Industrial CT parameter calibration method in offset scanning mode based on marker assistance

Abstract

The invention belongs to the field of industrial CT parameter calibration, particularly discloses an industrial CT parameter calibration method in an offset scanning mode based on marker assistance, provides a real-time interaction calibration method based on marker assistance, and aims to solve the problem of inaccuracy of automatic detection markers by introducing a manual interaction means. The distance SDD between the focus and the detector, the distance SOD between the focus and the original point of the space coordinate system and the offset distance d of geometric calibration parameters with higher precision are obtained. Meanwhile, according to the method, manual idle time during scanning is utilized, acquisition of accurate parameters is completed under the condition that the overall scanning frequency is not increased, and therefore detection time is saved, and application of an offset scanning mode is supported.

Description

Technical Field

[0001] The present invention relates to the field of industrial CT parameter calibration, and in particular to an industrial CT parameter calibration method in a marker-assisted offset scanning mode. Background Art

[0002] Industrial CT plays a vital role in modern industry and scientific research, used for nondestructive testing and analysis of the internal structure of materials. Cone-beam CT offset scanning mode, an imaging technique that expands the reconstruction field of view, offers unique advantages when inspecting large parts. Compared to traditional concentric scanning modes, offset scanning mode can scan and reconstruct the entire part in a single pass, significantly improving inspection efficiency and application flexibility.

[0003] In practical applications, for units that frequently inspect various unknown objects, operators must continuously adjust the CT system's mechanical parameters based on practical experience to achieve optimal imaging conditions. Reconstruction using offset scanning modes requires high accuracy in geometric calibration parameters. Whenever mechanical parameter adjustments are required, the modified geometric parameters (such as SOD and SDD) must be recalibrated.

[0004] Geometric calibration can be performed in two ways: independent scanning and simultaneous scanning. Independent scanning involves using a dedicated calibration model to scan and calibrate independently to obtain geometric calibration parameters. The calibration model is then removed, the object to be measured is placed, and the object is scanned without changing the geometric parameters. Reconstruction is performed using the obtained geometric calibration parameters. Simultaneous scanning, on the other hand, involves adding markers to obtain unique geometric features, or utilizing the inherent geometric features of the object to be measured. After scanning, the geometric calibration parameters are calculated by isolating the geometric features and then used for reconstruction.

[0005] However, existing technologies have some limitations in practical applications. In the independent scanning method, the company's existing traditional double-ball calibration method cannot be used in bias mode. If a dedicated calibration model is used, it will need to be recalibrated every time the parameters are adjusted, which will waste time and shorten the service life of the source. At the same time, the scanning mode also faces many challenges in practical applications. For example, not all objects to be measured have the same geometric features, and the general calibration method cannot be applied; factors such as the imaging field of view and magnification ratio limit the coexistence of the calibration model and the object to be measured; even when the density of the calibration object is different from that of the object to be measured, it will affect the imaging effect and lead to missed detection. Summary of the Invention

[0006] In order to solve the above technical problems, the present invention provides an industrial CT parameter calibration method in a marker-assisted offset scanning mode.

[0007] To achieve the above object, the present invention is implemented according to the following technical solutions:

[0008] The method for calibrating parameters of industrial CT in a marker-assisted offset scanning mode includes the following steps:

[0009] S1. Obtain a scanned image of the object to be measured and determine the magnitude of the parameters to be calibrated. The parameters to be calibrated include the distance SDD between the focus and the detector, the distance SOD between the focus and the origin of the spatial coordinate system, and the offset distance d.

[0010] S2. Select a solid sphere made of a single material as a marker for the scanned image of the object to be tested, and fix the object to be tested and the marker on the stage of the industrial CT.

[0011] S3, the industrial CT starts a bias scan to obtain a scanned image, determines the structure of the marker in the scanned image, and selects the marker in the scanned image;

[0012] S4. After the scanning is completed, the movement trajectory of the markers is obtained, and the markers in the selected scanned image are grouped;

[0013] S5. Perform ellipse fitting on the projection center coordinates of each group of markers and calculate the projection center of each group of markers. Then, compare the obtained projection center coordinates with the projection center coordinates of the actual acquired scan image and fit them to determine the optimal values of the distance SDD between the focus and the detector, the distance SOD between the focus and the origin of the spatial coordinate system, and the offset distance d.

[0014] Furthermore, in step S2, the solid spheres made of a single material include wooden balls, metal balls and plastic balls.

[0015] Furthermore, the density of the marker satisfies the following conditions:

[0016] 0.8*ρ R ≤ρ B ≤ρ R ;

[0017] Where: B represents the density of the marker; R represents the region of interest of the analyte; ρ R Indicates the density of the region of interest of the object to be tested;

[0018] If R is composed of materials with different densities, let the minimum density of the material to be tested in the region of interest be ρ Rmin , then the density of the marker satisfies the following conditions:

[0019] 0.8*ρ Rmin ≤ρ B ≤ρR min .

[0020] Furthermore, the diameter of the marker on the scanned image is not less than 8 pixels.

[0021] Furthermore, in step S2, when the object to be tested and the marker are fixed, for the same plane P perpendicular to the rotation axis, there is at most one marker; for different planes with markers, i , suppose two adjacent planes P i and P i+1 The spacing d i , then the spacing is more than 3 times the diameter of the marker d i ≥3D, total number of markers is 4-8.

[0022] Furthermore, in step S3, the strategy for selecting the marker in the scanned image is:

[0023] If the structure of the object to be tested is circular, manually delete the non-marked circular shape in the scanned image;

[0024] For markers that are visible to the naked eye but not successfully found in the scanned image, click inside the circle and select it using the region growing method or directly draw a circular marker with the mouse to select it;

[0025] Markers that are partially obscured by high-density areas in the scanned image are directly ignored;

[0026] For new markers appearing from the border of the scanned image, select them directly with the mouse;

[0027] In the case of automatic identification errors, delete them directly or make manual corrections.

[0028] Furthermore, the step S5 specifically includes:

[0029] S51, the projection center coordinates P(u ip , v ip ) to perform ellipse fitting and obtain the ellipse equation Where a and b are the major and minor axes of the ellipse respectively, O(x o ,y o ) are the coordinates of the center of the ellipse;

[0030] S52. Assume that there are m markers in total, n images are collected in total, and the total number of coordinates of the center of the k-th group of markers is m. k , assuming that the center coordinates of the kth group of markers on the first image are A k , then the vector The angle is set to δ k ; Let the spatial polar coordinates of the kth marker be If the offset distance is d, then:

[0031]

[0032] Among them: r, θ is an auxiliary fitting parameter, and the value range of r is 0.8-1.2; The value range is -180°-180°; the value range of θ is -90°-90°;

[0033] S53, set the detector offset vector D′(dx, o, dz) of the industrial CT, convert the polar coordinates of the k-th marker into the spatial coordinates (x k ,y k , z k ), the projection center coordinates Q(u iq , v iq ):

[0034]

[0035] S54. After calculating the circle center Q, compare it with the projection circle center coordinates P of the actual acquired scan image and perform fitting:

[0036]

[0037] When the cost is minimum, it is the optimal value of the distance SDD between the focus and the detector, the distance SOD between the focus and the origin of the spatial coordinate system, and the offset distance d.

[0038] Compared to existing technologies, this paper proposes a marker-assisted, real-time interactive calibration method. By introducing human interaction, this method corrects the inaccuracy of automatic marker detection and obtains more accurate geometric calibration parameters. Furthermore, this method utilizes manual idle time during scanning to obtain precise parameters without increasing the overall number of scans, thus saving detection time and supporting the use of offset scanning modes. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 Flowchart of the present invention.

[0040] Figure 2 To obtain the marker motion trajectory diagram after the scan is completed. DETAILED DESCRIPTION

[0041] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the embodiments. The specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0042] Reference Figure 1This embodiment exemplarily demonstrates a method for calibrating industrial CT parameters in a marker-assisted offset scanning mode. The specific steps are as follows:

[0043] S1. Obtain a scanned image of the object to be measured and determine the magnitude of the parameters to be calibrated. The parameters to be calibrated include the distance SDD between the focus and the detector, the distance SOD between the focus and the origin of the spatial coordinate system, and the offset distance d.

[0044] S2. Select a solid sphere made of a single material as a marker for the scanned image of the object to be tested, and fix the object to be tested and the marker on the stage of the industrial CT.

[0045] S3, the industrial CT starts a bias scan to obtain a scanned image, determines the structure of the marker in the scanned image, and selects the marker in the scanned image;

[0046] S4. After the scanning is completed, the movement trajectory of the markers is obtained, and the markers in the selected scanned image are grouped;

[0047] S5. Perform ellipse fitting on the projection center coordinates of each group of markers and calculate the projection center of each group of markers. Then, compare the obtained projection center coordinates with the projection center coordinates of the actual acquired scan image and fit them to determine the optimal values of the distance SDD between the focus and the detector, the distance SOD between the focus and the origin of the spatial coordinate system, and the offset distance d.

[0048] The above steps are described in detail later, specifically:

[0049] 1. Preparation before scanning

[0050] The preparation work before scanning is mainly divided into the following steps: determine the parameters to be calibrated and their magnitudes according to the scanning requirements, select appropriate markers, and fix the object to be measured and the markers.

[0051] Determine the magnitude of the parameter to be calibrated

[0052] When testing objects from different sources, the size and material of the objects often vary greatly; there will also be different parameters depending on the area of interest and the purpose of the test.

[0053] When inspecting different objects continuously, it's often necessary to change the distance between the focus and the detector (SDD) and the distance between the focus and the origin of the spatial coordinate system (SOD) to obtain the desired scanned image. Large objects can also require moving the detector or the rotation axis, requiring offset scanning to acquire images. Other geometric parameters often remain unchanged.

[0054] The parameters SDD, SOD, and offset distance d can all be obtained by measuring or calculating the mechanical parameters. These values can be used as initial values for subsequent fitting to speed up the fitting calculation.

[0055] Selection markers

[0056] In this embodiment, all candidate markers are solid spheres made of a single material, such as wooden balls, metal balls, plastic balls, etc. For the markers, the density ρ B (material), diameter D (size) two parameters.

[0057] For the marker density ρ B :Assume that the region of interest of the object to be tested is R and its density is ρ R In order to reduce the additional impact of the marker on the imaging effect of R, the marker should preferably meet the following requirements:

[0058] 0.8*ρ R ≤ρ B ≤ρ R ;

[0059] If R is composed of materials of various densities, let the minimum density of the material to be tested (non-air) in the region of interest be ρ Rmin , then the marker should preferably meet the following requirements:

[0060] 0.8*ρ Rmin ≤ρ B ≤ρ Rmin ;

[0061] However, as long as the markers can be identified by the naked eye in the scanned image, this method can be used normally. The selection of marker density is mainly to reduce the imaging impact on the area of interest. Therefore, the operator can also avoid interference with the imaging by placing markers in appropriate positions.

[0062] For the sphere diameter D:

[0063] The diameter of the sphere needs to be determined by the operator based on the scanning requirements, and it only needs to meet the conditions that it can be easily identified by the naked eye, such as the diameter of the marker on the scanned image is not less than 8 pixels.

[0064] Fix the analyte and marker

[0065] When the object to be tested and the marker are fixed, there is at most one marker on the same plane P perpendicular to the axis of rotation; for different planes with markers, i , suppose two adjacent planes P i and P i+1 The spacing d i , the spacing needs to be more than 3 times the diameter of the marker. i, to facilitate manual identification during interactive operations. The total number of markers should be 4-8, but not less than 3.

[0066] 2. Scan related

[0067] Initialize the marker structure

[0068] After turning on the radiation source and successfully obtaining a preview image, you need to select a marker template (such as material, size, etc.) in the software. This method will automatically calculate the shape of the possible markers on the preview image based on this.

[0069] The markers used in this method are all spheres, so the images displayed on the preview image are all circles. There are many methods for detecting circles. For example, directly using HoughCircles in OpenCV can obtain the coordinates and radius of possible circles.

[0070] Since the structure of the object to be tested may be circular, the operator needs to manually delete the circles that are not markers. The following situations should be noted:

[0071] Due to the offset scanning, not all markers may be displayed on the initial preview image. These markers that are temporarily out of the field of view will be interactively processed during real-time scanning.

[0072] For markers that are visible to the naked eye but not found by automatic detection, the operator can manually select them, such as by clicking inside the circle and selecting them through region growing, or by directly drawing a circular marker with the mouse.

[0073] Markers that are partially obscured by high-density areas can be ignored.

[0074] Start Scan

[0075] After starting the scan, this method will automatically identify the position of the circular marker in the next frame based on the circular marker in the previous frame. Assuming that the center coordinates of the circular marker found in the previous frame are O1, the search distance p can be set so that the center position O2 of the marker in the next frame satisfies:

[0076] |O2-O1|<p;

[0077] Also note the following:

[0078] New markers that appear from the image boundary can be directly selected using the mouse.

[0079] In the case of automatic identification errors, you can directly delete them or make manual corrections.

[0080] Marker grouping

[0081] After the scan is completed, the marker motion trajectory is obtained as follows Figure 2 As shown. Figure 2 It's clear that each marker's trajectory follows a roughly elliptical path. Markers that weren't correctly grouped together during automatic recognition can be manually regrouped. The human eye can usually identify markers that belong to the same elliptical path. If the operator can't determine this, they can simply delete the marker.

[0082] 3. Parameter calculation

[0083] The projection center coordinates P(u ip , v ip ) to perform ellipse fitting and obtain the ellipse equation Where a and b are the major and minor axes of the ellipse respectively, O(x o ,y o ) are the coordinates of the center of the ellipse;

[0084] Assume there are m markers in total, n images are collected, and the total number of center coordinates of the kth group of markers is m k , assuming that the center coordinates of the kth group of markers on the first image are A i , then the vector The angle is set to δ k ; Let the spatial polar coordinates of the kth marker be If the offset distance is d, then:

[0085]

[0086] Among them: r, θ is an auxiliary fitting parameter, the value range of r is 0.8-1.2, and the initial value of r in this embodiment is set to 1; The value range is -180°-180°; the value range of θ is -90°-90°;

[0087] Assume that the detector offset vector of industrial CT is D′(dx, 0, dz), and the polar coordinates of the kth marker are converted into spatial coordinates (x k ,y k , z k ), the projection center coordinates Q(u iq , v iq ):

[0088]

[0089] After calculating the circle center Q, compare it with the projection center coordinates P of the actual scanned image and perform fitting:

[0090]

[0091] When the cost is minimum, it is the optimal value of the distance SDD between the focus and the detector, the distance SOD between the focus and the origin of the spatial coordinate system, and the offset distance d.

[0092] The technical solution of the present invention is not limited to the above-mentioned specific embodiments. Any technical variations made according to the technical solution of the present invention fall within the protection scope of the present invention.

Claims

1. The industrial CT parameter calibration method in the marker-assisted offset scanning mode is characterized by: The following steps are involved: S1. Obtain a scanned image of the object to be measured and determine the magnitude of the parameters to be calibrated. The parameters to be calibrated include the distance SDD between the focus and the detector, the distance SOD between the focus and the origin of the spatial coordinate system, and the offset distance d. S2. Select a solid sphere made of a single material as a marker for the scanned image of the object to be tested, and fix the object to be tested and the marker on the stage of the industrial CT. S3, the industrial CT starts a bias scan to obtain a scanned image, determines the structure of the marker in the scanned image, and selects the marker in the scanned image; S4. After the scanning is completed, the movement trajectory of the markers is obtained, and the markers in the selected scanned image are grouped; S5. Perform ellipse fitting on the projection center coordinates of each group of markers and calculate the projection center of each group of markers. Then, compare the obtained projection center coordinates with the projection center coordinates of the actual acquired scan image and fit them to determine the optimal values of the distance SDD between the focus and the detector, the distance SOD between the focus and the origin of the spatial coordinate system, and the offset distance d.

2. The industrial CT parameter calibration method in the marker-assisted offset scanning mode according to claim 1 is characterized in that: In step S2, the single-material solid sphere includes a wooden ball, a metal ball, and a plastic ball.

3. The industrial CT parameter calibration method in the marker-assisted offset scanning mode according to claim 2, characterized in that: The density of the marker satisfies the following conditions: 0.8*p R ≤ρ B ≤ρ R ; Where: B represents the density of the marker; R represents the region of interest of the analyte; ρ R Indicates the density of the region of interest of the object to be tested; If R is composed of materials with different densities, let the minimum density of the material to be tested in the region of interest be ρ Rmin , then the density of the marker satisfies the following conditions: 0.8pR min ≤ρ B ≤ρ Rmin 。 4. The industrial CT parameter calibration method in the marker-assisted offset scanning mode according to claim 2, characterized in that: The diameter of the marker on the scanned image is not less than 8 pixels.

5. The industrial CT parameter calibration method in the marker-assisted offset scanning mode according to claim 1, characterized in that: In step S2, when the object to be tested and the marker are fixed, there is at most one marker on the same plane P perpendicular to the rotation axis; for different planes with markers, i , suppose two adjacent planes P i and P i+1 The spacing d i , then the spacing is more than 3 times the diameter of the marker d i ≥3D, total number of markers is 4-8.

6. The industrial CT parameter calibration method in the marker-assisted offset scanning mode according to claim 1, characterized in that: In step S3, the strategy for selecting the markers in the scanned image is: If the structure of the object to be tested is circular, manually delete the non-marked circular shape in the scanned image; For markers that are visible to the naked eye but not successfully found in the scanned image, click inside the circle and select it using the region growing method or directly draw a circular marker with the mouse to select it; Markers that are partially obscured by high-density areas in the scanned image are directly ignored; For new markers appearing from the border of the scanned image, select them directly with the mouse; In the case of automatic identification errors, delete them directly or make manual corrections.

7. The industrial CT parameter calibration method in a marker-assisted offset scanning mode according to claim 1, characterized in that: The step S5 specifically includes: S51, the projection center coordinates P(u ip , v ip ) to perform ellipse fitting and obtain the ellipse equation Where a and b are the major and minor axes of the ellipse respectively, O(x o ,y o ) are the coordinates of the center of the ellipse; S52. Assume that there are m markers in total, n images are collected in total, and the total number of coordinates of the center of the k-th group of markers is m. k , assuming that the center coordinates of the kth group of markers on the first image are A k , then the vector The angle is set to δ k ; Let the spatial polar coordinates of the kth marker be If the offset distance is d, then: Among them: r, θ is an auxiliary fitting parameter, and the value range of r is 0.8-1.2; The value range is -180°-180°; the value range of θ is -90°-90°; S53, set the detector offset vector D′(dx, 0, dz) of the industrial CT, and convert the polar coordinates of the kth marker into spatial coordinates (x k ,y k , z k ), the projection center coordinates Q(u iq , v iq ): S54. After calculating the circle center Q, compare it with the projection circle center coordinates P of the actual acquired scan image and perform fitting: When the cost is minimum, it is the optimal value of the distance SDD between the focus and the detector, the distance SOD between the focus and the origin of the spatial coordinate system, and the offset distance d.

Images