Liquid rocket engine structure deformation measurement method based on optical image observation

Abstract

The application relates to a liquid rocket engine structure deformation measurement method based on optical image observation. The method uses a high-precision industrial camera for observation, utilizes a binocular vision depth estimation principle to measure the displacement of multiple key points, and is calibrated on a displacement simulation platform, so that the liquid rocket engine structure deformation measurement is realized. The method has not only a faster detection speed, but also a higher detection accuracy, and is a liquid rocket engine structure deformation measurement method which has detection speed and accuracy and has higher practical application value.

Description

Technical Field

[0001] This invention belongs to the field of intelligent optoelectronic and optical measurement, specifically relating to a method for measuring structural deformation under static test scenarios of liquid rocket engines. Background Technology

[0002] The structural performance of a liquid rocket engine determines the payload and operating conditions of a launch vehicle. Therefore, before assembling a liquid rocket engine, a static test is required to verify its mechanical properties. Static tests of liquid rocket engines study and verify their strength characteristics under stress by applying static loads. Classic methods for measuring structural deformation include laser measurement and strain gauge measurement.

[0003] The above methods all achieve the purpose of structural deformation measurement by measuring the displacement and strain of key locations under static force. Most of them are contact measurement methods, which require setting up instruments and connecting systems on the surface of the object to be measured in advance. They have limitations such as cumbersome preparation work, limited number of measurement points, and interference from environmental factors, making it difficult to achieve all-round measurement of large-sized objects.

[0004] In recent years, with the continuous development of computing devices and intelligent optoelectronic technologies, measurement technologies based on optical image observation have been applied in various fields such as aerospace, healthcare, and intelligent manufacturing. Among them, stereo optical measurement systems composed of multiple camera groups have attracted widespread attention due to their high precision, ease of operation, and repeatability. Summary of the Invention

[0005] Technical problems to be solved

[0006] To avoid the shortcomings of existing technologies, this invention proposes a method for measuring the structural deformation of liquid rocket engines based on optical image observation.

[0007] Technical solution

[0008] A method for measuring structural deformation of a liquid rocket engine based on optical image observation is characterized by: designing multiple binocular vision systems to detect multi-point displacement based on the actual needs of multi-point displacement measurement of key points in the liquid rocket engine; when the binocular vision system is activated, firstly, the optical image of the object to be measured is imaged onto the image sensor of an industrial camera through the lens; then, the image sensor converts the light signal into an analog electrical signal, and the analog signal is converted into a digital signal by a digital-to-analog converter; then, the digital information is processed by an image processor and stored in a memory; finally, it is input through a digital interface or a video interface; the steps are as follows:

[0009] Step 1: The binocular vision system is calibrated using the Zhang Zhengyou calibration method;

[0010] Step 2: Acquire multiple field-of-view images simultaneously using hard triggering. Each field of view contains multiple target points, and the images acquired by different cameras are numbered and distinguished.

[0011] Step 3: Perform distortion correction and stereo correction based on the parameters obtained from the camera group calibration, so that the corresponding epipolar lines of the images captured by the left and right cameras are on the same horizontal line.

[0012] Step 4: For the corrected image, delineate the region of interest, calculate the disparity value of the target point, obtain the true coordinates, and calculate the displacement value based on the position changes at different times;

[0013] Step 5: By analyzing the static test scenario of the rocket engine, a multi-point displacement simulation system is constructed using multiple electric triaxial displacement platforms to simulate the displacement of key points in the static test. The system is calibrated and optimized. The system is tested in a real static test, and the displacement data measured by the system is integrated. Three-dimensional graphics are constructed using software, and the system is visualized using the platform to realize the measurement of structural deformation of the liquid rocket engine.

[0014] A further technical solution of the present invention: Step 1 is as follows:

[0015] 1a: A checkerboard calibration plate made of float glass is used, with a checkerboard size of 20mm and a side length manufacturing error of less than 0.01mm;

[0016] 1b: Use a fixed device to place the chessboard calibration board, keep the camera position fixed, move the calibration board and use the camera to take images of it from multiple angles to obtain the corresponding images from the left and right perspectives of the binocular camera;

[0017] 1c: The paired left and right images are input into the software system, and distortion correction is performed to obtain a preliminary intrinsic and extrinsic parameter matrix;

[0018] 1d: Remove calibration images with reprojection errors greater than 0.1 pixels to ensure the reprojection error is within 0.1 pixels, resulting in the intrinsic and extrinsic parameter matrix and distortion coefficient matrix as shown in formula (1):

[0019]

[0020] Where f x ,f y c represents focal length. x ,c y Indicates the coordinates of the principal point. The rotation matrix describes the orientation of the world coordinate system axes relative to the camera coordinate axes. The translation vector describes the coordinates of the origin in the camera coordinate system, where k1 and k2 are distortion coefficients.

[0021] A further technical solution of the present invention: Step 3 is as follows:

[0022] 3a: Perform distortion correction on the acquired left and right camera images respectively, and use the known image pixels... Substitute the coordinates into formulas (2) and (3) to calculate the corrected pixel coordinates (u,v) and obtain the image after distortion correction. The process is as follows:

[0023]

[0024]

[0025] Where (u,v) represents the pixel coordinates after distortion correction. (x, y) represents the pixel coordinates under actual radial distortion, (x, y) represents the continuous image coordinates under ideal distortion-free conditions, (u0, v0) represents the camera principal point, and k1, k2 represent the distortion coefficients.

[0026] 3b: Perform stereoscopic correction on the left and right camera images after distortion correction, in the following steps:

[0027] 1): Calculate the optical center positions c1 and c2 for the left and right cameras respectively, as shown in formulas (4) and (5). Substitute the intrinsic and extrinsic parameters R and t obtained from formula (1) into the equation:

[0028]

[0029]

[0030] Where c1 and c2 represent the optical center positions of the left and right cameras, respectively; R1 and R2 represent the rotation matrices of the left and right cameras, respectively; and K1 and K2 represent the camera intrinsic parameter matrices, respectively. These represent the translation matrices for the left and right cameras, respectively.

[0031] 2): Determine the rotation matrix of the new camera. Using the results obtained from formulas (4) and (5), substitute them into the following formula to calculate the rotation matrix R of the new camera. n :

[0032]

[0033] in r2=R1(3,:)×r1, r3=r1×r2;

[0034] 3): Find the intrinsic parameter matrix K of the new camera. n :

[0035]

[0036] 4): Find the transformation matrices T1 and T2:

[0037]

[0038]

[0039] 5): Transform the two images using transformation matrices T1 and T2 respectively to obtain the stereo-corrected image.

[0040] A further technical solution of the present invention: Step 4 is as follows:

[0041] 4a: Define the regions of interest on the corrected left and right camera images respectively, and determine the position of the corner target center in the image. The steps are as follows:

[0042] 1): Define the region of interest from the acquired images;

[0043] 2): Filter the region of interest to remove noise;

[0044] 3): Use the OTSU algorithm to detect the foreground region to further narrow down the search range, and use contour detection to draw the maximum bounding rectangle;

[0045] 4): Define the search mask and perform etching treatment;

[0046] 5): Perform sub-pixel corner detection on the image and select the mode of the disparity as the final disparity;

[0047] (4b): Obtain the pixel position of the target point based on the images from the left and right cameras, and then obtain the disparity(x). r ,x l The true coordinates (x, y, z) of the target point are calculated using the projection matrix.

[0048] The calculation process is as follows, and the formulas are shown in (10), (11), and (12):

[0049] [X,Y,Z,W] T =Q×[x,y,disparity(x)] r ,x l ),1] T (10)

[0050]

[0051] in:

[0052]

[0053] c x ,c y Let f be the coordinates of the principal point of the left camera in the image, and T be the focal length. xc′ represents the translation between the projection centers of the two cameras. x These are the coordinates of the right camera's principal point in the image. Thus, the keypoint coordinates (x, y, z) can be obtained from the above formula.

[0054] (4c): By taking images of the target point at different times, and using the already obtained true coordinates X = (x, y, z), the displacement value is calculated. Assuming that time t1 > t2, the displacement X can be obtained, as shown in formula (13):

[0055]

[0056] (4d): Repeat steps 1 to 3 for each target point of the camera group.

[0057] Beneficial effects

[0058] This invention proposes a method for measuring structural deformation of liquid rocket engines based on optical image observation. Addressing the requirements for structural deformation measurement in static testing scenarios of liquid rocket engines, it introduces computer vision methods to measure the displacement of key points in static tests. The main advantages are non-contact, repeatability, and ease of operation, significantly reducing manpower and material resources required in the testing environment. First, after calibrating the camera group using the Zhang Zhengyou calibration method, optical markers are attached to key points, and a spatial coordinate calculation algorithm is used to detect the movement of these markers, thereby simulating the displacement of the key points. This algorithm has two advantages: in terms of measurement efficiency, by designing a sparse parallax calculation algorithm and defining the user's region of interest to reduce the algorithm's computation time, the measurement time for a single camera group is optimized to 0.2s / test, meeting real-time measurement requirements. In terms of measurement accuracy, by improving the threshold segmentation algorithm and the sub-pixel positioning algorithm for the geometric center of the optical markers, a measurement resolution of 0.05mm / m can be achieved, meeting the displacement measurement requirements of static tests. Next, by designing an electric simulated displacement platform, the repeatability of the experiment is increased, and system calibration and optimization are achieved. Simultaneously, the obtained displacement data is visualized, making it easy to verify the accuracy of the structural deformation measurement method. This invention overcomes the shortcomings of traditional contact measurement methods, such as high operational difficulty, lack of repeatability, and cumbersome setup, enabling faster detection speeds and non-contact measurement. Attached Figure Description

[0059] The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Throughout the drawings, the same reference numerals denote the same parts.

[0060] Figure 1 This is a flowchart of the method of the present invention;

[0061] Figure 2 A diagram illustrating a binocular vision system;

[0062] Figure 3A schematic diagram of a multi-camera setup;

[0063] Figure 4 This is a 3D correction result diagram;

[0064] Figure 5 Define the region of interest map for the left camera image;

[0065] Figure 6 This is a diagram showing the corner detection results;

[0066] Figure 7 This is a diagram illustrating an electric simulation displacement platform.

[0067] Figure 8 The image shows the effect of verifying the deployment of a multi-camera group based on an electric displacement platform.

[0068] Figure 9 This is a diagram illustrating structural deformation. Detailed Implementation

[0069] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention. Furthermore, the technical features involved in the various embodiments of this invention described below can be combined with each other as long as they do not conflict with each other.

[0070] like Figure 1 This is a flowchart illustrating the overall process of measuring structural deformation of a liquid rocket engine based on optical image observation.

[0071] Step 1: Reduce repeated test errors by designing a standardized calibration process to obtain the internal and external parameters of the high-resolution industrial camera.

[0072] like Figure 2 The diagram shows the binocular vision system used in this invention. It employs an industrial camera, model MV-CH120-10UM, with a frame rate of 23 frames per second, a focal length of 16mm, and a resolution of 4096×3000 pixels. Based on the actual needs of multi-point displacement measurement at key points in liquid rocket engines, this invention designs multiple sets of binocular vision systems to detect multi-point displacement. When the binocular vision system is activated, the optical image of the object being measured is first imaged onto the image sensor of the industrial camera through the lens. Then, the image sensor converts the light signal into an analog electrical signal, which is then converted into a digital signal by a digital-to-analog converter. The digital information is then processed by an image processor and stored in a memory. Finally, the signal is input through a digital interface or video interface.

[0073] This invention employs the Zhang Zhengyou calibration method based on binocular vision, utilizing a checkerboard pattern for calibration. However, in actual calibration, issues such as checkerboard movement, uneven surface, and unclear grid lines can easily increase calibration errors, affecting the overall experimental results. Furthermore, if the image of the calibration board is partially missing or taken from a single angle during calibration, it will lead to significant errors in the calculated camera intrinsic and extrinsic parameters. Therefore, this invention designs a calibration process based on binocular optical image correction using the Zhang Zhengyou calibration method to reduce calibration errors. The specific steps are as follows:

[0074] (1a): A checkerboard calibration plate made of float glass is used, with a checkerboard size of 20mm and a side length manufacturing error of less than 0.01mm. Using this calibration plate can reduce the influence of diffuse reflection of light, and strict control of side length error can ensure the accuracy of calibration parameters;

[0075] (1b): Use a fixed device to place the chessboard calibration board, keep the camera position unchanged, move the calibration board and use the camera to take images of it from multiple angles. The purpose of this operation is to correct lens distortion. Therefore, it is expected to take images of the calibration board at different tilt angles. The calibration process requires the acquisition of 20 images to obtain the corresponding images from the left and right perspectives of the binocular camera.

[0076] (1c): The left and right images are input into the software system (Matlab) to perform distortion correction and obtain the preliminary intrinsic and extrinsic parameter matrix;

[0077] (1d): Reprojection error is an important parameter for judging the accuracy of calibration results. According to theoretical calculations, in order to meet the measurement accuracy requirements, the reprojection error of the calibration results must be less than 0.1 pixels. Therefore, calibration images with reprojection errors greater than 0.1 pixels need to be removed to ensure that the reprojection error is within the range of 0.1 pixels. The internal and external parameter matrix and distortion coefficient matrix are obtained as shown in formula (1). Take a camera as an example.

[0078]

[0079] Where f x ,f y c represents focal length. x ,c y Represents the coordinates of the principal point (relative to the coordinate plane). The rotation matrix describes the orientation of the world coordinate system axes relative to the camera coordinate axes. The translation vector describes the coordinates of the origin in the camera coordinate system, where k1 and k2 are distortion coefficients.

[0080] Step 2: Acquire multiple field-of-view images simultaneously using hard triggering. Each field of view contains multiple target points, and different field-of-view images are numbered to distinguish them.

[0081] like Figure 3 The diagram shows a multi-camera group setup constructed according to actual working conditions. This method uses multiple camera groups to build a joint camera network that can cover target points on multiple different planes. Each camera group captures all target points within the depth of field, meeting the requirements of key points on different planes in the static test of the rocket engine. A 15Hz square wave signal is generated using an FY6900 signal generator, and multiple cameras are simultaneously controlled to capture images using hard triggering. The acquired images are numbered and distinguished according to different fields of view and then read into a Python program.

[0082] Step 3: Perform distortion correction and stereo correction based on the parameters obtained from the camera group calibration, so that the corresponding epipolar lines of the images captured by the left and right cameras are on the same horizontal line.

[0083] This invention uses a set of cameras as an example, with each set of cameras employing the same steps to determine the target point displacement. The steps are as follows:

[0084] (3a): Perform distortion correction on the acquired left and right camera images respectively, and use the known image pixels Substitute the coordinates into formulas (2) and (3) to calculate the corrected pixel coordinates (u,v) and obtain the image after distortion correction. The process is as follows:

[0085]

[0086]

[0087] Where (u,v) represents the pixel coordinates after distortion correction. (x, y) represents the pixel coordinates under actual radial distortion, (x, y) represents the continuous image coordinates under ideal distortion-free conditions, (u0, v0) represents the camera principal point, and k1, k2 represent the distortion coefficients.

[0088] (3b): Stereoscopic correction is performed on the left and right camera images after distortion correction. The stereoscopically corrected left and right images are as follows: Figure 4 As shown, the steps are as follows:

[0089] 1): Calculate the optical center positions c1 and c2 for the left and right cameras respectively, as shown in formulas (4) and (5). Substitute the intrinsic and extrinsic parameters R and t obtained by formula (1) into the equation.

[0090]

[0091]

[0092] Where c1 and c2 represent the optical center positions of the left and right cameras, respectively; R1 and R2 represent the rotation matrices of the left and right cameras, respectively; and K1 and K2 represent the camera intrinsic parameter matrices, respectively. These represent the translation matrices for the left and right cameras, respectively.

[0093] 2): Determine the rotation matrix of the new camera. Using the results obtained from formulas (4) and (5), substitute them into the following formula to calculate the rotation matrix R of the new camera. n .

[0094]

[0095] in r2=R1(3,:)×r1, r3=r1×r2.

[0096] 3): Find the intrinsic parameter matrix K of the new camera. n As shown in formula (7).

[0097]

[0098] 4): Find the transformation matrices T1 and T2, as shown in formulas (8) and (9).

[0099]

[0100]

[0101] 5): Transform the two images using transformation matrices T1 and T2 respectively to obtain the stereo-corrected image.

[0102] Step 4: For the corrected image, delineate the region of interest, calculate the disparity value of the target point, solve for the spatial coordinates of the optical marker point, and calculate the displacement value based on the position changes at different times.

[0103] (4a): Define the regions of interest on the corrected left and right camera images respectively, and determine the position of the center of the corner target in the image. The steps are as follows:

[0104] 1): Define the region of interest from the acquired images. Taking the left camera as an example, for instance... Figure 5 As shown, the same principle applies to the right camera, defining the region of interest for the same target point.

[0105] 2): Filter the region of interest to remove noise.

[0106] 3): Use the OTSU algorithm to detect the foreground region to further narrow down the search range, and use contour detection to draw the largest bounding rectangle.

[0107] 4): Define the search mask and perform etching.

[0108] 5): Perform sub-pixel corner detection on the image, and select the mode of the disparity as the final disparity. The result is as follows: Figure 6 As shown in the image, the white dots represent testing centers.

[0109] (4b): Obtain the pixel position of the target point based on the images from the left and right cameras, and then obtain the disparity(x). r ,x l The true coordinates (x, y, z) of the target point are calculated using the projection matrix.

[0110] The calculation process is as follows, and the formulas are shown in (10), (11), and (12):

[0111] [X,Y,Z,W] T =Q×[x,y,disparity(x)] r ,x l ),1] T (10)

[0112]

[0113] in:

[0114]

[0115] c x ,c y Let f be the coordinates of the principal point of the left camera in the image, and T be the focal length. x c′ represents the translation (negative value) between the projection centers of the two cameras. x The coordinates of the right camera's principal point in the image are given, and the keypoint coordinates (x, y, z) can be obtained from the above formula.

[0116] (4c): By taking images of the target point at different times, and using the already obtained true coordinates X = (x, y, z), the displacement value is calculated. Assuming that time t1 > t2, the displacement X can be obtained, as shown in formula (13):

[0117]

[0118] (4d): Repeat steps 1 to 3 for each target point of the camera group.

[0119] Step 5: By analyzing the static test scenario of the liquid rocket engine, a multi-point displacement simulation system is constructed using multiple electric triaxial displacement platforms to simulate the displacement of key points in the static test, and the system is calibrated. The displacement data measured by the system is then integrated, and a 3D graphic is constructed using software, which is then visualized using the platform.

[0120] Due to the complexity and non-repeatability of static experiments on liquid rocket engines, and considering the measurability of visual displacement monitoring results and the economy of repeating experiments, a displacement platform was constructed in the laboratory to simulate the displacement of key points in the static experiment of liquid rocket engines.

[0121] In this invention, the testing requirements for liquid engines, such as... Figure 7 As shown, the left side represents the dedicated controller, and the right side represents the three-dimensional displacement platform. A solution using an electric displacement platform is proposed, employing an XYZ three-axis motion platform, electrically driven, and controlled in real-time using a Zolihan SC300 controller or a PLC controller, accurately recording the displacement along each axis. The XYZ axis range of this displacement stage is 20mm, and the repeatability positioning accuracy is less than or equal to 1.5μm, thus meeting the experimental requirements of this invention.

[0122] This invention verifies the accuracy of target point movement in different planes using an electric displacement stage, proving the effectiveness of the method. Due to the testing requirements of liquid engine systems, simultaneous detection of multiple target points on different planes is necessary; therefore, this method... Figure 8 As shown, the left camera group captures two target points on the same plane, while the right camera group captures three target points on the same plane. However, the target points captured by the left and right camera groups are not in the same plane. Deploying multiple camera groups and an electric displacement stage can simulate the experimental environment and meet actual needs. Therefore, by optimizing the measurement accuracy of the displacement stage, it can be transferred to real-world application scenarios.

[0123] This method uses an electric displacement stage to drive the target point to generate displacement, and synchronously records images of the displacement of key points in multiple sets of cameras through hardware pulse current triggering. After steps 1 to 4 for each set of cameras, visualization software is used to obtain a visualization of the final target point, such as... Figure 9 As shown, the horizontal axis represents time in seconds (s), and the vertical axis represents displacement distance in millimeters (mm). This is a diagram illustrating multi-point displacement. By visually analyzing the displacements of different key points, the structural deformation of the liquid rocket engine can be observed based on the displacements at these key locations, facilitating subsequent testing of the liquid rocket engine's stress characteristics.

[0124] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the scope of the technology disclosed in the present invention, and such modifications or substitutions should all be covered within the scope of protection of the present invention.

Claims

1. A method for measuring structural deformation of a liquid rocket engine based on optical image observation, characterized in that: Based on the actual needs of multi-point displacement measurement of key points in liquid rocket engines, multiple binocular vision systems are designed to detect multi-point displacement. When the binocular vision system is activated, the optical image of the object to be measured is first imaged onto the image sensor of an industrial camera through the lens; then the image sensor converts the light signal into an analog electrical signal, and the analog signal is converted into a digital signal by a digital-to-analog converter; then the digital information is processed by an image processor and stored in memory; finally, it is input through a digital interface or video interface; the steps are as follows: Step 1: The binocular vision system is calibrated using Zhang Zhengyou's calibration method; details are as follows: 1a: A checkerboard calibration plate made of float glass is used, with a checkerboard size of 20mm and a side length manufacturing error of less than 0.01mm; 1b: Use a fixed device to place the chessboard calibration board, keep the camera position fixed, move the calibration board and use the camera to take images of it from multiple angles to obtain the corresponding images from the left and right perspectives of the binocular camera; 1c: The paired left and right images are input into the software system, and distortion correction is performed to obtain a preliminary intrinsic and extrinsic parameter matrix; 1d: Remove calibration images with reprojection errors greater than 0.1 pixels to ensure the reprojection error is within 0.1 pixels, and obtain the intrinsic and extrinsic parameter matrix and distortion coefficient matrix as shown in formula (1): （1） in Indicates focal length. Indicates the coordinates of the principal point. The rotation matrix describes the orientation of the world coordinate system axes relative to the camera coordinate axes. This represents the translation vector, describing the coordinates of the origin in the camera coordinate system. The distortion coefficient; Step 2: Acquire multiple field-of-view images simultaneously using hard triggering. Each field of view contains multiple target points, and the images acquired by different cameras are numbered and distinguished. Step 3: Perform distortion correction and stereo correction based on the parameters obtained from the camera group calibration, so that the corresponding epipolar lines of the images captured by the left and right cameras are on the same horizontal line; specifically as follows: 3a: Perform distortion correction on the acquired left and right camera images respectively, and use the known image pixels... Substituting the coordinates into formulas (2) and (3), the corrected pixel coordinates are calculated. The process is as follows: to obtain the image after distortion correction. （2） （3） in, Represents the pixel coordinates after distortion correction. Represents the pixel coordinates under actual radial distortion. Represents the continuous image coordinates under ideal, distortion-free conditions. Represents the principal point of the camera. Indicates the distortion coefficient; 3b: Perform stereoscopic correction on the left and right camera images after distortion correction, in the following steps: 1): Determine the optical center positions for the left and right cameras respectively. As shown in formulas (4) and (5), the intrinsic and extrinsic parameters obtained using formula (1) Substitute: （4） （5） in These represent the optical center positions of the left and right cameras, respectively. These represent the rotation matrices for the left and right cameras, respectively. These represent the camera intrinsic parameter matrices, These represent the translation matrices for the left and right cameras, respectively. 2): Determine the rotation matrix of the new camera. Using the results obtained from formulas (4) and (5), substitute them into the following formula to calculate the rotation matrix of the new camera. : （6） in , , ; 3): Find the intrinsic parameter matrix of the new camera. : （7） 4): Find the transformation matrix : （8） （9） 5): Using transformation matrix The two images are transformed separately to obtain the stereo-corrected image; Step 4: For the corrected image, define the region of interest, calculate the disparity value of the target point, obtain the true coordinates, and calculate the displacement value based on the position changes at different times; details are as follows: 4a: Define the regions of interest on the corrected left and right camera images respectively, and determine the position of the corner target center in the image. The steps are as follows: 1): Define the region of interest from the acquired images; 2): Filter the region of interest to remove noise; 3): Use the OTSU algorithm to detect the foreground region to further narrow down the search range, and use contour detection to draw the maximum bounding rectangle; 4): Define the search mask and perform etching treatment; 5): Perform subpixel corner detection on the image and select the mode of the disparity as the final disparity; (4b): Obtain the pixel position of the target point based on the images from the left and right cameras, and then obtain the parallax. The true coordinates of the target point are calculated using the projection matrix. : The calculation process is as follows, and the formulas are shown in (10), (11), and (12): （10） （11） in: （12） The coordinates of the left camera principal point in the image. Focal length For the translation between the projection centers of the two cameras, These are the coordinates of the principal point of the right camera in the image. The keypoint coordinates can then be calculated from the above formula. ; (4c): By taking images of the target point at different times, and using the already calculated true coordinates... Calculate its displacement value, assuming time... The displacement can then be obtained. As shown in formula (13): （13） (4d): Repeat steps 1 to 3 for each target point of the camera group; Step 5: By analyzing the static test scenario of the rocket engine, a multi-point displacement simulation system is constructed using multiple electric triaxial displacement platforms to simulate the displacement of key points in the static test. The system is calibrated and optimized. The system is tested in a real static test, and the displacement data measured by the system is integrated. Three-dimensional graphics are constructed using software, and the system is visualized using the platform to realize the measurement of structural deformation of the liquid rocket engine.

Images