An armature attitude diagnostic method, system, device, medium, and program product.

Abstract

This invention discloses an armature attitude diagnosis method, system, device, medium, and program product, belonging to the field of electromagnetic launch technology. The invention proposes a method combining deep learning networks and contour fitting. Based on orthogonal two-axis projection images and combined with a three-dimensional point cloud model of the armature, a large amount of armature projection data is generated within the imaging system. The armature attitude information is predicted using a convolutional neural network, and the model attitude is adjusted based on the predicted information. Under this predicted attitude, the model attitude is fine-tuned within a small range. The attitude adjustment of the three-dimensional object is achieved by contour fitting of two-dimensional images in two directions. This solves the problems of time-consuming manual labeling and difficulty in quantifying errors.

Description

Technical Field

[0001] This invention belongs to the field of electromagnetic launch technology and relates to an armature attitude diagnosis method, system, device, medium, and program product. Background Technology

[0002] Electromagnetic orbital launch is a novel launch technology that uses electromagnetic force (energy) to propel projectiles to high or hypersonic speeds. It can break through the speed limits of traditional chemical launch methods, achieving muzzle velocities exceeding 2 km / s. It boasts significant advantages such as high launch kinetic energy, high system efficiency, high emission frequency, fast start-up time, strong sustained launch capability, and good security and stealth, making it a promising technology for military applications. However, the electromagnetic launch process is often accompanied by strong arc light and smoke interference, hindering image diagnostics based on visible light imaging. X-rays can penetrate arc light and smoke interference, enabling imaging diagnostics of the muzzle or internal armature during electromagnetic launch. Currently, researchers have used X-ray flash radiography to obtain armature images for observing armature deformation, cracking, and attitude changes. However, there are few case studies on processing armature images obtained through flash radiography.

[0003] Currently, the main method for imaging and attitude calculation of armatures during electromagnetic emission using flash imaging is feature point matching. This method works by imaging the armature from different perspectives, manually marking landmarks on the images at different angles, using the armature's 3D structure as a priori condition, and marking corresponding 3D landmarks on the object's 3D model. Through registration of multiple 2D images and 3D object landmarks, the object's rigidity transformation matrix is ​​calculated, enabling 3D attitude calculation for high-speed moving objects. However, this landmark-based cross-membrane registration method requires manual, experience-based landmark marking, which introduces significant errors that cannot be quantified. Furthermore, for complex objects, hundreds of landmarks are needed for registration, making the marking process time-consuming and labor-intensive. Additionally, imaging from different angles may result in complex structures covering landmarks or overlapping landmarks along different imaging directions, all of which affect the accuracy of attitude calculation. Summary of the Invention

[0004] The purpose of this invention is to solve the technical problems in the prior art and to provide an armature attitude diagnosis method, system, device, medium and program product.

[0005] To achieve the above objectives, the present invention employs the following technical solution:

[0006] An armature attitude diagnosis method includes the following steps:

[0007] Acquire actual captured images, and input these images into a pre-trained convolutional neural network to obtain the first pose of the armature point cloud model; the training method for the pre-trained convolutional neural network includes:

[0008] Obtain a training dataset, which is the projection of the armature point cloud model onto the two axes of a pre-established imaging system;

[0009] The training dataset is input into the convolutional neural network for training, resulting in a trained convolutional neural network.

[0010] Based on the first pose of the armature point cloud model, the pose of the armature point cloud model is adjusted by comparing the similarity between the projected contour of the armature point cloud model and the contour of the actual captured image, thus obtaining the second pose of the armature point cloud model.

[0011] The second pose of the armature point cloud model is obtained by comparing the scale difference between the projected contour of the armature point cloud model and the contour of the actual captured image, and adjusting the pose of the armature point cloud model.

[0012] Furthermore, the imaging system includes: two point sources, two imaging plates, and an object to be photographed; the lines connecting the two point sources to the centers of the corresponding imaging plates are perpendicular to each other; the centroid of the object to be photographed, the point sources, and the center of the imaging plates are arranged in a straight line; the origin of the three-dimensional coordinate system is the position of any point source.

[0013] Preferably, the step of acquiring the actual captured image and inputting the actual captured image into a pre-trained convolutional neural network to obtain the first pose of the armature point cloud model specifically involves:

[0014] The actual captured image is input into a pre-trained convolutional neural network to obtain the predicted armature angle information;

[0015] Based on the novel armature angle prediction, the rotation angle of the armature point cloud model is adjusted to obtain the first pose of the armature point cloud model.

[0016] Preferably, the first pose based on the armature point cloud model is obtained by adjusting the pose of the armature point cloud model by comparing the similarity between the projected contour of the armature point cloud model and the contour of the actual captured image, and the specific method is as follows:

[0017] Project the armature point cloud model into the imaging system and extract the projected contour of the armature point cloud model; extract the contour of the actual captured image based on the actual captured image.

[0018] Rotate the armature point cloud model around the X, Y, and Z axes with the same step size, and compare the similarity between the projected contour and the actual captured image contour until the closest projected contour is found.

[0019] Based on the armature angle corresponding to the closest projected profile, the rotation angle of the armature point cloud model is adjusted to obtain the second pose of the armature point cloud model.

[0020] Preferably, the second pose based on the armature point cloud model, which adjusts the pose of the armature point cloud model by comparing the scale difference between the projected contour of the armature point cloud model and the contour of the actual captured image, to obtain the optimal pose of the armature point cloud model, is specifically achieved as follows:

[0021] The armature point cloud model is projected into the imaging system, and the projected contour of the armature point cloud model is extracted; the contour of the actual captured image is extracted based on the actual captured image.

[0022] The two-dimensional coordinate point set of the reprojected contour and the actual captured image contour is transformed into a three-dimensional coordinate point set;

[0023] Based on the three-dimensional coordinate point set of the reprojected contour and the actual captured image contour, the three-dimensional coordinate point set of the reprojected contour that is closest to the actual captured image contour is found using the ICP algorithm, and the optimal rigid transformation matrix is ​​obtained.

[0024] The attitude of the armature point cloud model is adjusted based on the optimal rigid transformation matrix;

[0025] Calculate the scale difference between the reprojected contour and the actual captured image contour, where the scale difference is the area difference S between the two contours;

[0026] Iterate the above steps until S converges to S∈(0.99,1.01); obtain the pose of the optimal armature point cloud model, which is the pose of the armature point cloud model in the last iteration.

[0027] Furthermore, the rigid transformation matrix is:

[0028]

[0029] Where R is a 3*3 rotation matrix and T is a 3*1 translation vector.

[0030] A computer program product comprising computer instructions that instruct a computer to perform the aforementioned armature attitude diagnosis method.

[0031] An armature attitude diagnostic system, comprising:

[0032] An armature attitude prediction module is used to acquire actual captured images, input the actual captured images into a pre-trained convolutional neural network, and obtain the first attitude of the armature point cloud model; the training method of the pre-trained convolutional neural network includes:

[0033] Obtain a training dataset, which is the projection of the armature point cloud model onto the two axes of a pre-established imaging system;

[0034] The training dataset is input into the convolutional neural network for training, resulting in a trained convolutional neural network.

[0035] The armature attitude initial adjustment module, based on the first attitude of the armature point cloud model, adjusts the attitude of the armature point cloud model by comparing the similarity between the projected contour of the armature point cloud model and the contour of the actual captured image, and obtains the second attitude of the armature point cloud model.

[0036] The armature attitude fine-tuning module, based on the second attitude of the armature point cloud model, adjusts the attitude of the armature point cloud model by comparing the scale difference between the projected contour of the armature point cloud model and the contour of the actual captured image, thereby obtaining the optimal attitude of the armature point cloud model.

[0037] An electronic device includes a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor performs the steps of the armature attitude diagnosis method described above.

[0038] A computer-readable storage medium having a computer program stored thereon, which, when executed, implements the above-described armature attitude diagnosis method.

[0039] Compared with the prior art, the present invention has the following beneficial effects:

[0040] This invention combines convolutional neural networks and contour fitting methods. It predicts the pose information of the armature point cloud model through convolutional neural networks, and then uses contour fitting to make small-range adjustments to the pose of the armature point cloud model based on the pose. This eliminates the need for manual annotation of images from different angles, saving manpower. Furthermore, this invention uses two-axis contour constraints to adjust the pose of the armature point cloud model and achieves convergence through iterative calculation, which can reduce the errors caused by manual marking of landmarks.

[0041] Furthermore, by directly transforming two-dimensional contour points into three dimensions, the present invention can directly perform ICP iteration to obtain the rigid transformation matrix of the armature without the need for complex conversion from two-dimensional to three-dimensional transformation matrices, achieving pixel-level calculation accuracy and further reducing armature attitude calculation errors. Attached Figure Description

[0042] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0043] Figure 1 This is a block diagram of the armature attitude diagnosis method of the present invention;

[0044] Figure 2 This is a schematic diagram of the armature attitude diagnosis method of the present invention;

[0045] Figure 3 This is a graph showing the angle error curves of each axis in the armature attitude diagnosis method of the present invention.

[0046] Figure 4 This is the attitude reconstruction result of the electromagnetic transmission high-speed armature based on flash photography technology in this invention;

[0047] Figure 5 This is a block diagram of the armature attitude diagnosis system of the present invention. Detailed Implementation

[0048] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and marked in the accompanying drawings can generally be arranged and designed in various different configurations.

[0049] Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0050] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0051] In the description of the embodiments of the present invention, it should be noted that if terms such as "upper," "lower," "horizontal," or "inner" indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, or the orientation or positional relationship commonly used when the product of the invention is in use, they are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the present invention. Furthermore, terms such as "first" and "second" are only used to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0052] Furthermore, the use of the term "horizontal" does not imply that the component must be absolutely horizontal, but rather that it can be slightly tilted. For example, "horizontal" simply means that its direction is more horizontal than "vertical," and does not mean that the structure must be completely horizontal, but can be slightly tilted.

[0053] In the description of the embodiments of the present invention, it should also be noted that, unless otherwise explicitly specified and limited, the terms "set," "install," "connect," and "link" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in the present invention according to the specific circumstances.

[0054] The present invention will now be described in further detail with reference to the accompanying drawings:

[0055] See Figure 1 This invention discloses an armature attitude diagnosis method, comprising the following steps:

[0056] S1, acquire the actual captured image, input the actual captured image into a pre-trained convolutional neural network to obtain the first pose of the armature point cloud model; the convolutional neural network training method includes:

[0057] Obtain a training dataset, which is the projection of the armature point cloud model onto the two axes of a pre-established imaging system;

[0058] The training dataset is input into the convolutional neural network for training, resulting in a trained convolutional neural network.

[0059] S2, based on the first pose of the armature point cloud model, by comparing the similarity between the projected contour of the armature point cloud model and the contour of the actual captured image, the pose of the armature point cloud model is adjusted to obtain the second pose of the armature point cloud model.

[0060] S3, based on the second pose of the armature point cloud model, adjusts the pose of the armature point cloud model by comparing the scale difference between the projected contour of the armature point cloud model and the contour of the actual captured image, and obtains the optimal pose of the armature point cloud model.

[0061] In one embodiment, such as Figure 2 As shown, an armature attitude diagnosis method is provided, including the following steps:

[0062] S1. Based on the pre-established imaging system and three-dimensional coordinate system, establish the armature point cloud model;

[0063] An imaging system and armature point cloud model are pre-established: The imaging system consists of two point sources, two imaging plates, and an armature point cloud model. The initial state of the imaging system is as follows: the point sources, the centroid of the object being photographed, and the center of the imaging plate are aligned in a straight line, and the lines connecting the two sets of point sources and the corresponding imaging plate centers are perpendicular to each other; the entire imaging system is an orthogonal two-axis imaging setup.

[0064] The imaging system requires determining the distance between the point source and the armature point source model (SOD) and the distance between the point source and the imaging plate (SDD). The dimensions and resolution of the two imaging plates must be determined, and both plates must be identical in size. The SOD and SDD distances must be kept consistent along both axes to ensure the same magnification.

[0065] Select one of the point source locations as the origin of the imaging system coordinates, establish a three-dimensional coordinate system, and determine the coordinates of the two point sources, the corresponding imaging plate, and the object being photographed in the coordinate system; establish a three-dimensional point cloud model of the object being photographed in the three-dimensional coordinate system of the imaging system, and the object being photographed in this invention is the armature.

[0066] S2, acquire the actual captured image, input the actual captured image into a pre-trained convolutional neural network to obtain armature angle prediction information; adjust the pose of the armature point cloud model according to the armature angle prediction information to obtain the first pose of the armature point cloud model.

[0067] S201, First, the convolutional neural network is trained. The training method for the convolutional neural network is as follows:

[0068] Within the established imaging system, the positions of the point source and imaging plate are kept fixed along both axes. Based on the initial positions, the attitude of the armature point cloud model is adjusted by externally rotating the armature around the X, Y, and Z axes at arbitrary angles to generate projection data. The method for generating the projection data is as follows: code is written to calculate the projection coordinates of the armature point cloud model using point projection; then, code is written using the OpenCV library to convert the projection coordinates into pixel coordinates, generating a binary image with a size of 1600*1600 pixels and a pixel size of 40 pixels / mm. Then, the signal-to-noise ratio (SNR) of the actual experimentally captured image is calculated. Code is written using the OpenCV library to add Poisson noise and Gaussian noise to ensure that the generated image has the same SNR as the actual captured image, simulating the quality of the X-ray image obtained in the actual experiment.

[0069] The ResNet34 convolutional neural network parameters were configured as follows: The input layer was modified to accommodate the training data. The training data format consisted of two images per group, with labels representing the corresponding rotation angles around each axis (XYZ order), and the output being the predicted rotation angles around each axis. The loss function was set to the root mean square error between the predicted value and the label. Approximately 30,000 sets of data (i.e., binary images with the same signal-to-noise ratio as those obtained from actual experiments) were generated using the S2 method. These binary images were randomly divided into training and test sets at a ratio of 9:1 to train the convolutional neural network. During training, the initial learning rate was set to 0.01, decreasing to 0.001 after 30 epochs, and then to 0.0001 after the 50th epoch, for a total of 72 epochs. The batch size was set to 128.

[0070] S202, the actual captured X-ray image is input into the trained convolutional neural network. The actual captured X-ray image is the image obtained through orthogonal imaging. The convolutional neural network outputs three values: angles a, b, and c corresponding to the rotation of the armature point cloud model around the imaging system coordinate system in XYZ order. A rotation matrix R is calculated based on the output angle values. A program is written using the PCL library to multiply the armature point cloud model from its initial position by the rotation matrix R, adjusting the armature point cloud model's pose to obtain its first pose. The rotation matrix is ​​expressed as:

[0071]

[0072] In the formula, a, b, and c represent the angles of rotation of the armature point cloud model around the X, Y, and Z axes, respectively.

[0073] S3, based on the first pose of the armature point cloud model, the armature point cloud model is projected into the imaging system, and the projection contour of the armature point cloud model is extracted; the contour of the actual captured image is extracted; the similarity between the projection contour of the armature point cloud model and the contour of the actual captured image is compared, and the closest armature angle is found; based on the closest armature angle, the pose of the armature point cloud model is adjusted to obtain the second pose of the armature point cloud model.

[0074] The armature point cloud model in the first pose is projected from two directions, and the projected contours are extracted. The projected contours and the contours of the actual X-ray image are rotated in a small range around the XYZ axes with a certain step size. The similarity of the contours of the two-axis images is compared by the matchshape function in the OpenCV library to find the armature angle with the closest contour (determined according to the contour matching score). The pose of the armature point cloud model is adjusted to obtain the second pose of the armature point cloud model. The step size can be set according to the actual situation. The smaller the step size, the more accurate the result. However, the calculation speed will be slower as the amount of data increases.

[0075] S4. Based on the second pose of the armature point cloud model, the armature point cloud model is projected into the imaging system to obtain the reprojected outline of the armature point cloud model; the scale difference between the reprojected outline of the armature point cloud model and the outline of the actual captured image is compared to find the optimal rigid transformation matrix; based on the optimal rigid transformation matrix, the pose of the optimal armature point cloud model is obtained.

[0076] S401, the armature point cloud model is reprojected within the imaging system, and the reprojected outline of the armature point cloud model is extracted; based on the actual captured image, the outline of the actual captured image is extracted; the outline scale difference S between the reprojected outline and the actual X-ray image outline is calculated (by comparing the length or area of ​​the two outlines); then, the two-dimensional outline point set is converted into a three-dimensional coordinate point set in the imaging system coordinate system; the method for converting the two-dimensional outline point set into a three-dimensional coordinate point set is as follows: according to the position of the imaging plate, the missing third value of the coordinate point is assigned to obtain the three-dimensional coordinate point set of the two outlines.

[0077] S402, then ICP (Iterative Closest Point Algorithm) calculation is performed using the three-dimensional coordinate point set of the reprojected contour and the actual X-ray image contour to find the optimal matching parameters R and T of the three-dimensional point set that is closest to the actual X-ray image contour, and the rigid transformation matrix H is obtained (the translation vector needs to be adjusted according to the scale difference); the pose of the point cloud model is adjusted by multiplying the three-dimensional point set of the armature point cloud model by the rigid transformation matrix; the above process is iterated until the scale difference S between the reprojected contour of the armature point cloud model and the actual X-ray image contour converges within the set threshold (S∈(0.99, 1.01)).

[0078] Based on the above method, the same operation is performed in two orthogonal directions of the imaging system, and iterative calculations are performed sequentially to obtain the optimal rigid transformation matrix and the attitude of the optimal armature point cloud model.

[0079] The rigid transformation matrix is:

[0080]

[0081] Where R is a 3*3 rotation matrix and T is a 3*1 translation vector.

[0082] The error function of the ICP algorithm is:

[0083]

[0084] Where n is the number of contour coordinate points, q i p represents the coordinates of the actual wheel image contour points. i These are the coordinates of the projected contour points.

[0085] The optimal rigid transformation matrix is ​​the product of the initial rigid transformation matrix and all rigid transformation matrices calculated during the iteration process; the pose of the optimal armature point cloud model is the pose of the armature point cloud model in the last iteration.

[0086] To further illustrate the superiority of the armature attitude diagnosis method of this invention, firstly, Euler angle decomposition is performed on the optimal rigid transformation matrix obtained by this invention to obtain the angle parameters of the armature point cloud model's rotation around the X, Y, and Z axes, and the translation vector is extracted, which is the displacement of the armature point cloud model; then, the point cloud model reconstructed using the armature attitude diagnosis method of this invention is compared with a standard point cloud model; the angle error and displacement error of the armature point cloud model reconstructed by this invention are analyzed respectively, and the results are as follows. Figure 3 and Figure 4 As shown, the armature attitude diagnosis method of the present invention has a fast calculation speed and high accuracy.

[0087] In one embodiment, such as Figure 5 As shown, this invention discloses an armature attitude diagnostic system, comprising:

[0088] An armature attitude prediction module is used to acquire actual captured images, input the actual captured images into a pre-trained convolutional neural network, and obtain the first attitude of the armature point cloud model; the training method of the pre-trained convolutional neural network includes:

[0089] Obtain a training dataset, which is the projection of the armature point cloud model onto the two axes of a pre-established imaging system;

[0090] The training dataset is input into the convolutional neural network for training, resulting in a trained convolutional neural network.

[0091] The armature attitude initial adjustment module, based on the first attitude of the armature point cloud model, adjusts the attitude of the armature point cloud model by comparing the similarity between the projected contour of the armature point cloud model and the contour of the actual captured image, and obtains the second attitude of the armature point cloud model.

[0092] The armature attitude fine-tuning module, based on the second attitude of the armature point cloud model, adjusts the attitude of the armature point cloud model by comparing the scale difference between the projected contour of the armature point cloud model and the contour of the actual captured image, thereby obtaining the optimal attitude of the armature point cloud model.

[0093] In one embodiment, the present invention provides a computer program product or computer program including computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the steps described in the method embodiments above.

[0094] The computer program product or computer program provided by this invention is a software solution specifically designed for armature attitude diagnosis. It integrates advanced image processing, machine learning, and 3D point cloud processing technologies, enabling it to automatically analyze actually captured images and quickly and accurately determine the armature's attitude information based on these images. The computer program or computer program product of this invention is designed to be stored in various computer-readable storage media, including but not limited to hard disks, solid-state drives, USB flash drives, optical discs, and cloud storage. When the processor of a computer device reads and executes these computer instructions from the computer-readable storage media, it will automatically execute the steps in the above-described method embodiments, thereby achieving automated and intelligent diagnosis of the armature's attitude.

[0095] In one embodiment, the present invention provides an apparatus including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described method embodiments.

[0096] The device integrates a high-performance memory and processor to ensure efficient execution of complex image processing, machine learning, and 3D point cloud processing tasks. The memory stores key data such as pre-trained convolutional neural network models, armature point cloud models, training datasets, and actual captured images. The processor has powerful computing capabilities, enabling it to quickly execute complex image processing and machine learning algorithms, support multi-task parallel processing, and process multiple image data simultaneously.

[0097] In another example, the device also includes an image input interface and a data transmission interface. The image input interface supports importing actually captured images from external devices such as cameras and scanners. The data transmission interface is used for data exchange with computers, other devices, or networks.

[0098] In one embodiment, the present invention provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0099] The computer-readable storage medium of the present invention includes: read-only memory (ROM), random access programmable read-only memory (PROM), electrically erasable programmable read-only memory (EEPROM), random access memory (RAM), dynamic random access memory (DRAM), static random access memory (SRAM), flash memory, non-volatile memory, CD-ROM, CD-R, CD+R, CD-RW, CD+RW, DVD-ROM, DVD-R, DVD+R, DVD-RW, DVD+RW, DVD-RAM, BD-ROM, BD-R, BD-RLTH, BD-RE, Blu-ray or optical disc storage, hard disk drive (HDD), solid-state drive (SSD), card memory (such as multimedia cards or security digital (SD) cards), magnetic tape, floppy disk, magneto-optical data storage device, optical data storage device, hard disk, solid-state drive, and any other device configured to store a computer program and any associated data, data files, and data structures in a non-transitory manner and to provide the computer program and any associated data, data files, and data structures to a processor or computer such that the processor or computer can execute the computer program. The computer program in the aforementioned computer-readable storage medium can run in an environment deployed in computer devices such as clients, hosts, agent devices, and servers. Furthermore, the computer program, along with any associated data, data files, and data structures, is distributed across a networked computer system, enabling the computer program, along with any associated data, data files, and data structures, to be stored, accessed, and executed in a distributed manner via one or more processors or computers.

[0100] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for armature attitude diagnosis, characterized in that, Includes the following steps: Acquire the actual captured image, and input the actual captured image into a pre-trained convolutional neural network to obtain the first pose of the armature point cloud model; The pre-trained convolutional neural network training method includes: Obtain a training dataset, which is the projection of the armature point cloud model onto the two axes of a pre-established imaging system; The training dataset is input into the convolutional neural network for training, resulting in a trained convolutional neural network. Based on the first pose of the armature point cloud model, the pose of the armature point cloud model is adjusted by comparing the similarity between the projected contour of the armature point cloud model and the contour of the actual captured image, thus obtaining the second pose of the armature point cloud model. The second pose based on the armature point cloud model involves adjusting the pose of the armature point cloud model by comparing the scale difference between the projected contour of the armature point cloud model and the contour of the actual captured image, to obtain the optimal pose of the armature point cloud model. This includes: The armature point cloud model is projected into the imaging system, and the projected contour of the armature point cloud model is extracted; the contour of the actual captured image is extracted based on the actual captured image. The two-dimensional coordinate point set of the reprojected contour and the actual captured image contour is transformed into a three-dimensional coordinate point set; Based on the three-dimensional coordinate point set of the reprojected contour and the actual captured image contour, the three-dimensional coordinate point set of the reprojected contour that is closest to the actual captured image contour is found using the ICP algorithm, and the optimal rigid transformation matrix is ​​obtained. The attitude of the armature point cloud model is adjusted based on the optimal rigid transformation matrix; Calculate the scale difference between the reprojected contour and the actual captured image contour, where the scale difference is the area difference S between the two contours; Iterate through the above steps until S converges to The pose of the optimal armature point cloud model is obtained, wherein the pose of the optimal armature point cloud model is the pose of the armature point cloud model in the last iteration.

2. The armature attitude diagnosis method according to claim 1, characterized in that, The imaging system includes: two point sources, two imaging plates, and the object being photographed; the lines connecting the two point sources to the centers of the corresponding imaging plates are perpendicular to each other; the centroid of the object being photographed, the point sources, and the center of the imaging plates are arranged in a straight line.

3. The armature attitude diagnosis method according to claim 1, characterized in that, The process of acquiring the actual captured image and inputting it into a pre-trained convolutional neural network to obtain the first pose of the armature point cloud model is as follows: The actual captured image is input into a pre-trained convolutional neural network to obtain the predicted armature angle information; Based on the predicted armature angle information, the rotation angle of the armature point cloud model is adjusted to obtain the first pose of the armature point cloud model.

4. The armature attitude diagnosis method according to claim 1, characterized in that, The first pose based on the armature point cloud model is obtained by comparing the similarity between the projected contour of the armature point cloud model and the contour of the actual captured image, and then adjusting the pose of the armature point cloud model. The specific method is as follows: Project the armature point cloud model into the imaging system and extract the projected contour of the armature point cloud model; extract the contour of the actual captured image based on the actual captured image. Rotate the armature point cloud model around the X, Y, and Z axes with the same step size, and compare the similarity between the projected contour and the actual captured image contour until the closest projected contour is found. Based on the armature angle corresponding to the closest projected profile, the rotation angle of the armature point cloud model is adjusted to obtain the second pose of the armature point cloud model.

5. The armature attitude diagnosis method according to claim 1, characterized in that, The rigid transformation matrix is: in, R It is a 3x3 rotation matrix. T It is a 3*1 translation vector.

6. A computer program product, said computer program product comprising computer instructions, characterized in that, The computer instructions instruct the computer to execute the armature attitude diagnosis method according to any one of claims 1-5.

7. An armature attitude diagnosis system, based on the armature attitude diagnosis method of claim 1, characterized in that, include: The armature attitude prediction module is used to acquire the actual captured image and input the actual captured image into a pre-trained convolutional neural network to obtain the first attitude of the armature point cloud model. The pre-trained convolutional neural network training method includes: Obtain a training dataset, which is the projection of the armature point cloud model onto the two axes of a pre-established imaging system; The training dataset is input into the convolutional neural network for training, resulting in a trained convolutional neural network. The armature attitude initial adjustment module, based on the first attitude of the armature point cloud model, adjusts the attitude of the armature point cloud model by comparing the similarity between the projected contour of the armature point cloud model and the contour of the actual captured image, and obtains the second attitude of the armature point cloud model. The armature attitude fine-tuning module, based on the second attitude of the armature point cloud model, adjusts the attitude of the armature point cloud model by comparing the scale difference between the projected contour of the armature point cloud model and the contour of the actual captured image, thereby obtaining the optimal attitude of the armature point cloud model.

8. An electronic device, characterized in that, The method includes a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor causes the processor to perform the steps of the armature attitude diagnosis method as described in any one of claims 1-5.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed, it implements the armature attitude diagnosis method as described in any one of claims 1-5.

Images