Method for reconstructing three-dimensional pressure / temperature field of model surface based on contour feature matching

Abstract

The application relates to a three-dimensional pressure / temperature field reconstruction method based on contour feature matching, which comprises the following steps: S1, obtaining three-dimensional data of a model; S2, obtaining a two-dimensional pressure / temperature cloud chart of the model surface by using pressure-sensitive / temperature-sensitive paint technology; S3, extracting a three-dimensional contour of the model surface to obtain three-dimensional and two-dimensional point clouds; S4, estimating an initial value of a projection matrix; S5, obtaining a two-dimensional re-projection contour point cloud and calculating an average re-projection error; S6, if the average re-projection error is not greater than a threshold value, directly taking the result of S4 as an optimal projection matrix, executing S7, otherwise calculating a new projection matrix, then circulating S5 and S6 until the average re-projection error is not greater than the threshold value, and executing S7; S7, projecting the three-dimensional point cloud onto the pressure / temperature cloud chart according to the projection matrix, extracting pressure and temperature information, and completing three-dimensional pressure / temperature field reconstruction. Compared with the prior art, the application has the advantages of high reconstruction precision and strong robustness.

Description

Technical Field

[0001] This invention relates to the field of fluid dynamics testing technology, and in particular to a method for reconstructing the three-dimensional pressure / temperature field of a model surface based on contour feature matching. Background Technology

[0002] Pressure-sensitive paint (PSP) is an optical pressure measurement technique. This technique first involves spraying a pressure-sensitive paint onto the surface of the model under test, then using a specific wavelength of light to excite a special luminescent material within the coating, and finally capturing the luminescence image with a camera. Due to the "oxygen quenching" effect, the photoluminescence process of PSP is affected by the partial pressure of oxygen in the environment. Based on this effect, a functional relationship can be established between the emitted light intensity or lifetime of PSP and the surrounding air pressure. Compared with traditional contact pressure measurement methods, PSP technology has many advantages, including high spatial resolution, low cost, and good model adaptability, and is currently widely used in aerodynamic experiments. Temperature-sensitive paint (TSP) is a sister technology to PSP. Certain coatings exhibit a "thermal quenching" effect during their luminescence process, thus demonstrating temperature sensitivity. The measurement principle and system of TSP are very similar to those of PSP, enabling the measurement of the entire surface temperature of the model.

[0003] For test models with complex three-dimensional shapes, PSP / TSP technology directly obtains a two-dimensional pressure / temperature contour map of the model. This results in the loss of one dimension of information during imaging. Furthermore, for experiments with limited optical windows, a relatively tilted shooting angle is necessary, leading to significant projection distortion in the two-dimensional image. While the pressure or temperature distribution on the model surface can be observed from the two-dimensional pressure / temperature contour map, extracting this information is extremely difficult. The pixels on the two-dimensional pressure / temperature contour map are also difficult to align with the computational grid of numerical simulation, making result comparison impossible. Therefore, there is an urgent need to map the two-dimensional pressure / temperature results to the pressure / temperature information of the three-dimensional model surface to achieve accurate reconstruction of the three-dimensional pressure / temperature field.

[0004] Currently, existing technologies both domestically and internationally utilize feature points on the model surface to reconstruct three-dimensional pressure / temperature fields. A paper published in the journal *Experiments in Fluids*, titled "Image registration for pressure-sensitive paint applications," demonstrates how to use the projection equations of photogrammetry to correlate the three-dimensional coordinates of an experimental model with the two-dimensional coordinates of an image. The authors applied this method to pressure measurements on a scaled-down model of the XB-70 Valkyrie bomber. Since the aircraft model surface lacked sufficient natural feature points, artificial markers were placed on the model. The authors point out that the positioning accuracy of these feature points is the main source of error in solving the transformation coefficients of the projection equations. Furthermore, the contrast of the artificial markers changes during the experiment due to variations in the luminescence of the pressure-sensitive paint, making automatic feature point recognition unreliable. Sometimes, manual marking of feature points in the image is still necessary, making it difficult to guarantee recognition accuracy.

[0005] A paper published in the *AIAA Journal*, titled "Three-Dimensional Boundary-Layer Transition on a Swept Wing at Mach 3.5," developed a data processing system that can map a two-dimensional temperature field measured by a temperature-sensitive coating onto a three-dimensional model surface. This system uses photogrammetry to reconstruct the three-dimensional temperature field of a swept-wing model surface at Mach 3.4, facilitating the observation of the transition line position. The system also employs a feature-point-based reconstruction algorithm, selecting pressure-measuring holes and thermocouple placement points on the model surface to calculate the projection correspondence between the three-dimensional model and the two-dimensional image. The model, only 40 cm long, is relatively flat and wide, allowing for a large number of pressure-measuring holes and thermocouples to be placed on its surface. However, some complex models with thin walls and curved geometries cannot provide a large number of identifiable feature points, leading to a significant decrease in the accuracy of the three-dimensional pressure / temperature field reconstruction.

[0006] The paper "Image Data Processing Technology for Pressure-Sensitive Paint" published in the journal *Experimental Fluid Mechanics* also employs a three-dimensional pressure / temperature field reconstruction method based on feature points on the model surface. It utilizes the Moravec operator to extract feature points based on the grayscale variance of the image, and then determines the relationship between pixels and spatial points on the model by solving the projection transformation equation in photogrammetry. This method is applied to a delta wing model with a simple three-dimensional structure, placed in a wind tunnel during experiments, allowing the camera to capture the luminous image of the PSP from directly above, resulting in relatively high accuracy and precision in feature point recognition. However, in some PSP experiments where optical path arrangement is difficult (e.g., planar blade cascade PSP experiments), the shape and grayscale of the feature points change due to the tilted viewing angle, severely affecting the accuracy of feature point recognition; in such cases, this method cannot be applied.

[0007] Chinese patent CN 103217238 A discloses a high-precision display method for pressure-sensitive coating measurement results. To improve the accuracy of three-dimensional pressure / temperature field reconstruction, this method proposes to refine the model surface into partitions, dividing it into at least five large regions, each of which is further uniformly divided into at least five smaller regions. This method requires at least six marker points to be uniformly marked within each smaller region, and then calculates coordinate transformation coefficients for each region to finally reconstruct the pressure distribution on the model's three-dimensional surface. However, this method requires a very large number of feature points, has high requirements for feature point distribution, and does not consider the decrease in feature point recognition accuracy under tilted viewing angles.

[0008] In summary, the currently widely used feature-point-based 3D reconstruction methods cannot achieve accurate reconstruction of 3D pressure / temperature fields when the number of feature points is small, their distribution is uneven, and the recognition accuracy is reduced due to tilted viewpoints. Summary of the Invention

[0009] The purpose of this invention is to provide a more accurate and robust method for reconstructing the three-dimensional pressure / temperature field of a model surface based on contour feature matching, under conditions such as a limited number of feature points, imperfect distribution, and tilted viewing angle.

[0010] The objective of this invention can be achieved through the following technical solutions:

[0011] A method for reconstructing the three-dimensional pressure / temperature field of a model surface based on contour feature matching, the method comprising the following steps:

[0012] S1. Obtain the three-dimensional data of the model;

[0013] S2. Spray pressure-sensitive or temperature-sensitive coatings onto the model surface, activate the coatings under experimental conditions, and obtain a two-dimensional pressure cloud map corresponding to the pressure-sensitive coating or a two-dimensional temperature cloud map corresponding to the temperature-sensitive coating after calibration and image post-processing.

[0014] S3. Extract the three-dimensional contour of the model surface and discretize the contour lines to form a three-dimensional point cloud P. At the same time, find the two-dimensional contour corresponding to the three-dimensional point cloud on the cloud map to obtain the two-dimensional point cloud p.

[0015] S4. Estimate the initial value of the projection matrix based on the feature points on the model surface;

[0016] S5. Project the 3D point cloud onto the image coordinate system based on the projection matrix to obtain the 2D reprojected contour point cloud. Traverse the points of the 2D reprojected contour point cloud, find the point in the 2D point cloud that is closest to the current point in the 2D reprojected contour point cloud, and form a point cloud. Calculate the average reprojection error, where the projection matrix represents the projection transformation relationship between 3D points on the model and their corresponding points in the 2D image, and the number of contour points in the 3D point cloud P and the number of 2D reprojected contour points are calculated. The number of contour points is the same, n;

[0017] S6. If the average reprojection error is not greater than the threshold, then the projection matrix of S4 is directly used as the optimal projection matrix, and S7 is executed; otherwise, the nearest point cloud obtained by searching is used. Calculate a new projection matrix with the 3D point cloud P, then loop S5 and S6 until the average reprojection error is no greater than the threshold, and then execute S7.

[0018] S7. Based on the obtained optimal projection matrix, project the 3D point cloud of the model surface onto the cloud map to complete the reconstruction.

[0019] Furthermore, the average reprojection error of S5 is:

[0020]

[0021] Where L represents the projection matrix. This represents a two-dimensional reprojected contour point cloud. This represents the two-dimensional point cloud p and the current two-dimensional reprojected contour point cloud. The point cloud is formed by the nearest point, where n represents the point cloud. The number of contour points contained therein.

[0022] Furthermore, the optimization process for solving the projection matrix is ​​as follows:

[0023]

[0024]

[0025] Among them, P iLet (u, v) be the i-th point in the 3D point cloud, (u, v) be the 2D coordinates of the contour feature point, and l be a 9-dimensional column vector, in which the constituent elements l i It is the i-th row of the projection matrix L. Represents the i-th reprojection point in the two-dimensional contour point cloud p. The two-dimensional coordinates of the nearest point.

[0026] Furthermore, after introducing weights, the optimization process for solving the projection matrix is ​​as follows:

[0027]

[0028]

[0029] Among them, P i Let (u, v) be the i-th point in the 3D point cloud, (u, v) be the 2D coordinates of the feature point, and l be a 9-dimensional column vector, in which the constituent elements l i This is the i-th row of the projection matrix L, where L represents the projection matrix and n represents the point cloud. The number of contour points contained therein, where W is the weight matrix. Represents the i-th reprojection point in the two-dimensional contour point cloud p. The two-dimensional coordinates of the nearest point.

[0030] Furthermore, the elements w of the weight matrix W i The weight of the i-th point in the 3D point cloud is as follows:

[0031] w i =exp(-q i / q max )

[0032] Where, q i Let q be the curvature at the i-th point. max This represents the maximum curvature.

[0033] Furthermore, the specific steps for estimating the initial value of the projection matrix are as follows:

[0034] Select at least 6 feature points, with their 2D image coordinates denoted as (u, v) and their 3D spatial coordinates denoted as (x, v). w ,y w ,z w By solving the constraint equations of all feature points, a system of linear equations is obtained. The elements in the projection matrix are solved by the least squares method, and the solution is used as the initial value of the projection matrix.

[0035] Furthermore, the simultaneous system of linear equations is as follows:

[0036]

[0037] Among them, L1 all the way to L 12 is an element in the projection matrix L.

[0038] Furthermore, the selected feature points include pressure measurement orifice points and corner points on the blade profile.

[0039] Furthermore, the pressure-sensitive / temperature-sensitive coating is excited by using an excitation light source.

[0040] Furthermore, in S7, after projection, the pressure or temperature value at the corresponding pixel is taken to reconstruct the corresponding pressure or temperature field.

[0041] Compared with the prior art, the present invention has the following beneficial effects:

[0042] (1) The present invention can reconstruct the three-dimensional pressure field / temperature field of the model surface from the two-dimensional pressure / temperature cloud map measured by pressure / temperature sensitive coating technology, which can effectively improve the visualization of surface pressure / temperature measurement results, and is especially suitable for models with complex three-dimensional geometric shapes.

[0043] (2) This invention uses the characteristic contour of the model as the basis for calculating the reprojection matrix, without relying on the feature points on the model surface. Even when the number of feature points on the model surface is small, the accuracy is not high, and the recognition of feature points is difficult due to factors such as viewing angle tilt and lighting changes, resulting in decreased accuracy, it is still possible to achieve high-precision three-dimensional pressure / temperature field reconstruction. Attached Figure Description

[0044] Figure 1 This is a flowchart of the present invention;

[0045] Figure 2 This is a three-dimensional model of the leaf cascade.

[0046] Figure 3 This is a calibrated and post-processed two-dimensional pressure cloud map;

[0047] Figure 4 It is a two-dimensional outline of a leaf shape;

[0048] Figure 5 A comparison of the effects of reprojection using the projection matrix calculated using feature points and the projection matrix optimized using feature contours;

[0049] Figure 6 The reconstructed three-dimensional pressure field;

[0050] Figure 7 This is a pressure-location diagram. Detailed Implementation

[0051] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. These embodiments are based on the technical solution of the present invention and provide detailed implementation methods and specific operating procedures. However, the scope of protection of the present invention is not limited to the following embodiments.

[0052] This invention proposes a method for reconstructing the three-dimensional pressure / temperature field of a model surface based on contour feature matching. The flowchart of the method is as follows: Figure 1 As shown. This invention includes the following steps: First, acquire the three-dimensional data of the model surface. Then, spray pressure-sensitive coating or temperature-sensitive coating onto the model surface. Under experimental conditions, use an excitation light source to excite the pressure-sensitive coating or temperature-sensitive coating on the model surface. Use a camera to capture the luminous image of the model surface under experimental conditions. Calculate the light intensity ratio between the experimental image and the reference image pixel by pixel. Calculate a two-dimensional pressure / temperature cloud map based on the light intensity ratio-pressure / temperature calibration curve. Subsequently, extract the feature contour from the model's three-dimensional data and reproject it onto the two-dimensional image. Iterate the projection matrix continuously to gradually reduce the difference between the reprojected contour and the model's surface contour in the reference image, thereby obtaining the optimal projection matrix. Finally, use the iterated projection matrix to reproject the model's three-dimensional point cloud onto the two-dimensional pressure / temperature cloud map, extract the pressure or temperature values ​​of the corresponding pixels, and display the results in three-dimensional form, realizing the reconstruction of the three-dimensional pressure / temperature field. Compared with the prior art, this invention can accurately reconstruct the three-dimensional pressure / temperature field of the model surface even when the model surface lacks feature points or the feature point recognition accuracy is low due to factors such as tilted viewing angle.

[0053] Example 1:

[0054] In Example 1, the pressure field was reconstructed by using pressure-sensitive coating technology to measure the pressure on a turbine blade cascade model with high viscosity and severe shading, and then reconstructing the three-dimensional pressure field on its surface.

[0055] This invention includes the following steps:

[0056] Step S1: Obtain the 3D data of the model. In this embodiment, the 3D model of the cascade can be exported from CAD software. For example... Figure 2 As shown, the blade cascade model contains 6 turbine blades, which block each other, so the blade surface can only be photographed from the side.

[0057] Step S2: Pressure-sensitive coating is sprayed onto the surfaces of the two middle blades (blade 3 and blade 4) of the blade cascade model. Under experimental conditions, an excitation light source is used to excite the pressure-sensitive coating on the model surface. A camera is used to capture the luminescent image of the model surface under experimental conditions. After calibration and image post-processing, two-dimensional pressure cloud maps of the surfaces of blades 3 and 4 are obtained. The results are as follows: Figure 3 As shown;

[0058] Step S3: Extract the three-dimensional contour of the blade surface and discretize the contour lines to form a three-dimensional point cloud representation as follows:

[0059] P = {P w1 ,P w2 …P wn}

[0060] At the same time, find the corresponding two-dimensional outline of the leaf shape on the PSP image (such as...). Figure 4 As shown), the contour pixels form a two-dimensional point cloud:

[0061] p = {p1, p2, ... p} m}

[0062] Step S4: Estimate the initial value of the projection matrix based on the feature points on the blade surface. In this example, pressure measurement holes on blade surfaces 3 and 4 can be used as feature points, but these points are located on the same spatial plane, which would result in the coefficient matrix not being of full rank. Therefore, additional feature points are needed. Corner points on the blade profile are selected as supplementary points. All used feature points are... Figure 4 The feature points are marked with a cross symbol. The two-dimensional image coordinates of the feature points are denoted as (u, v), and their three-dimensional spatial coordinates are denoted as (x, v). w ,y w ,z w The relationship between two-dimensional and three-dimensional coordinates can be described using a projection matrix:

[0063]

[0064] Solving the constraint equations for all feature points simultaneously yields the following system of linear equations:

[0065]

[0066] The 12 elements of the projection matrix can be solved using the least squares method. However, since the identification of these feature points has a large error, the resulting projection matrix can only be used as an initial value, denoted as L. (0) ;

[0067] Step S5: Based on the initial values ​​of the projection matrix, project the 3D model contour point cloud onto the image coordinate system to obtain the 2D reprojected contour point cloud. Find the two-dimensional contour point p that is closest to the current reprojection point. And calculate the average reprojection error of n contour points:

[0068]

[0069] Step S6: If the average reprojection error is greater than the threshold, then use the current two-dimensional contour points. The new projection matrix L is calculated using the three-dimensional contour points P, and they satisfy the following relationship:

[0070]

[0071] Here, l is a 9-dimensional column vector, and its constituent elements are l i This is the i-th row of the projection matrix L. In this example, the blade end face profile is relatively complex. Although the number of points at the turning points of the profile curve is small, they determine the trend of the curve. Points on the flat profile do not contribute much to solving the projection matrix. To further improve the accuracy of solving the projection matrix, points at different positions on different profiles are assigned different weights. Combining the constraint equations formed by the weighted points, we can obtain the following system of equations:

[0072]

[0073] The weight matrix W is a 2n×2n diagonal matrix:

[0074] W = diag([w1 w1 w2 w2…w n w n ])

[0075] element w in the matrix i The weight corresponding to the i-th contour point is the ratio of the curvature of the current point to the maximum curvature in the point cloud, and it is mapped to the interval [0,1] using an exponential function with base e. The specific definition is as follows:

[0076] w i =exp(-q i / q max )

[0077] The curvature q of the contour points can be derived from the 3D data of the model. The optimized projection matrix L can be calculated using the least squares method. Steps 5 and 6 are repeated until the error between the reprojected point cloud and the contour point cloud is less than a threshold. The effects of reprojection using the projection matrix calculated with feature points and the projection matrix optimized with feature contours are compared, for example... Figure 5 As shown, the reprojection effect of the projection matrix obtained by solving the feature contour in this invention is more consistent with the actual leaf shape contour obtained by photography. Statistical calculations show that the average reprojection error of the contour points of blades 3 and 4 is reduced from 11.85 pixels and 9.02 pixels to 1.63 pixels and 0.74 pixels, respectively.

[0078] Step S7: Based on the projection matrix L obtained through iterative optimization, project the point cloud of the model surface onto the two-dimensional pressure cloud map measured by the pressure-sensitive coating technology, extract the pressure values ​​at the corresponding pixels, and realize the reconstruction of the three-dimensional pressure field. The reconstructed three-dimensional pressure field is as follows: Figure 6As shown. The pressure values ​​at any location on the model surface can be easily viewed from the reconstructed three-dimensional pressure field of this invention, and continuous pressure distributions can be extracted along different locations, such as... Figure 7 As shown, the pressure curve at 50% leaf height was extracted, and the results are in good agreement with the data from the pressure measurement well.

[0079] Example 2:

[0080] In Example 2, the temperature field was reconstructed by using temperature-sensitive coating technology to measure the surface temperature of the turbine blade cascade and reconstruct its three-dimensional temperature field, so as to observe the accurate location of the boundary layer transition on the blade surface.

[0081] This invention includes the following steps:

[0082] Step S1: Obtain the 3D data of the model. The 3D model of the cascade can be obtained by exporting from CAD software or using 3D scanning technology.

[0083] Step S2: Spray a temperature-sensitive coating onto the surface of the blade to be tested on the blade model. Under experimental conditions, use an excitation light source to excite the temperature-sensitive coating on the model surface. Use a camera to capture the luminescent image of the model surface under experimental conditions. Then, after calibration and image post-processing, obtain a two-dimensional temperature cloud map of the blade surface.

[0084] Step S3: Extract the three-dimensional contour of the blade surface and discretize the contour lines to form a three-dimensional point cloud representation as follows:

[0085] P = {P w1 ,P w2 …P wn}

[0086] Simultaneously, the corresponding two-dimensional contour of the leaf shape is found on the TSP image, and the contour pixels form a two-dimensional point cloud:

[0087] p = {p1, p2, ... p} m}

[0088] Step S4: Estimate the initial value of the projection matrix based on a small number of feature points on the blade surface. The relationship between the two-dimensional and three-dimensional coordinates of the blade can be described by the projection matrix:

[0089]

[0090] Solving the constraint equations for all feature points simultaneously yields the following system of linear equations:

[0091]

[0092] The 12 elements of the projection matrix can be solved using the least squares method. Since there are errors in the identification of these feature points, the resulting projection matrix can only be used as its initial value, denoted as L. (0) ;

[0093] Step S5: Based on the initial values ​​of the projection matrix, project the 3D model contour point cloud onto the image coordinate system to obtain the 2D reprojected contour point cloud. Find the two-dimensional contour point p that is closest to the current reprojection point. And calculate the average reprojection error of n contour points:

[0094]

[0095] Step S6: If the average reprojection error is greater than the threshold, then use the current two-dimensional contour points. The new projection matrix L is calculated using the three-dimensional contour points P, and they satisfy the following relationship:

[0096]

[0097] Here, l is a 9-dimensional column vector, and its constituent elements are l i This is the i-th row of the projection matrix L. The turbine blade end face profile is quite complex. Although the number of points where the profile curve inflects is small, they determine the trend of the curve. Points on flat profiles contribute little to solving the projection matrix. To further improve the accuracy of solving the projection matrix, points at different positions on different profiles are assigned different weights. Combining the constraint equations formed by the weighted points, we can obtain the following system of equations:

[0098]

[0099] The weight matrix W is a 2n×2n diagonal matrix:

[0100] W = diag([w1 w1 w2 w2…w n w n ])

[0101] element w in the matrix i The weight corresponding to the i-th contour point is the ratio of the curvature of the current point to the maximum curvature in the point cloud, and it is mapped to the interval [0,1] using an exponential function with base e. The specific definition is as follows:

[0102] w i =exp(-q i / q max )

[0103] The curvature q of the contour points can be derived from the 3D data of the model. The optimized projection matrix L can be calculated using the least squares method. Repeat steps 5 and 6 until the error between the reprojected point cloud and the contour point cloud is less than a threshold. Reproject using the projection matrix calculated from the feature points and the projection matrix optimized using the feature contours.

[0104] Step S7: Based on the projection matrix L obtained through iterative optimization, the point cloud of the model surface is projected onto the two-dimensional temperature cloud map measured by the temperature-sensitive coating technology. The temperature values ​​at the corresponding pixels are extracted to achieve three-dimensional temperature field reconstruction. From the three-dimensional temperature field reconstructed by this invention, the temperature values ​​at any location on the model surface can be easily viewed. The heat transfer rates of the laminar and turbulent regions on the turbine blade surface are different, resulting in changes in the blade surface temperature. This allows observation of the boundary layer transition location. Furthermore, information such as heat flow can be further inferred from the three-dimensional temperature field.

[0105] The preferred embodiments of the present invention have been described in detail above. It should be understood that those skilled in the art can make numerous modifications and variations based on the concept of the present invention without creative effort. Therefore, all technical solutions that can be obtained by those skilled in the art based on the concept of the present invention through logical analysis, reasoning, or limited experimentation on the basis of existing technology should be within the scope of protection defined by the claims.

Claims

1. A method for reconstructing the three-dimensional pressure / temperature field of a model surface based on contour feature matching, characterized in that, The method includes the following steps: S1. Obtain the three-dimensional data of the model; S2. Spray pressure-sensitive or temperature-sensitive coatings onto the model surface, activate the coatings under experimental conditions, and obtain a two-dimensional pressure cloud map corresponding to the pressure-sensitive coating or a two-dimensional temperature cloud map corresponding to the temperature-sensitive coating after calibration and image post-processing. S3. Extract the three-dimensional contour of the model surface and discretize the contour lines to form a three-dimensional point cloud P. At the same time, find the two-dimensional contour corresponding to the three-dimensional point cloud on the cloud map to obtain the two-dimensional point cloud p. S4. Estimate the initial value of the projection matrix based on the feature points on the model surface; S5. Project the 3D point cloud onto the image coordinate system based on the projection matrix to obtain the 2D reprojected contour point cloud. Traverse the points of the 2D reprojected contour point cloud, find the point in the 2D point cloud that is closest to the current point in the 2D reprojected contour point cloud, and form a point cloud. Calculate the average reprojection error, where the projection matrix represents the projection transformation relationship between 3D points on the model and their corresponding points in the 2D image, and the number of contour points in the 3D point cloud P and the number of 2D reprojected contour points are calculated. The number of contour points is the same, n; S6. If the average reprojection error is not greater than the threshold, then the projection matrix of S4 is directly used as the optimal projection matrix, and S7 is executed; otherwise, the nearest point cloud obtained by searching is used. Calculate a new projection matrix with the 3D point cloud P, then loop S5 and S6 until the average reprojection error is no greater than the threshold, and then execute S7. S7. Based on the obtained optimal projection matrix, project the 3D point cloud of the model surface onto the cloud map to complete the reconstruction.

2. The method for reconstructing a three-dimensional pressure / temperature field on a model surface based on contour feature matching according to claim 1, characterized in that, The average reprojection error of S5 is: Where L represents the projection matrix. Represents a two-dimensional reprojected contour point cloud. This represents the two-dimensional point cloud p and the current two-dimensional reprojected contour point cloud. The point cloud is formed by the nearest point, where n represents the point cloud. The number of contour points contained therein.

3. The method for reconstructing a three-dimensional pressure / temperature field on a model surface based on contour feature matching according to claim 1, characterized in that, The optimization process for solving the projection matrix is ​​as follows: Among them, P i Let (u, v) be the i-th point in the 3D point cloud, (u, v) be the 2D coordinates of the contour feature point, and l be a 9-dimensional column vector, in which the constituent elements l i It is the i-th row of the projection matrix L. Represents the i-th reprojection point in the two-dimensional contour point cloud p. The two-dimensional coordinates of the nearest point.

4. The method for reconstructing a three-dimensional pressure / temperature field on a model surface based on contour feature matching according to claim 1, characterized in that, After introducing weights, the optimization process for solving the projection matrix is ​​as follows: Among them, P i Let (u, v) be the i-th point in the 3D point cloud, (u, v) be the 2D coordinates of the feature point, and l be a 9-dimensional column vector, in which the constituent elements l i This is the i-th row of the projection matrix L, where L represents the projection matrix and n represents the point cloud. The number of contour points contained therein, where W is the weight matrix. Represents the i-th reprojection point in the two-dimensional contour point cloud p. The two-dimensional coordinates of the nearest point.

5. The method for reconstructing a three-dimensional pressure / temperature field on a model surface based on contour feature matching according to claim 4, characterized in that, The elements w of the weight matrix W i The weight of the i-th point in the 3D point cloud is as follows: w i =exp(-q i / q max ) Where, q i Let q be the curvature at the i-th point. max This represents the maximum curvature.

6. The method for reconstructing a three-dimensional pressure / temperature field on a model surface based on contour feature matching according to claim 1, characterized in that, The specific steps for estimating the initial value of the projection matrix are as follows: Select at least 6 feature points, with their 2D image coordinates denoted as (u, v) and their 3D spatial coordinates denoted as (x, v). w ,y w ,z w By solving the constraint equations of all feature points, a system of linear equations is obtained. The elements in the projection matrix are solved by the least squares method, and the solution is used as the initial value of the projection matrix.

7. The method for reconstructing a three-dimensional pressure / temperature field on a model surface based on contour feature matching according to claim 6, characterized in that, The simultaneous system of linear equations is as follows: Among them, L1 all the way to L 12 is an element in the projection matrix L.

8. The method for reconstructing a three-dimensional pressure / temperature field on a model surface based on contour feature matching according to claim 1, characterized in that, The selected feature points include pressure measurement orifice points and corner points on the blade profile.

9. The method for reconstructing a three-dimensional pressure / temperature field on a model surface based on contour feature matching according to claim 1, characterized in that, The pressure-sensitive / temperature-sensitive coating is excited by using an excitation light source.

10. The method for reconstructing a three-dimensional pressure / temperature field on a model surface based on contour feature matching according to claim 1, characterized in that, In S7, after projection, the pressure or temperature value at the corresponding pixel is taken, and the corresponding pressure or temperature field is reconstructed.

Images