Image processing method and image processing device

Abstract

To provide an image processing method and an image processing device which can generate a highly accurate image.SOLUTION: A plurality of pixel values nH, nM, nL obtained from a target region 3 are composed of one reference pixel value and other pixel values, in which a plurality of correction patterns are set in advance. An image processing method comprises: a correction pattern determination step for determining a correction pattern applied to the target region 3 based on the reference pixel value; a correction step for correcting the reference pixel value and the other pixel values by the determined correction patterns A, B, C; and a generation step for generating an image of the target to be imaged based on the corrected pixel values of the plurality of target region 3.SELECTED DRAWING: Figure 3

Description

[Technical field]

[0001] The present invention relates to an image processing method and an image processing device that processes images obtained by an X-ray device or the like, and more particularly to an image processing method and an image processing device that are capable of generating highly accurate images. [Background technology]

[0002] Various image processing methods for X-ray devices and the like have been proposed (see, for example, Patent Document 1). Patent Document 1 discloses an image processing method that generates an image with emphasized edges by processing one image with three filters and then combining the results. This image processing method makes it possible to generate an image in which the boundaries between objects made of different materials are clearly defined. Because the image is clearer, visibility is improved.

[0003] The applicant has proposed an energy discrimination type X-ray device that acquires pixel values ​​for each X-ray energy range (see, for example, Patent Document 2). The X-ray device described in Patent Document 2 acquires images for each energy range from an object to be photographed, and synthesizes these multiple images to generate one image.

[0004] In the X-ray device described in Patent Document 2, there was a problem in that quantitativeness was lost when image processing was performed to emphasize edges using the method described in Patent Document 1. When image processing was performed to emphasize edges, a deviation occurred in the estimated effective atomic number. There was a problem in that the accuracy of the X-ray device as an energy analyzer was reduced. [Prior art documents] [Patent documents]

[0005] [Patent Document 1] Japanese Patent Publication No. 2009-164 [Patent Document 2] International Publication No. 2019 / 083014 Summary of the Invention [Problem to be solved by the invention]

[0006] The present invention has been made in consideration of the above problems, and an object of the present invention is to provide an image processing method and an image processing device that are capable of generating a highly accurate image. [Means for solving the problem]

[0007] An image processing method for achieving the above-mentioned object, in which a plurality of pixel values ​​are obtained from a target area constituting a part of a photographed object, a composite of the plurality of pixel values ​​is determined as a pixel value of the target area, and an image of the photographed object is generated based on the plurality of target areas for which pixel values ​​have been determined, is characterized in that the plurality of pixel values ​​obtained from the target area are composed of one reference pixel value and other pixel values, a plurality of correction patterns are set in advance, and the method includes a correction pattern determination step in which the correction pattern to be applied to the target area is determined based on the reference pixel value, a correction step in which the reference pixel value and other pixel values ​​are corrected by the determined correction pattern, and a generation step in which an image of the photographed object is generated based on the corrected pixel values ​​of the plurality of target areas.

[0008] An image processing device for achieving the above-mentioned object acquires a plurality of pixel values ​​from a target area constituting a part of a photographed object, determines a pixel value of the target area by combining the plurality of pixel values, and generates an image of the photographed object based on the plurality of target areas for which pixel values ​​have been determined, and is characterized in that the image processing device comprises an acquisition mechanism that acquires one reference pixel value and other pixel values ​​for each of the target areas, a processing mechanism that processes the pixel values ​​acquired by the acquisition mechanism, and a memory mechanism that stores a plurality of correction patterns used by the processing mechanism, and the processing mechanism has a correction pattern determination unit that determines the correction pattern to be applied in the target area based on the reference pixel value, a correction unit that corrects the reference pixel value and other pixel values ​​using the determined correction pattern, and a generation unit that generates an image of the photographed object based on the corrected pixel values ​​of the plurality of target areas. Effect of the Invention

[0009] According to the present invention, multiple pixel values ​​obtained from one target region are corrected with the same correction pattern. It is possible to avoid a state in which multiple pixel values ​​obtained from one target region are corrected with different correction patterns, resulting in inconsistency between the multiple pixel values. The present invention is advantageous for generating a highly accurate image. [Brief description of the drawings]

[0010] [Figure 1] FIG. 2 is an explanatory diagram illustrating an example of a photographing object and a plurality of images obtained. [Diagram 2] 2 is an explanatory diagram illustrating an example of an image generated by combining the multiple images in FIG. 1; [Diagram 3] 10A and 10B are explanatory diagrams illustrating the states of acquired pixel values. [Figure 4] FIG. 1 is an explanatory diagram illustrating a flow of an image processing method. [Diagram 5] FIG. 1 is an explanatory diagram illustrating a schematic example of an image processing method. [Figure 6] 10 is an explanatory diagram illustrating an example of a correction range set for a pixel; [Figure 7] FIG. 11 is an explanatory diagram illustrating a correction pattern. [Figure 8] FIG. 11 is an explanatory diagram illustrating an example of an uncorrected image. [Figure 9] 1 is an explanatory diagram illustrating an example of an image corrected by an image processing method. [Figure 10] FIG. 1 is an explanatory diagram illustrating an example of an image corrected by a conventional image processing method. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0011] The image processing method and image processing device will be described below based on the embodiment shown in the drawings. In the drawings, the horizontal direction is indicated by arrow x, the vertical direction perpendicular to the horizontal direction x is indicated by arrow y, and the height direction perpendicular to the horizontal direction x and the vertical direction y is indicated by arrow z.

[0012] The image processing method of the present invention will be described by taking as an example a case where an object 1 is photographed with an energy discrimination type X-ray device. As shown in the left side of Fig. 1, the object 1 is made of an experimental material, for example, a combination of aluminum 1a and acrylic 1b. For the sake of explanation, the area of ​​aluminum 1a is shaded in Fig. 1. The object 1 is configured to have a size of, for example, 40 mm in the horizontal direction x, 30 mm in the vertical direction y, and 10 mm in the height direction z. The atomic number of aluminum 1a is 13, and the effective atomic number of acrylic 1b is 6.5.

[0013] When X-rays are irradiated from the X-ray device to the object 1 along the height direction z, the X-rays that pass through the object 1 are detected by a detector. The detector has a detection layer made of, for example, a semiconductor (CdTe, CZT (CdZnTe), etc.) that directly converts X-rays into an electrical signal. This detection layer is made up of, for example, 200 μm x 200 μm pixels arranged in the horizontal direction x and vertical direction y. When imaging, a pixel value is obtained for each pixel.

[0014] The energy discrimination type X-ray device counts the number of photons in a state where the number of photons is discriminated for each predetermined energy range according to the magnitude of the energy of the photons detected by the pixels of the detector. For example, three energy ranges are set: a high energy BIN, a medium energy BIN, and a low energy BIN. For example, when 100 photons are incident on a pixel, the photons are counted for each energy range, such as 20 in the high energy BIN, 50 in the medium energy BIN, and 30 in the low energy BIN. As shown in the right side of FIG. 1, an image is generated for each energy range. In FIG. 1, an image corresponding to the high energy BIN is shown as (1), an image corresponding to the medium energy BIN as (2), and an image corresponding to the low energy BIN as (3). The number of energy ranges set in advance is not limited to three, and may be set to two or four or more.

[0015] As shown in the example of Figure 2, one X-ray image is generated by synthesizing three images generated for each energy range. This X-ray image is an effective atomic number image that reflects the atomic numbers of the materials that make up the photographed object 1. In the effective atomic number image, the effective atomic number is expressed by color tone or black and white shading. Here, synthesis includes performing calculations using pixel values ​​of multiple images to generate new pixel values.

[0016] The above is the process of obtaining an X-ray image using a conventional energy discrimination type X-ray device. Next, the image processing method of the present invention will be described. FIG. 3 shows a schematic diagram of pixel values ​​n obtained from 3×3 pixels 2. In photographing an object 1 to be photographed, pixel values ​​n are obtained for each object region 3. In this embodiment, one pixel 2 is set as one object region 3. A plurality of pixels 2 may be set as one object region 3.

[0017] In one target region 3, pixel values ​​n are obtained for each energy range. For example, in a 3×3 target region 3, 3×3×3=27 pixel values ​​n are obtained. For the sake of explanation in FIG. 3, the three pixel values ​​n obtained in one target region 3 are expressed by stacking the target region 3 in the height direction in FIG. 3. In this specification, the height direction in FIG. 3 may be referred to as a virtual height direction z'. The virtual height direction is a different concept from the actual height z. The stacking of target regions 3 in the virtual height direction is virtual, and merely represents that three pixel values ​​n are obtained from one target region 3. The actual target regions 3 are formed by lining up in the horizontal directions x and y, and are not formed in multiple in the height direction z. In a certain target region 3, the pixel value n corresponding to a high-energy BIN is H and the pixel value n corresponding to the medium energy BIN M and n corresponding to the low energy BIN L These three pixel values ​​can be said to be pixel values ​​when the same location (same coordinates) on the object 1 to be photographed is photographed.

[0018] Three pixel values ​​n H , n M , n Lis composed of one reference pixel value and other pixel values. For example, the pixel value n H When the pixel value n corresponding to the medium energy BIN is set as the reference pixel value, M and the pixel value n corresponding to the low energy BIN L The reference pixel value can be preset. Specifically, for example, the pixel value n H It is possible to set in advance the reference pixel value.

[0019] When the image processing method is started as illustrated in Fig. 4, one correction pattern to be applied to one target region 3 is determined from a plurality of preset correction patterns (hereinafter, sometimes referred to as a correction pattern determination step S10). Specifically, for example, three correction patterns, pattern A, pattern B, and pattern C, are preset. In the correction pattern determination step S10, one correction pattern, for example, pattern A, is determined based on the reference pixel value. As illustrated in Fig. 3, a correction pattern is determined for each target region 3.

[0020] The reference pixel value and other pixel values ​​are corrected by the determined correction pattern (hereinafter, sometimes referred to as correction step S20). When pattern A is determined in a certain target area 3, the reference pixel value and other pixel values ​​in this target area 3 are corrected by pattern A. In other words, multiple pixel values ​​n obtained from the same coordinates of the shooting target 1 are corrected by pattern A. H , n M , n L are corrected with a common pattern in correction step S20. All three target regions 3 stacked in the imaginary height direction z' are corrected with the same correction pattern. In the multiple target regions 3, the pixel value n is corrected. In the example of FIG. 3, 27 pixel values ​​n are corrected.

[0021] Based on the pixel values ​​n' corrected in the multiple target regions 3, an image of the object 1 is generated (hereinafter, sometimes referred to as generation step S30). The pixel value n is corrected for each target region 3, and one image is generated based on the corrected pixel value n'. Specifically, for example, H , n M , n L are corrected to obtain three corrected pixel values ​​n'. In the embodiment illustrated in Figure 3, 27 corrected pixel values ​​n' are obtained. The 27 corrected pixel values ​​n' are combined to generate an image.

[0022] In one target region 3 in FIG. 3, three pixel values ​​n H , n M , n L are corrected with the same correction pattern. This makes it possible to avoid a situation in which multiple pixel values ​​n obtained from one target region 3 are corrected with different correction patterns, resulting in inconsistency between the multiple pixel values ​​n. It can be said that consistency can be achieved among the corrected pixel values ​​n in the virtual height direction z'. This image processing method is advantageous for generating highly accurate images.

[0023] The imaging to which the image processing method is applied is not limited to imaging using an energy discrimination type X-ray device. This image processing method can also be applied to MRI (magnetic resonance imaging) and CT (computed tomography). The image processing method described above can be applied to any imaging method that acquires multiple pixel values ​​n (reference pixel value and other pixel values) from one target region 3 specified by three-dimensional coordinates in an imaging target 1.

[0024] It is desirable that the reference pixel value be composed of pixel values ​​n obtained under the same conditions for a plurality of target regions 3. For example, pixel values ​​n of high energy BINs in all target regions 3 HIt is possible to set in advance the pixel value n' as the reference pixel value. Since the pixel value n acquired under the same conditions is set as the reference pixel value, it is possible to avoid a problem in which the correction patterns are not consistent between the target regions 3. It can be said that the corrected pixel values ​​n' are consistent in the horizontal directions x and y in FIG. 3. The image processing method having the above configuration is advantageous for generating images with even higher accuracy.

[0025] Here, the same conditions refer to the same method or the same time. The same method refers to a method of acquiring pixel values ​​in the same energy range or under the same measurement conditions. The same time refers to a method of acquiring pixel values ​​n at the same time when taking multiple images.

[0026] The reference pixel value may be composed of pixel values ​​n obtained under different conditions depending on the target region 3. Specifically, the reference pixel value of a certain target region 3 may be a pixel value n of a high energy BIN. H The reference pixel value of the other target area 3 adjacent to this is the pixel value n L In this case, one of the acquired pixel values ​​n is determined as the reference pixel value. For example, the largest pixel value n may be determined as the reference pixel value. In this case, the conditions for determining the reference pixel value are set in advance.

[0027] Next, an image processing device 4 for implementing the above-mentioned image processing method will be described. The image processing device 4 can be configured with various known computers. This computer has a central processing unit (CPU), a main storage unit (memory), and an auxiliary storage unit (e.g., HDD). The computer may be configured to be connected to an input unit (keyboard, mouse) and an output unit (display, printer).

[0028] 5, the image processing device 4 includes an acquisition mechanism 5 that acquires pixel values ​​n acquired by a detector or the like separately for each target region 3, a processing mechanism 6 that processes the pixel values ​​n acquired by the acquisition mechanism 5, and a storage mechanism 7 that stores a plurality of correction patterns used by the processing mechanism 6. The processing mechanism 6 includes a correction pattern determination unit 8 that executes a correction pattern determination step S10, a correction unit 9 that executes a correction step S20, and a generation unit 10 that executes a generation step S30.

[0029] A correction pattern determination unit 8 acquires a plurality of pixel values ​​n from the acquisition mechanism 5, and determines a correction pattern based on a reference pixel value among the pixel values ​​n. The determined correction pattern, for example, pattern A, is read from the storage mechanism 7 to the processing mechanism 6. A correction unit 9 corrects the pixel value n based on the pixel value n and the correction pattern read from the storage mechanism 7. A generation unit 10 generates an image of the object 1 to be photographed based on the pixel value n' corrected by the correction unit 9. The generated image is displayed, for example, on a display connected to a computer.

[0030] Next, a specific example of the correction pattern determination step S10 will be described. As illustrated in Fig. 4, the correction pattern determination step S10 may be composed of a range setting step S11 for setting a predetermined range P1 centered on the target area 3, an average calculation step S12 for calculating an average pixel value M within this range P1, a comparison step S13 for comparing the pixel value n of the target area 3 with the average pixel value M, and a selection step S14 for selecting a correction pattern according to the comparison result. For the sake of explanation, the specific processes included in the correction pattern determination step S10 are enclosed by dashed lines in Fig. 4.

[0031] As shown in Fig. 6, in the range setting step S11, a rectangular range (hereinafter sometimes referred to as correction range P1) including one target area 3 to be corrected (e.g., C3,3) is set as the center, and includes 5 x 5 target areas 3. 25 target areas 3 including the target area 3 to be corrected (C3,3) are set as the correction range P1. For the sake of explanation, the correction range P1 is indicated by a dashed line in Fig. 6.

[0032] The size of the correction range P1 is not limited to 5×5, and may be increased to, for example, 7×7. The more target areas 3 included in the correction range P1, the more accurate the correction can be. On the other hand, the greater the number, the greater the amount of calculation required to correct the pixel value n. If the number of target areas 3 is too large, the effect of the correction becomes too great, and the deviation between the actual object 1 to be photographed and the generated image may become large.

[0033] The size of the correction range P1 may be reduced, for example to 3 × 3. The fewer the number of target regions 3 included in the correction range P1, the less the amount of calculation required to correct the pixel value n. On the other hand, the smaller the number, the greater the influence of noise, which may result in a decrease in the accuracy of the correction.

[0034] The image processing device 4 may have a configuration that allows the correction range P1 to be changed by an operation. By changing the correction range P1 according to the size of the object 1 to be photographed and the size of the range to be photographed in particular, appropriate photographing can be realized.

[0035] The shape of the correction range P1 is not limited to a square. The shape of the correction range P1 may be a rectangle, a circle, a triangle, or the like. It may be a polygon having pentagons or more sides. The correction range P1 may be formed in a cross or diamond shape with the target area 3 at its center.

[0036] In the average calculation step S12, first, the pixel values ​​n of the target region 3 (C1,1-C5,5) included in the correction range P1 are sorted in order of magnitude. The pixel values ​​n are n1, n2, ... n25 in ascending order. Here, when the reference pixel value in the target region 3 (C3,3) is the pixel value ni of the high-energy BIN, the pixel values ​​n1-25 are all used as the pixel values ​​of the high-energy BIN. When the reference pixel value in the target region 3 (C3,3) is the pixel value ni of the low-energy BIN, the pixel values ​​n1-25 are all used as the pixel values ​​of the low-energy BIN.

[0037] Next, the average pixel value M in the correction range P1 is calculated. The average pixel value M can be expressed, for example, as the average value of the three smallest pixel values ​​(n1-3) and the three largest pixel values ​​(n23-25) in the correction range P1. Specifically, the average pixel value M can be expressed as M = (n1 + n2 + n3 + n23 + n24 + n25) / 6. The average pixel value M is not limited to the above, and may be the average value of four or more pixel values ​​n on each of the smallest and largest sides. The average pixel value M may also be expressed as the average value of all pixel values ​​n in the correction range P1. Specifically, the average pixel value M can be expressed as M = (n1 + ··· + n25) / 25.

[0038] In a comparison step S13, the average pixel value M is compared with pixel values ​​ni of the target region 3 (C3,3) to be corrected. In a selection step S14, a correction pattern is determined according to the comparison result.

[0039] The correction pattern is determined according to how the pixel values ​​ni of the target area 3 (C3,3) to be corrected compare with the pixel values ​​n of the surrounding target areas 3. Therefore, the pixel values ​​ni of the target area 3 (C3,3) to be corrected can be corrected while effectively utilizing the pixel values ​​n of the surrounding target areas 3. This configuration is advantageous for generating a highly accurate image.

[0040] For the target region 3 (C3,4) to be corrected, a 5×5 correction range P1 is set centered on this target region 3 in the same manner as above, and the pixel values ​​ni of the target region 3 (C3,4) are corrected. If the reference pixel value in the target region 3 (C3,4) is a medium energy BIN, the pixel value n used to calculate the average pixel value M is the pixel value n of the medium energy BIN. By repeating the above, the pixel values ​​ni of multiple target regions 3 are corrected.

[0041] As illustrated in FIG. 6, in the range setting step S11, there are cases where the correction range P1 goes out of the image, such as when the target area 3 (C5,9) to be corrected hits the edge of the image, and the 5×5 correction range P1 cannot be set. For target areas 3 where the correction range P1 cannot be set in this way, the correction step S20 can be omitted. When taking an image using an X-ray device or the like, the object 1 to be photographed is almost always positioned so that a particularly important part is at the center of the image. The edges of the image obtained by photographing are rarely required to be accurate. Therefore, for target areas 3 where the correction range P1 cannot be set, even if no correction is performed, there is almost no problem with the accuracy of the image. The shape of the correction range P1 may be different between the center and the edges of the image.

[0042] The correction step S20 may be performed using the pixel value n that is within the range set in the range setting step S11 and that is actually obtained. In this case, as shown in FIG. 6, the correction can be performed for the target region 3 (C5,9) using the pixel values ​​n in the correction range P1 surrounded by a dashed line. In this case, the pixel values ​​n included in the correction range P1 are n1-15. The average pixel value M can be expressed, for example, as M=(n1+n2+n3+n13+n14+n15) / 6.

[0043] 4, the correction pattern determination unit 8 that executes the correction pattern determination step S10 of the above embodiment has a range setting unit 11 that sets a predetermined range centered on the target region 3, an average calculation unit 12 that calculates an average pixel value M within this range, a comparison unit 13 that compares the pixel values ​​ni of the target region 3 with the average pixel value M, and a selection unit 14 that selects a correction pattern depending on the comparison result. For the sake of explanation, the range setting unit 11 and other components that constitute the correction pattern determination unit 8 are shown by dashed lines in FIG.

[0044] Next, a specific example of the correction pattern will be described. As shown in FIG. 7, a numerical range P2 is set in advance based on the average pixel value M. The numerical range P2 is set to a range of an allowable error ±d with the average pixel value M as the center. The allowable error d is set to d=10, for example. If the number of photons detectable by one pixel 2 is, for example, 255, the pixel value n obtained from this pixel 2 will be 0-255. 4% of the detectable pixel value n is 10. This number can be set in advance as the allowable error d. The maximum value detectable by pixel 2 is set to d. max In this case, the allowable error d is d=0.04d max It can be expressed as:

[0045] The allowable error d is the maximum value d max It is recommended to set it in the range of 1-10% of 0.01d. max ≦d≦0.10d max The allowable error d is set so that it satisfies the following: More preferably, the allowable error d is set to the maximum value d max Set it within the range of 1-3% of 0.01d. max ≦d≦0.03d max If the target area 3 is made up of multiple pixels 2, the maximum value that can be detected in one target area 3 is set to d max And this maximum value d max The allowable error d is preset based on

[0046] 7, when the pixel value ni of the target area 3 to be corrected is included in the numerical range P2, the pixel value ni of the target area 3 is adopted as the corrected pixel value n' (hereinafter, sometimes referred to as a no-correction pattern). In other words, when Md≦ni≦M+d is satisfied, the pixel value ni is not corrected.

[0047] As shown in the middle section of FIG. 7, when the pixel value ni of the target area 3 to be corrected is smaller than the numerical range P2, the average value of a predetermined number (for example, three) on the smaller side among the pixel values n1-25 within the range set in the range setting step S11 is adopted as the corrected pixel value n' (hereinafter sometimes referred to as a decreasing correction pattern). That is, when ni < M - d is satisfied, n' = (n1 + n2 + n3) / 3. As shown in the middle section of FIG. 7, the pixel value ni of the target area 3 is corrected to the corrected pixel value n'. The pixel value ni is corrected to a value smaller than the original value.

[0048] As shown in the lower section of FIG. 7, when the pixel value ni of the target area 3 to be corrected is larger than the numerical range P2, the average value of a predetermined number (for example, three) on the larger side among the pixel values n1-25 within the range set in the range setting step S11 is adopted as the corrected pixel value n' (hereinafter sometimes referred to as an increasing correction pattern). That is, when M + d < ni is satisfied, n' = (n23 + n24 + n25) / 3. As shown in the lower section of FIG. 7, the pixel value ni of the target area 3 is corrected to the corrected pixel value n'. The pixel value ni is corrected to a value larger than the original value.

[0049] The number of pixel values n (for example, n1-3) used to calculate the corrected pixel value n' is not limited to three. The number of pixel values n may be less than three, such as two or one. It may also be larger than three, such as four or five. If the number of pixel values n is too small, it is likely to be affected by noise, and the accuracy of the corrected pixel value n' may decrease. If the number of pixel values n is too large, the change amount of the corrected pixel value n' becomes small, and it becomes difficult to correct blurring.

[0050] When the pixel value ni of the target area 3 is close to the average pixel value M, in order to adopt a non-correction pattern, the state (atomic number) of the imaging object 1 can be accurately reproduced in the image. When the pixel value ni of the target area 3 is slightly smaller or slightly larger than the surroundings, since it is corrected to a sufficiently small or sufficiently large value, blurring of the image can be suppressed. This configuration is advantageous for generating a high-precision image.

[0051] The preset correction pattern is not limited to the above. The correction pattern can be appropriately set according to the type of imaging device such as an X-ray device or an MRI device and the purpose of imaging. For example, if the pixel value ni of the target area 3 is not included in the numerical range P2, the pixel value ni may be increased or decreased at a preset ratio. If the pixel value ni of the target area 3 is smaller than the numerical range P2, the pixel value ni may be corrected to, for example, 20%, and if it is larger, the pixel value ni may be corrected to, for example, 180%. Specifically, if the average pixel value M=100, d=10, and the pixel value ni of the target area 3 is 80, the corrected pixel value n'=80×0.2=16. If the pixel value ni of the target area 3 is 120, the corrected pixel value n'=120×1.8=216.

[0052] The specific content of the correction pattern determination step S10 is not limited to the configuration shown in FIG. 4. The specific content of the correction pattern determination unit 8 is not limited to the configuration exemplified in FIG. 5. For example, the correction pattern may be determined by comparing the pixel value ni of the target area 3 (C3,3) with the pixel value of one adjacent target area 3 (C2,3). When the pixel value ni of the target area 3 (C3,3) is within a predetermined range of the pixel value n to be compared, a correction pattern without correction may be used; when the pixel value ni of the target area 3 (C3,3) is smaller than the predetermined range, a correction pattern in which the pixel value ni is decreased may be used; and when the pixel value ni is larger than the predetermined range, a correction pattern in which the pixel value ni is increased may be used. In this case, the specific content of the correction pattern determination step S10 is different from the configuration exemplified in FIG. 4. The specific content of the correction pattern determination unit 8 is different from the configuration exemplified in FIG. 5.

[0053] The detector may be configured with a line sensor in which multiple pixels 2 are arranged in a line. In this case, the object 1 to be photographed is photographed by the X-ray device as it passes the line sensor while moving in one direction on a conveyor or the like. The image acquired by the line sensor is the same as when it is photographed by a detector in which multiple pixels 2 are arranged in the horizontal direction x and vertical direction y. Therefore, even when a line sensor is used, the image processing method described above is applicable.

[0054] Next, an image obtained by the image processing method will be described. Fig. 8 shows an effective atomic number image obtained when an object 1 is imaged using an energy discrimination X-ray device. In Fig. 8, the effective atomic number is expressed by black and white shading. This image corresponds to the enlarged area P3 indicated by the dashed line in Fig. 1. The image in Fig. 8 is an uncorrected image that has not been subjected to correction.

[0055] As shown in Fig. 8, at the boundary between aluminum 1a with atomic number 13 and acrylic 1b with effective atomic number 6.5, the image may become blurred due to the effects of charge sharing and the like. A phenomenon (charge sharing) may occur in which a photon incident on a pixel 2 adjacent to this pixel 2 is detected as a photon detected by the pixel 2. The effects of charge sharing and the like may cause the image to become blurred, reducing the accuracy of the image.

[0056] The image in Figure 8 can be interpreted as meaning that there is a substance with a different atomic number, such as effective atomic number 9, between the aluminum 1a and the acrylic 1b. Also, the image in Figure 8 makes it impossible to accurately grasp the length L1 of the acrylic 1b in the horizontal direction x. The actual length L1 of the acrylic 1b in the photographed object 1 is 1 mm.

[0057] FIG. 9 shows an effective atomic number image generated after imaging the object 1 with an energy discrimination type X-ray device and correcting the pixel value n by the image processing method described above. By the image processing method described above, the pixel value ni of the target area 3 is corrected to a state where it is sufficiently smaller or larger than the surrounding pixel value n at the boundary between materials with different atomic numbers. In other words, when the change in pixel value n within the correction range P1 is large and the pixel value n is smaller or larger than the numerical range P2, it is corrected to a state where it is sufficiently smaller or larger than the average pixel value M. By correcting the blur, an image with a clear boundary between materials can be obtained. Since quantitativeness is not lost even if the blur is corrected, an image showing the effective atomic number with high accuracy can be obtained. In addition, the image in FIG. 9 allows the length L1 of the acrylic 1b in the horizontal direction x to be accurately grasped.

[0058] Fig. 10 shows an effective atomic number image generated after imaging the object 1 with an energy discrimination X-ray device and performing image processing to enhance edges using the method described in Patent Document 1. Specifically, an image corresponding to a high-energy BIN was subjected to image processing to enhance edges using the method described in Patent Document 1, and similarly, images corresponding to a medium-energy BIN and a low-energy BIN were each subjected to image processing using the method described in Patent Document 1. After that, the three corrected images were combined to generate the effective atomic number image shown in Fig. 10.

[0059] Since multiple pixel values ​​n obtained from one target region 3 may be corrected with different correction patterns, there is a lack of consistency between the multiple pixel values ​​n. In the virtual height direction z', the pixel values ​​n after correction are inconsistent. As a result, quantitativeness is greatly lost, and the estimated atomic number is significantly different from reality. It is clear that with conventional image processing methods, the accuracy of the X-ray device as an energy analyzer is significantly reduced. [Explanation of symbols]

[0060] 1. Subject of the photo 1a Aluminum 1b Acrylic 2 pixels 3. Target Area 4. Image Processing Device 5 Acquisition mechanism 6 Processing mechanism 7 Memory mechanism 8. Correction pattern determination section 9. Correction Section 10 Generation part 11 Range setting section 12 Average calculation section 13 Comparison section 14 Selection section x Horizontal y Vertical direction z Height direction z' Virtual height direction n pixel value n' corrected pixel value n H Pixel values ​​corresponding to high energy BINs n M Pixel value corresponding to medium energy BIN n L Pixel value corresponding to low energy BIN ni (pixel value to be corrected) M average pixel value P1 correction range P2 Numeric range P3 Expanded Range d Tolerance d max Maximum L1 (acrylic) length S10 Correction pattern determination step S11 Range setting step S12 Averaging step S13 Comparison step S14 Selection step S20 Correction Step S30 Generation Step

Claims

1. An image processing method in which a plurality of pixel values ​​are acquired from a target area constituting a part of a photographed object, a composite of the plurality of pixel values ​​is determined as a pixel value of the target area, and an image of the photographed object is generated based on the plurality of target areas whose pixel values ​​have been determined, The plurality of pixel values ​​acquired from the target region are composed of one reference pixel value and other pixel values, a correction pattern determination step in which a plurality of correction patterns are set in advance, and the correction pattern to be applied to the target region is determined based on the reference pixel value; a correction step in which the reference pixel value and other pixel values ​​are corrected using the determined correction pattern; and generating an image of the object based on the corrected pixel values ​​of a plurality of the object regions.

2. The image processing method according to claim 1 , wherein the reference pixel value is made up of pixel values ​​obtained under the same conditions for a plurality of the target regions.

3. The correction pattern determination step includes:

3. The image processing method according to claim 1, further comprising: a range setting step for setting a predetermined range centered on the target area; an average calculation step for calculating an average pixel value within this range; a comparison step for comparing the pixel values ​​of the target area with the average pixel value; and a selection step for selecting the correction pattern in accordance with a result of the comparison.

4. The correction pattern is a no-correction pattern in which, when a pixel value of the target region is included within a predetermined numerical range that is set based on the average pixel value, the pixel value of the target region is adopted; a decreasing correction pattern that adopts an average value of a predetermined number of pixel values ​​on the smaller side among the pixel values ​​within the range set in the range setting step when the pixel value of the target region is smaller than the numerical range; 4. The image processing method according to claim 3, further comprising an increasing correction pattern for adopting an average value of a predetermined number of pixel values ​​on the larger side among the pixel values ​​within the range set in the range setting step when the pixel value of the target region is larger than the numerical range.

5. 1. An image processing device that acquires a plurality of pixel values ​​from a target area constituting a part of a photographed object, determines a pixel value of the target area by combining the plurality of pixel values, and generates an image of the photographed object based on the plurality of target areas whose pixel values ​​have been determined, The method includes: an acquisition mechanism that acquires one reference pixel value and other pixel values ​​for each of the target regions; a processing mechanism that processes the pixel values ​​acquired by the acquisition mechanism; and a storage mechanism that stores a plurality of correction patterns used by the processing mechanism, The processing mechanism comprises: a correction pattern determination unit that determines the correction pattern to be applied to the target region based on the reference pixel value; a correction unit that corrects the reference pixel value and other pixel values ​​using the determined correction pattern; and a generation unit that generates an image of the object to be photographed based on the corrected pixel values ​​of the plurality of object regions.

6. The image processing device according to claim 5 , wherein the reference pixel value is made up of pixel values ​​obtained under the same conditions for a plurality of the target regions.

7. The correction pattern determination unit 7. The image processing device according to claim 5, further comprising: a range setting unit that sets a predetermined range centered on the target area; an average calculation unit that calculates an average pixel value within this range; a comparison unit that compares the pixel values ​​of the target area with the average pixel value; and a selection unit that selects the correction pattern depending on a result of the comparison.

8. The correction pattern is a no-correction pattern in which, when a pixel value of the target region is included within a predetermined numerical range that is set based on the average pixel value, the pixel value of the target region is adopted; a decreasing correction pattern that adopts an average value of a predetermined number of pixel values ​​on the smaller side among the pixel values ​​in the range set by the range setting unit when the pixel value of the target region is smaller than the numerical range; 8. The image processing device according to claim 7, further comprising an increasing correction pattern for adopting an average value of a predetermined number of pixel values ​​on the larger side among the pixel values ​​within the range set by the range setting unit when the pixel value of the target region is larger than the numerical range.

Images