CN115828339A - Industrial CT image-based three-dimensional CAD solid model reconstruction method for workpiece with internal defects - Google Patents

Abstract

The invention discloses a method for reconstructing a three-dimensional CAD solid model of a workpiece with internal defects based on industrial CT images. The steps include: 1) acquiring a CT sequence image of the workpiece with defects; 2) processing the CT sequence images of the workpiece with defects to obtain the surface of the workpiece STL model and internal defect STL model; 3) the original CAD entity model of the workpiece and the workpiece STL model are registered, and the workpiece STL model is carried out space transformation, thereby obtaining the workpiece STL model after the space transformation; 4) according to the workpiece of step 3) STL model, generate CAD solid model of the workpiece with internal defects. The invention can realize the reconstruction of the CAD entity model of the workpiece with internal defects, and further qualitatively and quantitatively analyze the influence of defects on the overall performance of the workpiece.

Description

A 3D CAD solid model reconstruction of workpieces with internal defects based on industrial CT images construction method

technical field

The invention relates to the field of mechanical technology, in particular to a method for reconstructing a three-dimensional CAD solid model of a workpiece with internal defects based on industrial CT images.

Background technique

Some workpieces inevitably produce various internal defects in the process of manufacturing and use. In order to evaluate the impact of internal defects on the performance of the workpiece, a three-dimensional CAD solid model of the workpiece with internal defects can be established, and then digitally simulated Quantitative analysis of defective workpieces. To reconstruct the 3D CAD solid model of the workpiece with internal defects, the internal defects of the workpiece must be detected first. Industrial computed tomography (Computed Tomography, CT) technology can clearly, accurately and intuitively display the internal structure, composition, material and defects of the object by reconstructing the two-dimensional tomographic image or three-dimensional model of the detected object under non-destructive conditions.

At present, the research on workpieces with internal defects using industrial CT technology mainly focuses on three-dimensional defect visualization, defect measurement and statistics, and defect evolution analysis.

Some scholars use high-resolution CT to scan the cracks generated by carbon fiber reinforced composite materials under local impact, and analyze the damage mechanism and defect distribution to quantitatively identify the potential damage and failure mechanism of the workpiece.

Some scholars use industrial CT technology to scan and reconstruct 2A12 aluminum alloy cylindrical specimens that have undergone low-cycle fatigue experiments, and use 3D volume rendering to reconstruct them, and obtain 3D visualization images and their size information that are in good agreement with the actual crack situation. This reconstruction method is inefficient, and the small internal defects of the workpiece are easily lost during the reconstruction process.

Some scholars have used micro-focus CT to perform three-dimensional surface reconstruction and three-dimensional visualization of cracks, and studied the three-dimensional distribution of crack front growth increments in 2A50 wrought aluminum notched samples, but this method is limited to visualization of internal defects and cannot be quantified analyze. The above-mentioned studies are only to reconstruct the three-dimensional surface of the internal defects of the workpiece, and then observe and measure the spatial position and size of the defects. However, these studies cannot be simulated through digital simulation because the three-dimensional CAD solid model of the workpiece with internal defects cannot be obtained. Evaluate the impact of internal defects on the overall performance of the workpiece.

Some studies have been carried out on the reconstruction of the CAD entity model of the workpiece with defects. In order to improve the repair efficiency and reliability of the worn parts, some scholars proposed a method based on reverse engineering to reconstruct the CAD entity model of the worn workpiece. Acquisition and reconstruction of workpiece surface wear defects, this method is only used for reconstruction of 3D CAD solid model of workpieces with surface defects, and there is still no better method for reconstruction of workpieces with internal defects. Some scholars have proposed a new CAD solid model reconstruction method, which fits the parameter description (CAD template) containing the reconstructed object to the grid data to generate a CAD solid model with higher accuracy. This method relies on the CAD template, and Not suitable for reconstruction of CAD solid models with defective workpieces.

It can be seen that the current direct reconstruction of the 3D CAD solid model of the workpiece with internal defects has the following problems:

(1) There are a lot of manual operations in the CAD entity model reconstruction process, the process is complicated and cumbersome, and the efficiency is low;

(2) The reconstruction results are closely related to personal professional ability, and the reconstruction efficiency and accuracy of workpieces with more complex structures cannot be guaranteed;

(3) The proportion of defect volume is low, and there are no obvious geometric features. It is difficult to retain small features in the reconstruction process, and the fidelity of the model is reduced.

Contents of the invention

The object of the present invention is to provide a kind of reconstruction method of three-dimensional CAD solid model of the band internal defect workpiece based on industrial CT image, comprising the following steps:

1) Obtain CT sequence images of workpieces with defects;

2) Process the CT sequence images of the workpiece with defects to obtain the STL model of the workpiece;

The steps of processing the CT sequence images of workpieces with defects include:

Using the three-dimensional CV model and MC algorithm to process the volume data and reconstruct the STL model of the CT sequence images of the workpiece with defects, the STL model of the workpiece with defects is obtained;

The steps of volume data processing and STL model reconstruction of CT sequence images of workpieces with defects using 3D CV model and MC algorithm include:

2.1) sequentially read the CT sequence images of the workpiece to obtain the three-dimensional volume data to be processed;

2.2) Utilize the three-dimensional CV model to iteratively calculate the energy function to minimize the energy function;

The energy function F(C,c ₁ ,c ₂ ) of the three-dimensional CV model is as follows:

F(C,c ₁ ,c ₂ )＝μArea(C)+νVolume(C)+λ ₁ ∫ _in(c) |u ₀ (x,y,z)-c ₁ | ² dxdydz (1)

+λ ₂ ∫ _out(C) |u ₀ (x,y,z)-c ₂ | ² dxdydz

Among them, Area (C) represents the surface area of the evolution surface, Volume (C) represents the volume surrounded by the evolution surface; c ₁ , c ₂ represent the average gray value of pixels inside and outside the evolution surface C respectively; the constant μ, ν≥0, Constant λ ₁ , λ ₂ >0; u ₀ (x, y, z) is the gray value of any point on the constituent space area.

The steps of iteratively calculating the energy function using the three-dimensional CV model include:

2.2.1) Set parameter μ, parameter ν, parameter λ ₁ , parameter λ ₂ , parameter Δt, and initialize the level set function φ ₀ (x,y,z);

2.2.2) Establish the level set iteration expression of the energy function, namely:

为梯度算子；u₀指代u₀(x,y,z)；In the formula, φ is the signed distance function; H is the Heaviside function; δ is the Dirac function;

is the gradient operator; u ₀ refers to u ₀ (x,y,z);

2.2.3) Obtain the discrete iterative expression of the level set iterative expression (2) by using the finite difference method, and use the discrete iterative expression to invert the curve C, so as to iteratively calculate the energy function.

2.3) Every iteration T period, extract the zero level set of the three-dimensional CV model, and use the MC algorithm to reconstruct the isosurface of the zero level set;

2.4) judge whether the number of iterations reaches the number of termination iterations, if so, output the STL model with the defective workpiece reconstructed by the MC algorithm, otherwise, return to step 3);

The steps of extracting the STL model of the workpiece with defects include:

2.2.1) Establish an edge information table according to the vertex coordinates of each surface of the STL model of the workpiece with defects;

2.2.2) Select an unmarked triangular patch as a seed and mark it; add the marked triangular patch to the patch queue;

2.2.3) Take out the marked triangles from the queue, use the three sides of the marked triangles as the starting point, search for other unmarked faces that contain any side of the marked triangles, and mark these faces with the marked triangles. the same mark for the patch;

2.2.4) Repeat step 2.2.3) until there is no unmarked facet containing any edge of the marked triangular facet;

2.2.5) judge whether the queue is empty, if so, then enter step 2.2.6), otherwise return to step 2.2.2);

2.2.6) Extract the specified area, and output the patches with the same mark as a new STL model. Extract the specified area, and output the same-labeled patches as a new STL model; the specified area is a connected area composed of the same-labeled patches, including the surface area of the workpiece and several internal defect areas.

3) Register the original CAD entity model of the workpiece and the STL model of the workpiece, and perform spatial transformation on the STL model of the workpiece, so as to obtain the STL model of the workpiece after the spatial transformation;

The steps of registering the original CAD solid model of the workpiece and the STL model of the workpiece include:

3.1) Use the Delaunay triangulation method to convert the original CAD entity model of the workpiece into a point cloud to obtain the original point cloud model of the workpiece;

3.2) The steps of performing rough point cloud registration on the original CAD entity model of the workpiece and the STL model of the workpiece using the fast point feature histogram algorithm include:

3.2.1) For the target point cloud Q and the source point cloud P, use the formula (3) to calculate the FPFH of the point cloud set, namely:

In the formula, ω _i is the measure of the distance between the query point p _q and the adjacent point p _i , k is the average number of points in the neighborhood of each point, SPFH(p _i ) and SPFH(p _q ) are based on p _i and p _q is the value corresponding to the simplified feature histogram of the query point;

3.2.2) Obtain sampling points: Sampling the source point cloud P to obtain m points, in order to make the sampling point space as large as possible to cover point clouds, limit the minimum distance between sampling points;

3.2.3) Find point pairs: search for one or more corresponding points in the target point cloud with similar FPFH according to the sampling points in the source point cloud, and randomly select one as a space transformation point pair;

3.2.4) Space transformation: Calculate the rigid body transformation matrix according to the selected point pairs, and perform space transformation on the source point cloud, and complete the point cloud rough registration of the original point cloud model of the workpiece and the STL model of the workpiece.

3.3) Use the iterative closest point algorithm to perform point cloud fine registration on the original point cloud model of the workpiece and the STL model of the workpiece after the first rigid body transformation, and obtain the rigid body transformation matrix;

The steps of performing point cloud fine registration on the original point cloud model of the workpiece and the STL model of the workpiece after the first rigid body transformation by using the iterative closest point algorithm include:

工件原始点云模型的点云集合为

3.3.1) Record the point cloud set of the workpiece STL model after the first rigid body transformation as

The point cloud set of the original point cloud model of the workpiece is

中；N_q为点数；N_p、N_x为点数；3.3.2) For point P _i , search for the nearest point in the point cloud set X as the corresponding point, and write the corresponding point set

middle; N _q is the number of points; N _p and N _x are the number of points;

3.3.3) Calculate the rotation transformation matrix R and the translation transformation matrix T of fine registration according to the corresponding point set Q;

3.3.4) Perform rigid body transformation on the internal defect STL model after the first rigid body transformation according to the precisely registered rotation transformation matrix R and translation transformation matrix T;

3.3.5) Calculate the error change E _k of two adjacent transformations, if the error change E _k is less than a given threshold τ, then end the registration process, otherwise judge whether the number of iterations k reaches the maximum number of iterations, if so, then End the registration process, otherwise, set the number of iterations k=k+1, and return to step 3.3.1);

Among them, the error measurement function E(R,T) after the current rigid body transformation is as follows:

In the formula, R represents the rotation transformation matrix of fine registration, and T represents the translation transformation matrix.

3.4) Use the rigid body transformation matrix to perform space transformation on the internal defect STL model after the first rigid body transformation, so as to obtain the workpiece STL model after space transformation.

4) According to the STL model of the workpiece in step 3), a CAD solid model of the workpiece with internal defects is generated.

The steps for generating a CAD solid model of a workpiece with internal defects include:

4.1) Carry out topological reconstruction to the workpiece STL model after space transformation, and obtain the defect CAD entity model, the steps include:

4.1.1) Using the three-axis block sorting algorithm to delete duplicate vertices, the steps include:

4.1.1.1) Create vertex arrays X, Y, Z, input the internal defect STL model, and read three coordinate values (x, y, z) of a vertex each time;

4.1.1.2) Determine whether the vertex has been saved according to the coordinate value: judge by the coordinate x, y, z value in turn, if there is no value in the corresponding vertex array, then enter step 4.1.1.3), if the coordinate x value exists , then turn to the Y array pointed to by the coordinate x to judge whether the coordinate y exists, otherwise enter step 4.1.1.3);

If the coordinate y exists, turn to the Z array pointed to by the coordinate y to judge whether the coordinate z exists, otherwise enter step 4.1.1.3);

If the coordinate z exists, do not save the current coordinate value (x, y, z), and return to 4.1.1.1); otherwise, go to step 4.1.1.3);

4.1.1.3) Save the coordinate values (x, y, z) into the corresponding X, Y, Z arrays, and sort them;

The sorting rules are: after saving the coordinate x, sort the X array, and correspondingly transform the Y and Z arrays pointed to by the coordinate x; after saving the coordinate y, sort the Y array, and correspondingly transform the Z array pointed to by the coordinate xy; after saving the coordinate z, , to sort the Z array.

4.1.2) Establish vertex, edge and surface data tables;

Wherein, the step of creating the vertex data table includes: starting from the X array, determining the coordinate y and the coordinate z corresponding to each coordinate x, and adding them to the vertex table as new vertex coordinates (x, y, z);

The steps of creating an edge data table include: creating edges in pairs according to the coordinates of the three vertices of each grid, judging whether the created edge is in the edge data table, if not, establishing the index of the edge and the corresponding vertex in the vertex table, Create side information and add the created side to the side data table;

The step of creating the surface data table includes: adding the number of the surface and its corresponding normal vector, the indexes of the three vertices and the three edges forming the surface together as surface information to the surface data table;

4.1.3) Using the reconstructed topological relationship between the triangles, take the edge as the core, reconstruct all the triangle vertices, generate triangles from the points and edges, merge all the triangles into a whole, and get Closed space boundary, generate defect CAD solid model according to the closed boundary surface;

4.2) Carry out Boolean operations on the CAD entity model of internal defects and the original CAD entity model of the workpiece to generate a CAD entity model of the workpiece with defects.

It is worth noting that this patent aims at the problem that it is difficult to obtain a 3D CAD solid model of a workpiece with internal defects. On the basis of industrial CT technology, a method for reconstructing a 3D CAD solid model of a workpiece with internal defects based on industrial CT images is proposed. , the method is mainly divided into three steps: (1) 3D segmentation and extraction of internal defects. (2) The CAD entity model of the workpiece is registered with the reconstructed model. (3) The CAD solid model of the workpiece with internal defects is generated.

Technical effect of the present invention is beyond doubt, and beneficial effect of the present invention is as follows:

Aiming at the problem that it is difficult to obtain the three-dimensional CAD solid model of the workpiece with internal defects, the present invention combines industrial CT technology, fully utilizes the original CAD solid model existing in the workpiece design stage, and directly stores the internal defect information extracted from the industrial CT image in the original CAD of the workpiece Refactored out of the solid model.

The present invention first reconstructs the STL model of the workpiece and the STL model of the defect from the industrial CT sequence image of the workpiece with internal defects; The solid model is registered; finally, the internal defect STL model is solidified, and the original CAD solid model of the workpiece is performed with Boolean operations to generate a 3D CAD solid model of the workpiece with internal defects. The actual effectiveness of the invention is verified by reconstructing the three-dimensional CAD solid model of the actual combustion chamber insert.

The present invention mainly aims at the problem that it is difficult to obtain the CAD entity model in the digital analysis of workpieces with internal defects, and invents a method for reconstructing the three-dimensional CAD entity model of workpieces with defects based on industrial CT images, which provides a digital model simulation for workpieces with internal defects The direct and accurate model basis enables more accurate evaluation of the service performance and service life of the workpiece.

The present invention obtains defect information through industrial CT sequence images of internal defect workpieces, fully utilizes the 3D CAD entity model existing in the design stage, and solves the problem of internal defect workpieces through defect segmentation and extraction, model registration, and 3D CAD entity model reconstruction. The problem that the 3D CAD solid model is difficult to obtain. The effectiveness of the invention is verified by reconstructing the three-dimensional CAD solid model of the combustion chamber insert and performing finite element simulation, and the invention is also applicable to the reconstruction of the solid model of the workpiece containing small volume or multi-target internal defects. The invention can provide direct and accurate model support for subsequent digital simulation of specific workpieces with defects, and design a technical route based on comprehensive consideration of the method's practicability, adaptability and optimization convenience.

The invention can realize the reconstruction of the CAD entity model of workpieces with internal defects, and can further qualitatively and quantitatively analyze the influence of defects on the overall performance of the workpiece; the invention can effectively control the reconstruction time and avoid the time-consuming artificial reverse engineering method , The small features of the workpiece disappear and other problems.

Description of drawings

Fig. 1 is the flowchart that method realizes;

Figure 2 is the industrial CT sequence image of the combustion chamber insert;

Fig. 3 is the STL model of the combustion chamber insert with air holes;

Figure 4(a) is the STL model of the combustion chamber insert, and (b) is the STL model of the air hole (enlarged about 9 times);

Fig. 5 is the point cloud of the CAD solid model of the combustion chamber insert, Fig. 5(a) is the CAD solid model, Fig. 5(b) is the point cloud model (surface display), and Fig. 5(c) is the point cloud model (network grid display);

Figure 6 shows the registration of the combustion chamber insert CAD solid model and the reconstructed model, Figure 6(a) is the initial position, Figure 6(b) is the rough registration result, and Figure 6(c) is the fine registration result;

Figure 7 shows the registration detection effect of the combustion chamber insert CAD solid model and the reconstructed model;

Figure 8 shows the position change of the air hole and the combustion chamber insert (the arrow points to the air hole), Figure 8(a) is the initial position, and Figure 8(b) is the position after space transformation;

Fig. 9 is a pore CAD entity model, Fig. 9 is (a) an appearance view, and Fig. 9(b) is a cross-sectional view;

Figure 10 is a CAD solid model of a combustion chamber insert with air holes, Figure 10(a) is an appearance view (the arrow points to air holes), Figure 10(b) is a longitudinal section view, and Figure 10(c) is a partial view;

Fig. 11 is the pressure cloud diagram of the thermodynamic simulation of the combustion chamber insert.

Detailed ways

The present invention will be further described below in conjunction with the examples, but it should not be understood that the scope of the subject of the present invention is limited to the following examples. Without departing from the above-mentioned technical ideas of the present invention, various replacements and changes made according to common technical knowledge and conventional means in this field shall be included in the protection scope of the present invention.

Example 1:

Referring to Figures 1 to 11, a method for reconstructing a three-dimensional CAD solid model of a workpiece with internal defects based on industrial CT images includes the following steps:

1) Obtain CT sequence images of workpieces with defects;

2) Process the CT sequence images of the workpiece with defects to obtain the STL model of the surface of the workpiece and the STL model of internal defects;

The steps of processing the CT sequence images of workpieces with defects include:

2.1) Use the 3D CV model (Chan-Vese model, level set model) and MC algorithm (moving cube algorithm, marchingcube algorithm) to process the volume data and reconstruct the STL model of the CT sequence images of the workpiece with defects, and obtain the STL model of the workpiece with defects ;

The steps of volume data processing and STL model reconstruction of CT sequence images of workpieces with defects using 3D CV model and MC algorithm include:

2.1.1) sequentially read the CT sequence images of the workpiece to obtain the three-dimensional volume data to be processed;

2.1.2) Use the three-dimensional CV model to iteratively calculate the energy function to minimize the energy function;

The energy function F(C,c ₁ ,c ₂ ) of the three-dimensional CV model is as follows:

F(C,c ₁ ,c ₂ )＝μArea(C)+νVolume(C)+λ ₁ ∫ _in(c) |u ₀ (x,y,z)-c ₁ | ² dxdydz (1)

+λ ₂ ∫ _out(C) |u ₀ (x,y,z)-c ₂ | ² dxdydz

Among them, Area (C) represents the surface area of the evolution surface, Volume (C) represents the volume surrounded by the evolution surface; c ₁ , c ₂ represent the average gray value of pixels inside and outside the evolution surface C respectively; the constant μ, ν≥0, Constant λ ₁ , λ ₂ >0; u ₀ (x, y, z) is the gray value of any point on the constituent space area.

The steps of iteratively calculating the energy function using the three-dimensional CV model include:

2.1.2.1) Set parameter μ, parameter ν, parameter λ ₁ , parameter λ ₂ , parameter Δt, and initialize level set function φ ₀ (x,y,z);

2.1.2.2) Establish the level set iterative expression of the energy function, namely:

In the formula, φ is the signed distance function; H is the Heaviside function; δ is the Dirac function; ▽ is the gradient operator; u ₀ refers to u ₀ (x, y, z);

2.1.2.3) Obtain the discrete iterative expression of the level set iterative expression (2) by using the finite difference method, and use the discrete iterative expression to invert the curve C, so as to iteratively calculate the energy function.

2.1.3) Every iteration T period, extract the zero level set of the three-dimensional CV model, and use the MC algorithm to reconstruct the isosurface of the zero level set;

2.1.4) Judging whether the number of iterations reaches the number of termination iterations, if so, output the STL model of the workpiece with defects reconstructed by the MC algorithm, otherwise, return to step 3);

2.2) Extract the STL model of the workpiece with defects to obtain the workpiece STL model and internal defect STL model.

The steps of extracting the STL model of the workpiece with defects include:

2.2.1) Establish an edge information table according to the vertex coordinates of each surface of the STL model of the workpiece with defects;

2.2.2) Select an unmarked triangular patch as a seed and mark it; add the marked triangular patch to the patch queue;

2.2.3) Take out the marked triangles from the queue, use the three sides of the marked triangles as the starting point, search for other unmarked faces that contain any side of the marked triangles, and mark these faces with the marked triangles. the same mark for the patch;

2.2.4) Repeat step 2.2.3) until there is no unmarked facet containing any edge of the marked triangular facet;

2.2.5) judge whether the queue is empty, if so, then enter step 2.2.6), otherwise return to step 2.2.2);

2.2.6) Extract the specified area, and output the same-labeled patches as a new STL model; the specified area is a connected area composed of the same-labeled patches, including the surface area of the workpiece and several internal defect areas. There are two kinds of patch marks, one of which corresponds to the workpiece STL model, and the other corresponds to the internal defect STL model.

3) Register the original CAD entity model of the workpiece and the complete STL model of the workpiece, and perform space transformation on the STL model of the workpiece, so as to obtain the internal defect STL model after space transformation;

The steps of registering the original CAD solid model of the workpiece and the STL model of the workpiece include:

3.1) Use the Delaunay triangulation method to convert the point cloud of the original CAD solid model of the workpiece (such as the CAD solid model of the combustion chamber insert) to obtain the original point cloud model of the workpiece;

3.2) The steps of performing rough point cloud registration on the original CAD entity model of the workpiece and the STL model of the workpiece using the fast point feature histogram algorithm include:

3.2.1) For the target point cloud Q and the source point cloud P, use the formula (3) to calculate the FPFH of the point cloud set, namely:

In the formula, ω _i is the measure of the distance between the query point p _q and the adjacent point p _i , k is the average number of points in the neighborhood of each point, SPFH(p _i ) and SPFH(p _q ) are based on p _i and p _q is the value corresponding to the simplified feature histogram of the query point;

3.2.2) Obtain sampling points: Sampling the source point cloud P to obtain m points, in order to make the sampling point space as large as possible to cover point clouds, limit the minimum distance between sampling points;

3.2.3) Find point pairs: search for one or more corresponding points in the target point cloud with similar FPFH according to the sampling points in the source point cloud, and randomly select one as a space transformation point pair;

3.2.4) Space transformation: Calculate the rigid body transformation matrix according to the selected point pairs, and perform space transformation on the source point cloud, and complete the point cloud rough registration of the original point cloud model of the workpiece and the STL model of the workpiece.

3.3) Use the iterative closest point algorithm to perform point cloud fine registration on the original point cloud model of the workpiece and the STL model of the workpiece after the first rigid body transformation, and obtain the rigid body transformation matrix;

The steps of performing point cloud fine registration on the original point cloud model of the workpiece and the STL model of the workpiece after the first rigid body transformation by using the iterative closest point algorithm include:

工件原始点云模型的点云集合为

3.3.1) Record the point cloud set of the workpiece STL model after the first rigid body transformation as

The point cloud set of the original point cloud model of the workpiece is

中；N_q为点数；N_p、N_x为点数；3.3.2) For point P _i , search for the nearest point in the point cloud set X as the corresponding point, and write the corresponding point set

middle; N _q is the number of points; N _p and N _x are the number of points;

3.3.3) Calculate the rotation transformation matrix R and the translation transformation matrix T of fine registration according to the corresponding point set Q;

3.3.4) Perform rigid body transformation on the workpiece STL model after the first rigid body transformation according to the precisely registered rotation transformation matrix R and translation transformation matrix T;

3.3.5) Calculate the error change E _k of two adjacent transformations, if the error change E _k is less than a given threshold τ, then end the registration process, otherwise judge whether the number of iterations k reaches the maximum number of iterations, if so, then End the registration process, otherwise, set the number of iterations k=k+1, and return to step 3.3.1);

Among them, the error measurement function E(R,T) after the current rigid body transformation is as follows:

In the formula, R represents the rotation transformation matrix of fine registration, and T represents the translation transformation matrix. qi and pi are represented in the form of coordinates.

3.4) Use the rigid body transformation matrix to perform space transformation on the workpiece STL model after the first rigid body transformation, so as to obtain the space transformed workpiece STL model.

4) According to the STL model of the workpiece in step 3), a CAD solid model of the workpiece with internal defects is generated.

The steps for generating a CAD solid model of a workpiece with internal defects include:

4.1) Carry out topological reconstruction to the workpiece STL model after space transformation, and obtain the defect CAD entity model, the steps include:

4.1.1) Using a three-axis block sorting algorithm (refer to Cui Shubiao, Zhang Yisheng, Liang Shuyun, Li Dequn. A fast filtering algorithm for redundant vertices in STL patches and its application. China Mechanical Engineering, 2001(02): 173-175 +117) Delete duplicate vertices, the steps include:

4.1.1.1) Create vertex arrays X, Y, Z, input the internal defect STL model, and read three coordinate values (x, y, z) of a vertex each time;

4.1.1.2) Determine whether the vertex has been saved according to the coordinate value: judge by the coordinate x, y, z value in turn, if there is no value in the corresponding vertex array, then enter step 4.1.1.3), if the coordinate x value exists , then turn to the Y array pointed to by the coordinate x to judge whether the coordinate y exists, otherwise enter step 4.1.1.3);

If the coordinate y exists, turn to the Z array pointed to by the coordinate y to judge whether the coordinate z exists, otherwise enter step 4.1.1.3);

If the coordinate z exists, do not save the current coordinate value (x, y, z), and return to 4.1.1.1); otherwise, go to step 4.1.1.3);

4.1.1.3) Save the coordinate values (x, y, z) into the corresponding X, Y, Z arrays, and sort them;

The sorting rules are: after saving the coordinate x, sort the X array, and correspondingly transform the Y and Z arrays pointed to by the coordinate x; after saving the coordinate y, sort the Y array, and correspondingly transform the Z array pointed to by the coordinate x y; after saving the coordinate z, , to sort the Z array.

4.1.2) Establish vertex, edge and surface data tables;

Wherein, the step of creating the vertex data table includes: starting from the X array, determining the coordinate y and the coordinate z corresponding to each coordinate x, and adding them to the vertex table as new vertex coordinates (x, y, z);

The steps of creating an edge data table include: creating edges in pairs according to the coordinates of the three vertices of each grid, judging whether the created edge is in the edge data table, if not, establishing the index of the edge and the corresponding vertex in the vertex table, Create side information and add the created side to the side data table;

The step of creating the surface data table includes: adding the number of the surface and its corresponding normal vector, the indexes of the three vertices and the three edges forming the surface together as surface information to the surface data table;

4.1.3) Using the reconstructed topological relationship between the triangles, take the edge as the core, reconstruct all the triangle vertices, generate triangles from the points and edges, merge all the triangles into a whole, and get Closed space boundary, generate defect CAD solid model according to the closed boundary surface;

4.2) Carry out Boolean operations on the CAD entity model of internal defects and the original CAD entity model of the workpiece to generate a CAD entity model of the workpiece with defects.

Example 2:

A method for reconstructing a three-dimensional CAD solid model of a workpiece with internal defects based on industrial CT images, comprising the following steps:

1. Three-dimensional segmentation and extraction of internal defects

(1) Input the CT sequence image of the workpiece with internal defects from industrial CT scanning;

(2) Use the 3D CV model for volume data processing and STL model reconstruction to obtain the STL model of the workpiece with internal defects, as shown in Figure 3;

(3) STL model extraction of combustion chamber inserts and internal defects, the results are shown in Figure 4.

2. Registration of workpiece CAD solid model and reconstruction model

(1) Use Delaunay triangulation to convert the CAD solid model of the combustion chamber into a point cloud, as shown in Figure 5;

(2) Use the Fast Point Feature Histogram (FPFH) algorithm to perform coarse registration of point clouds;

(3) Perform point cloud fine registration through the Iterative Closest Point (ICP) algorithm, the result is shown in Figure 6, and the registration accuracy is detected and displayed, the result is shown in Figure 7.

(4) Perform rigid body transformation on internal defects to obtain the positional relationship between defects and the original CAD entity model of the workpiece, as shown in Figure 8.

3. Generation of CAD solid model of workpiece with internal defects

(1) The topology reconstruction of the internal defect STL model, the CAD entity model of the internal defect reconstruction is shown in Figure 9;

(2) Carry out Boolean operation with the internal defect CAD entity model and the original CAD entity model of the workpiece to generate the CAD entity model of the defective workpiece, as shown in Figure 10;

(3) Import the reconstructed CAD solid model of the workpiece with internal defects into the ANSYS finite element simulation software to verify the availability of the reconstructed model. The results are shown in Figure 11.

Example 3:

A method for reconstructing a three-dimensional CAD solid model of a workpiece with internal defects based on industrial CT images. The main steps are shown in Embodiment 2, wherein the three-dimensional segmentation and extraction steps of internal defects include:

The 3D CV model performs defect segmentation directly from the volume data, and obtains an internally closed defect-surrounding surface. The spatial region composed of industrial CT sequence images is regarded as composed of background and target whose gray values are approximately constant, and their gray values are u ₀ ^b and u ₀ ^t respectively. At this time, there is a closed evolution surface C to divide the space region, and the space regions inside and outside the surface are denoted by in(C) and out(C) respectively. Assuming that the gray value of any point on the constituent space area is u ₀ (x, y, z), the energy function of the three-dimensional CV model is:

F(C,c ₁ ,c ₂ )＝μArea(C)+νVolume(C)+λ ₁ ∫ _in(c) |u ₀ (x,y,z)-c ₁ | ² dxdydz (1)

+λ ₂ ∫ _out(C) |u ₀ (x,y,z)-c ₂ | ² dxdydz

Among them, Area (C) represents the surface area of the evolution surface, Volume (C) represents the volume surrounded by the evolution surface, c ₁ and c ₂ represent the average gray value of the pixels inside and outside the evolution surface C, constant μ, ν≥0, λ ₁ , λ ₂ >0, where λ ₁ , λ ₂ always take 1.

The process of surface evolution of the 3D CV model is to solve the formula (1) to minimize its value. By introducing the signed distance function φ, the Heaviside function H, the Dirac function δ, and the time variable t, the level set iteration expression of the formula (1) is obtained as :

By using the finite difference method to obtain the discrete iterative expression of formula (2), the three-dimensional volume data composed of industrial CT sequence slices can be segmented through programming.

The experimental steps for volume data processing and STL model reconstruction of the sample using the 3D CV model and MC algorithm are as follows:

Step Ⅰ: sequentially read the industrial CT sequence images of the workpiece with internal defects to obtain the three-dimensional volume data to be processed;

Step II: Set parameters μ, ν, λ ₁ , λ ₂ , Δt, and initialize the level set function φ ₀ (x, y, z);

Step Ⅲ: Continuously iterate the level set function through discrete iteration until the termination condition is reached;

Step Ⅳ: Use MC algorithm to reconstruct the isosurface of the zero level set for display every time period of iteration;

Step Ⅴ: When the iteration terminates, save the triangular mesh STL model reconstructed by the MC algorithm for further analysis.

For the STL model of the workpiece with internal defects obtained through triangular mesh surface reconstruction, the STL model of internal defects needs to be extracted from it. The surface of the workpiece and the surface of the internal defect are respectively surrounded by different triangular faces to form a closed surface, forming different connected regions, and the connected region analysis method can be used to extract the defect STL model. In the STL model composed of triangular patches, any side must be shared by two triangles, and the area connected by the equivalent triangular patches containing adjacent sides is considered as a whole, and the entire isosurface is divided into Connected areas that are independent of each other, through this method, the calibration and extraction of the defect area and the workpiece itself are completed. The specific implementation steps are as follows:

Step Ⅰ: Input the STL model of the workpiece with defects, and establish the edge information table according to the vertex coordinates in the patch;

Step II: Use a triangle patch as a seed, mark it and add it to the patch queue;

Step Ⅲ: Take out the marked triangular patches from the queue one by one, take the three edges of the patch as the starting point, find other unmarked patches containing this edge as a connected set, mark them and add them to the patch queue;

Step Ⅳ: When the queue is empty, all marked patches are regarded as a connected region;

Step Ⅴ: Repeat steps II-Ⅳ until all patches are marked;

Step Ⅵ: Extract the specified area, and export the patches with the same mark as a new STL model.

Analyzing the STL model of the combustion chamber insert with pores, the proportion of the defect area of the pores is 1.26%, and the proportion of the volume is 0.27%. This experiment verifies that the present invention has a better effect in three-dimensional segmentation of small-volume defects.

Example 4:

A method for reconstructing a three-dimensional CAD solid model of a workpiece with internal defects based on industrial CT images, the main steps of which are shown in Embodiment 2, wherein the steps of registering the CAD solid model of the workpiece with the reconstructed model include:

(1) Realization of coarse registration

The purpose of coarse registration is to transform the source point cloud through rigid body transformation through initial registration, so that the pose of the source point cloud and the target point cloud are as close as possible. The fast point feature histogram (FPFH) algorithm has a fast solution speed and is applied For convenience, this paper uses the FPFH algorithm for rough matching.

FPFH is optimized from the computational complexity of the point feature histogram (PointFeatureHistogram, PFH). PFH is based on the relationship between a central point and its neighbors. By calculating the relationship between normals, use <α, φ, θ, d> four feature element values constitute a descriptor, where <α, φ, θ> represents the angle deviation of the normal between two points, d=||p ₂ -p ₁ || ₂ represents the Euclidean Finally, a multi-dimensional histogram is formed by calculating the descriptors between all two points in the neighborhood to express the feature change of the model surface.

For point cloud data with n points, if the number of points in the neighborhood of each point is k, the theoretical computational complexity of PFH is o(nk ² ), and FPFH only calculates the features between the center point and the adjacent point in the neighborhood Elements form a simplified point feature histogram (SimplifiedPointFeatureHistogram, SPFH), which improves the calculation speed. The calculation process is as follows:

Step Ⅰ: Only calculate the feature elements between the query point P _q neighborhood and its immediate neighbor point P _i to form SPFH;

Step II: For each adjacent point P _i in the neighborhood of P _q , determine the SPFH with P _i as the query point;

Step Ⅲ: Calculate the FPFH of P _q by weighting the SPFH of P _q and all P _i . The weighted calculation formula is as follows:

Among them, ω _i is the measure of the distance between the query point P _q and the adjacent point P _i , and represents the weight of (P _q , P _i ) occupied by the point pair.

The FPFH feature of a point in a point cloud is quickly obtained through the FPFH algorithm, and the application to point cloud registration is to compare whether the FPFH features of point pairs in two point clouds are close. The rough registration algorithm process is as follows:

Step Ⅰ: Calculate FPFH. For the target point cloud Q and the source point cloud P, calculate the FPFH of the point cloud set respectively;

Step II: Obtain sampling points. Sampling the source point cloud P to obtain m points, in order to make the sampling point space as large as possible to cover the point cloud set, limit the minimum distance between sampling points;

Step Ⅲ: Find point pairs. According to the sampling points in the source point cloud, one or more corresponding points close to the FPFH are found in the target point cloud, and one is randomly selected as a space transformation point pair;

Step Ⅳ: Space transformation. The rigid body transformation matrix is calculated according to the selected point pairs, and the source point cloud is spatially transformed to make it close to the target point cloud.

(2) Realization of fine registration

Fine registration is based on rough registration to make the poses of the two point cloud models closer to get the best matching position. Currently the most widely used is the Iterative Closest Point (ICP) algorithm ^[22] . The essence of the algorithm is the optimal matching based on the least squares method, and it executes the iterative process from determining the corresponding point set to calculating the optimal rigid body transformation until the measurement error meets the set convergence accuracy or reaches the maximum number of iterations, thus ending the entire registration process.

和目标点云集合

对于P中的每个点P_i搜索在X中的最近点作为对应点，设在X中搜索到的P的对应点集为

则误差测度的函数表达式为：Assume that there is a collection of source point clouds

and target point cloud set

For each point _Pi in P, search for the nearest point in X as the corresponding point, and set the corresponding point set of P searched in X as

Then the functional expression of the error measure is:

Among them, R represents the rotation transformation matrix of fine registration, and T represents the translation transformation matrix.

For the target point cloud X and the source point cloud P, the registration process of the ICP algorithm is as follows:

Step Ⅰ: Select the corresponding point set. Calculate the closest corresponding point set Q of the source point cloud P in the target point cloud X, so that it satisfies P: Q=C(P,X), and C means looking for the nearest point

Step II: Calculate the rigid body transformation matrices R and T from the corresponding point set.

Step Ⅲ: Position information conversion. According to the matrix in II, the source point cloud is rigidly transformed to a new position.

Step IV: Calculate the error change E _k for two adjacent transformations, if it is less than a given threshold τ, the iteration termination condition is reached, and the registration process ends. Otherwise, let the number of iterations k=k+1, if k does not reach the maximum number of iterations, turn to step I to continue iterating

Example 5:

A method for reconstructing a three-dimensional CAD solid model of a workpiece with internal defects based on industrial CT images. The main steps are shown in Embodiment 2, wherein the steps for generating the CAD solid model of a workpiece with internal defects include:

(1) STL model topology reconstruction of internal defects

For the STL file format that expresses the triangular mesh model, it stores all the triangular patch information used to approximate the CAD solid model, including the coordinates of the three vertices that make up each triangular patch and the normal vector information of the plane where it is located. The space formed The area is closed and bounded. Compared with the CAD solid model expressed by the boundary representation (Boundary/Representation, B-Rep), the STL model does not have the topological relationship between the triangular meshes, such as the relationship between edges and faces. Therefore, the key step in transforming from STL model to B-Rep solid model is to reconstruct the topological relationship between triangular meshes. The reconstruction of topological relationship mainly includes two steps: deleting duplicate vertices and establishing data tables of vertices, faces and edges.

Step Ⅰ: Delete duplicate vertices. The STL file repeatedly stores the vertex information of the contact surface. The information storage of these repeated vertices causes the STL file to occupy a large memory and reduces the efficiency of topological relationship reconstruction. The three-axis block sorting algorithm ^[22] is used to delete duplicate vertices. The specific implementation process is as follows:

(a) Create vertex arrays X, Y, Z, input the STL file, and read three coordinate values (x, y, z) of a vertex each time;

(b) Determine whether the vertex has been saved according to the coordinate value. Judgment is made sequentially by the values of x, y, and z. If the value does not exist in the corresponding vertex array, enter (c). If the value of x exists, turn to the Y array pointed to by x to judge y. If y exists, turn to y The Z array pointed to judges z, if z exists, enter (a);

(c) Save the unsaved x, y, and z values transferred into this item to the corresponding X, Y, and Z arrays, and sort them. The rule is, if you save x, sort the X array, and change the Y and Z arrays it points to accordingly; if you save y, sort the Y array, and change the Z array it points to accordingly; if you save z, then change the Z array Sort.

Step II: Create vertex, edge, face data tables

(a) Create a vertex data table. According to the three coordinate value arrays X, Y, Z obtained in step I, starting from the X array, determine the y coordinate and z coordinate corresponding to each x, and add it as a new vertex coordinate (x, y, z) to in the vertex table;

(b) Create an edge data table. In (a) the vertex coordinate table after deduplication, create edges in pairs according to the three vertex coordinates of each grid, and judge whether it is in the edge data table, if not, establish the relationship between this edge and the corresponding vertex in the vertex table Index, create side information, and add it to the side data table;

(c) Create a surface data table. The number of the surface and its corresponding normal vector, the indices of the three vertices and the three edges that make up the surface are added to the surface data table as surface information.

After completing the topology reconstruction of the STL model, in order to retain the defect space information as much as possible, the topological relationship between the reconstructed triangles is directly used, and the edges are used as the core to reconstruct the entities of all triangle vertices, and the points and edges are generated. Triangular faces, and then merge all the triangular faces into a whole to form a closed space boundary like the STL model, and then generate a B-Rep solid model from the closed boundary surface.

(2) Boolean operation

The present invention uses the Boolean operation function provided in the OpenCasCade open source library to complete the Boolean operation between three-dimensional entities, and finally reconstructs to obtain the three-dimensional CAD entity model of the workpiece with internal defects.

Embodiment 6:

A method for reconstructing a three-dimensional CAD solid model of a workpiece with internal defects based on industrial CT images, comprising the following steps:

1) Obtain CT sequence images of workpieces with defects;

2) Process the CT sequence images of the workpiece with defects to obtain the STL model of the workpiece;

3) Register the original CAD entity model of the workpiece and the STL model of the workpiece, and perform spatial transformation on the STL model of the workpiece, so as to obtain the STL model of the workpiece after the spatial transformation;

4) According to the STL model of the workpiece after space transformation, a CAD solid model of the workpiece with internal defects is generated.

Embodiment 7:

A method for reconstructing a three-dimensional CAD solid model of a workpiece with internal defects based on industrial CT images, the main content of which is shown in Embodiment 6, wherein the steps of processing the CT sequence images of the workpiece with defects include:

Using the 3D CV model and MC algorithm to process the volume data and reconstruct the STL model of the CT sequence images of the workpiece with defects, the STL model of the workpiece with defects is obtained.

Embodiment 8:

A method for reconstructing a three-dimensional CAD solid model of a workpiece with internal defects based on industrial CT images, the main content of which is shown in Embodiment 6, wherein, the three-dimensional CV model and the MC algorithm are used to perform volume data processing and STL model reconstruction on the CT sequence images of the workpiece with defects The steps include:

1) sequentially read the CT sequence images of the workpiece to obtain the three-dimensional volume data to be processed;

2) Using the three-dimensional CV model to iteratively calculate the energy function to minimize the energy function;

3) Extract the zero-level set of the three-dimensional CV model every iteration T period, and use the MC algorithm to reconstruct the isosurface of the zero-level set;

4) Judging whether the number of iterations reaches the number of termination iterations, if so, output the STL model with the defective workpiece reconstructed by the MC algorithm, otherwise, return to step 3);

Embodiment 9:

A method for reconstructing a three-dimensional CAD solid model of a workpiece with internal defects based on industrial CT images, the main content of which is shown in Embodiment 6, wherein the energy function F(C,c ₁ ,c ₂ ) of the three-dimensional CV model is as follows:

F(C,c ₁ ,c ₂ )＝μArea(C)+νVolume(C)+λ ₁ ∫ _in(c) |u ₀ (x,y,z)-c ₁ | ² dxdydz (1)

+λ ₂ ∫ _out(C) |u ₀ (x,y,z)-c ₂ | ² dxdydz

Among them, Area (C) represents the surface area of the evolution surface, Volume (C) represents the volume surrounded by the evolution surface; c ₁ , c ₂ represent the average gray value of pixels inside and outside the evolution surface C respectively; the constant μ, ν≥0, Constant λ ₁ , λ ₂ >0; u ₀ (x, y, z) is the gray value of any point on the constituent space area.

Example 10:

A method for reconstructing a three-dimensional CAD solid model of a workpiece with internal defects based on industrial CT images, the main content of which is shown in Embodiment 6, wherein the step of using the three-dimensional CV model to iteratively calculate the energy function includes:

1) Set parameter μ, parameter ν, parameter λ ₁ , parameter λ ₂ , parameter Δt, and initialize the level set function φ ₀ (x,y,z);

2) Establish the level set iterative expression of the energy function, namely:

为梯度算子；u₀指代u₀(x,y,z)；In the formula, φ is the signed distance function; H is the Heaviside function; δ is the Dirac function;

is the gradient operator; u ₀ refers to u ₀ (x,y,z);

3) Obtain the discrete iterative expression of the level set iterative expression (2) by using the finite difference method, and use the discrete iterative expression to invert the curve C, so as to iteratively calculate the energy function.

Example 11:

A method for reconstructing a three-dimensional CAD solid model of a workpiece with internal defects based on industrial CT images, the main content of which is shown in Embodiment 6, wherein the step of extracting the STL model of the workpiece with defects includes:

1) Establish an edge information table according to the vertex coordinates of each surface of the STL model of the workpiece with defects;

2) Select an unmarked triangular patch as a seed and mark it; add the marked triangular patch to the patch queue;

3) Take out the marked triangles from the queue, use the three sides of the marked triangles as a starting point, search for other unmarked faces that contain any side of the marked triangles, and mark these faces with the marked triangles the same mark;

4) Repeat step 3) until there is no unmarked facet containing any edge of the marked triangular facet;

5) judge whether the queue is empty, if so, then enter step 6), otherwise return to step 2);

6) Extract the specified area, and output the same-labeled patch as a new STL model; the specified area is a connected area composed of the same-labeled patch, including the surface area of the workpiece and several internal defect areas.

Example 12:

A method for reconstructing a three-dimensional CAD solid model of a workpiece with internal defects based on industrial CT images, the main content of which is shown in Embodiment 6, wherein the steps of registering the original CAD solid model of the workpiece and the STL model of the workpiece include:

1) Use the Delaunay triangulation method to convert the point cloud of the original CAD entity model of the workpiece to obtain the original point cloud model of the workpiece;

2) Use the fast point feature histogram algorithm to perform rough point cloud registration on the original point cloud model of the workpiece and the STL model of the workpiece, and obtain the STL model of the workpiece after the first rigid body transformation;

3) Use the iterative closest point algorithm to perform point cloud fine registration on the original point cloud model of the workpiece and the STL model of the workpiece after the first rigid body transformation, and obtain the rigid body transformation matrix;

4) Use the rigid body transformation matrix to perform space transformation on the workpiece STL model after the first rigid body transformation, so as to obtain the space transformed workpiece STL model.

Example 13:

A method for reconstructing a three-dimensional CAD solid model of a workpiece with internal defects based on industrial CT images, the main content of which is shown in Embodiment 6, wherein the original CAD solid model of the workpiece and the STL model of the workpiece are roughly matched with point clouds using the fast point feature histogram algorithm Standard steps include:

1) For the target point cloud Q and the source point cloud P, use the formula (3) to calculate the FPFH of the point cloud set, namely:

In the formula, ω _i is the measure of the distance between the query point p _q and the adjacent point p _i , k is the average number of points in the neighborhood of each point, SPFH(p _i ) and SPFH(p _q ) are based on p _i and p _q is the value corresponding to the simplified feature histogram of the query point;

2) Obtain sampling points: Sampling the source point cloud P to obtain m points, in order to make the sampling point space as large as possible to cover the point cloud set, limit the minimum distance between sampling points;

3) Look for point pairs: Find one or more corresponding points in the target point cloud according to the sampling points in the source point cloud with similar FPFH, and randomly select one as a space transformation point pair;

4) Space transformation: Calculate the rigid body transformation matrix according to the selected point pair, and perform space transformation on the source point cloud, and complete the point cloud rough registration of the original point cloud model of the workpiece and the STL model of the workpiece.

Example 14:

A method for reconstructing a three-dimensional CAD solid model of a workpiece with internal defects based on industrial CT images, the main content of which is shown in Embodiment 6, wherein the original point cloud model of the workpiece and the STL model of the workpiece after the first rigid body transformation are reconstructed using the iterative closest point algorithm The steps for point cloud fine registration include:

工件原始点云模型的点云集合为

N_p、N_x为点数；1) Record the point cloud set of the workpiece STL model after the first rigid body transformation as

The point cloud set of the original point cloud model of the workpiece is

N _p and N _x are points;

中；N_q为点数；2) For point P _i , search for the nearest point in the point cloud set X as the corresponding point, and write the corresponding point set

middle; N _q is the number of points;

3) Calculate the rotation transformation matrix R and the translation transformation matrix T of fine registration according to the corresponding point set Q;

4) Perform rigid body transformation on the internal defect STL model after the first rigid body transformation according to the finely registered rotation transformation matrix R and translation transformation matrix T;

5) Calculate the error change E _k of two adjacent transformations, if the error change E _k is less than a given threshold τ, then end the registration process, otherwise judge whether the iteration number k reaches the maximum number of iterations, if so, end the registration Quasi-process, otherwise, make the number of iterations k=k+1, and return to step 1);

Among them, the error measurement function E(R,T) after the current rigid body transformation is as follows:

In the formula, R represents the rotation transformation matrix of fine registration, and T represents the translation transformation matrix.

Example 15:

A method for reconstructing a three-dimensional CAD solid model of a workpiece with internal defects based on industrial CT images, the main content of which is shown in Embodiment 6, wherein the step of generating a CAD solid model of a workpiece with internal defects includes:

1) Carry out topology reconstruction on the workpiece STL model after space transformation, and obtain the internal defect CAD entity model, the steps include:

1.1) Use the three-axis block sorting algorithm to delete duplicate vertices. The steps include:

1.1.1) Create vertex arrays X, Y, Z, input the internal defect STL model, and read three coordinate values (x, y, z) of a vertex each time;

1.1.2) Determine whether the vertex has been saved according to the coordinate value: judge by the coordinate x, y, z value in turn, if there is no value in the corresponding vertex array, then enter step 1.1.3), if the coordinate x value exists , then turn to the Y array pointed to by the coordinate x to judge whether the coordinate y exists, otherwise enter step 1.1.3);

If the coordinate y exists, turn to the Z array pointed to by the coordinate y to judge whether the coordinate z exists, otherwise enter step 1.1.3);

If the coordinate z exists, do not save the current coordinate value (x, y, z), and return to 1.1.1), otherwise, go to step 3);

1.1.3) Save the coordinate values (x, y, z) into the corresponding X, Y, Z arrays and sort them;

The sorting rules are: after saving the coordinate x, sort the X array, and correspondingly transform the Y and Z arrays pointed to by the coordinate x; after saving the coordinate y, sort the Y array, and correspondingly transform the Z array pointed to by the coordinate x y; after saving the coordinate z, , to sort the Z array.

1.2) Establish vertex, edge and surface data tables;

Wherein, the step of creating the vertex data table includes: starting from the X array, determining the coordinate y and the coordinate z corresponding to each coordinate x, and adding them to the vertex table as new vertex coordinates (x, y, z);

The steps of creating an edge data table include: creating edges in pairs according to the coordinates of the three vertices of each grid, judging whether the created edge is in the edge data table, if not, establishing the index of the edge and the corresponding vertex in the vertex table, Create side information and add the created side to the side data table;

The steps of creating the surface data table include: adding the number of the surface and its corresponding normal vector, the indexes of the three vertices and the three edges forming the surface together as surface information to the surface data table;

1.3) Utilize the topological relationship between the reconstructed triangular faces, take the edge as the core, and reconstruct all the triangular vertices, generate triangular faces from the points and edges, and merge all the triangular faces into a whole to obtain a closed Spatial boundary, generate defect CAD solid model according to the closed boundary surface;

2) Perform Boolean operations on the CAD entity model of internal defects and the original CAD entity model of the workpiece to generate a CAD entity model of the workpiece with defects.

Claims (10)

1. A three-dimensional CAD solid model reconstruction method of a workpiece with an internal defect based on an industrial CT image is characterized by comprising the following steps:

1) Acquiring a CT sequence image of a workpiece with defects;

2) And processing the CT sequence image of the workpiece with the defect to obtain the STL model of the workpiece.

3) Registering the workpiece original CAD solid model and the workpiece STL model, and carrying out spatial transformation on the workpiece STL model so as to obtain a workpiece STL model after spatial transformation;

4) And generating a CAD entity model of the workpiece with the internal defects according to the workpiece STL model after the space transformation.

2. The method for reconstructing the three-dimensional CAD solid model of the workpiece with the internal defect based on the industrial CT image as recited in claim 1, wherein the step of processing the CT sequence image of the workpiece with the defect comprises the following steps:

and carrying out volume data processing and STL model reconstruction on the CT sequence image of the workpiece with the defect by using a three-dimensional CV model and an MC algorithm to obtain the STL model of the workpiece with the defect.

3. The industrial CT image-based three-dimensional CAD solid model reconstruction method for workpieces with internal defects, as recited in claim 2, is characterized in that the steps of utilizing a three-dimensional CV model and an MC algorithm to carry out volume data processing and STL model reconstruction on CT sequence images of workpieces with defects comprise:

1) Sequentially reading CT sequence images of the workpiece to obtain three-dimensional volume data to be processed;

2) Iteratively calculating an energy function by using the three-dimensional CV model to minimize the energy function;

3) Extracting a zero level set of the three-dimensional CV model every iteration T period, and performing isosurface reconstruction on the zero level set by using an MC algorithm;

4) And (4) judging whether the iteration times reach the termination iteration times, if so, outputting the STL model of the workpiece with the defect reconstructed by the MC algorithm, and otherwise, returning to the step 3).

4. The method for reconstructing the three-dimensional CAD solid model of the workpiece with the internal defect based on the industrial CT image as recited in claim 3, characterized in that the energy function F (C, C) of the three-dimensional CV model ₁ ,c ₂ ) As follows:

wherein, area (C) represents the surface Area of the evolution curved surface, volume (C) represents the Volume surrounded by the evolution curved surface; c. C ₁ ,c ₂ Respectively representing the average value of the pixel gray levels inside and outside the evolution curved surface C; constant mu, v is more than or equal to 0 and constant lambda ₁ ，λ ₂ >0；u ₀ (x, y, z) is the gray scale value of any point on the component spatial region.

5. The method for reconstructing the three-dimensional CAD solid model of the workpiece with the internal defect based on the industrial CT image as recited in claim 4, characterized in that the step of iteratively calculating the energy function by using the three-dimensional CV model comprises the following steps:

1) Setting a parameter mu, a parameter v and a parameter lambda ₁ Parameter λ, parameter ₂ Parameter Δ t, initialization level set function φ ₀ (x,y,z)；

2) Establishing a level set iterative expression of an energy function, namely:

wherein φ is a symbol distance function; h is a Heaviside function; δ is a Dirac function;

is a gradient operator; u. of ₀ Reference to u ₀ (x,y,z)；

3) And (3) obtaining a discrete iteration expression of the level set iteration expression (2) by using a finite difference method, and inverting the curve C by using the discrete iteration expression, thereby performing iterative computation on the energy function.

6. The method for reconstructing the three-dimensional CAD solid model of the workpiece with the internal defect based on the industrial CT image as recited in claim 2, characterized in that the step of extracting the STL model of the workpiece with the defect comprises the following steps:

1) Establishing a side information table according to the vertex coordinates of each patch of the STL model of the workpiece with the defect;

2) Selecting an unmarked triangular patch as a seed and marking; adding the marked triangular patch into a patch queue;

3) Taking out the marked triangular patch from the queue, taking three edges of the marked triangular patch as starting points, searching other unmarked patches containing any edge of the marked triangular patch, and marking the patches with the same mark as the marked triangular patch;

4) Repeating the step 3) until no unmarked patch containing any edge of the marked triangular patch exists;

5) Judging whether the queue is empty, if so, entering the step 6), otherwise, returning to the step 2);

6) Extracting a designated area, and outputting the patches marked with the same marks as a new STL model; the designated area is a communication area formed by the same marked surface patches and comprises a workpiece surface area and a plurality of internal defect areas.

7. The industrial CT image-based three-dimensional CAD solid model reconstruction method for workpieces with internal defects as recited in claim 1, wherein the step of registering the original CAD solid model of the workpiece and the STL model of the workpiece comprises:

1) Performing point cloud of the original CAD entity model of the workpiece by using a Delaunay triangulation method to obtain an original point cloud model of the workpiece;

2) Carrying out point cloud rough registration on the workpiece original point cloud model and the workpiece STL model by using a fast point feature histogram algorithm to obtain a workpiece STL model after the first rigid body transformation;

3) Performing point cloud fine registration on the workpiece original point cloud model and the workpiece STL model after the first rigid body transformation by using an iterative closest point algorithm to obtain a rigid body transformation matrix;

4) And carrying out space transformation on the workpiece STL model after the first rigid body transformation by using the rigid body transformation matrix so as to obtain the workpiece STL model after the space transformation.

8. The industrial CT image-based three-dimensional CAD solid model reconstruction method for workpieces with internal defects as recited in claim 7, wherein the step of performing point cloud coarse registration on the original CAD solid model of the workpiece and the STL model of the workpiece by using a fast point feature histogram algorithm comprises:

1) For the target point cloud Q and the source point cloud P, using formula (3), calculating the FPFH of the point cloud set, that is:

in the formula, ω _i Is a query point p _q And the adjacent point p _i K is the average number of points in each point neighborhood, SPFH (p) _i ) And SPFH (p) _q ) Is respectively p _i And p _q The value corresponding to the simplified characteristic histogram of the query point is obtained;

2) Acquiring sampling points: sampling the source point cloud P to obtain m points, and limiting the minimum distance between sampling points in order to enable the sampling point space to cover the point cloud set as much as possible;

3) Finding point pairs: searching one or more corresponding points close to the FPFH (field programmable gate flash) in the target point cloud according to the sampling points in the source point cloud, and randomly selecting one corresponding point as a spatial transformation point pair;

4) Spatial transformation: and performing rigid body transformation matrix calculation according to the selected point pairs, and performing space transformation on the source point cloud to complete point cloud rough registration of the workpiece original point cloud model and the workpiece STL model.

9. The industrial CT image-based three-dimensional CAD solid model reconstruction method for workpieces with internal defects according to claim 7, characterized in that the step of performing point cloud fine registration on the workpiece original point cloud model and the workpiece STL model after the first rigid body transformation by using the iterative closest point algorithm comprises:

1) The point cloud set of the workpiece STL model after the first rigid body transformation is recorded as

The point cloud set of the original point cloud model of the workpiece is

N _p 、N _x The number is counted;

2) For point P _i Searching the nearest point in the point cloud set X as a corresponding point, and writing the corresponding point set

The preparation method comprises the following steps of (1) performing; n is a radical of _q Counting the number of points;

3) Calculating a rotation transformation matrix R and a translation transformation matrix T of the fine registration according to the corresponding point set Q;

4) Performing rigid body transformation on the internal defect STL model after the first rigid body transformation according to the rotation transformation matrix R and the translation transformation matrix T of the fine registration;

5) Calculating error change quantity E of two adjacent transformations _k If the error changes the quantity E _k If the value is less than the given threshold value tau, finishing the registration process, otherwise, judging whether the iteration number k reaches the maximum iteration number, if so, finishing the registration process, otherwise, enabling the iteration number k = k +1, and returning to the step 1);

wherein, the error measure function E (R, T) after the current rigid body transformation is as follows:

in the formula, R represents a rotation transformation matrix of the fine registration, and T represents a translation transformation matrix.

10. The industrial CT image-based three-dimensional CAD solid model reconstruction method for workpieces with internal defects according to claim 7, wherein the step of generating CAD solid models for workpieces with internal defects comprises:

1) Carrying out topological reconstruction on the workpiece STL model after space transformation to obtain an internal defect CAD entity model, wherein the steps comprise:

1.1 Using a three-axis block sorting algorithm to delete the repeated vertexes, comprising the steps of:

1.1.1 Create vertex array X, Y, Z, input internal defect STL model, read three coordinate values (x, y, z) of one vertex at a time;

1.1.2 Determine whether the vertex has been saved based on the coordinate values: sequentially judging the values of the coordinates x, Y and z, if the value does not exist in the corresponding vertex array, entering the step 1.1.3), if the value of the coordinate x exists, turning to the Y array pointed by the coordinate x to judge whether the coordinate Y exists, and otherwise, entering the step 1.1.3);

if the coordinate y exists, judging whether the coordinate Z exists by turning to a Z array pointed by the coordinate y, otherwise, entering the step 1.1.3);

if the coordinate z exists, the current coordinate value (x, y, z) is not saved, and 1.1.1 is returned, otherwise, the step 3 is carried out;

1.1.3 Storing the coordinate values (X, Y, Z) in corresponding X, Y, Z arrays and sorting;

the sequencing rule is as follows: after the coordinate X is saved, sorting the X arrays, and correspondingly transforming the Y, Z array pointed by the coordinate X; after saving the coordinate Y, sorting the Y arrays, and correspondingly converting the Z arrays pointed by the coordinate xy; after saving the coordinates Z, the Z array is sorted.

1.2 Create vertex, edge, face data tables;

wherein the step of creating the vertex data table comprises: determining a coordinate y and a coordinate z corresponding to each coordinate X according to the starting X array, and adding the coordinates as new vertex coordinates (X, y, z) into a vertex table;

the step of creating the edge data table includes: creating edges in pairs according to the coordinates of the three vertexes of each grid, judging whether the created edges are in an edge data table, if not, establishing indexes of the edges and the corresponding vertexes in the vertex table, creating edge information, and adding the created edges into the edge data table;

the step of creating a surface data table includes: adding the serial number of the surface, the corresponding normal vector thereof, three vertexes forming the surface and indexes of the three edges into a surface data table together as surface information;

1.3 Using the reconstructed topological relation among the triangular surface patches and taking the edges as the core, performing entity reconstruction on all triangular vertexes, generating triangular surfaces by using the points and the edges, combining all the triangular surfaces into a whole to obtain a closed space boundary, and generating a defect CAD entity model according to the closed boundary surface;

2) And performing Boolean operation on the internal defect CAD entity model and the original CAD entity model of the workpiece to generate the CAD entity model of the workpiece with the defects.

Images