Binocular vision measurement method and device for three-dimensional combustion surface of solid propellant

Abstract

This invention discloses a binocular reconstruction measurement method for the three-dimensional combustion surface of solid propellants: A binocular vision imaging system with a beam-splitting structure is established, and binocular images of the combustion surface during the solid propellant combustion process are acquired; the combustion surface region of the binocular image is extracted as a region of interest (ROI) image and enhanced; feature point identification and matching of the solid propellant combustion surface are performed on the enhanced ROI image to obtain matched feature points; a three-dimensional combustion surface point cloud is reconstructed based on the matched feature points obtained in step 3 using triangulation; the reconstructed three-dimensional surface point cloud is post-processed to obtain a three-dimensional combustion surface mesh of the solid propellant, thus obtaining the three-dimensional morphology information of the solid propellant combustion surface. This invention also provides a device for the binocular reconstruction measurement method of the three-dimensional combustion surface of solid propellants. This method and device can realize the reconstruction of the three-dimensional combustion surface of solid propellants, obtaining visualization of the three-dimensional combustion surface and three-dimensional morphology measurement results.

Description

Technical Field

[0001] This invention relates to the field of solid propellant combustion visualization and measurement, specifically to a binocular vision measurement method and apparatus for the three-dimensional combustion surface of solid propellants. Background Technology

[0002] Combustion instability is a key factor affecting the operational performance of solid rocket engines. Instability can lead to abnormal vibrations in the rocket body, decreased accuracy, and even thrust oscillations, localized heat surges, and alterations in the internal trajectory, causing the engine to fail to complete its intended mission and potentially even resulting in explosion and disintegration. The core factor of engine instability is the unsteady combustion response of the solid propellant; therefore, the development of sophisticated measurement techniques for unsteady combustion of solid propellants is of great significance.

[0003] Solid propellants are characterized by multi-component and heterogeneous properties, and their combustion process involves multiphase coupling. Currently, combustion response measurements in this field mainly rely on T-type burner experiments. Although this method can obtain combustion response data to a certain extent, it suffers from low accuracy and large dispersion. Refined combustion analysis requires advanced combustion diagnostic methods, but due to the high-temperature, high-pressure, and smoky combustion environment of solid propellants, general commercial combustion diagnostic equipment is difficult to apply directly. Therefore, new methods need to be developed to acquire refined experimental data, providing data support for a deeper understanding of unsteady-state combustion mechanisms.

[0004] In the three-dimensional measurement of solid propellant combustion, some work has already conducted research on three-dimensional measurement techniques and applications for solid propellant combustion plumes. Typical and commonly used techniques include digital holography, tomography, and defocused interferometric imaging. These techniques, through three-dimensional measurement methods, enable the measurement of parameters such as the three-dimensional position, velocity, distribution, and particle size of typical condensed phase products in the propellant plume, such as aluminum agglomerates. For example, Chinese patent CN120213755A discloses a panoramic microscopic imaging measurement method for the three-dimensional surface of burning metal particles, including: acquiring dynamic panoramic microscopic images of burning metal particles; preprocessing the acquired panoramic microscopic images to separately crop the burning metal particles from the panoramic microscopic images and obtain binary images of the burning metal particles and their surface oxides; modeling the burning metal particles using a long ellipsoidal geometric model to obtain the mapping relationship between the three-dimensional surface of the burning metal particles and the two-dimensional information of the acquired panoramic microscopic images; and calculating the three-dimensional surface area of ​​the burning metal particles and their surface oxides using the binary images of the burning metal particles and their surface oxides, as well as the mapping relationship. Chinese patent CN113091917A discloses a method for measuring the three-dimensional gas-phase flame of 100-micron-sized aluminum combustion particles in solid propellants. The method includes the following steps: Step 1: Before ignition of the aluminum-containing solid propellant, acquire imaging images of the propellant in nine directions. Obtain the precise projection angle corresponding to each imaging image. Step 2: After ignition of the aluminum-containing solid propellant, acquire dynamic imaging images of the flame at nine angles during the dynamic combustion process. Step 3: Select a reconstruction area to obtain an image of the gas-phase flame of a single aluminum particle combustion. Step 4: Reconstruct the point cloud from the image of the target flame to obtain an initial three-dimensional digital matrix. Use ART iteration to obtain the matrix of the three-dimensional spatial distribution of the aluminum particle combustion flame. This method yields the three-dimensional morphology and three-dimensional dynamic changes of the flame of a single 100-micron-sized aluminum particle combustion under aluminum-containing solid propellant combustion conditions.

[0005] However, due to occlusion and resolution issues, these imaging techniques have not yet been applied to the visualization and measurement of solid propellant combustion surfaces. Currently, most methods for visualizing and measuring solid propellant combustion surfaces rely on two-dimensional methods such as high-speed microscopy and infrared imaging. These methods cannot directly obtain three-dimensional information about the combustion surface. In conclusion, the three-dimensional measurement of solid propellant combustion surfaces still faces significant challenges. Summary of the Invention

[0006] The purpose of this invention is to provide a binocular imaging reconstruction method and apparatus for the three-dimensional combustion surface of solid propellants; the method and apparatus can realize the reconstruction of the three-dimensional combustion surface of solid propellants and obtain visualization of the three-dimensional combustion surface and three-dimensional morphology measurement results.

[0007] To achieve the above-mentioned objectives, the present invention provides the following technical solution: A binocular vision measurement method for the three-dimensional combustion surface of solid propellants, the measurement method comprising the following steps: Step 1: Establish a binocular vision imaging system with a spectral structure and acquire binocular images of the burning surface during the combustion process of solid propellant; Step 2: Extract the burning surface region from the binocular image of the burning surface as the region of interest image and enhance it; Step 3: Perform feature point identification and matching on the solid propellant burning surface of the enhanced region of interest image to obtain the matched feature points; Step 4: Reconstruct the three-dimensional point cloud of the combustion surface based on the matching feature points obtained in Step 3 using the triangulation method; Step 5: The reconstructed 3D surface point cloud is post-processed to obtain a 3D combustion surface mesh for solid propellant, thus obtaining the 3D morphology information of the solid propellant combustion surface.

[0008] In step 1, a binocular vision imaging system with a beam-splitting structure is established. The binocular system is calibrated using the Zhang calibration method to obtain camera intrinsic parameters, camera extrinsic parameters, and distortion parameters. The method for acquiring binocular imaging images of solid propellant combustion using the system mainly includes: Step 1-1: Arrange the optical axes of the two cameras vertically and place a beam splitter at the intersection of the camera optical axes (approximately 45°). Adjust the three-dimensional position of the two cameras so that their fields of view overlap at the same position. Then, adjust the position of the two cameras by translating left and right to control the baseline of the binocular imaging system, so that there is parallax between the two cameras and the area to be measured still appears completely in the fields of view of the two cameras. Steps 1-2: Calculate the camera intrinsic parameters and the rotation matrix and translation vector between the binocular cameras using Zhang's calibration method and checkerboard images, and obtain the camera distortion parameters. Specifically, arrange a checkerboard calibration board in the common field of view of the binocular imaging system, and take no fewer than three checkerboard images in different poses. Calculate the camera intrinsic parameters using Zhang's calibration method and the checkerboard images. and the rotation matrix between the binocular cameras Translation vector And obtain the camera distortion parameters; Steps 1-3: Use a synchronization signal to control the camera to take pictures simultaneously and use an external trigger source to ensure that the starting frames of the binocular imaging system are the same, and take binocular images of the burning surface during the solid propellant combustion process.

[0009] In step 2, the method of extracting the burning surface region from the binocular image of the burning surface as the region of interest image and enhancing it to obtain the enhanced region of interest image for low illumination and weak feature texture of the burning surface includes: Step 2-1: Correct the image using distortion parameters and flip the image from one of the cameras horizontally; Step 2-2: Extract the burning surface portion separately from the binocular image of the burning surface using image segmentation methods. Obtain the burning surface segmentation mask and the detection box position of the burning surface portion on the image, denoted as . ,in This indicates the coordinates of the top-left corner of the detection box on the image. This indicates the number of pixels in length and width corresponding to the detection box; Steps 2-3: Use a segmentation mask to remove image information from the non-flammable area, and use an image detection box to crop the flammable part from the image to obtain the region of interest image; Steps 2-4: Enhance the low-light and weak texture of the combustion surface in the binocular imaging image of solid propellant combustion using a deep learning image low-light enhancement algorithm to obtain the enhanced region of interest image.

[0010] In step 2, the acquired solid propellant burning surface original image (burning surface binocular image) is subjected to the image preprocessing in steps 2-1 to 2-3 and the enhancement in step 2-4. This can obtain the region of interest image after low illumination and weak feature texture enhancement of the burning surface. The algorithm realizes the enhancement of dark areas and weak texture of the burning surface.

[0011] In step 3, solid propellant combustion surface feature point identification and matching are performed on the enhanced region of interest image. This involves identifying and matching solid propellant combustion surface feature points based on dense or semi-dense feature point identification and matching methods. Specific methods include: Step 3-1: Based on the dense or semi-dense feature point recognition and matching method, solid propellant combustion surface feature points are identified and matched in the enhanced region of interest image to obtain the feature points on the enhanced region of interest image and the matching sequence on the region of interest image; Step 3-2: Remap the coordinates of the feature points to the binocular imaging image to obtain the pixel coordinate information of the matching feature point sequence; specifically: through the detection box information The coordinates of the identified and matched feature points By adding the coordinates of the top left corner of the detection box Remapping back to the original image location yields the coordinates of the feature points in the original image. .

[0012] In step 4, based on the camera intrinsic parameters, camera rotation matrix, and translation vector of the binocular vision imaging system obtained in step 1, the three-dimensional point cloud of the combustion surface is reconstructed using triangulation based on the matched feature points obtained in step 3. The specific method is as follows: The camera intrinsic parameters obtained in step 1 and camera rotation matrix and translation vector And the results of image feature point recognition and matching of a set of image pixels obtained in step 3. and The three-dimensional position reconstruction equations for the corresponding feature points can be listed as follows: ; In the formula, Let be the three-dimensional coordinates of the point in the world coordinate system to be solved, where is the principal coordinate system. It is the identity matrix, and After the dimensionless equation is obtained, solving this equation can realize the solution of the three-dimensional point cloud.

[0013] In step 5, the method for obtaining a three-dimensional combustion surface mesh of solid propellant through post-processing of the reconstructed three-dimensional surface point cloud includes: Step 5-1: Delete point clouds with abnormal surface curvature and point clouds that are abnormally separated from the group; Step 5-2: Calculate the normal vector of the point cloud; Step 5-3: Use the Poisson reconstruction method to reconstruct the 3D point cloud into a 3D mesh surface.

[0014] The present invention also provides a binocular vision measurement device for the three-dimensional combustion surface of solid propellants, the measurement device comprising: A binocular vision imaging system with a beam-splitting structure is used to perform step 1; The synchronous trigger signal system is connected to the binocular vision imaging system with a beam splitting structure, and is used to control the synchronous acquisition of binocular images of the burning surface of solid propellant by the two cameras in the binocular vision imaging system. The data acquisition system is used to perform steps 2-5.

[0015] A binocular vision imaging system with a beam-splitting structure includes two cameras, a beam splitter, and a camera support, used to record binocular imaging images of the burning surface of solid propellant; furthermore, the binocular vision imaging system with a beam-splitting structure uses a beam splitter to avoid possible mechanical interference between the cameras, one camera directly images the burning surface through the beam splitter, and the other camera images through the reflection of the beam splitter.

[0016] The synchronous trigger signal system includes a trigger switch, a trigger signal line, a synchronization signal generator, and a synchronization signal line. The trigger signal and synchronization signal of the synchronous trigger signal system are respectively connected to two cameras in the binocular imaging system. Furthermore, the synchronous trigger signal system uses the synchronization signal generator to provide two adjustable delay pulse TTL signals to achieve simultaneous acquisition of the cameras at the same time and the same frame rate, and the trigger signal controls the starting frames of the two cameras to be the same.

[0017] The data acquisition system includes a network cable and a data acquisition computer. The data acquisition system connects the data acquisition computer and the camera on the same network segment for data interaction, and uploads the data captured by the camera to the data acquisition computer for subsequent processing.

[0018] To address the challenges of visualizing and measuring the three-dimensional combustion surface of solid propellants, this invention proposes a binocular imaging measurement method for the three-dimensional combustion surface of solid propellants. This method enables the reconstruction of the three-dimensional point cloud and mesh of the combustion surface of burning solid propellants, acquiring three-dimensional morphological information of the combustion surface. The method provided by this invention enables the visualization and measurement of the three-dimensional reconstruction of the solid propellant combustion surface, providing a technical means for characterizing and measuring key parameters such as combustion stability and burning rate of solid propellants. This provides data support for addressing unsteady combustion issues in the solid propellant combustion process and is of great significance for a deeper understanding of the unsteady combustion mechanism of solid propellants and key issues in aerospace propulsion such as engine combustion instability.

[0019] The binocular vision measurement device for three-dimensional combustion surfaces of solid propellants provided by this invention can effectively avoid mechanical interference problems caused by insufficient lens imaging distance and excessive camera size in dual cameras, and effectively reduce the imaging baseline of the established binocular imaging system, reducing the difficulty of subsequent matching. The device provided by this invention takes into account the consistency of acquisition time during measurement in high-speed dynamic scenes, ensuring the acquisition of dynamic combustion surface images. Attached Figure Description

[0020] Figure 1 A schematic diagram of a binocular vision imaging system with a spectral structure for solid propellant combustion; Figure 2 Flowchart for binocular reconstruction data processing of solid propellant combustion; Figure 3 Images of the solid propellant combustion surface captured in Example 1 (left image is the field of view of the left camera, right image is the field of view of the right camera); Figure 4 The image shown is from Example 1, after segmentation and cropping and low-light enhancement (left image is the unenhanced result, right image is the enhanced result). Figure 5 The image shows a three-dimensional reconstruction of the solid propellant combustion surface provided in Example 1. Wherein: 1. Right camera lens; 2. Right camera; 3. Left camera lens; 4. Left camera; 5. Beam splitter; 6. Trigger source; 7. Synchronization signal generator; 8. Computer; 9. Solid propellant strip to be tested; 10. Baseline of the binocular system; 11. Equivalent position of the right camera lens; 12. Equivalent position of the right camera. Detailed Implementation

[0021] The specific implementation of the technical solution of the present invention will be further described below through examples and in conjunction with the accompanying drawings.

[0022] like Figure 2 As shown, the binocular imaging measurement method for three-dimensional surface imaging of solid propellant combustion provided in this embodiment includes the following steps: Step 1: Establish a binocular vision imaging system with a beam splitting structure, calibrate the binocular system using Zhang's calibration method, obtain camera intrinsic parameters, camera extrinsic parameters and distortion parameters, and acquire binocular imaging images of solid propellant combustion. like Figure 1 As shown, a binocular vision imaging system with a beam-splitting structure includes: Two cameras, a right camera 2 and a left camera 4, each equipped with a lens: a right camera lens 1 and a left camera lens 3, resulting in a system equivalent pixel size better than 10 μm per pixel. The right camera 2 and left camera 4 are placed vertically, with the solid propellant combustion platform positioned on the optical axis of one of the cameras. A beam splitter 5 is placed at the intersection of the two cameras, forming an angle of approximately 45° with each camera. The right camera lens 1 and right camera 2 are positioned at equivalent positions 11 and 12, respectively, relative to the horizontal offset distance of the left camera 4 from the baseline 10 of the binocular system, for image capture. Preferably, in this embodiment, the binocular system captures binocular images of the burning surface of the solid propellant at a baseline 10 of approximately 30 mm. Trigger source 6 is connected to right camera 2 and left camera 4, and simultaneously activates both cameras to start shooting, thereby controlling the frame start positions of the two cameras to be the same. Synchronizer 7, connected to right camera 2 and left camera 4, is used to control the shooting sequence of the two cameras, so that the two cameras can capture images of burning metal particles at the same frame rate and at the same time. Computer 8 is used to receive and record image data transmitted by the right camera 2 and the left camera 4, and process it to obtain the three-dimensional surface measurement results of solid propellant combustion.

[0023] The solid propellant strip 9 to be tested is placed in the common field of view of the two cameras, that is, on the optical axis of one of the cameras.

[0024] In this embodiment, the right camera 2 and the left camera 4 are arranged at a 90° angle. The beam splitter is located at the intersection of the optical axes of the two cameras and is placed at approximately a 45° angle to the cameras. The maximum frame rate of the cameras is 20 Hz, and the resolution is 2048×2048. Both the right camera 2 and the left camera 4 are equipped with fixed-focus lenses with a focal length of 105 mm. The cameras are positioned approximately 15 cm from the combustion platform, resulting in an equivalent pixel size of 8.4 μm for the system imaging. The synchronizer achieves nanosecond-level accuracy, with a timing signal frequency of 20 Hz and a pulse width of 10 μs, and is connected to the cameras using BNC cables. The external trigger source uses a manual switch connected to the cameras via a BNC cable, and simultaneous triggering of the cameras is achieved by providing a short-circuit signal. The data collected by the cameras is transmitted to computer 8 via a network cable. The computer has 16 GB of memory and a storage capacity of no less than 512 GB.

[0025] In this embodiment, a checkerboard is used as a calibration board for system calibration. First, the focus positions of the cameras are adjusted to be consistent, and the images in the fields of view of the two cameras are adjusted to be mirror images. Then, a certain displacement is generated between one camera and the other camera to create parallax in the image. A synchronizer is used to control the camera shooting sequence. A trigger source is used to control the camera activation time. At least three checkerboard calibration images are taken for the calibration calculation of the camera intrinsic and extrinsic parameters of the binocular imaging system.

[0026] The binocular imaging system needs to achieve synchronization and simultaneous triggering. The specific operation is as follows: connect the synchronization signal generator 7 to the two cameras, set the cameras to be controlled by the external synchronization timing signal, and use the synchronizer to give two identical timing signals to control the cameras; connect the external trigger source to the two cameras, set the cameras to be activated by the external trigger signal, and realize that the cameras start acquiring data simultaneously.

[0027] The calibration of the camera's intrinsic and extrinsic parameters using a checkerboard calibration plate involves the following steps: A calibration plate of known length is placed within the common field of view of the two cameras in the binocular imaging system. In this embodiment, the checkerboard calibration plate used is a 12×9 square plate, with each small square having a side length of 500 μm. At least three images of different poses are captured, and then Zhang's calibration algorithm is used to calibrate the intrinsic and extrinsic parameters of the binocular cameras. The solid propellant used in this embodiment is burned under normal temperature and pressure. The propellant block is a 4×4×5 mm cuboid, placed within the common field of view of the binocular imaging cameras to ensure that both cameras can clearly image the three-dimensional burning surface of the propellant block.

[0028] The solid propellant combustion surface image captured in this embodiment is as follows: Figure 3 As shown, the left image is the field of view of the left camera, and the right image is the field of view of the right camera.

[0029] In step 2, the original image of the solid propellant burning surface is preprocessed to obtain an enhanced region of interest image for low illumination and weak texture features of the burning surface. The specific steps are as follows: Step 2-1: Correct the image using distortion parameters and flip the image of one of the cameras horizontally; in this embodiment, the field of view of the camera that is imaged after being reflected 90 degrees by the beam splitter is flipped. Step 2-2: Extract the burning surface from the image using image segmentation methods to obtain a binary image of the burning surface and the location of the detection box for the burning surface in the image, denoted as . ,in This indicates the coordinates of the top-left corner of the detection box on the image. This indicates the number of pixels in length and width corresponding to the detection box; In this embodiment, the image segmentation method used is to obtain a segmentation mask for the burning area using a thresholding method. Morphological operations, closing and opening operations, are then used to optimize the segmentation mask result, reducing holes and noise in the binary image. Preferably, the burning area mask can also be extracted using the Segment Anything Model (SAM) series or deep learning segmentation methods trained and fine-tuned on a dataset constructed with expert annotations.

[0030] Steps 2-3: Enhance the cropped solid propellant burning surface region of interest image using a deep learning low-light enhancement algorithm to obtain the enhanced region of interest image.

[0031] In this embodiment, as a preferred approach, Self-Calibrated Illumination (SCI) based on deep learning is used as an image enhancement algorithm for low-light enhancement, thereby enhancing the low-light and weak feature texture of the burning surface in the image.

[0032] In this embodiment, the image that has been segmented, cropped, and enhanced under low light conditions is as follows: Figure 4 As shown, the left image represents the unenhanced result, and the right image represents the enhanced result.

[0033] In step 3, the method for identifying and matching solid propellant combustion surface feature points in the region of interest image based on dense or semi-dense feature point recognition and matching methods, and obtaining pixel coordinate information of the matching feature point sequence on the original image, includes: Step 3-1: Use dense or semi-dense image feature point recognition and matching methods to obtain feature points and matching sequences on the cropped image; In this embodiment, a Robust DenseFeature Matching (RoMa) semi-dense deep learning model based on the DINO v2 image pre-training model is used to identify and match the burning surface feature points in the image, and obtain the matched feature point sequence in the burning surface region of the image. Step 3-2: Detect box information The coordinates of the identified and matched feature points By adding the coordinates of the top left corner of the detection box Remapping back to the original image location yields the coordinates of the feature points in the original image. .

[0034] Step 4: Reconstruct the three-dimensional point cloud of the combustion surface from the matching feature point sequence using triangulation methods; The camera intrinsic parameters obtained in step 1 and camera rotation matrix and translation vector And the results of image feature point recognition and matching of a set of image pixels obtained in step 3. and The three-dimensional position reconstruction equations for the corresponding feature points can be listed as follows: ; In the formula, Let be the three-dimensional coordinates of the point in the world coordinate system to be solved, where is the principal coordinate system. It is the identity matrix, and After the dimensionless equation is obtained, solving this equation can realize the solution of the three-dimensional point cloud.

[0035] In step 5, the method for obtaining a three-dimensional combustion surface mesh of solid propellant from the post-processed reconstructed three-dimensional surface point cloud includes: Step 5-1: Delete point clouds with abnormal surface curvature and point clouds that are abnormally out of the group; In this embodiment, point clouds with abnormal surface curvature and point clouds that are abnormally out of the group are deleted by manual annotation using MeshLab software; Preferably, the Statistical Outlier Removal (SOR) method can also be used to remove abnormal point clouds. Step 5-2: Calculate the normal vector of the point cloud; Step 5-3: Use the Poisson reconstruction method to reconstruct the point cloud into a 3D mesh surface.

[0036] The specific results of the solid propellant combustion three-dimensional mesh reconstruction are as follows: Figure 4 As shown.

[0037] This embodiment successfully realizes the visualization reconstruction and measurement of the three-dimensional surface of solid propellant combustion using a binocular imaging measurement method, proving the feasibility of the method and system.

[0038] The above is a detailed description of the present invention in conjunction with the embodiments. However, the implementation of the present invention is not limited to the above embodiments. Any changes, substitutions, combinations and simplifications made under the core guiding idea of ​​the present invention are included within the protection scope of the present invention.

Claims

1. A binocular vision measurement method of a solid propellant three-dimensional combustion surface, characterized in that, The measurement method includes the following steps: Step 1: Establish a binocular vision imaging system with a spectral structure and acquire binocular images of the burning surface during the combustion process of solid propellant; Step 2: Extract the burning surface region from the binocular image of the burning surface as the region of interest image and enhance it; Step 3: Perform feature point identification and matching on the solid propellant burning surface of the enhanced region of interest image to obtain the matched feature points; Step 4: Reconstruct the three-dimensional point cloud of the combustion surface based on the matching feature points obtained in Step 3 using the triangulation method; Step 5: The reconstructed 3D surface point cloud is post-processed to obtain a 3D combustion surface mesh for solid propellant, thus obtaining the 3D morphology information of the solid propellant combustion surface.

2. The binocular vision measurement method for the three-dimensional combustion surface of solid propellants according to claim 1, characterized in that, In step 1, specifically: Step 1-1: Arrange the optical axes of the two cameras vertically, place a beam splitter at the intersection of the camera optical axes, and adjust the three-dimensional position of the two cameras so that the camera fields of view overlap at the same position. Then, adjust the position of the two cameras left and right to control the baseline of the binocular imaging system so that there is parallax between the two cameras and the area to be measured still appears completely in the fields of view of the two cameras. Steps 1-2: Use Zhang's calibration method and checkerboard image to calculate the camera intrinsic parameters and the rotation matrix and translation vector between the two cameras, and obtain the camera distortion parameters; Steps 1-3: Use a synchronization signal to control the camera to take pictures simultaneously and use an external trigger source to ensure that the starting frames of the binocular imaging system are the same, and take binocular images of the burning surface during the solid propellant combustion process.

3. The binocular vision measurement method for the three-dimensional combustion surface of solid propellants according to claim 1, characterized in that, In step 2, the method for extracting the burning surface region of the binocular image as the region of interest and enhancing it is as follows: 2-1. Extract the burning surface part from the binocular image of the burning surface using image segmentation methods to obtain the burning surface segmentation mask and the detection box position of the burning surface part on the binocular image of the burning surface; Step 2-2: Use a segmentation mask to remove image information from the non-flammable surface area, and use a detection box to crop the flammable surface part from the binocular image of the flammable surface to obtain the region of interest image; Steps 2-3: Enhance the burning surface texture in the region of interest image using a deep learning low-light enhancement algorithm to obtain the enhanced region of interest image.

4. The binocular vision measurement method for the three-dimensional combustion surface of solid propellants according to claim 1, characterized in that, In step 3, solid propellant combustion surface feature points are identified and matched on the enhanced region of interest image. The method for obtaining the matched feature points is as follows: Step 3-1: Based on the dense or semi-dense feature point recognition and matching method, solid propellant combustion surface feature points are identified and matched in the enhanced region of interest image to obtain the feature points on the enhanced region of interest image and the matching sequence on the region of interest image; Step 3-2: Remap the coordinates of the feature points to the binocular imaging image to obtain the pixel coordinate information of the matching feature point sequence.

5. The binocular vision measurement method for the three-dimensional combustion surface of solid propellants according to claim 1, characterized in that, In step 4, based on the camera intrinsic parameters and camera rotation matrix and translation vector of the binocular vision imaging system in step 1, the three-dimensional point cloud of the combustion surface is reconstructed using triangulation based on the matched feature points obtained in step 3.

6. The binocular vision measurement method for the three-dimensional combustion surface of solid propellants according to claim 1, characterized in that, In step 5, the post-processing method includes: Step 5-1: Delete point clouds with abnormal surface curvature and point clouds that are abnormally separated from the group; Step 5-2: Calculate the normal vector of the point cloud; Step 5-3: Use the Poisson reconstruction method to reconstruct the 3D point cloud into a 3D mesh surface.

7. A binocular vision measurement device for the three-dimensional combustion surface of a solid propellant using the method described in any one of claims 1-6, characterized in that, The measuring device includes: A binocular vision imaging system with a beam-splitting structure is used to perform step 1; The synchronous trigger signal system is connected to the binocular vision imaging system with a beam splitting structure, and is used to control the synchronous acquisition of binocular images of the burning surface of solid propellant by the two cameras in the binocular vision imaging system. The data acquisition system is used to perform steps 2-5.

8. The binocular vision measurement device for the three-dimensional combustion surface of solid propellants according to claim 7, characterized in that, The binocular vision imaging system with a beam splitting structure includes two cameras and imaging lenses, a beam splitter and camera support; wherein, one camera directly images the burning surface through the beam splitter, and the other camera images the surface through the reflection of the beam splitter.

9. The binocular vision measurement device for the three-dimensional combustion surface of solid propellants according to claim 7, characterized in that, The system includes a trigger switch, a trigger signal line, a synchronization signal generator, and a synchronization signal line. The trigger signal line and synchronization signal line of the synchronization trigger signal system are respectively connected to two cameras in the binocular imaging system. The synchronization signal generator provides two adjustable delay pulse TTL signals to achieve simultaneous acquisition of the cameras at the same time and the same frame rate. The trigger signal controls the starting frames of the two cameras to be the same.

Images