will occur on the high-brightness side of the edge, and an undershoot will occur on the low-brightness side.

As shown in the lower figure of FIG6 , the brightness of the overshoot portion is higher than that of portion Z1, and the brightness of the undershoot portion is lower than that of portion Z2. Therefore, when obtaining brightness statistics for each region, obtaining statistics at the overshoot portion and the undershoot portion can make the brightness difference of each region on both sides of the edge more obvious than obtaining statistics at portion Z1 or portion Z2. Therefore, in this embodiment, CPU 27 sets the range of brightness overshoot and undershoot from the boundary to the inner side of the region on both sides of the boundary as the statistical range, and obtains brightness statistics according to the statistical range.

The statistical range is a range of a predetermined width from the boundary toward the inner side of the region, and can be defined by various methods. In this embodiment, the range of width obtained by using a predetermined image processing is set as the statistical range. Specifically, in this embodiment, CPU27 uses the shrinkage processing of the morphology processing to calculate the difference between the regional image before the shrinkage processing and the regional image after the shrinkage processing, and specifies the statistical range.

To perform this processing, CPU27 generates a binary image in the processing object, with the pixels of area number L being 1 and the other pixels being 0. FIG. 4D shows an image generated by the binarization for the area number 2 shown in FIG. 4C. CPU27 performs a shrinkage process on the image obtained by the binarization. The shrinkage process can be implemented using a known algorithm. FIG. 4E shows an image after the shrinkage process is performed on the image shown in FIG. 4D. However, in FIG. 4E, the boundary between the area number 1 and the area number 2, and the boundary between the area number 2 and the area number 3 are indicated by white lines. In actual image processing, the boundary does not appear (the pixel value is 0).

Then, CPU27 specifies the statistical range by removing the image after the shrinkage processing from the image before the shrinkage processing. FIG. 4F shows the image after removing the image shown in FIG. 4E from the image shown in FIG. 4D. According to the above processing, as shown in FIG. 4F, the pixel value of the predetermined width range from the boundary toward the inner side of the area numbered 2 becomes 1. Therefore, the pixel with a value of 1 belongs to the statistical range used for statistically calculating the brightness of the area numbered 2. In addition, in FIG. 6, the lower figure of FIG. 6 represents the statistical range of the insulating part as Zu and the statistical range of the conductive part as Zo. Moreover, the statistical range only needs to belong to the range where there is an overshoot or undershoot of the brightness from the boundary toward the inner side of the area. The shrinkage degree can be preset or determined in accordance with the area size, brightness size, etc. In addition, the statistical range can be determined by processing other than shrinkage processing of the shape processing, for example, a certain width range from the boundary line can be used as the composition of the statistical range, etc.

As described above, if the statistical range of the area with area number L is obtained, CPU27 obtains the statistical value of the brightness (step S220). That is, CPU27 specifies the statistical range of the area with area number L from the slice image with slice number k, and obtains the statistical value of the brightness of the statistical range. For example, as shown in FIG. 4F, CPU27 extracts the pixels represented by white in FIG. 4F from the slice image shown in FIG. 4A, and obtains the statistical value of the brightness of each pixel. After obtaining the statistical value of the brightness as described above, CPU27 assigns the corresponding statistical value of the brightness to the area with area number L. In addition, the inner side of the area refers to the opposite side of the outer side when the boundary line is regarded as the outermost side. Therefore, if the area with area number 2 is surrounded by the boundary lines of two areas (area numbers 1 and 3), it is defined as two statistical ranges toward the inside of each boundary line. On the other hand, if the area with area number 4 only has a boundary line with one area (area number 3) and there are no other boundaries, it is defined as one statistical range toward the inside of one boundary line.

If the brightness statistics are obtained for the area with area number L, CPU27 increments the area number L (step S225). Then, CPU27 determines whether the area number L is greater than the total number of areas captured by step S205 (step S230). In step S230, when the area number L is not determined to be greater than the total number of areas, CPU27 repeats the processing after step S215.

On the other hand, in step S230, when it is determined that the region number L is greater than the total number of regions, CPU27 obtains the threshold value that should be applied to the slice image of the slice number k (step S235). The classification process of the conductive part and the insulating part shown in Figure 3B is a process of classifying the object in the slice image into either the conductive part or the insulating part. The data representing the slice image is data representing the brightness value of each pixel, so if the brightness value is analyzed, any pixel can be classified as belonging to either the conductive part or the insulating part. In this embodiment, the classification is performed by comparing the brightness statistics of each region contained in the slice image with the threshold value.

Therefore, CPU27 obtains the threshold value that should be applied to the slice image of slice number k. The threshold value can be used as long as it can distinguish each area into two types such as a conductor part and an insulating part based on the brightness statistics of each area. The threshold value can be obtained by various methods. This embodiment uses the discriminant analysis method. Specifically, CPU27 performs the following calculations.

1. Set the total number of regions obtained in step S205 as Zmax

2. For the Zmax regions, sort the statistical values obtained from step S220 in ascending order

3. Divide the statistical values into data sets from No. 1 to No. i and data sets from No. i+1 to No. Zmax, and calculate the inter-group variance (i is any integer from 1 to Zmax)

4. The i (=i ₀ ) with the largest inter-group variance value in the loop survey

5. Set the i _0th data value as the threshold

According to the above processing, the threshold value for classification is obtained in such a way that the brightness statistics obtained in step S220 can maximize the inter-group variance.

Then, CPU27 classifies the region according to the threshold value (step S240). That is, CPU27 compares the brightness statistics obtained by step S220 with the threshold value obtained by step S235, and regards the region where the brightness statistics value is greater than the threshold value as the conductor part. In addition, CPU27 regards the region where the brightness statistics value is below the threshold value as the insulating part. In addition, various methods can be used to specify the threshold value using the discriminant analysis method. For example, the threshold value can be specified by maximizing the separation degree (=inter-group variation/intra-group variation).

Next, CPU27 stores the value representing the classification obtained in step S240 (step S245). That is, CPU27 assigns a corresponding value representing the classification result of each region (e.g., the conductive part is 1, the insulating part is -1, etc.) to each position (pixel) of each region. According to the above processing, each region in the slice image with slice number k is classified as either a conductive part or an insulating part.

FIG. 5A shows the brightness statistics (average) calculated according to the statistical range shown in FIG. 4F for the area numbers 1 to 4 divided as shown in FIG. 4C. In this example, the brightness statistics of the area numbers 1 to 4 are 125.1, 69.0, 114.8, and 249.9. In this example, the threshold value obtained according to the discriminant analysis method is 125.1. Therefore, in this example, the area numbers 1 to 3 are classified as insulating parts, and the area number 4 is classified as a conductive part. CPU 27 associates the information representing the classification result with each area. FIG. 5A illustrates an example of a corresponding association in which the conductive part is assigned 1 and the insulating part is assigned -1. In addition, in this embodiment, classification is implemented for each area, but the boundaries of the relevant areas are not classified, indicating that the value of the classification result is assigned to the corresponding association as 0.

If the value representing the classification has been stored, CPU27 increments the slice number k (step S250). Then, CPU27 determines whether the slice number k is greater than the total number of slice images (step S255). In step S255, unless it is determined that the slice number k is greater than the total number of slice images, CPU27 repeats the processing after step S205. In step S255, when it is determined that the slice number k is greater than the total number of slice images, the classification is terminated for all slice images.

The processing shown in FIG3B is implemented for multiple slice images respectively, and the classification result is assigned to each position (pixel) of each slice image. Each position of the slice image exists in a two-dimensional plane parallel to the first direction, and the slice image can be obtained by cutting and reconstructing information at different positions in the Z direction. Therefore, if the processing shown in FIG3B is completed, each position in the relevant three-dimensional space is classified as a conductive part or an insulating part.

When the classification is performed as described above, CPU27 returns to the processing shown in FIG3A. Then, CPU27 uses the function of the contrast emphasis unit 27e to perform contrast emphasis (step S120). That is, CPU27 refers to the value representing the associated classification result corresponding to each position of the reconstruction information, and adjusts the brightness contrast of the position according to whether the position belongs to the conductor part or the insulation part. Specifically, CPU27 maintains the brightness when the position is the conductor part, and reduces the brightness when the position is the insulation part, thereby emphasizing the contrast. In addition, the degree of brightness reduction can be preset, such as setting it to half the original brightness value.

Next, the CPU 27 performs boundary interpolation (step S130). The boundary interpolation is a process for preventing the boundary line from being seen unnaturally. The boundary interpolation of this embodiment is a process for preventing the brightness of the boundary line from changing abruptly. FIG. 5B shows a state diagram after assigning values representing the classification results to the corresponding associations of each region shown in FIG. 4C. In this example, only the corresponding association of region number 4 in the center is assigned 1, while the corresponding associations of the remaining region numbers 1 to 3 are assigned -1. The corresponding associations of the unclassified boundary lines are assigned 0. In FIG. 5B , the pixel whose corresponding association is assigned to 1 is colored with a white icon, the pixel whose corresponding association is assigned to -1 is colored with a gray icon, and the pixel whose corresponding association is assigned to 0 is colored with a black icon.

In this embodiment, the width of the boundary line is set to 1 pixel. In addition, for each area, contrast emphasis is used to make the brightness difference greater than the original image. Therefore, in a boundary line with a width of 1 pixel, if the brightness on both sides of the boundary line is used for interpolation processing, the brightness of the area boundary will change rapidly, resulting in an unnatural image. However, in the embodiment, CPU27 performs correction by increasing the width of the boundary line, and then performs interpolation processing based on the brightness of each area on both sides of the corrected boundary line to obtain the brightness of the corrected boundary line. The interpolation processing can use various known processing, for example, various smoothing filters can be applied. In addition, various calculations such as bilinear interpolation and bicubic interpolation can also be used for interpolation processing. FIG5C shows a slice taken from the reconstructed information after interpolation at the same position as FIG4A, and specifically shows the same portion as FIG4A. As shown in FIG5C, although the brightness difference of the insulating portion is visible, the brightness of the insulating portion is reduced, while the brightness of the conductive portion is not reduced, so the conductive portion and the insulating portion can be clearly distinguished. In addition, the boundary between the conductive portion and the insulating portion will not be seen unnaturally.

Through the above processing, a state in which the reconstruction information is corrected in a manner that can clearly distinguish the conductor part from the insulating part is presented. The X-ray inspection device of this embodiment can display the slice image in any direction based on the corrected reconstruction information. To perform the display, the CPU 27 uses the function of the display control unit 27f to accept the section (step S140). Specifically, the user specifies the section of the substrate W using the input unit 24. The section of this embodiment is a section of the substrate W cut in the second direction (Z direction) perpendicular to the first direction. The section can be specified by various methods, for example, the user can specify: the position of the penetration hole of the analysis object and the rotation direction of the rotation axis passing through the position and parallel to the Z direction, etc.

Next, CPU27 uses the function of the display control unit 27f to display the analysis target image (step S150). That is, CPU27 obtains information indicating the position penetration amount on the section accepted by step S140 from the corrected reconstruction information, and generates a slice image as the analysis target image. Then, CPU27 controls the output unit 25 to display the analysis target image. This result is, for example, that the image shown in FIG. 2E is regarded as the analysis target image and displayed on the output unit 25. There is almost no visible artifact in the image, and the user can easily distinguish the conductor part containing the through hole and the wiring pattern from the insulating part. Therefore, the user can easily analyze the quality of the conductor part based on the analysis target image.

Furthermore, in this embodiment, the first direction is a direction perpendicular to the stacking direction of the substrate W of the stacked substrate having multiple layers. Therefore, in this embodiment, the slice image cut in the first direction parallel to the layer of the substrate W becomes the object of analysis. Within the layer, the wiring pattern is expanded in a planar manner to form a relatively simple shape. On the other hand, in the vertical direction of the layer, the wiring pattern becomes more complex. For example, the slice image shown in FIG. 2C is simpler than the slice image shown in FIG. 2B. Therefore, by analyzing the slice image cut in the first direction, the conductor part and the insulating part can be classified using the image with a relatively simple shape.

Furthermore, in this embodiment, the first direction is the perpendicular direction of the rotation axis Ax. Therefore, the conductive part and the insulating part can be classified according to the slice image that is cut in a direction parallel to the circle of the rotation trajectory of the X-ray detector 12 and in the X-Y plane in the moving direction of the substrate W. Therefore, the cutting direction of the slice image can be easily specified from the spatial configuration of the X-ray generator 11, the X-ray detector 12, the positioning mechanism, etc. Therefore, the user can easily and intuitively grasp the direction from which the slice image for classifying the conductive part and the insulating part of the substrate W is cut.

Furthermore, in this embodiment, the conductor part and the insulating part are classified according to the multiple slice images of different cutting positions in the second direction. Therefore, the reconstruction information expanded in the height direction of the substrate W can be classified into the conductor part and the insulating part. In addition, if the cutting position in the second direction is set along the entire height of the substrate W, the conductor part and the insulating part can be classified for the reconstruction information covering the entire height range of the substrate W.

Furthermore, in this embodiment, since the conductive part and the insulating part are classified according to the slice image cut in the first direction, the classification can be performed according to the simple shape conductive part image contained in the slice image. Then, after the classification is performed, according to the reconstruction information emphasized by contrast, by setting the slice image cut in the second direction as the analysis target image, the analysis target image can be analyzed in a state that the conductive part and the insulating part can be clearly distinguished.

Furthermore, in this embodiment, for each of the multiple regions captured from the slice image, the brightness statistics representing the region values are obtained. Therefore, the classification of each region can be implemented according to the brightness statistics corresponding to each region, and the classification can be performed by analyzing a small number of statistics. Therefore, the conductor part and the insulation part can be classified by simple processing.

Furthermore, in this embodiment, the brightness statistics are obtained by statistically calculating the brightness within a statistical range of a predetermined width adjacent to the boundary line instead of targeting the entire area. Therefore, the value reflecting the brightness characteristics near the boundary line can be set as the brightness statistics value.

Furthermore, in this embodiment, the statistical range is a range of a predetermined width from the boundary toward the inner side of the region, and there is brightness overshoot or undershoot in this range. Therefore, according to this embodiment, the brightness can be emphasized (or suppressed) compared to the brightness of the entire region.

Furthermore, according to this structure, the difference between the brightness of the conductor part and the brightness of the insulating part can be made more obvious. Therefore, compared with the structure of statistically calculating the overall brightness of the area, it is easier to set the threshold for classifying the conductor part and the insulating part. FIG5D shows the brightness diagram of each position along the arrow A of the slice image shown in FIG4A. The upper figure is the same slice image as FIG4A, and the lower figure is a brightness diagram. In addition, in the slice image shown in the upper figure, each area is associated with the same area number as FIG4C, and the figure shown in the lower figure also corresponds to the area number.

As shown in the upper figure of FIG. 5D, the area with area number 4 is a closed area surrounded by a boundary line, and the brightness is greater than the area around area number 4. Therefore, there is only overshoot in the area with area number 4, and if the statistical value of the brightness within the statistical range adjacent to the boundary line is obtained, the statistical value becomes very large. As a result, the brightness corresponding to the area with area number 4 can be made very greater than the surrounding area. Accordingly, the closed area with brightness greater than the surrounding area can well present the image of the wiring pattern and the through hole. Therefore, according to this embodiment, the brightness value can be obtained in a manner that is very likely to classify the typical image of the wiring pattern and the through hole as the conductor part.

Furthermore, for example, in the area with brightness lower than the surrounding area, there is only undershoot in the area. If the brightness statistics within the statistical range of the adjacent boundary are obtained, the statistics become very small. As a result, the brightness of the area corresponding to the area with brightness lower than the surrounding area is very small. Accordingly, the area with brightness lower than the surrounding area is very likely to be an insulating part. Therefore, according to this embodiment, the brightness value can be obtained in a manner that the typical image of the insulating part is very likely to be classified as an insulating part.

Furthermore, in this embodiment, the threshold for classifying the region into the conductor part and the insulation part is obtained according to the discriminant analysis method. Therefore, the threshold for classifying the region can be easily obtained.

(3) Other embodiments:

The above embodiment is one embodiment of the present invention, and other embodiments may also be adopted. For example, the form of the X-ray inspection device is not limited to the form shown in FIG. 1. For example, a plurality of X-ray imaging mechanisms 10 may be installed in a factory, and the control unit 20 is implemented by a computer, so that the control unit 20 of a computer is used to control the plurality of X-ray imaging mechanisms 10.

Furthermore, at least a part of the X-ray image acquisition unit 27a, the reconstruction calculation unit 27b, the slice image acquisition unit 27c, the classification unit 27d, the contrast emphasis unit 27e, and the display control unit 27f may also be separately present in multiple devices. For example, the function of the display control unit 27f may be used to obtain the function of the analyzed object image by using the CPU 27, and then transmit the analyzed object image to other devices, and use the function of the display control unit 27f provided in other devices to display the analyzed object image on the display.

Of course, some of the structures of the above-mentioned implementation methods can be omitted, and the processing order can also be changed or omitted. For example, the pre-processing of step S100 shown in FIG. 3A introduces noise when capturing the edge so that the edge cannot be seen. Therefore, if the edge can be captured without being affected by noise even without pre-processing, the pre-processing can be omitted. In addition, for example, if the interpolation of the relevant boundary portion does not appear unnatural even if it is interpolated by the boundary line of one pixel, it can be interpolated by the boundary line of one pixel. In addition, by being able to see that the brightness of the pixel adjacent to the boundary line is the same or the brightness of the reconstructed information at the boundary position can be used without correction, interpolation can be omitted. Furthermore, in the above-mentioned embodiment, the threshold value used for classifying the conductive part and the insulating part is obtained based on the slice image of each slice number k, but a common threshold value can also be obtained from a plurality of slice images.

The first direction of cutting the substrate by the slice image acquisition unit is not limited to the direction parallel to the substrate surface as in the above-mentioned embodiment, but can be various directions. However, because the slice image obtained by cutting in a direction parallel to the first direction is an image for performing analysis for classifying the conductor part and the insulating part, the direction parallel to the first direction is set to be able to perform the analysis. For example, when cutting in the first direction, it is best to select the first direction in such a way that the conductor part does not become a small area that cannot be analyzed.

The classification part only needs to be able to classify the slice image into the conductor part and the insulation part. Therefore, in addition to the above-mentioned embodiments, various methods can also be used to implement classification. For example, the statistical range for obtaining the statistical value of the brightness of each area can also be used, and the composition that is not limited to a part of the area can also be used. In addition, for the boundary of a specific area, the design information of the substrate (the design position of the conductor part and the insulation part) can also be considered. In addition, various known processes such as noise reduction processing, boundary extraction, reduction processing, magnification processing, morphology processing, etc. can also be used. In addition, some of the processing (for example: reduction processing, magnification processing, etc.) can also be omitted. Furthermore, it is not necessary to use topographic processing during shrinkage or expansion processing. For example, an area within a certain distance from the boundary can be set as the statistical range and the capture process can be performed.

The contrast emphasis part can be any part that can emphasize the contrast between the conductor part and the insulating part. Therefore, the method for emphasizing the contrast is not limited to the above method. For example, the brightness of the insulating part can be reduced while the brightness of the conductor part can be increased. In addition, the brightness of the conductor part can be reduced as long as the contrast can be emphasized by reducing the brightness of the insulating part. In addition, even if the brightness of the insulating part is maintained (or increased), the contrast can be emphasized by adjusting the brightness of the conductor part. That is, by emphasizing the contrast, it is less likely to identify the false image than before emphasizing the contrast, and the conductor part and the insulating part can be easily distinguished.

The statistical range is the range where the brightness from the boundary to the inside of the area has overshoot or undershoot. Therefore, as long as either overshoot or undershoot is included in the statistical range, the statistical values of the brightness of the conductor and the insulation are more likely to differ compared to the configuration in which only the brightness of the flatter portion in the area is used for statistics. The statistical range is the range that includes overshoot or undershoot, and it is sufficient to include the flat portion after the overshoot or undershoot is attenuated, but in order to make the brightness statistical value more likely to differ, it is better not to include the flat portion.

Furthermore, the classifying of the conductive part and the insulating part according to the slice image of the reconstructed information and the contrast and emphasis technique can also be applied to the program or method. Also, part of it can be appropriately changed to software, or part of it can be changed to hardware, etc. In addition, the recording medium of the program of the control device is also an invention. Of course, the recording medium of the software can be a magnetic recording medium, a semiconductor memory, or any other recording medium developed in the future.

10: X-ray imaging device

11: X-ray generator

11a: X-ray output unit

12: X-ray detector

12a: Detection surface

20: Control Department

21: Generator control unit

22: Detector control unit

23: Positioning mechanism control unit

24: Input section

25: Output section

26: Memory

26a: Program data

26b: X-ray image data

27:CPU

27a: X-ray image acquisition unit

27b: Reconstruction calculation department

27c: Slice image acquisition unit

27d: Classification Department

27e: Contrast emphasis section

27f: Display control unit

W: Substrate

Claims (10)

An X-ray inspection device includes: an X-ray image acquisition unit, which acquires a plurality of X-ray images taken by irradiating a substrate with X-rays at an angle inclined to a predetermined direction; a reconstruction calculation unit, which performs a reconstruction calculation based on the X-ray images to acquire reconstruction information; a slice image acquisition unit, which acquires a slice image of the substrate cut in a preset first direction based on the reconstruction information, and acquires a plurality of slice images at different cutting positions; a classification unit, which classifies the slice images into a conductive portion and an insulating portion, and classifies the slice images based on the reconstruction information. The classification results of the slice images are used to classify each position representing the reconstruction information as the conductive part or the insulating part; the contrast emphasis part emphasizes the contrast between the conductive part and the insulating part, and emphasizes the brightness contrast of each position according to whether each position representing the reconstruction information belongs to the conductive part or the insulating part; and the display control part obtains the analysis target image of the substrate cut in a second direction perpendicular to the first direction according to the reconstruction information after contrast emphasis, and displays the analysis target image on the display part. An X-ray inspection device as described in claim 1, wherein the contrast emphasis portion reduces the brightness of the insulating portion. The X-ray inspection device as described in claim 1, wherein the classification unit captures the boundary contained in the slice image, captures the area divided by the captured boundary, obtains the brightness statistics of the captured area, and classifies the area into the conductive part or the insulating part according to the brightness statistics of the area. An X-ray inspection device as described in claim 3, wherein the brightness statistical value of the region is a value obtained by statistically obtaining the brightness within a statistical range, and the statistical range is a predetermined width range from the boundary toward the inside of the region. An X-ray inspection device as described in claim 4, wherein the statistical range is the range of brightness overshoot or undershoot from the boundary toward the inner side of the region. An X-ray inspection device as described in claim 4, wherein the classification section classifies the region into the conductor portion or the insulating portion by comparing the brightness statistical value of the region with a threshold value; the threshold value is a value that is specifically binarized using a discriminant analysis method based on the plurality of brightness statistical values obtained from the plurality of statistical ranges contained in the slice image. An X-ray inspection device as described in claim 1, wherein the substrate is a stacked substrate having a plurality of layers; and the first direction is a direction perpendicular to the stacking direction of the layers. The X-ray inspection device as described in claim 1, wherein the X-ray image acquisition unit rotates at least one of the X-ray detector, the X-ray generator, and the substrate around a rotation axis and captures the X-ray image at multiple positions at different rotation angles; and the first direction is a direction perpendicular to the rotation axis. An X-ray inspection method includes: an X-ray image acquisition step, which is to obtain a plurality of X-ray images taken by irradiating a substrate with X-rays at an angle inclined to a predetermined direction; a reconstruction calculation step, which is to perform a reconstruction calculation based on the X-ray images to obtain reconstruction information; a slice image acquisition step, which is to obtain a slice image of the substrate cut in a preset first direction according to the reconstruction information, and obtain a plurality of slice images at different cutting positions; a classification step, which is to classify the slice images into a conductive part and an insulating part, and According to the classification results of the slice images, each position representing the reconstruction information is classified as the conductor part or the insulating part; the contrast emphasis step is to emphasize the contrast between the conductor part and the insulating part, which is to emphasize the brightness contrast of each position according to whether each position representing the reconstruction information belongs to the conductor part or the insulating part; and the display control unit is to obtain the analysis object image of the substrate cut in a second direction perpendicular to the first direction according to the reconstruction information after contrast emphasis, and display the analysis object image on the display unit. An X-ray inspection program enables a computer to have the following functions: An X-ray image acquisition unit, which acquires a plurality of X-ray images taken by irradiating a substrate with X-rays at an angle inclined to a predetermined direction; a reconstruction calculation unit, which performs a reconstruction calculation based on the X-ray images to acquire reconstruction information; a slice image acquisition unit, which acquires a slice image of the substrate cut in a preset first direction based on the reconstruction information, and acquires a plurality of slice images at different cutting positions; a classification unit, which classifies the slice images into conductive parts and insulating parts , and according to the classification results of the slice images, each position representing the reconstruction information is classified as the conductor part or the insulating part; a contrast emphasis part, which emphasizes the contrast between the conductor part and the insulating part, emphasizes the brightness contrast of each position according to whether each position representing the reconstruction information belongs to the conductor part or the insulating part; and a display control part, which obtains an analysis target image of the substrate cut in a second direction perpendicular to the first direction according to the reconstruction information after contrast emphasis, and displays the analysis target image on the display part.

Images