Irregular particle three-dimensional shape measuring device and method based on speckle imaging

Abstract

The invention discloses an irregular particle three-dimensional shape measurement device and method based on speckle imaging, and belongs to the field of irregular particle three-dimensional shape measurement, and the irregular particle three-dimensional shape measurement device comprises an irregular particle shakeout device which is used for enabling irregular particles to fall in a sparse free falling body form, and is used for forming lamellar laser vertical to the falling direction of the irregular particles, the laser sheet generating module is used for forming complete speckle distribution on the surfaces of the irregular particles; the laser sheet generating module is used for generating a laser sheet, the multi-view image collecting module is used for collecting speckle distribution on the surfaces of irregular particles in a multi-view mode and forming speckle images, the synchronous control system is used for controlling the laser sheet generating module and the multi-view image collecting module and synchronizing the time of the laser sheet generating module and the multi-view image collecting module, and the three-dimensional reconstruction processing unit is used for obtaining three-dimensional shape data of the irregular particles according to the speckle images. According to the invention, the irregular particles can be rapidly, non-destructively and in-situ measured so as to meet the requirements of aeroengine erosion research on the precision, efficiency and integrity of irregular particle data.

Description

Technical Field

[0001] This invention relates to the field of three-dimensional morphology measurement of irregular particles, and particularly to a device and method for measuring the three-dimensional morphology of irregular particles based on speckle imaging. Background Technology

[0002] In complex dusty environments (such as sandstorms), aero-engines are prone to ingesting sand particles during operation, leading to erosion damage on the blade surface caused by high-speed sand impacts. This severely affects engine performance and service life, and may even threaten flight safety. To conduct research on blade erosion mechanisms and develop protective designs, it is necessary to obtain three-dimensional morphological information of sand particles to support high-precision numerical simulation and erosion simulation.

[0003] Current technologies primarily rely on focused ion beam scanning electron microscopy (FIE) to acquire the three-dimensional morphology of sand grains. Specifically, this involves layer-by-layer cutting of the sample using a focused ion beam, followed by high-resolution imaging of each cross-section using scanning electron microscopy. Three-dimensional surface reconstruction is then achieved through tomographic image sequence registration and processing, combined with a moving cubes algorithm, to obtain high-precision three-dimensional morphology data for the sand grains. However, this method is complex and cumbersome in its measurement and modeling process, has a long operation cycle, is difficult to implement for large-scale sample statistics, and is destructive to the sand grains. Furthermore, this method cannot perform in-situ measurements, and cannot obtain key experimental data such as sand grain impact points and velocities, limiting the real-time analysis and accurate simulation of the erosion process. Summary of the Invention

[0004] The purpose of this invention is to provide a device and method for measuring the three-dimensional morphology of irregular particles based on speckle imaging, so as to overcome the defects of the existing technology. This invention can perform rapid, non-destructive, in-situ measurement of sand grains (irregular particles), realize the efficient acquisition of the three-dimensional morphology of large-scale irregular particles, and simultaneously capture key impact test data to meet the accuracy, efficiency and integrity requirements of aero-engine erosion research on irregular particle data.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: A device for measuring the three-dimensional morphology of irregular particles based on speckle imaging, comprising: Irregular particle sand removal device: used to make irregular particles fall in a sparse free-fall manner; Laser sheet generation module: used to form a thin sheet of laser perpendicular to the falling direction of irregular particles, so that a complete speckle distribution is formed on the surface of the irregular particles; Multi-view image acquisition module: used to acquire speckle distribution on irregular particle surfaces from multiple perspectives to form speckle images; Synchronization control system: used to control the laser sheet generation module and the multi-view acquisition module, and to synchronize their time; The 3D reconstruction processing unit is used to process speckle images, obtain the ID, timestamp, and viewpoint information corresponding to each irregular particle in the speckle image; based on the pre-trained speckle-to-2D morphology mapping model and the ID, timestamp, and viewpoint information corresponding to each irregular particle, it maps the acquired speckle images to obtain a multi-view 2D morphology dataset of particles; based on the multi-view 2D morphology dataset of particles, it completes the 3D morphology reconstruction and optimization of irregular particles through a 3D morphology reconstruction algorithm to obtain 3D morphology data of irregular particles.

[0006] Furthermore, the irregular particle sand removal device is a funnel-shaped sand removal device made of transparent quartz or acrylic material, and the outlet diameter of the funnel-shaped channel of the funnel-shaped sand removal device is 2-10mm.

[0007] Furthermore, the laser sheet generation module includes a pulsed laser, a reflector, and a sheet light generator. The reflector is used to reflect the incident laser beam emitted by the pulsed laser to the sheet light generator, and the sheet light generator expands the incident laser beam into a thin sheet-like laser perpendicular to the falling direction of the irregular particles.

[0008] Furthermore, the multi-view image acquisition module includes several cameras and telephoto macro lenses mounted along both sides of the thin-film laser and at different pitch angles, with each camera and telephoto macro lens corresponding to one another, and each camera and its corresponding telephoto macro lens forming an image acquisition unit.

[0009] Furthermore, the image acquisition unit is provided with five units, including a first image acquisition unit, a second image acquisition unit, a third image acquisition unit, a fourth image acquisition unit, and a fifth image acquisition unit; The first image acquisition unit, the second image acquisition unit, and the third image acquisition unit are arranged in the same vertical reference plane, and are distributed at 120° angles on the circumference with the central reference point as the center. The first image acquisition unit, the fourth image acquisition unit, and the fifth image acquisition unit are located in the same horizontal reference plane and are distributed at 120° angles on the circumference with the central reference point as the center. The central reference point is the geometric center of an equilateral triangle formed by three image acquisition units within the same vertical reference plane or three image acquisition units within the same horizontal reference plane.

[0010] Furthermore, the image acquisition unit is supported and fixed by a multi-view shooting and fixing device, which includes a fixed base plate and four support rods fixed on the fixed base plate, namely the first support rod, the second support rod, the third support rod and the fourth support rod, and a camera fixing plate is installed above each support rod; The camera mounting plate on the first support rod is used to fix the camera of the first image acquisition unit, the camera mounting plate on the second support rod is used to fix the camera of the fifth image acquisition unit, the camera mounting plate on the third support rod is used to fix the camera of the fourth image acquisition unit, and the camera mounting plate on the fourth support rod is used to fix the cameras of the second and third image acquisition units.

[0011] Furthermore, the three-dimensional reconstruction processing unit includes a speckle image processing module, a two-dimensional topography reconstruction module, and a three-dimensional topography reconstruction module; The speckle image processing module is used to automatically crop and extract features from the speckle image, and obtain the ID, timestamp and viewpoint information corresponding to each irregular particle in the speckle image; The two-dimensional morphology reconstruction module is used to automatically group speckle images according to the IDs corresponding to irregular particles, so that speckle images of the same irregular particle under different viewpoint information are classified into the same dataset, and arranged in order according to viewpoint information within each group to form a multi-view speckle image sequence. The multi-view speckle image sequence is input into a pre-trained speckle-to-two-dimensional morphology mapping model to obtain the two-dimensional morphology image corresponding to the multi-view speckle image sequence. The two-dimensional morphology image is then grouped again according to the IDs corresponding to irregular particles, maintaining the same viewpoint order as when it was input, and is bound to the timestamp and projection parameters of the multi-view acquisition module to form a particle multi-view two-dimensional morphology dataset. The three-dimensional morphology reconstruction module is used to uniformly register the two-dimensional morphology images from various perspectives in the multi-view two-dimensional morphology dataset of particles to the same three-dimensional coordinate system, perform binarization and edge detection on each two-dimensional morphology image, extract the two-dimensional projection contours of irregular particles under the perspective, backproject the two-dimensional projection contours to three-dimensional space, generate several view frustums containing all possible shapes, perform spatial Boolean intersection operation on several view frustums to obtain a three-dimensional envelope consistent with the two-dimensional projection contours of all perspectives, further discretize the obtained three-dimensional envelope into a regular voxel mesh, label voxels according to the regular voxel mesh to obtain a preliminary three-dimensional voxel model, use a voxel-to-surface conversion algorithm to convert the three-dimensional voxel model into a three-dimensional mesh surface, optimize the three-dimensional mesh surface, and obtain the three-dimensional morphology data of irregular particles. The pre-trained speckle-to-two-dimensional morphology mapping model is trained based on a deep generative network and irregular particle samples with known morphologies.

[0012] A method for measuring the three-dimensional morphology of irregular particles based on speckle imaging includes the following steps: Irregular particles fall in a sparse free-fall manner; A thin sheet of laser light is formed perpendicular to the falling direction of the irregular particles, so that a complete speckle distribution is formed on the surface of the irregular particles. The speckle distribution on the surface of the irregular particles is acquired from multiple perspectives to form a speckle image. The speckle image is processed to obtain the ID, timestamp, and viewpoint information of each irregular particle in the speckle image. Based on the pre-trained speckle-to-2D morphology mapping model and the ID, timestamp, and viewpoint information of each irregular particle, the collected speckle image is mapped to obtain a multi-view 2D morphology dataset of particles. Based on the multi-view 2D morphology dataset of particles, the 3D morphology reconstruction algorithm is used to complete the 3D morphology reconstruction and optimization of irregular particles, and obtain the 3D morphology data of irregular particles.

[0013] Furthermore, the process of acquiring speckle distribution on the surface of irregular particles from multiple perspectives to form a speckle image specifically involves: When irregular particles fall freely through the area of ​​the thin-film laser, the speckle distribution on the surface of the irregular particles is simultaneously acquired from different perspectives. During the acquisition process, background removal and gray-level normalization based on a sliding window are automatically completed, and high-frequency noise is suppressed by median filtering and wavelet denoising methods to form a speckle image.

[0014] Further, the speckle image is processed to obtain the ID, timestamp, and viewpoint information corresponding to each irregular particle in the speckle image. Based on the pre-trained speckle-to-2D morphology mapping model and the ID, timestamp, and viewpoint information corresponding to each irregular particle, the acquired speckle image is mapped to obtain a multi-view 2D morphology dataset of particles. Based on the multi-view 2D morphology dataset of particles, a 3D morphology reconstruction algorithm is used to complete the 3D morphology reconstruction and optimization of irregular particles, resulting in 3D morphology data of irregular particles. Specifically, this includes: Automatic cropping and feature extraction are performed on speckle images to obtain the ID, timestamp, and viewpoint information corresponding to each irregular particle in the speckle image; The speckle images are automatically grouped according to the IDs of the irregular particles, so that speckle images of the same irregular particle under different viewpoints are classified into the same dataset. Within each group, they are arranged in order according to the viewpoint information to form a multi-view speckle image sequence. The multi-view speckle image sequence is input into a pre-trained speckle-to-2D topography mapping model to obtain the 2D topography image corresponding to the multi-view speckle image sequence. The 2D topography image is then grouped again according to the IDs of the irregular particles, maintaining the same viewpoint order as when it was input. At the same time, it is bound to the timestamp and the projection parameters of the multi-view acquisition module to form a particle multi-view 2D topography dataset. The 2D topography images from various perspectives in the multi-view 2D topography dataset of particles are uniformly registered into the same 3D coordinate system. Each 2D topography image is binarized and edge detected to extract the 2D projection contours of irregular particles under the perspective. The 2D projection contours are back-projected into 3D space to generate several view frustums containing all possible shapes. Spatial Boolean intersection operation is performed on several view frustums to obtain a 3D envelope consistent with the 2D projection contours of all perspectives. The obtained 3D envelope is further discretized into a regular voxel mesh. Voxels are labeled according to the regular voxel mesh to obtain a preliminary 3D voxel model. A voxel-to-surface conversion algorithm is used to convert the 3D voxel model into a 3D mesh surface. The 3D mesh surface is optimized to obtain the 3D topography data of irregular particles. The pre-trained speckle-to-two-dimensional morphology mapping model is trained based on a deep generative network and irregular particle samples with known morphologies.

[0015] Compared with the prior art, the present invention has the following beneficial technical effects: (1) Non-destructive measurement: This invention does not require physical processing such as cutting or embedding of irregular particles, and can directly obtain the three-dimensional morphology data of irregular particles, which can completely preserve the original structure and surface features of irregular particles; (2) Rapid detection: Through the synchronous cooperation of the laser sheet generation module and the multi-view camera, the present invention can realize real-time imaging and three-dimensional reconstruction of free-falling irregular particle flow, with high measurement efficiency, and is suitable for the rapid establishment of a large-scale irregular particle sample database. (3) Key data acquisition: During the measurement process, the present invention can bind timestamps and perspective information to each irregular particle, thereby acquiring key experimental data such as the movement trajectory of irregular particles and potential impact points, providing important support for numerical simulation and erosion research; (4) Adapting to complex morphology: By combining deep generative networks with multi-view three-dimensional morphology reconstruction, this invention can accurately capture the complex contours and surface details of irregular particles, and is applicable to a wide range of particle sizes, especially suitable for the characterization of irregular samples such as natural sand. (5) Modular and scalable: The device of the present invention adopts a modular design, and each functional unit can be flexibly adjusted and expanded according to experimental needs. It can not only meet the needs of aero-engine sand and dust test, but also be applied to abrasive testing, pharmaceutical particle analysis and other fields. Attached Figure Description

[0016] The accompanying drawings are provided to further understand the invention and constitute a part of this invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.

[0017] Figure 1This is a schematic diagram of the irregular particle three-dimensional morphology measurement device based on speckle imaging according to the present invention; Figure 2 This is a schematic diagram of the multi-view measurement camera layout of the present invention, wherein (a) is a schematic diagram of the arrangement of three cameras in the vertical plane, and (b) is a schematic diagram of the arrangement of three cameras in the horizontal plane. Figure 3 This is a structural diagram of the camera multi-view shooting fixing device of the present invention, wherein (a) is a side view and (b) is a bottom view; Figure 4 This is a schematic diagram of the mapping model from speckle to two-dimensional morphology of the present invention; Figure 5 This is a schematic diagram of the irregular particle three-dimensional morphology reconstruction algorithm of the present invention.

[0018] The components include: 1. Pulsed laser; 2. Incident laser beam; 3. Reflector; 4. Sheet light generator; 5. Sheet-shaped laser; 6. Irregular particles; 7. Camera; 8. Telephoto macro lens; 9. Post-processing unit; 10. Horizontal reference plane; 11. Vertical reference plane; 12. Center reference point; 13. First image acquisition unit; 14. Second image acquisition unit; 15. Third image acquisition unit; 16. Fourth image acquisition unit; 17. Fifth image acquisition unit; 18. Fixed base plate; 19. First support rod; 20. Second support rod; 21. Third support rod; 22. Fourth support rod; 23. Camera mounting plate. Detailed Implementation

[0019] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0020] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0021] Example 1 This invention provides a rapid measurement device for the three-dimensional morphology of sand grains based on speckle imaging, including a laser sheet generation module, a multi-view image acquisition module, an irregular particle falling device (for making irregular particles fall in a sparse free-fall form), a synchronous control system, and a three-dimensional reconstruction processing unit.

[0022] The laser sheet generation module is used to form a thin sheet laser 5 perpendicular to the falling direction of the irregular particles 6, and to uniformly distribute the incident laser beam 2 on the falling irregular particles 6, so that the surface of the irregular particles 6 forms high-contrast speckle or contour features, which facilitates subsequent image analysis and two-dimensional morphology extraction.

[0023] During the measurement process, irregular particles 6 fall in a sparse free-fall manner. Irregular particles 6, only illuminated by the thin-film laser 5, can be captured for imaging and 3D topography reconstruction. The multi-view image acquisition module consists of at least five cameras 7 and matching lenses (this invention uses a telephoto macro lens 8), which acquire speckle images of irregular particles 6 from different pitch angles and side views, capturing the topography information of irregular particles 6 in different directions, providing rich projection data for 3D topography reconstruction. The irregular particle falling device is a funnel-shaped falling device made of transparent material, used to guide irregular particles 6 to fall freely from the inlet to the outlet, ensuring that irregular particles 6 do not collide or rotate severely during the fall, while controlling the sparse distribution of irregular particles 6, ensuring that each imaging captures only a single irregular particle 6 or a small number of irregular particles. Particle 6 reduces image overlap interference; the synchronous control system coordinates the irradiation of the laser sheet generation module and the triggering acquisition of the multi-view acquisition module to achieve precise time synchronization; the 3D reconstruction processing unit is used to automatically crop and extract features from the acquired multi-view speckle images to obtain the ID (Identification), timestamp and view information corresponding to each irregular particle 6 in the speckle image, providing the necessary spatiotemporal correspondence for subsequent 2D morphology generation and 3D reconstruction. In addition, generative 2D morphology calculation is performed to obtain a multi-view 2D morphology dataset of particles. Finally, by combining the multi-view 2D morphology dataset of particles, the 3D morphology reconstruction and optimization of irregular particles is completed through a multi-view 3D morphology reconstruction algorithm, and the output is 3D morphology data of irregular particles that can be used for numerical simulation or erosion research.

[0024] The specific measurement procedure is as follows: 1. Activate the laser sheet generation module and the irregular particle sand removal device. The laser sheet generation module includes a pulsed laser 1, a reflector 3, and a sheet light generator 4. The incident laser beam 2 emitted by the pulsed laser 1 is transmitted to the sheet light generator 4 through the reflector 3. The sheet light generator 4 expands the incident laser beam 2 into a thin sheet laser 5 with controllable thickness, adjustable from 500 to 3000 μm, to cover the typical sand particle size (i.e., irregular particles 6). The thin sheet laser 5 is perpendicular to the falling direction of the irregular particles 6, ensuring that the irregular particles 6 form a complete speckle distribution when passing through the thin sheet laser 5. The irregular particle falling device is a funnel-shaped falling device made of transparent quartz or acrylic material. Its outlet diameter can be adjusted within the range of 2–10 mm according to experimental requirements to control the sparsity of the irregular particles 6. The inside of the funnel-shaped channel is smoothly treated to avoid deflection or rotation caused by collision between the irregular particles 6 and the wall, thus stabilizing the falling trajectory of the irregular particle stream (formed by several irregular particles 6).

[0025] 2. Installation and Synchronous Debugging of Multi-View Image Acquisition Module A multi-view image acquisition module consisting of at least five cameras 7 and telephoto macro lenses 8 is installed along both sides of the thin-film laser 5 and at different elevation angles. The camera 7 is arranged with horizontal viewing angles symmetrically installed on both sides, and three side viewing directions distributed at different elevation angles to ensure complete coverage of the three-dimensional morphology information of the irregular particles 6. A synchronous control system is responsible for triggering the laser sheet generation module and the cameras 7, synchronizing them in sub-millisecond time to ensure that the speckle image does not undergo motion blur. Before acquisition, the intrinsic and extrinsic parameters of each camera 7 are jointly calibrated using a calibration board to obtain accurate projection geometry. After debugging, the optical axis positions of the laser sheet generation module and each camera 7 are fixed to ensure accurate and reliable image acquisition.

[0026] 3. Speckle Image Acquisition When the irregular particle stream enters the area where the sheet-like laser 5 is located, the irregular structure on the surface of the irregular particles 6, when irradiated by the sheet-like laser 5, causes a phase difference in the scattered light, thus generating a speckle image of the irregular particles 6 at the imaging end. The multi-view image acquisition module simultaneously acquires speckle images of the irregular particles 6, capturing only the irregular particles 6 irradiated by the sheet-like laser 5. The multi-view image acquisition module automatically removes the background from the speckle image, uses a sliding window-based grayscale normalization algorithm to eliminate uneven illumination, and suppresses high-frequency noise through median filtering or wavelet denoising. Each irregular particle 6 has a unique ID, timestamp, and viewpoint information to ensure that particle information is not lost during subsequent reconstruction.

[0027] 4. Two-dimensional topography reconstruction After acquiring speckle images of irregular particles 6, the speckle images are first processed to obtain the ID, timestamp, and viewpoint information corresponding to each irregular particle 6. Then, all speckle images are automatically grouped according to the unique ID of the irregular particle 6, so that speckle images of the same irregular particle 6 under different camera views are classified into the same dataset. Within each group, they are arranged in an orderly manner according to the viewpoint information of camera 7, forming a multi-view speckle image sequence. Subsequently, the speckle images in the multi-view speckle image sequence are input one by one into a pre-trained speckle-to-two-dimensional morphology mapping model. The pre-trained speckle-to-2D topography mapping model is trained based on a deep generative network. It can directly infer the 2D topography corresponding to irregular particles 6 from the speckle image distribution and output a 2D topography image under this viewpoint information. The 2D topography image contains the outline of irregular particles 6. Finally, the output 2D topography image is grouped again according to the ID of irregular particles 6, while maintaining the same viewpoint order as the input. At the same time, it is bound to the projection parameters and timestamp of camera 7 to form a complete multi-view 2D topography dataset of particles, providing an ordered and traceable input for subsequent 3D topography reconstruction.

[0028] 5. Three-dimensional topography reconstruction and optimization After obtaining a complete dataset of multi-view 2D topography of particles, the 2D topography images from different perspectives were first spatially registered with the projection parameters and timestamps of camera 7 to ensure that the results from each perspective corresponded to the true position and orientation of the same irregular particle 6 in a unified coordinate system. Subsequently, the registered 2D topography images were input into a 3D topography reconstruction algorithm. The 2D topography images from each perspective were back-projected, and their intersection was calculated in 3D space to obtain a preliminary 3D voxel model of the irregular particle 6. To improve the completeness and accuracy of the reconstruction results, the 3D voxel model was further meshed, and surface discontinuities, burrs, and voids were eliminated using smoothing filtering and hole filling methods, ultimately forming geometrically continuous and topologically complete 3D topography data of irregular particles.

[0029] Example 2 Reference Figure 1 This invention proposes a rapid measurement device for the three-dimensional morphology of sand grains based on speckle imaging, comprising a pulsed laser 1, an incident laser beam 2, a reflecting mirror 3, a sheet light generator 4, a thin-film laser 5, a camera 7, a telephoto macro lens 8, and a post-processing unit 9 (i.e., a three-dimensional reconstruction processing unit). The pulsed laser 1, reflecting mirror 3, and sheet light generator 4 are combined to generate the thin-film laser 5, while the camera 7 and telephoto macro lens 8 form an image acquisition unit used to collect the laser speckle signal generated after the thin-film laser 5 irradiates irregular particles 6.

[0030] Reference Figure 2This invention employs a multi-view measurement camera layout, comprising five image acquisition units: a first image acquisition unit 13, a second image acquisition unit 14, a third image acquisition unit 15, a fourth image acquisition unit 16, and a fifth image acquisition unit 17. The first image acquisition unit 13, the second image acquisition unit 14, and the third image acquisition unit 15 are arranged within the same vertical reference plane 11, and are distributed at 120° equiangular intervals on the same circumference with the central reference point 12 as the center. The first image acquisition unit 13, the fourth image acquisition unit 16, and the fifth image acquisition unit 17 are located within the same horizontal reference plane 10, and are also arranged at 120° equiangular intervals on the same circumference with the central reference point 12 as the center.

[0031] Reference Figure 3 This invention employs a multi-view camera shooting fixing device to support and fix the image acquisition unit. The multi-view camera shooting fixing device uses a fixed base plate 18 as the main fixing and supporting component. Four support rods are welded on the fixed base plate 18, namely the first support rod 19, the second support rod 20, the third support rod 21, and the fourth support rod 22. A camera fixing plate 23 is installed above each support rod for fixing the image acquisition unit. The specific matching relationship is as follows: the first support rod 19 corresponds to the first image acquisition unit 13, the second support rod 20 corresponds to the fifth image acquisition unit 17, the third support rod 21 corresponds to the fourth image acquisition unit 16, and the fourth support rod 22 corresponds to the second image acquisition unit 14 and the third image acquisition unit 15.

[0032] Reference Figure 1 , Figure 2 and Figure 3 The implementation idea of ​​multi-view synchronous acquisition involved in this invention is as follows: a. A pulsed laser 1 generates an incident laser beam 2, which, after being oriented by a reflector 3, enters a sheet laser generator 4 and expands into a thin, controllable-thickness sheet laser 5. The thickness of this sheet laser 5 is adjustable from 500 to 3000 μm, capable of covering the particle size of common sand grains (i.e., irregular particles 6). The sheet laser 5 is arranged perpendicular to the falling direction of the irregular particles 6. When the irregular particles 6 fall freely through the sheet laser 5, the surface microstructure of the sheet laser 5 causes light scattering and phase difference effects, thereby generating a speckle distribution at the imaging end of the camera 7. The falling irregular particles 6 enter the measurement area through a funnel-shaped sand-falling device made of transparent quartz material. The outlet diameter of this funnel-shaped sand-falling device can be adjusted within the range of 2–10 mm to control the sparsity of the particle flow of irregular particles 6. The inner wall of the funnel channel of the funnel-type sand dropping device is smoothed to prevent irregular particles 6 from colliding with or rotating violently with the funnel channel wall, thus ensuring the stability of the falling trajectory of the irregular particle flow (formed by the aggregation of several irregular particles 6).

[0033] b. In Figure 2 In the camera layout shown, the device is equipped with five image acquisition units: a first image acquisition unit 13, a second image acquisition unit 14, a third image acquisition unit 15, a fourth image acquisition unit 16, and a fifth image acquisition unit 17. The first image acquisition unit 13, the second image acquisition unit 14, and the third image acquisition unit 15 are arranged within the same vertical reference plane 11, with a central reference point 12 as the center, and are distributed at 120° equiangular intervals on the circumference. The first image acquisition unit 13, the fourth image acquisition unit 16, and the fifth image acquisition unit 17 are located within the same horizontal reference plane 10, also with a central reference point 12 as the center, and are arranged at 120° equiangular intervals. This arrangement allows for the simultaneous acquisition of speckle information of irregular particles 6 from different viewing angles, ensuring complete coverage of the three-dimensional morphological features. Figure 3 The multi-view shooting fixing device shown uses a fixed base plate 18 as the main supporting component. Four support rods are welded on the fixed base plate 18, and a camera fixing plate 23 is installed above each support rod to install the corresponding image acquisition unit, so as to achieve stable and reliable positioning of the camera 7.

[0034] c. During the specific testing process, the sheet-like laser 5, along with the first image acquisition unit 13, the second image acquisition unit 14, the third image acquisition unit 15, the fourth image acquisition unit 16, and the fifth image acquisition unit 17, are all triggered by a synchronous control system to ensure that they work synchronously within a sub-millisecond time, thereby avoiding motion blur during the rapid fall of irregular particles 6. Before acquisition, the internal and external parameters of the camera 7 are jointly calibrated using a calibration board to establish accurate projection geometry. In the experiment, when the irregular particles 6 fall freely through the area of ​​the sheet-like laser 5, the first image acquisition unit 13, the second image acquisition unit 14, the third image acquisition unit 15, the fourth image acquisition unit 16, and the fifth image acquisition unit 17 synchronously acquire speckle images from different perspectives. During the acquisition process, background removal and grayscale normalization based on a sliding window are automatically performed, and high-frequency noise is suppressed through median filtering and wavelet denoising methods. Each irregular particle is assigned a unique ID during image acquisition, along with a timestamp and viewpoint information (the viewpoint information is the camera viewpoint label), thus ensuring that the data is traceable during subsequent 2D shape generation and 3D reconstruction, and that no particle information is lost.

[0035] Reference Figure 4 and Figure 5 The implementation idea of ​​the irregular particle three-dimensional morphology measurement involved in this invention (i.e., the implementation idea of ​​post-processing unit 9) is as follows: a. To address the nonlinear mapping relationship between speckle images and two-dimensional morphology of irregular particle 6 surfaces, a generative image translation model with a deep generative network as its backbone is constructed. The training data for this generative image translation model consists of a large number of irregular particle samples with known morphologies. Each irregular particle sample contains a corresponding speckle image and a real two-dimensional morphology image. During training, the speckle image is used as input, and the target two-dimensional morphology image is used as supervised output. The generative image translation model learns the spatial structure information contained in the speckle distribution through adversarial training and joint optimization of reconstruction error, and establishes a mapping relationship with the surface morphology of irregular particle 6, thus obtaining a speckle-to-two-dimensional morphology mapping model. The final speckle-to-two-dimensional morphology mapping model can directly accept speckle images of any irregular particle 6 as input during the inference stage and automatically generate the corresponding two-dimensional morphology image (which is a grayscale image).

[0036] b. After acquiring speckle images of irregular particles 6, the speckle images are first processed to obtain the ID, timestamp, and viewpoint information corresponding to each irregular particle 6 in the speckle images. Then, all speckle images are automatically grouped according to the unique ID of the irregular particle 6, so that speckle images of the same irregular particle 6 under different camera 7 views are classified into the same dataset. Within each group, they are arranged in an orderly manner according to the viewpoint information of camera 7, forming a multi-view speckle image sequence. Subsequently, the speckle images in the multi-view speckle image sequence are input one by one into the speckle-to-2D morphology mapping model, and the 2D morphology image corresponding to the multi-view speckle image sequence is output. The 2D morphology image contains the outline of the irregular particle 6. Finally, the output 2D morphology image is grouped again according to the ID of the irregular particle 6, maintaining the same viewpoint order as when it was input, and is bound to the projection parameters and timestamp of camera 7 to form a complete particle multi-view 2D morphology dataset.

[0037] c. After obtaining a complete multi-view 2D topography dataset of particles, 2D topography images from five different viewpoints are first used as input. Combined with the intrinsic parameters (focal length, principal point coordinates, distortion parameters) and extrinsic parameters (position and pose information) of camera 7, the 2D topography images from each viewpoint are uniformly registered to the same 3D coordinate system, ensuring that all data correspond to the same irregular particle 6. Subsequently, each 2D topography image is binarized and edge-detected to extract the 2D projection contour of the irregular particle 6 under the viewpoint, forming the corresponding contour mask. Based on this, the 2D projection contour is back-projected into 3D space using the camera projection matrix, generating a view frustum containing all possible shapes. By performing a spatial Boolean intersection operation on the five view frustums, a 3D envelope consistent with the 2D projection contours of all viewpoints is obtained. The resulting 3D envelope (i.e., the intersection region) is further discretized into a regular voxel mesh, and voxels are labeled according to their internal and external relationships, thus obtaining a preliminary 3D voxel model. Subsequently, a voxel-to-surface conversion algorithm was used to transform the 3D voxel model into a 3D mesh surface. This was then optimized using post-processing techniques such as smoothing filtering, hole filling, and redundant vertex removal to obtain continuous, smooth, and geometrically complete 3D morphological data of irregular particles. The final output of this irregular particle 3D morphological data can be directly applied to numerical simulations, erosion mechanism research, or 3D database construction.

[0038] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit its scope of protection. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that after reading the present invention, they can still make various changes, modifications or equivalent substitutions to the specific implementation of the invention, but these changes, modifications or equivalent substitutions are all within the scope of protection of the pending claims of the invention.

Claims

1. A device for measuring the three-dimensional topography of irregular particles based on speckle imaging, characterized in that, The system comprises: an irregular particle sanding device for making irregular particles (6) fall in a sparse free fall form; a laser sheet generating module for forming a thin sheet laser (5) perpendicular to the falling direction of the irregular particles (6) to form a complete speckle distribution on the surface of the irregular particles (6); a multi-view image acquisition module for multi-view acquisition of the speckle distribution on the surface of the irregular particles (6) to form a speckle image; a synchronous control system for controlling the laser sheet generating module and the multi-view image acquisition module and synchronizing the two; a three-dimensional reconstruction processing unit for processing the speckle image to obtain the ID, timestamp and perspective information of each irregular particle (6) in the speckle image; based on a pre-trained mapping model of speckle to two-dimensional morphology and the ID, timestamp and perspective information of each irregular particle (6), the collected speckle image is mapped to obtain a particle multi-view two-dimensional morphology data set; based on the particle multi-view two-dimensional morphology data set, the three-dimensional morphology reconstruction and optimization of the irregular particles are completed through a three-dimensional morphology reconstruction algorithm to obtain the three-dimensional morphology data of the irregular particles.

2. The device for measuring the three-dimensional topography of irregular particles based on speckle imaging according to claim 1, characterized in that, The irregular particle sanding device is a funnel-shaped sanding device made of transparent quartz or acrylic material, and the funnel-shaped channel outlet diameter of the funnel-shaped sanding device is 2-10 mm.

3. The device for measuring the three-dimensional topography of irregular particles based on speckle imaging according to claim 1, characterized in that, The laser sheet generating module comprises a pulse laser (1), a reflector (3) and a sheet light generator (4), the reflector (3) is used for reflecting the incident laser beam (2) emitted by the pulse laser (1) to the sheet light generator (4), and the sheet light generator (4) expands the incident laser beam (2) into a thin sheet laser (5) perpendicular to the falling direction of the irregular particles (6).

4. The device for measuring the three-dimensional topography of irregular particles based on speckle imaging according to claim 1, characterized in that, The multi-view image acquisition module comprises a plurality of cameras (7) and long focus micro lenses (8) installed along both sides of the thin sheet laser (5) and at different pitch angles, and the camera (7) and the long focus micro lens (8) correspond to each other, and each camera (7) and the corresponding long focus micro lens (8) form an image acquisition unit.

5. The device for measuring the three-dimensional topography of irregular particles based on speckle imaging according to claim 4, characterized in that, The image acquisition unit is provided with five image acquisition units, including a first image acquisition unit (13), a second image acquisition unit (14), a third image acquisition unit (15), a fourth image acquisition unit (16) and a fifth image acquisition unit (17); The first image acquisition unit (13), the second image acquisition unit (14) and the third image acquisition unit (15) are arranged in the same vertical reference surface (11), and the center reference point (12) is taken as the center, and they are distributed at an angle of 120° on the circumference; The first image acquisition unit (13), the fourth image acquisition unit (16) and the fifth image acquisition unit (17) are located in the same horizontal reference surface (10), and the center reference point (12) is taken as the center, and they are distributed at an angle of 120° on the circumference; The center reference point (12) is the geometric center of the equilateral triangle composed of the three image acquisition units in the same vertical reference surface (11) or the three image acquisition units in the same horizontal reference surface (10).

6. The device for measuring the three-dimensional topography of irregular particles based on speckle imaging according to claim 5, characterized in that, The image acquisition unit is supported and fixed by a multi-view shooting fixing device, the multi-view shooting fixing device comprises a fixed bottom plate (18), four support rods are fixed on the fixed bottom plate (18), which are a first support rod (19), a second support rod (20), a third support rod (21) and a fourth support rod (22), and a camera fixing plate (23) is installed above each support rod; The camera fixing plate (23) on the first support rod (19) is used for fixing the camera (7) of the first image acquisition unit (13), the camera fixing plate (23) on the second support rod (20) is used for fixing the camera (7) of the fifth image acquisition unit (17), the camera fixing plate (23) on the third support rod (21) is used for fixing the camera (7) of the fourth image acquisition unit (16), and the camera fixing plate (23) on the fourth support rod (22) is used for fixing the cameras (7) of the second image acquisition unit (14) and the third image acquisition unit (15).

7. The device for measuring the three-dimensional topography of irregular particles based on speckle imaging according to claim 1, characterized in that, The three-dimensional reconstruction processing unit comprises a speckle image processing module, a two-dimensional topography reconstruction module and a three-dimensional topography reconstruction module; The speckle image processing module is used for automatically cropping and feature extraction of the speckle image, and obtaining the ID, timestamp and perspective information corresponding to each irregular particle (6) in the speckle image; The two-dimensional topography reconstruction module is used for automatically grouping the speckle images according to the ID of the irregular particle (6), so that the speckle images of the same irregular particle (6) under different perspective information are classified into the same data set, and the speckle images are sequentially arranged in each group according to the perspective information, forming a multi-perspective speckle image sequence, inputting the multi-perspective speckle image sequence into a pre-trained speckle-to-two-dimensional topography mapping model, obtaining a two-dimensional topography image corresponding to the multi-perspective speckle image sequence, grouping the two-dimensional topography image again according to the ID of the irregular particle (6), and keeping the same perspective order as when inputting, binding with the timestamp and the projection parameters of the multi-perspective acquisition module, and forming a particle multi-perspective two-dimensional topography data set; The three-dimensional topography reconstruction module is used for uniformly registering the two-dimensional topography images of each perspective in the particle multi-perspective two-dimensional topography data set into the same three-dimensional coordinate system, binarizing and edge detecting each two-dimensional topography image, extracting the two-dimensional projection contour of the irregular particle (6) under the perspective, back-projecting the two-dimensional projection contour to the three-dimensional space, generating several view pyramids containing all possible shapes, performing a spatial Boolean intersection operation on the several view pyramids, obtaining a three-dimensional envelope body consistent with all two-dimensional projection contours of the perspective, further discretizing the obtained three-dimensional envelope body into a regular voxel grid, marking the voxels according to the regular voxel grid, and obtaining a preliminary three-dimensional voxel model, converting the three-dimensional voxel model into a three-dimensional mesh surface by using a voxel-to-surface conversion algorithm, and optimizing the three-dimensional mesh surface to obtain the three-dimensional topography data of the irregular particle. The pre-trained speckle-to-two-dimensional topography mapping model is trained based on a deep generation network and known topography irregular particle samples.

8. A method for measuring the three-dimensional topography of irregular particles based on speckle imaging, based on the device for measuring the three-dimensional topography of irregular particles based on speckle imaging according to any one of claims 1-7, characterized in that, The method comprises the following steps: The irregular particles (6) are in the form of sparse free fall; Forming a sheet-shaped laser (5) perpendicular to the falling direction of the irregular particles (6) to form a complete speckle distribution on the surface of the irregular particles (6), and collecting the speckle distribution on the surface of the irregular particles (6) through multi-view to form a speckle image; Processing the speckle image to obtain the ID, timestamp and view information corresponding to each irregular particle (6) in the speckle image, mapping the collected speckle image based on the pre-trained mapping model from speckle to two-dimensional morphology and the ID, timestamp and view information corresponding to each irregular particle (6), obtaining a particle multi-view two-dimensional morphology data set, and completing the three-dimensional morphology reconstruction and optimization of the irregular particles through a three-dimensional morphology reconstruction algorithm based on the particle multi-view two-dimensional morphology data set to obtain three-dimensional morphology data of the irregular particles.

9. The method according to claim 8, wherein, The speckle distribution on the surface of the irregular particles (6) is collected through multi-view to form a speckle image, specifically as follows: When the irregular particles (6) fall freely through the area of the sheet-shaped laser (5), the speckle distribution on the surface of the irregular particles (6) is collected through different views synchronously, and the background is automatically removed, the grayscale is normalized based on a sliding window, and high-frequency noise is suppressed through median filtering and wavelet denoising during the collection process to form a speckle image.

10. The method according to claim 8, wherein, The speckle image is processed to obtain the ID, timestamp and view information corresponding to each irregular particle (6) in the speckle image, the collected speckle image is mapped based on the pre-trained mapping model from speckle to two-dimensional morphology and the ID, timestamp and view information corresponding to each irregular particle (6), a particle multi-view two-dimensional morphology data set is obtained, and three-dimensional morphology reconstruction and optimization of the irregular particles are completed through a three-dimensional morphology reconstruction algorithm based on the particle multi-view two-dimensional morphology data set to obtain three-dimensional morphology data of the irregular particles, specifically including: The speckle image is automatically cropped and feature-extracted to obtain the ID, timestamp and view information corresponding to each irregular particle (6) in the speckle image; The speckle image is automatically grouped according to the ID of the irregular particles (6), so that the speckle images of the same irregular particle (6) under different view information are classified into the same data set, and are sequentially arranged in each group according to the view information to form a multi-view speckle image sequence, the multi-view speckle image sequence is input into the pre-trained mapping model from speckle to two-dimensional morphology to obtain the two-dimensional morphology image corresponding to the multi-view speckle image sequence, the two-dimensional morphology image is again grouped according to the ID of the irregular particles (6) and maintains the same view order as when inputting, and is bound with the timestamp and the projection parameters of the multi-view collection module to form a particle multi-view two-dimensional morphology data set; The two-dimensional morphology images of each view in the particle multi-view two-dimensional morphology data set are uniformly registered in the same three-dimensional coordinate system, each two-dimensional morphology image is binarized and edge detected, the two-dimensional projection contour of the irregular particle (6) under the view is extracted, the two-dimensional projection contour is back projected to the three-dimensional space, a plurality of view cones containing all possible shapes are generated, the plurality of view cones are subjected to spatial Boolean intersection operation, a three-dimensional envelope body consistent with the two-dimensional projection contours of all views is obtained, the obtained three-dimensional envelope body is further discretized into a regular voxel grid, the voxel is marked according to the regular voxel grid, and a preliminary three-dimensional voxel model is obtained, the three-dimensional voxel model is converted into a three-dimensional grid surface by using a voxel-to-surface conversion algorithm, the three-dimensional grid surface is subjected to optimization processing, and a three-dimensional morphology data of the irregular particle is obtained. The pre-trained speckle-to-two-dimensional morphology mapping model is obtained based on a deep generation network and irregular particle samples with known morphologies.

Images