Industrial ct multi-dimension size measurement method for complex workpiece

Abstract

This invention discloses a multi-dimensional dimensional measurement method for complex workpieces using industrial CT. First, the method automatically calculates motion parameters based on the CT scan section. Then, a multi-dimensional platform executes attitude adjustment commands according to these parameters, adjusting the workpiece to the posture required for CT scanning, thus improving the positioning accuracy and efficiency of industrial CT scanning. Next, the reconstructed image from the CT scan data is vectorized and its features extracted. Finally, it is matched with a CAD model, and error analysis and calculation are performed on both. This invention can directly calculate the spatial coordinate transformation parameters of complex workpieces, ensuring consistency in workpiece inspection while reducing noise interference and improving the accuracy of error analysis. It solves the problems of complex workpiece scanning coordinate calculation, low scanning positioning efficiency, and poor accuracy of scanning error analysis in industrial CT inspection.

Description

Technical Field

[0001] This invention relates to the field of industrial CT measurement technology, and more specifically, to a method for measuring the multi-dimensional dimensions of complex workpieces using industrial CT. Background Technology

[0002] Industrial CT inspection relies on the attenuation and absorption characteristics of radiation in the inspected object. The ability of a substance to absorb radiation is related to its properties. Therefore, by utilizing the attenuation patterns and distribution of X-rays or gamma rays with a certain energy and intensity emitted by radioactive nuclides or other radiation sources within the inspected object, it is possible to obtain detailed information about the object's interior using detector arrays. Finally, computer information processing and image reconstruction techniques are used to display this information as an image. In industrial CT inspection of complex workpieces, multiple cross-sections need to be inspected. Therefore, multiple clamping operations or multi-axis motion devices are required to adjust the workpiece's spatial posture. During inspection, operators often need to manually adjust the workpiece's inspection posture and perform multiple trial adjustments to the inspected area to basically locate structural defects. The manual operation of the inspection process is cumbersome, time-consuming, and inefficient, and the inspection results may be distorted, resulting in low reliability. Furthermore, the complexity of the workpiece leads to poor accuracy in industrial CT inspection, thus affecting the judgment of the dimensional accuracy of complex workpieces.

[0003] CN112903730A discloses a multi-dimensional attitude adjustment device including a tilting unit, a rotating unit, a translating unit, and a clamping unit. The tilting unit includes a mounting frame, a first shaft, and a first motor. A support block is fixed on the first shaft and is laterally rotatably connected to the mounting frame. The first motor is connected to the first shaft to drive its rotation. The rotating unit includes a connecting block, a second shaft, and a second motor. The connecting block is movably connected to the support block. The second shaft is vertically arranged and its top end is fixedly connected to the connecting block. The second motor is connected to the bottom end of the second shaft to drive its rotation. The second motor is mounted on the first shaft. The translating unit is connected to the mounting frame to drive its lateral movement. The clamping unit is mounted on the top end of the connecting block and is used to clamp the workpiece. This patent can adjust the test section of the workpiece to the required position through the interaction of the rotating unit, tilting unit, and translating unit, which can conveniently realize the switching of multiple detection sections of the workpiece, avoid repeated clamping of the workpiece, and thus improve detection efficiency. However, it does not disclose how to quickly adjust the attitude for scanning. In addition, CN 112669460B reconstructs a 3D mesh image from the CT image data of the workpiece under test and registers it with the CAD triangular mesh image, thereby quickly obtaining the deviation of the workpiece under test. However, due to the interference of image noise during the reconstruction of the 3D mesh, the vertex positions of the 3D mesh contain noise errors, which leads to a large error in the workpiece deviation measured when comparing the slice image with the CAD, especially for high-precision parts, the test error is even greater. Summary of the Invention

[0004] The main technical problem to be solved by this invention is to address the shortcomings of existing industrial CT inspection technologies, such as complex calculation of scanning coordinates for complex workpieces, low scanning positioning efficiency, and poor accuracy of scanning error analysis. This invention provides a method for multi-dimensional dimensional measurement of complex workpieces using industrial CT.

[0005] The objective of this invention is achieved through the following technical solution:

[0006] A method for multi-dimensional dimensional measurement of complex workpieces using industrial CT, comprising the following steps:

[0007] S1. Based on a multi-dimensional platform, a scanning plane is preset with three-point spatial coordinates, posture adjustment parameters are calculated, and CT scans are performed on complex workpieces to obtain industrial two-dimensional CT slice images.

[0008] S2. Perform precise vectorization on industrial CT images based on the error-compensated spatial moment algorithm, and then extract features;

[0009] S21. Preprocess industrial CT images;

[0010] S22. Extract image edges, search edge point sets to construct contour linked list sequence, perform probabilistic circle detection on non-empty linked lists, and accurately vectorize circles that meet the whole circle constraint; then insert them into the circle linked list.

[0011] S23. Use the set intersection method to fit lines to other contour lists, and determine whether short line segments are adjacent. If they are adjacent, merge them into long line segments.

[0012] S24. Use the perpendicular bisector method to find the arc and determine whether it forms an arc; perform precise vectorization of the contour points that form an arc and insert them into the arc linked list; calculate the least squares error of non-arc contour points and perform precise vectorization of the line and insert them into the line linked list.

[0013] S3. Edit the vectorized graphic and then save it as a DXF file;

[0014] S4. Match the vectorized graphic primitives of CT with the CAD design model and calculate and analyze the error.

[0015] Furthermore, the coordinates of the three points M1, M2, and M3 on the scanning plane in step S1 are:

[0016] M1(X1, Y1, Z1)

[0017] M2(X2, Y2, Z2)

[0018] M3(X3, Y3, Z3)

[0019] Its general equation is: Ax + By + Cz + D = 0, where:

[0020] A=(y2-y1)×(z3-z2)-(z2-z1)×(y3-y2)

[0021] B = (z2 - z1) × (x3 - x2) - (z3 - z2) × (x2 - x1)

[0022] C=(x2-x1)×(y3-y2)-(y2-y1)×(x3-x2)

[0023] D=x1×((z2-z1)×(y3-y2)-(y2-y1)×(z3-z2))+y1×((z3-z2)×(x2-x1)-(z2-z1)×(x3-x2))+z1×((y2-y1)×(x3-x2)-(x2-x1)×(y3-y2)).

[0024] Furthermore, the attitude adjustment in step S1 includes rotating the detection surface around the Z-axis to be parallel to the X-axis and rotating the detection surface around the X-axis to be parallel to the XY plane.

[0025] Furthermore, when the detection surface is rotated around the Z-axis to be parallel to the X-axis, let the angle when the detection plane is rotated around the Z-axis to be parallel to the X-axis be θ′. After rotation, the coordinates of M1′, M2′, and M3′ are:

[0026] M1′(X1′、Y1′、Z1)

[0027] M2′(X2′、Y2′、Z2)

[0028] M3′(X3′、Y3′、Z3)

[0029] In the formula:

[0030]

[0031] According to the general form of the plane equation: Ax + By + Cz + D = 0, when the plane is parallel to the X-axis, A = 0.

[0032]

[0033] The value of the rotation angle θ' around the X-axis is obtained.

[0034] Furthermore, when the detection surface is rotated around the X-axis to be parallel to the XY plane, let the angle be θ” when the detection surface is rotated around the X-axis to be parallel to the XY plane. The coordinates of M1″, M2″, and M3″ after the rotation are:

[0035] M1″(X1′、Y1″、Z1′)

[0036] M2″(X2′、Y2″、Z2′)

[0037] M3″(X3′、Y3″、Z3′)

[0038] In the formula:

[0039]

[0040] According to the general form of the plane equation: Ax + By + Cz + D = 0, when the plane is parallel to the XY plane, A = 0 and B = 0.

[0041] A=(Y2”-Y1”)×(Z3’-Z2’)-(Z2’-Z1’)×(Y3”-Y2”)=0

[0042] B=(Z2'-Z1')×(X3'-X2')-(Z3'-Z2')×(X2'-X1')=0

[0043] The value of the rotation angle θ around the X-axis is obtained.

[0044] Furthermore, the preprocessing described in step S21 includes image enhancement, removal of image artifacts and noise.

[0045] Furthermore, the editing process described in step S3 includes connecting adjacent line segments, deleting redundant line segments, and isolating points.

[0046] Furthermore, in step S4, the idea of ​​minimum bounding box is used to introduce centroid, and a set of primitive features is selected through human-computer interaction to achieve registration between models.

[0047] Furthermore, step S4 adds the function of fitting short straight lines into long straight lines and fitting short straight lines into arcs as a reference for registration.

[0048] Furthermore, step S4 specifically includes:

[0049] S41.CT vectorized graphic primitive pairing CAD design model primitive information;

[0050] S42. Discretize the vectorized primitives of the CT image;

[0051] S43. Calculate the normal error to the CAD model.

[0052] Compared with existing technologies, the beneficial effects are:

[0053] This invention can automatically calculate motion parameters based on CT scan sections. A multi-dimensional platform executes attitude adjustment commands according to these parameters, adjusting the workpiece to the posture required for CT scanning, thus improving the positioning accuracy and efficiency of industrial CT scans. Simultaneously, this invention reduces noise interference and improves the accuracy of error analysis by vectorizing the reconstructed images from CT scan data, including linear and circular vectorization. Attached Figure Description

[0054] Figure 1 This is a schematic diagram of a multi-dimensional platform.

[0055] Figure 2 This is a vectorized flowchart.

[0056] Figure 3 To accurately vectorize the image.

[0057] Figure 4 The result of registering the vectorized model with the CAD model.

[0058] Figure 5 This is a comparison diagram showing the error between the vectorized model and the CAD model. Detailed Implementation

[0059] The following examples further explain and clarify the invention, but the specific examples do not limit the invention in any way.

[0060] Example 1

[0061] This embodiment provides a method for multi-dimensional dimensional measurement of complex workpieces using industrial CT, the steps of which include:

[0062] S1. As Figure 1 Based on a multi-dimensional platform, the scanning plane is preset with three-point spatial coordinates, the posture adjustment parameters are calculated, and CT scans are performed on complex workpieces to obtain industrial two-dimensional CT slice images.

[0063] S2. Perform precise vectorization of industrial CT images based on an error-compensated spatial moment algorithm, such as... Figure 2 Then extract features;

[0064] S21. Preprocess industrial CT images, including image enhancement, removal of image artifacts and noise;

[0065] S22. Extract image edges, search edge point sets to construct contour linked list sequence, perform probabilistic circle detection on non-empty linked lists, and accurately vectorize circles that meet the whole circle constraint; then insert them into the circle linked list.

[0066] S23. Use the set intersection method to fit lines to other contour lists, and determine whether short line segments are adjacent. If they are adjacent, merge them into long line segments.

[0067] S24. Use the perpendicular bisector method to find the arc and determine whether it forms an arc; perform precise vectorization of the contour points that form an arc and insert them into the arc linked list; calculate the least squares error of non-arc contour points and perform precise vectorization of the line and insert them into the line linked list.

[0068] S3. Editing and processing of vectorized graphics, such as... Figure 3 This includes connecting adjacent line segments, deleting redundant line segments and isolated points, and saving the vectorized graphics as a DXF format file.

[0069] S4. Select primitive features and match them with the model, then calculate and analyze the error.

[0070] Example 2

[0071] This embodiment provides a calculation method for adjusting the attitude parameters of a multi-dimensional platform, using the longer side of the multi-dimensional platform as the X-axis, the shorter side as the Y-axis, the vertical direction as the Z-axis, and the center of the multi-dimensional platform as the zero point. The calculation steps include:

[0072] S11. Set the coordinates of three points M1, M2, and M3 within the detection surface as follows:

[0073] M1(X1, Y1, Z1)

[0074] M2(X2, Y2, Z2)

[0075] M3(X3, Y3, Z3)

[0076] The detection surface function is calculated, and the determinant of the plane equation is as follows:

[0077] x-x1 y-y1 z-z1

[0078] x2-x1 y2-y1 z2-z1=0

[0079] x3-x2 y3-y2 z3-z2

[0080] have:

[0081] (x-x1)×(y2-y1)×(z3-z2)+(y-y1)×(z2-z1)×(x3-x2)+(z-z1)×(x2-x1)×(y3-y2)- (z-z1)×(y2-y1)×(x3-x2)-(z2-z1)×(y3-y2)×(x-x1)-(z3-z2)×(y-y1)×(x2-x1)=0

[0082] X×(y2-y1)×(z3-z2)-x1×(y2-y1)×(z3-z2)+Y×(z2-z1)×(x3-x2)-y1×(z2-z1)×(x3-x2)+Z×(x2-x1)×(y3-y2)-z1×(x2-x1)×(y3-y2)- Z×(y2-y1)×(x3-x2)+z1×(y2-y1)×(x3-x2)-X×(z2-z1)×(y3-y2)+x1×(z2-z1)×(y3-y2)-Y×(z3-z2)×(x2-x1)+y1×(z3-z2)×(x2-x1)=0

[0083] X×((y2-y1)×(z3-z2)-(z2-z1)×(y3-y2))+Y×((z2-z1)×(x3-x2)-(z3-z2)×(x2-x1))+Z×((x2-x1)×(y3-y2)-(y2-y1)×(x3-x2))+x1 ×((z2-z1)×(y3-y2)-(y2-y1)×(z3-z2))+y1×((z3-z2)×(x2-x1)-(z2-z1)×(x3-x2))+z1×((y2-y1)×(x3-x2)-(x2-x1)×(y3-y2))=0

[0084] The general form of the plane equation is Ax + By + Cz + D = 0, which gives us...

[0085] A=(y2-y1)×(z3-z2)-(z2-z1)×(y3-y2)

[0086] B = (z2 - z1) × (x3 - x2) - (z3 - z2) × (x2 - x1)

[0087] C=(x2-x1)×(y3-y2)-(y2-y1)×(x3-x2)

[0088] D=x1×((z2-z1)×(y3-y2)-(y2-y1)×(z3-z2))+y1×((z3-z2)×(x2-x1)-(z2-z1)×(x3-x2))+z1×((y2-y1)×(x3-x2)-(x2-x1)×(y3-y2)).

[0089] S12. Rotate the detection surface around the Z-axis to a position parallel to the X-axis;

[0090] Let the angle be θ′ when the detection plane rotates around the Z-axis to an angle parallel to the X-axis. After rotation, the coordinates of M1′, M2′, and M3′ are:

[0091] M1′(X1′、Y1′、Z1)

[0092] M2′(X2′、Y2′、Z2)

[0093] M3′(X3′、Y3′、Z3)

[0094] In the formula:

[0095]

[0096] According to the general form of the plane equation: Ax + By + Cz + D = 0, when the plane is parallel to the X-axis, A = 0.

[0097]

[0098] Based on the above formula, the value of the rotation angle θ' around the X-axis can be calculated.

[0099] S13. Rotate the detection surface around the X-axis until it is parallel to the XY plane.

[0100] Suppose that when the detection plane rotates around the X-axis to be parallel to the XY plane, the angle is θ”. The coordinates of M1″, M2″, and M3″ after the rotation are:

[0101] M1″(X1′、Y1″、Z1′)

[0102] M2″(X2′、Y2″、Z2′)

[0103] M3″(X3′、Y3″、Z3′)

[0104] In the formula:

[0105]

[0106] According to the general form of the plane equation: Ax + By + Cz + D = 0, when the plane is parallel to the XY plane, A = 0 and B = 0.

[0107] A=(Y2”-Y1”)×(Z3’-Z2’)-(Z2’-Z1’)×(Y3”-Y2”)=0

[0108] B=(Z2'-Z1')×(X3'-X2')-(Z3'-Z2')×(X2'-X1')=0

[0109] Substituting the values ​​into the coordinate formulas above, the rotation angle θ” around the X-axis can be calculated.

[0110] Example 3

[0111] This embodiment provides a method for multi-dimensional dimensional measurement of complex workpieces using industrial CT, the steps of which include:

[0112] S1. As Figure 1Based on a multi-dimensional platform, the scanning plane is preset with three-point spatial coordinates, the posture adjustment parameters are calculated, and CT scans are performed on complex workpieces to obtain industrial two-dimensional CT slice images.

[0113] S2. Perform precise vectorization of industrial CT images based on an error-compensated spatial moment algorithm, such as... Figure 2 Then extract features;

[0114] S21. Preprocess industrial CT images, including image enhancement, removal of image artifacts and noise;

[0115] S22. Extract image edges, search edge point sets to construct contour linked list sequence, perform probabilistic circle detection on non-empty linked lists, and accurately vectorize circles that meet the whole circle constraint; then insert them into the circle linked list.

[0116] S23. Use the set intersection method to fit lines to other contour lists, and determine whether short line segments are adjacent. If they are adjacent, merge them into long line segments.

[0117] S24. Use the perpendicular bisector method to find the arc and determine whether it forms an arc; perform precise vectorization of the contour points that form an arc and insert them into the arc linked list; calculate the least squares error of non-arc contour points and perform precise vectorization of the line and insert them into the line linked list.

[0118] S3. Editing and processing of vectorized graphics, including connecting similar line segments, deleting redundant line segments and isolated points, and saving the vectorized graphics as a DXF format file.

[0119] S4. Select primitive features and match them with the model, then calculate and analyze the error.

[0120] S41. Construct a CAD design model based on the original workpiece data;

[0121] S42. The CAD design model primitive information paired with the vectorized graphic primitives of CT is introduced with a centroid using the concept of minimum bounding box, and a set of primitive features is selected through human-computer interaction to achieve registration between models.

[0122] S42. Discretize the vectorized primitives of the CT image into 48 points;

[0123] S43. Calculate the normal error to the CAD model.

[0124] Example 4

[0125] This embodiment provides a method for multi-dimensional dimensional measurement of complex workpieces using industrial CT, the steps of which include:

[0126] S1. As Figure 1Based on a multi-dimensional platform, the scanning plane is preset with three-point spatial coordinates, the posture adjustment parameters are calculated, and CT scans are performed on complex workpieces to obtain industrial two-dimensional CT slice images.

[0127] S2. Perform precise vectorization of industrial CT images based on an error-compensated spatial moment algorithm, such as... Figure 2 Then extract features;

[0128] S21. Preprocess industrial CT images, including image enhancement, removal of image artifacts and noise;

[0129] S22. Extract image edges, search edge point sets to construct contour linked list sequence, perform probabilistic circle detection on non-empty linked lists, and accurately vectorize circles that meet the whole circle constraint; then insert them into the circle linked list.

[0130] S23. Use the set intersection method to fit lines to other contour lists, and determine whether short line segments are adjacent. If they are adjacent, merge them into long line segments.

[0131] S24. Use the perpendicular bisector method to find the arc and determine whether it forms an arc; perform precise vectorization of the contour points that form an arc and insert them into the arc linked list; calculate the least squares error of non-arc contour points and perform precise vectorization of the line and insert them into the line linked list.

[0132] S3. Editing and processing of vectorized graphics, such as... Figure 3 This includes connecting adjacent line segments, deleting redundant line segments and isolated points, and saving the vectorized graphics as a DXF format file.

[0133] S4. For example Figures 4-5 Select primitive features and match them with the model, such as Figures 4-5 The error was calculated and analyzed.

[0134] S41. Construct a CAD design model based on the original workpiece data;

[0135] S42. Pair CAD design model primitive information with CT vectorized graphic primitives.

[0136] The concept of minimum bounding boxes is used to introduce a centroid, and a set of primitive features is selected through human-computer interaction to achieve registration between models. Additionally, the function of fitting short straight lines to long straight lines and short straight lines to arcs is added as a registration benchmark. Two line segments are defined as a single long line segment and refitted as a long straight line; or two straight lines are defined as belonging to the same arc and refitted as a new arc.

[0137] S42. Discretize the vectorized primitives of the CT image into 48 points;

[0138] S43. Calculate the normal error to the CAD model.

[0139] Obviously, the above embodiments of the present invention are merely examples for clearly illustrating the present invention, and are not intended to limit the implementation of the present invention. Those skilled in the art will recognize that other variations or modifications can be made based on the above description. It is neither necessary nor possible to exhaustively describe all embodiments here. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the claims of the present invention.

Claims

1. A complex workpiece industrial CT multi-dimensional dimensional measurement method, characterized by the steps of Comprise: S1. Based on multi-dimensional platform, preset scanning plane with three-point space coordinates, multi-dimensional platform performs posture adjustment command according to posture adjustment parameter to carry out CT scanning on complex workpiece, and industrial two-dimensional CT slice image is obtained; Wherein the posture adjustment comprises detecting that the surface rotates around Z axis to be parallel to X axis and the surface rotates around X axis to be parallel to XY plane; S2. The industrial CT image is accurately vectorized based on error compensation space matrix algorithm, and then features are extracted; S21. The industrial CT image is preprocessed; S22. The image edge is extracted, the edge point set is searched, the contour linked list sequence is constructed, the non-empty linked list is subjected to probability method circle detection, the circle that meets the whole circle constraint condition is accurately vectorized, and then the circle linked list is inserted; S23. The set intersection method is used for straight line fitting on other contour linked list, whether adjacent short straight line segments are judged, if adjacent, the long straight line segment is merged; S24. The perpendicular bisector method is used for circular arc, whether the circular arc is judged; the contour point of the circular arc is accurately vectorized, and the circular arc linked list is inserted; The least square error is calculated for the non-circular arc contour point, the straight line is accurately vectorized, and the straight line linked list is inserted; S3. The edited graph is saved as a DXF format file after the vectorized graph is edited; S4. The CT vectorized graph element features are matched with the CAD design model, and error is calculated and analyzed.

2. The method according to claim 1, wherein, The three-point M1, M2, M3 coordinates of the scanning plane in step S1 are: M1 (X1, Y1, Z1) M2 (X2, Y2, Z2) M3 (X3, Y3, Z3) The general equation is: Ax+By+Cz+D=0, wherein: A= (Y2-Y1) × (Z3-Z2) - (Z2-Z1) × (Y3-Y2) B= (Z2-Z1) × (X3-X2) - (Z3-Z2) × (X2-X1) C= (X2-X1) × (Y3-Y2) - (Y2-Y1) × (X3-X2) D=X1× ( (Z2-Z1) × (Y3-Y2) - (Y2-Y1) × (Z3-Z2) ) + Y1× ( (Z3-Z2) × (X2-X1) - (Z2-Z1) × (X3-X2) ) + Z1× ( (Y2-Y1) × (X3-X2) - (X2-X1) × (Y3-Y2) ).

3. The method of claim 2, wherein the complex workpiece is a three-dimensional object. When the detection surface rotates around the Z axis to be parallel to the X axis, the angle when the detection surface rotates around the Z axis to be parallel to the X axis is denoted as , and the coordinates after rotation are , , ​ （ ） （ ） （ ） In the formula: According to the general equation of the plane: Ax+By+Cz+D=0, when the plane is parallel to the X axis, A=0 A= (Y2'-Y1') × (Z3-Z2) - (Z2-Z1) × (Y3'-Y2') =0 obtaining a rotation angle around the z-axis value.

4. The method of claim 2, wherein the complex workpiece is a multi-dimensional size measurement method of an industrial CT, characterized in that, When the detection surface rotates about the X axis to be parallel to the XY plane, the angle is θ", and the coordinates of the detection surface after rotation are: 、 、 coordinates are: （ ） （ ） （ ） In the formula: According to the general equation of the plane: Ax+By+Cz+D=0, when the plane is parallel to the XY plane, A=0, B=0 A= (Y2"-Y1") × (Z3'-Z2') - (Z2'-Z1') × (Y3"-Y2") =0 B= (Z2'-Z1') × (X3'-X2') - (Z3'-Z2') × (X2'-X1') =0 The rotation angle θ" around the X axis is obtained.

5. The method of claim 1, wherein the complex workpiece is a multi-dimensional size measurement method of an industrial CT, characterized in that, The preprocessing in step S21 comprises image enhancement, removing image artifacts and noise.

6. The method of claim 1, wherein the complex workpiece is a multi-dimensional size measurement method of an industrial CT. The editing process in step S3 includes connecting similar line segments, deleting redundant line segments and isolated points.

7. The method of claim 1, wherein the complex workpiece is a multi-dimensional size measurement method of an industrial CT. The centroid is introduced in step S4 using the idea of minimum bounding box, and a set of feature points are selected by human-computer interaction to realize the registration between models.

8. The method of claim 1, wherein the complex workpiece is an automobile part. In step S4, the functions of fitting short straight lines into long straight lines and fitting short straight lines into circular arcs are added as the reference for registration.

9. The method of claim 1, wherein the complex workpiece is a multi-dimensional size measurement method of an industrial CT, characterized by, Step S4 includes: S41. CAD design model information of the paired CT vectorized graphics primitives; S42. Discretize the CT image vectorized primitives; S43. Calculate the normal error to the CAD model.

Images