Industrial ct beam hardening correction method

Abstract

The present application relates to an industrial CT beam hardening correction method, comprising: taking the jth row of pixel points in each circumferential DR image to form a projection image; CT reconstruction is carried out on the projection image to obtain a tomographic image; a homogeneous material region in the detected sample is selected in the tomographic image; a function relationship formula corresponding to the minimum standard deviation of all corrected pixel values in the region is obtained, and the obtained function relationship formula is recorded as the optimal function relationship formula; the pixel value of each pixel point in each circumferential DR image is substituted into the optimal function relationship formula to obtain the corrected pixel value of each pixel point, and then the DR image after beam hardening correction is obtained; finally, the DR image after beam hardening correction is reconstructed to obtain the corrected CT image of the detected sample. The method does not need additional processing of contrast test blocks or test samples with known structures, the process is simple, especially for homogeneous material parts, the full-automatic interactive link can be realized, and the image quality and efficiency of industrial CT are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of CT detection, in particular to an industrial CT beam hardening correction method. BACKGROUND

[0002] With the development of detector technology, three-dimensional cone beam industrial CT technology based on area array detector has been rapidly popularized. Industrial CT uses an X-ray machine as a ray source, which emits wide-spectrum X-rays with continuous spectral distribution. When the wide-spectrum rays pass through the material, the lower-energy rays are preferentially absorbed, that is, the attenuation coefficient of low-energy rays is larger than that of high-energy rays, so that the energy spectrum of the rays transmitted through the object changes, which is manifested as an increase in the average energy of the X-rays passing through the material, which is the beam hardening phenomenon. The beam hardening phenomenon seriously affects the CT detection effect, and the direct manifestation is that the density uniform slice shows different brightness on the reconstructed image, and the pixel value distribution on the image shows "cup-shaped" or "striped" artifacts, which directly affects the defect judgment, detection or disease diagnosis.

[0003] At present, there are various hardening correction methods on the market. For example, the applicant's prior application for a Chinese invention patent with the patent number ZL202110402097.9 discloses a face array industrial CT beam hardening correction method. This method uses the attenuation data of X-rays passing through different thickness of material to calculate the energy spectrum of the current X-rays, and on the basis of accurate estimation of the energy spectrum, the wide-spectrum calculation result is mapped to a single energy spectrum to realize hardening correction. However, this method needs to manufacture homogeneous material test blocks of different thicknesses, and the same detection process is used to detect the test blocks and the detected objects, which is relatively cumbersome. For example, the method disclosed in the paper "X-ray spectrum estimation method based on computed tomography scanning data" is to use a known structure phantom to estimate the energy spectrum, that is, to perform CT scanning on the phantom with known structure and material, so as to obtain the corresponding relationship between the length of the rays passing through the material and the attenuation data, that is, the attenuation curve. Similarly, this method needs to manufacture homogeneous material test blocks to ensure that the structure is known, and the same detection process is used to detect the test blocks and the detected objects, which is also relatively cumbersome. Therefore, further improvement is needed for the prior art. SUMMARY

[0004] The technical problem to be solved by the present application is to provide an industrial CT beam hardening correction method which is simple to operate.

[0005] The technical solution adopted by the present application to solve the above technical problem is: an industrial CT beam hardening correction method, characterized by comprising the following steps:

[0006] Step 1, performing a circumferential DR scan on the sample to be detected to obtain n circumferential DR images of the sample to be detected; each circumferential DR image of the sample to be detected has the same size of r*s, and r and s are both positive numbers;

[0007] Step 2, taking the jth row of pixel point data in each circumferential DR image of the sample to be detected, and arranging all the pixel point data in sequence to form a projection image with a size of n*s;

[0008] Step 3, performing CT reconstruction on the projection image obtained in step 2 to obtain a tomographic image of the jth row of the sample to be detected;

[0009] Step 4, selecting at least a partial region S inside the sample to be detected in the tomographic image of the jth row of the sample to be detected, and the region S corresponds to a region where the homogeneous material inside the sample to be detected is located;

[0010] Step 5, setting a correction formula y m =f(x m ), x m is the pixel value of the mth pixel point in the region S, m∈{1,2,…M}, M is the total number of pixel points in the region S, y m is the corrected pixel value of x m , and f(.) is a function relationship corresponding to the correction formula; obtaining a function relationship formula corresponding to the minimum standard deviation of all the corrected pixel values in the region S, and recording the obtained function relationship formula as the optimal function relationship formula;

[0011] Step 6, substituting the pixel value of each pixel point in each circumferential DR image of the sample to be detected into the optimal function relationship formula to obtain the corrected pixel value of each pixel point in each circumferential DR image of the sample to be detected, and then obtaining n DR images of the sample to be detected after beam hardening correction;

[0012] Step 7, reconstructing the n DR images of the sample to be detected after beam hardening correction to obtain a corrected CT image of the sample to be detected.

[0013] In order to improve the correction effect, the value of j in step 2 is in the range of r / 2-k≤j≤r / 2+q, and r / 2-k≥1 and r / 2+q≤r, and k and q are both constants. The closer the value of j is to r / 2, the better the correction effect is.

[0014] Preferably, in step 2, the value of j is is a floor function.

[0015] Preferably, in step 3, FDK algorithm or iterative method is used to perform CT reconstruction on the projection image.

[0016] Preferably, the region S in step 4 is a line region or a surface region.

[0017] The function relationship satisfied by the correction formula in the application can be various, preferably, the correction formula in step 5 is y m =a+bx m +cx m 2 +dx m 3 , a, b, c and d are all constants, then the optimal function relationship in step 5 corresponds to the best a, b, c and d values.

[0018] Preferably, the value range of a is [-2 p-1 , 0], and p is the bit number of the industrial CT detector for collecting images.

[0019] Preferably, the value range of b is [0, 20].

[0020] Preferably, the value range of c is [-1x10 -3 , -1x10 -6 ].

[0021] Preferably, the value range of d is [1x10 -8 , 1x10 -11 ].

[0022] Compared with the prior art, the application has the advantages that: the method does not need additional processing of contrast test blocks or test samples with known structures, only needs to obtain the circumferential DR image of the scanned product, and the CT image calibration can be realized through image processing, the process is simple, the method avoids human intervention as much as possible, especially for homogeneous material parts, full automation and unmanned machine interaction can be realized, the industrial CT image quality and efficiency are improved, the method is suitable for most types of industrial CT products on the market, and has industry promotion value. BRIEF DESCRIPTION OF DRAWINGS

[0023] Figure 1 is a circumferential DR image schematic diagram of a first detected sample in the embodiment of the application;

[0024] Figure 2 is a circumferential DR image schematic diagram of a 100th detected sample in the embodiment of the application;

[0025] Figure 3 is a projection diagram schematic diagram in the embodiment of the application;

[0026] Figure 4 is a tomographic image schematic diagram in the embodiment of the application;

[0027] Figure 5 Fig. 1 is a schematic diagram of selecting a line region in a tomographic image in an embodiment of the present application;

[0028] Figure 6 Fig. 2 is a schematic diagram of selecting a surface region in a tomographic image in an embodiment of the present application;

[0029] Figure 7 Fig. 3 is a contrast diagram of CT images before and after correction in an embodiment of the present application. DETAILED DESCRIPTION

[0030] The present application will be further described in detail below with reference to the embodiments of the drawings.

[0031] The industrial CT beam hardening correction method in the embodiment includes the following steps:

[0032] Step 1, performing circumferential DR scanning on the detected sample to obtain n circumferential DR images of the detected sample; each circumferential DR image of the detected sample has the same size, which is r*s, and r and s are both positive numbers;

[0033] In the embodiment, a better scanning process needs to be set for the current industrial CT before circumferential DR scanning, so as to ensure that the subsequent image correction effect is better. The detected sample in the embodiment can be a homogeneous complex structure or can not be a homogeneous complex structure. If the detected sample is a homogeneous complex structure, the region in the following step 4 can be randomly selected. If the detected sample is not a homogeneous complex structure, the region in the following step 4 needs to be selected from the region of the homogeneous material when the region is selected.

[0034] As shown in Figure 1 and Figure 2 , r = 912 and s = 1156 in the embodiment.

[0035] Step 2, taking the jth row of pixel point data in each circumferential DR image of the detected sample, and sequentially grouping all the taken pixel point data to form a projection image, and the image size of the projection image is n*s.

[0036] The value range of j is r / 2-k≤j≤r / 2+q, and r / 2-k≥1 and r / 2+q≤r, and k and q are both constants; the k and q are basically close to 0, and the closer to 0, the better the correction method in the present application, and the present application is preferably, is a floor function, and in the embodiment, The projection image is shown in Figure 3 .

[0037] Step 3, performing CT reconstruction on the projection image obtained in step 2 to obtain the tomographic image of the jth row of the detected sample.

[0038] FDK algorithm or the iterative method is used to perform CT reconstruction on the projection image in the embodiment, and the reconstruction technology of the FDK algorithm or the iterative method is prior art, which will not be described here; the tomographic image of the jth row of the detected sample is as shown in FIG. 8; Figure 4

[0039] Step 4, selecting at least a partial region S inside the detected sample in the tomographic image of the jth row of the detected sample, the region S corresponding to a region where the homogeneous material inside the detected sample is located;

[0040] The region S in the embodiment is a line region or a surface region, as shown in FIG. 9, the region S is a line region, that is, the region S is composed of lines; as shown in FIG. 10, the region S is a surface region, that is, the region S is composed of square regions; Figure 5 Figure 6

[0041] Step 5, setting a correction formula y m =f(x n ), x m is a pixel value of an mth pixel point in the region S, m∈{1, 2, …M}, M is a total number of pixel points in the region S, y m is a pixel value after correction on x m ; f(.) is a function relationship corresponding to the correction formula; a function relationship formula corresponding to a minimum standard deviation of all corrected pixel values in the region S is obtained, and the obtained function relationship formula is recorded as an optimal function relationship formula;

[0042] The function relationship satisfied by the above correction formula can be confirmed according to experiments, and a function relationship to be initially selected can be selected, for example, a linear function, a quadratic function and a cubic function, etc. The correction formula in the embodiment is y m =a+bx m +cx m 2 +dx m 3 a, b, c and d are all constants, and then the optimal function relationship formula obtained in step 5 corresponds to the best a, b, c and d values;

[0043] The value range of a is [-2 p-1 , 0], p is the bit number of the industrial CT detector for collecting an image; the value range of b is [0, 20]; the value range of c is [-1×10 -3 , -1×10 -6 ]; and the value range of d is [1×10 -8 , 1×10 -11 ];

[0044] ​​​Step 6: Substitute the pixel value of each pixel in the circumferential DR image of each tested sample into the optimal function relationship to obtain the corrected pixel value of each pixel in the circumferential DR image of each tested sample, and then obtain n DR images of the tested sample after beam hardening correction.

[0045] Step 7: Reconstruct the DR images of the n tested samples after beam hardening correction to obtain the corrected CT images of the tested samples.

[0046] like Figure 7 As shown, Figure 7 (a) corresponds to the CT image of the sample before correction. Figure 7 (b) Corresponds to the CT image of the tested sample after correction. Before correction, the gray values ​​were inconsistent, and after correction, the gray values ​​were consistent.

[0047] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the technical principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. An industrial CT beam hardening correction method, characterized by The method comprises the following steps: Step 1: performing a circumferential DR scan on a sample to be detected to obtain n circumferential DR images of the sample to be detected; each circumferential DR image of the sample to be detected has the same size of r*s, and r and s are positive numbers; Step 2: taking jth row of pixel point data in each circumferential DR image of the sample to be detected, and sequentially combining all the taken pixel point data to form a projection image, wherein the projection image has an image size of n*s; Step 3: performing CT reconstruction on the projection image obtained in step 2 to obtain a tomographic image of the jth row of the sample to be detected; Step 4: selecting at least a partial region S in the tomographic image of the jth row of the sample to be detected, wherein the region S corresponds to a region of homogeneous material in the sample to be detected; Step 5, setting correction formula y m = f(x m ), x m is the pixel value of the mth pixel point in the region S, m ∈ {1, 2, …M}, M is the total number of pixel points in the region S, y m is the pixel value after correction on x m ; f(.) is a function relationship corresponding to the correction formula; a function relationship formula corresponding to a minimum standard deviation of all corrected pixel values in the region S is obtained, and the obtained function relationship formula is recorded as an optimal function relationship formula; Step 6: substituting the pixel value of each pixel point in each circumferential DR image of the sample to be detected into the optimal function relationship formula to obtain the corrected pixel value of each pixel point in each circumferential DR image of the sample to be detected, thereby obtaining n DR images of the sample to be detected after beam hardening correction; Step 7: performing reconstruction on the n DR images of the sample to be detected after beam hardening correction to obtain a corrected CT image of the sample to be detected.

2. The industrial CT beam hardening correction method of claim 1, wherein: In step 2, the value of j is in the range of r / 2-k≤j≤r / 2+q, and r / 2-k≥1 and r / 2+q≤r, wherein k and q are constants.

3. The industrial CT beam hardening correction method of claim 2, wherein: The step 2 in is a floor function.

4. The industrial CT beam hardening correction method of claim 1, wherein: In step 3, the FDK algorithm or the iterative method is used to perform CT reconstruction on the projection image.

5. The industrial CT beam hardening correction method of claim 1, wherein: In step 4, the region S is a line region or a surface region.

6. The industrial CT beam hardening correction method according to any one of claims 1 to 5, characterized in that: The correction formula in step 5 is y m = a + bx m + cx m 2 + dx m 2 a, b, c and d are all constants, then the optimal function relationship in step 5 corresponds to obtaining the best a, b, c and d values.

7. The industrial CT beam hardening correction method of claim 6, wherein: a is in the range of [-2 p-1 ,0], and p is the bit number of the image collected by the industrial CT detector.

8. The industrial CT beam hardening correction method of claim 6, wherein: The value of b is in the range of [0, 20].

9. The industrial CT beam hardening correction method of claim 6, wherein: The value range of c is [-1x10 -3 , 1x10 -6 ].

10. The industrial CT beam hardening correction method of claim 6, wherein: The value of d is in the range of [1 x 10 -8 , 1 x 10 -11 ].

Images