Solid engine grain defect intelligent detection and visual characterization system

Abstract

The invention provides an intelligent detection and visual representation system for solid engine grain defects. The system comprises a two-dimensional slice data preprocessing module, a three-dimensional visual modeling module and a model post-processing and analysis module. According to the system, by fusing a multi-modal image processing algorithm and a three-dimensional modeling technology, sub-voxel-level three-dimensional reconstruction and spatial quantitative analysis of defects such as air holes and cracks are realized, and coordinate distribution, geometric dimensions and quantity statistics of the defects in a grain three-dimensional space can be accurately positioned; the technical bottleneck of a traditional two-dimensional detection method in the aspects of defect deep analysis and multi-parameter coupling analysis can be solved. All the core modules cooperate to improve the accuracy and efficiency of detecting the defects of the air holes in the solid engine, accurately detect the spatial positions and shapes of the air holes in the grain, and ensure the integrity and safety of the SRM grain structure.

Description

Technical Field

[0001] The present invention relates to the technical field of intersection of industrial detection and computer vision, and relates to, but is not limited to, an intelligent detection and visualization characterization system for defects in solid motor grains. Background Art

[0002] Solid rocket motor (SRM) grains are subject to a variety of complex loads during production, storage, transportation, and use. These include temperature loads caused by curing cooling, environmental changes, and aerodynamic heating; gravity loads during long-term storage; high overloads during launch and maneuvering; vibration loads during transportation and onboard flight; and impact and internal pressure loads during handling and ignition. Under the action of single or combined loads, while also constrained by the casing, the grain generates internal stress and strain fields. If these loads exceed the allowable range of its mechanical properties, defects such as internal porosity, cracks, and debonding can develop in the grain, posing a significant risk to engine operational safety and, in severe cases, leading to engine test failure. Therefore, grain structural integrity analysis is a crucial aspect of the solid rocket motor lifecycle and a pressing research priority for practical applications.

[0003] Industrial Computed Tomography (ICT) inspection technology has been gradually applied to SRM quality inspection and has played a significant role. However, 3D ICT technology in SRM currently faces several challenges in its practical application. These include data processing complexity. ICT scanning generates enormous amounts of data, requiring efficient data processing algorithms for processing and analysis. Automated defect recognition algorithms can be affected by noise, artifacts, and material properties, leading to false positives or missed negatives. Furthermore, 3D image visualization can lack intuitiveness for certain shapes and locations, making it difficult to understand complex internal structures and defects. Summary of the Invention

[0004] The present invention provides a solid motor grain defect intelligent detection and visualization characterization system to solve the technical problems existing in the existing solid motor grain defect detection technology, such as lack of accurate three-dimensional information, low data processing efficiency and reliance on manual labor for defect quantification.

[0005] The technical method of the embodiment of the present invention is implemented as follows:

[0006] In a first aspect, an embodiment of the present invention provides a solid motor grain defect intelligent detection and visualization characterization system, the system comprising a two-dimensional slice data preprocessing module, a three-dimensional visualization modeling module, and a model post-processing and analysis module;

[0007] The two-dimensional slice data preprocessing module is used to perform structural part identification and positioning on the two-dimensional ICT image in the acquired solid motor ICT file based on a preset circular structural part target automatic identification and detection algorithm, and determine the circular structure area image of the solid motor;

[0008] An improved anisotropic diffusion denoising algorithm based on adaptive median filtering is used to remove pulse interference from the circular structure area image multiple times to obtain a circular structure area image after multiple denoising;

[0009] Based on the robust adaptive Canny algorithm of defect edges, circular structure boundary detail extraction and noise interference elimination processing are sequentially performed on the circular structure area image after multiple denoising to obtain a high-precision edge image;

[0010] The three-dimensional visualization modeling module is used to perform image registration, normalization, and ROI extraction on the original ICT slice image based on a preset deep learning ICT slice image interpolation algorithm to obtain an integrated continuous slice sequence;

[0011] Use 3D convolutional neural networks to advance deep features of adjacent slices and align features through optical flow estimation or deformable convolution;

[0012] Using the generative adversarial network architecture, the U-Net generator generates intermediate interpolated slices from two adjacent frame slices, and the PatchGAN discriminator is used to evaluate the authenticity of the generated image and optimize it through a multi-objective loss function;

[0013] Anisotropic diffusion filtering is used to smooth the edges of the interpolated slices, and then the interpolated data and the original data are integrated through voxel fusion technology to obtain the processed slice data;

[0014] Importing the processed slice data into a three-dimensional modeling module to construct a three-dimensional voxel grid and perform model reconstruction;

[0015] Perform edge smoothing and voxel fusion optimization on the reconstructed 3D model to generate a high-precision interactive 3D visualization model;

[0016] The model post-processing and analysis module is used to perform three-dimensional reconstruction of the pores on the pore image in the two-dimensional ICT image to obtain a three-dimensional reconstruction model of the deformed pores and a volume estimation result;

[0017] Based on the volume estimation result, a fault diagnosis result of the solid motor is determined.

[0018] In some embodiments, the two-dimensional slice data pre-processing module further includes a circular region extraction sub-module;

[0019] The circular region extraction submodule is used to perform binary background segmentation, morphological processing and connected domain analysis on the two-dimensional ICT image to obtain multiple candidate regions; each candidate region carries the basic features of the solid motor; the basic features include area features, boundary features and geometric features;

[0020] Eliminate non-target areas whose areas are smaller than a preset area threshold from the plurality of candidate areas to obtain retained areas; number and segment the retained areas to obtain a plurality of image subsets;

[0021] Performing Sobel operator edge extraction on each of the image subsets to obtain an edge point set; establishing a rectangular coordinate system based on the center of gravity of each of the edge point sets as the origin, and calculating the polar angles of the four vertices of the minimum circumscribed rectangle relative to the center;

[0022] Based on the edge point set, calculating the circularity index of each of the image subsets, and eliminating areas that do not meet the circular feature to obtain areas that meet the circular feature;

[0023] Randomly selecting a subset of edge points in the reserved area for circle fitting, using an optimized sampling strategy to reduce computational complexity, and generating candidate circles containing center coordinates and radius parameters;

[0024] Each candidate circle is verified. If a sufficient number of edge points fall within a preset radius tolerance range of each candidate circle, it is determined to be a valid circle and the circular structure area image is output.

[0025] In some embodiments, the two-dimensional slice data preprocessing module further includes a filtering and denoising submodule;

[0026] The filtering and denoising submodule is used to divide the circular structure area image into local areas to obtain the circular area image after division;

[0027] Adopting an adaptive median filter to filter the divided circular region image, remove impulse noise, and obtain a preliminary denoised circular structure region image;

[0028] Calculating the gradient modulus of the circular structure region image after preliminary denoising, replacing the gradient modulus of the original image with the gradient modulus, and dynamically adjusting the diffusion coefficient in the anisotropic diffusion model;

[0029] Based on the adjusted diffusion coefficient, anisotropic diffusion denoising is performed on the circular structure region image after the preliminary denoising, retaining image edge details to obtain the circular structure region image after multiple denoising.

[0030] In some embodiments, the two-dimensional slice data preprocessing module further includes an adaptive Canny submodule;

[0031] The adaptive Canny submodule is used to perform grayscale processing on the selected area in the circular structure area image after multiple denoising to obtain a grayscale processed image;

[0032] Constructing a Gaussian filter to perform two-dimensional convolution on the grayscale processed image to reduce fluctuations in local details of the image, thereby obtaining a convolved image;

[0033] Calculating the gradient magnitude and direction of the convolved image using the finite difference of the first-order partial derivative; performing non-maximum suppression on the gradient magnitude and retaining the local gradient maximum point;

[0034] A dual threshold algorithm is used, with a high threshold used to segment the background and the target, and a low threshold used to connect the broken edges;

[0035] The edge area is refined, the defect outline is extracted and the potential defect features are identified to obtain the high-precision edge image.

[0036] In some embodiments, the model post-processing and analysis module includes a pore ellipsoid model building module, a pore shape parameter extraction module, a three-dimensional display module, and a volume calculation module;

[0037] The pore ellipsoid model building module is used to simulate the deformed pore into two upper and lower triaxial ellipsoids. By determining the three-dimensional coordinates of the ellipsoid center, the three semi-axis lengths and the three-axis rotation angles, a total of nine parameters, a three-dimensional ellipsoid model is constructed. The origin of the ellipsoid model is the center of the solid rocket engine base.

[0038] The pore shape parameter extraction module is used to preprocess the pore image to obtain a pore edge binary image, first directly calculate the ellipse center coordinates, the semi-major axis length, and the angle between the major axis and the horizontal axis, then divide the pore edge image according to the major axis, use Hough transform to detect the semi-minor axis lengths of the upper and lower halves of the ellipse respectively, and determine the optimal semi-minor axis parameter from multiple solutions using the area minimum error method;

[0039] The three-dimensional display module is used to visualize the upper and lower half ellipsoids of the pore based on the constructed three-dimensional regular volume data field and the ellipsoid parameters;

[0040] The volume calculation module is used to perform three-dimensional reconstruction on the pore edge binary image, calculate the pore volume using a volume calculation formula, and calculate the average value of the pore volume in multiple images.

[0041] In some embodiments, the system further includes a centralized management platform; the centralized management platform includes a front-end interface module, a confidentiality management module, a data security management module, and a flaw detection report management module;

[0042] The front-end interface module is used to create a cross-platform operation interface based on the PyQt library and QtDesigner. The interface includes an interactive bar on the left and a preview box on the right. The interactive bar on the left is used to receive user operation instructions, and the preview box on the right is used to load CT images and 3D reconstructed images. The front-end interface module supports users to import CT slice images, generate 3D images, analyze and interact with 3D images, and output reconstruction results and defect statistics.

[0043] The confidentiality management module is used to encrypt sensitive data using the AES-GCM algorithm of the Crypto.js library, store encryption keys on the back end, and ensure data transmission security through HTTPS; based on the RBAC model, access control is implemented by storing user access control lists in the database and recording sensitive data access logs;

[0044] The data security management module includes a data encryption submodule and a data backup and recovery submodule. The data encryption submodule uses a database engine that supports transparent data encryption to encrypt database and file system data and implements a key rotation strategy. The data backup and recovery submodule uses Veeam Backup & Replication to perform full and incremental backups and manage the storage and access rights of backup data.

[0045] The flaw detection report management module is used to store document template information based on the Django backend framework and MySQL / PostgreSQL database, and combines the React / Vue.js front-end framework and Ant Design / Bootstrap for interface design; uses Python's ReportLab library to generate PDF format reports, and uses Django to process report data logic; uses the RESTful API provided by Django to handle report viewing and modification operations, and realizes report interaction through state management; stores flaw detection report files on a storage server, uses the MySQL / PostgreSQL database to store report metadata, and combines with Elasticsearch to realize report retrieval.

[0046] In a second aspect, an embodiment of the present invention provides a method for intelligent detection and visual characterization of defects in solid motor grains, the method comprising:

[0047] Based on a preset circular structural component target automatic recognition and detection algorithm, the structure component recognition and positioning are performed on the two-dimensional ICT image in the acquired solid motor ICT file to determine the circular structure area image of the solid motor;

[0048] An improved anisotropic diffusion denoising algorithm based on adaptive median filtering is used to remove pulse interference from the circular structure area image multiple times to obtain a circular structure area image after multiple denoising;

[0049] Based on the robust adaptive Canny algorithm of defect edges, circular structure boundary detail extraction and noise interference elimination processing are sequentially performed on the circular structure area image after multiple denoising to obtain a high-precision edge image;

[0050] Based on the preset deep learning ICT slice image interpolation algorithm, the original ICT slice images are processed with image registration, normalization, and ROI extraction to obtain an integrated continuous slice sequence;

[0051] Use 3D convolutional neural networks to advance deep features of adjacent slices and align features through optical flow estimation or deformable convolution;

[0052] Using the generative adversarial network architecture, the U-Net generator generates intermediate interpolated slices from two adjacent frame slices, and the PatchGAN discriminator is used to evaluate the authenticity of the generated image and optimize it through a multi-objective loss function;

[0053] Anisotropic diffusion filtering is used to smooth the edges of the interpolated slices, and then the interpolated data and the original data are integrated through voxel fusion technology to obtain the processed slice data;

[0054] Importing the processed slice data into a three-dimensional modeling module to construct a three-dimensional voxel grid and perform model reconstruction;

[0055] Perform edge smoothing and voxel fusion optimization on the reconstructed 3D model to generate a high-precision interactive 3D visualization model;

[0056] Performing three-dimensional reconstruction of the pores on the pore image in the two-dimensional ICT image to obtain a three-dimensional reconstruction model of the deformed pores and a volume estimation result;

[0057] Based on the volume estimation result, a fault diagnosis result of the solid motor is determined.

[0058] Compared with the prior art, the technical solution of the present invention has the following beneficial effects:

[0059] The intelligent detection and visualization system for solid rocket grain defects provided by the embodiments of the present invention achieves the following: First, the geometric error between the three-dimensional defect model and the actual structure is ≤5% (the international leading level is 8-12%), ensuring that the morphological restoration of defects such as pores and cracks meets the requirements of engineering diagnosis; in addition, the spatial coordinate positioning error of the defect is ≤0.1% (traditional methods are 0.5-1%), which can accurately identify the distribution of defects at the shell / grain interface and provide reliable data support for structural integrity assessment. Second, the present invention requires only ≤5 ICT tomographic images to complete three-dimensional reconstruction, reducing data acquisition costs while avoiding the risk of radiation accumulation caused by multiple scans; in addition, the automatic fitting time of the defect three-dimensional model of the present invention is ≤10 minutes, and combined with GPU parallel acceleration technology, it realizes the engineering application of SRM batch detection. Third) The present invention supports virtual sectioning, material delamination, and stress-defect coupling display, allowing inspectors to intuitively analyze the spatial relationship between defects and the shell / insulation layer, improving diagnostic efficiency. Furthermore, the present invention automatically generates risk assessment reports that comply with the MIL-STD-3022 standard based on defect volume, equivalent diameter, and spatial density parameters, thereby reducing the misjudgment rate of defects within solid motor grains.

[0060] In addition, the present invention can also realize a security and resource optimization system: the present invention adopts encrypted storage and blockchain evidence storage technology, combined with permission hierarchical management, to realize the three-dimensional permission isolation of "personnel-data-operation", and support real-time backtracking of audit logs; broad application prospects and market: relying on the technical advantages of "high precision-fast response-full dimension", the present invention has built a full-stack software platform of "three-dimensional intelligent detection-multi-field coupling simulation-autonomous decision-making closed loop", forming an autonomous and controllable industrial software technology ecosystem. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] Figure 1 This is a schematic diagram of the structure of the solid motor grain defect intelligent detection and visual characterization system provided by the present invention;

[0062] Figure 2 This is a schematic diagram of the ICT file processing process in the slice data preprocessing module provided by the present invention; wherein, Figure 2 (a) is the original image; Figure 2 (b) is the circular area extraction map; Figure 2 (c) is the shadow removal map; Figure 2 (d) is the multi-layer structure identification diagram; Figure 2 (e) is the edge detection map; Figure 2 (f) is the noise removal map;

[0063] Figure 3 is a three-dimensional visualization result view provided by the present invention; wherein, Figure 3 (a) is the X-axis view; Figure 3 (b) is the Y-axis view; Figure 3 (c) is the Z-axis view;

[0064] Figure 4 This is a fine defect display diagram provided by the present invention; wherein, Figure 4 (a) Detailed display of the lower defect; (b) Detailed display of the side defect; (c) Detailed display of the upper defect;

[0065] Figure 5 It is the hierarchical structure and internal mold diagram provided by the present invention; wherein, Figure 5 (a) is a hierarchical structure diagram; Figure 5 (b) is the internal mold diagram;

[0066] Figure 6 : is a schematic diagram of the projection of the deformed bubble model and the ellipsoid model in spherical coordinates provided by the present invention; wherein, Figure 6 (a) is the projection diagram of the deformed bubble model. Figure 6 (b) is the ellipsoid model diagram in spherical coordinates;

[0067] Figure 7 is a schematic diagram of the three-dimensional ellipsoid and its parameters provided by the present invention, wherein: Figure 7 (a) is a schematic diagram of a three-dimensional ellipsoid in space; Figure 7 (b) Schematic diagram of spatial three-dimensional ellipsoid parameters;

[0068] Figure 8 This is a schematic diagram of the pore ellipse model reconstructed by Hough transform provided by the present invention;

[0069] Figure 9 This is a schematic diagram of pore visualization obtained by Hough transform provided by the present invention;

[0070] Figure 10 This is a schematic diagram of the system operation interface provided by the present invention;

[0071] Figure 11 This is a three-dimensional reconstruction effect display diagram provided by the present invention;

[0072] Figure 12 This is a diagram showing the interface effect of defect 1 detailed information provided by the present invention;

[0073] Figure 13 This is a diagram showing the interface effect of defect 2 detailed information provided by the present invention;

[0074] Figure 14 This is a diagram showing the interface effect of defect 3 detailed information provided by the present invention;

[0075] Figure 15 This is a three-dimensional engine model interface effect display diagram provided by the present invention;

[0076] Figure 16 It is the confidentiality management interface provided by the present invention;

[0077] Figure 17 This is a schematic diagram of the overall architecture of the data security management submodule provided by the present invention;

[0078] Figure 18 It is a schematic diagram of the overall architecture of the flaw detection report management submodule provided by the present invention. DETAILED DESCRIPTION

[0079] In order to make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings. The described embodiments should not be regarded as limiting the present invention. All other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0080] In the following description, references to "some embodiments" describe a subset of all possible embodiments. However, it is understood that "some embodiments" may be the same subset or different subsets of all possible embodiments, and may be combined with each other without conflict. Unless otherwise defined, all technical and scientific terms used in the embodiments of the present invention have the same meaning as commonly understood by those skilled in the art to which the embodiments of the present invention pertain. The terms used in the embodiments of the present invention are for the purpose of describing the embodiments of the present invention only and are not intended to limit the present invention.

[0081] The embodiment of the present invention provides a solid motor grain defect intelligent detection and visualization characterization system, which includes: a two-dimensional slice data preprocessing module 100, a three-dimensional visualization modeling module 200, and a model post-processing and analysis module 300. Figure 1 , Figure 1 It is a structural diagram of a solid motor grain defect intelligent detection and visual characterization system provided by an embodiment of the present invention.

[0082] The two-dimensional slice data preprocessing module 100 is used to perform structural part identification and positioning on the two-dimensional ICT image in the acquired solid motor ICT file based on a preset circular structural part target automatic identification and detection algorithm, and determine the circular structure area image of the solid motor;

[0083] An improved anisotropic diffusion denoising algorithm based on adaptive median filtering is used to remove pulse interference from the circular structure area image multiple times to obtain a circular structure area image after multiple denoising;

[0084] Based on the robust adaptive Canny algorithm of defect edges, circular structure boundary detail extraction and noise interference elimination processing are sequentially performed on the circular structure area image after multiple denoising to obtain a high-precision edge image;

[0085] The three-dimensional visualization modeling module 200 is used to perform image registration, normalization, and ROI extraction on the original ICT slice image based on a preset deep learning ICT slice image interpolation algorithm to obtain an integrated continuous slice sequence;

[0086] Use 3D convolutional neural networks to advance deep features of adjacent slices and align features through optical flow estimation or deformable convolution;

[0087] Using the generative adversarial network architecture, the U-Net generator generates intermediate interpolated slices from two adjacent frame slices, and the PatchGAN discriminator is used to evaluate the authenticity of the generated image and optimize it through a multi-objective loss function;

[0088] Anisotropic diffusion filtering is used to smooth the edges of the interpolated slices, and then the interpolated data and the original data are integrated through voxel fusion technology to obtain the processed slice data;

[0089] Importing the processed slice data into a three-dimensional modeling module to construct a three-dimensional voxel grid and perform model reconstruction;

[0090] Perform edge smoothing and voxel fusion optimization on the reconstructed 3D model to generate a high-precision interactive 3D visualization model;

[0091] The model post-processing and analysis module 300 is used to perform three-dimensional reconstruction of the pores on the pore image in the two-dimensional ICT image to obtain a three-dimensional reconstruction model of the deformed pores and a volume estimation result;

[0092] Based on the volume estimation result, a fault diagnosis result of the solid motor is determined.

[0093] The intelligent detection and visualization system for solid rocket grain defects includes the following:

[0094] (1) By studying the detection method and imaging principle of ICT, and combining signal processing and other related technologies to analyze the data information of the engine grain ICT file, data processing technology is used to extract data from the ICT file and form a two-dimensional sequence tomographic image of the grain. Based on image segmentation technology, the boundary detection and extraction algorithm of the two-dimensional image is studied to realize the automatic recognition and extraction of the defect boundary scanned in the two-dimensional tomographic image of the grain.

[0095] (2) Using ICT two-dimensional tomographic images as the initial data source, digital image processing, computer graphics and other technologies are used to study the interpolation and three-dimensional reconstruction algorithms of tomographic images, and the processed two-dimensional or three-dimensional data are converted into a three-dimensional model that can accurately describe the spatial characteristics of regular bodies, so that the reconstructed three-dimensional model can intuitively display the shape and size of defects inside the grain.

[0096] (3) Research on the precise visualization of 3D defect models, enabling visualization of 3D defect models at any position and in any direction, with a particular focus on the precise visualization of defects within the shell, insulation layer, liner, and grain. Interactive operations will enable the display model to be rotated, scaled, and translated, achieving a multi-angle intuitive display of internal engine defects.

[0097] (4) Accurately locate, measure, and count defects in 3D models. Design efficient and accurate measurement methods to accurately measure the spatial distance, angle, and volume of defect structures in 3D models. Obtain information on the size, shape, and location of defects through digital descriptions such as coordinates and volume. Also, interval statistics of defects within SRMs can be achieved based on parameters such as the equivalent diameter or maximum projected area of ​​the defect.

[0098] (5) Realize the management of 3D visual modeling results. The constructed 3D visual modeling results are stored uniformly. Management and operation personnel can view and export the modeling results and flaw detection reports by logging into the system. The system can also provide modeling result files and flaw detection reports in different formats from any perspective. The system has security management and other functions to protect data security.

[0099] In some embodiments, the two-dimensional slice data preprocessing module 100 also includes a circular area extraction submodule 110; the circular area extraction submodule 110 is used to perform binary background segmentation, morphological processing and connected domain analysis on the two-dimensional ICT image to obtain multiple candidate areas; each of the candidate areas carries the basic features of the solid engine; the basic features include area features, boundary features and geometric features; non-target areas with an area less than a preset area threshold are eliminated from the multiple candidate areas to obtain retained areas; the retained areas are numbered and segmented to obtain multiple image subsets; each of the image subsets is subjected to Sobel operator edge extraction. Take, and obtain an edge point set; establish a rectangular coordinate system based on the center of gravity of each edge point set as the origin, and calculate the polar angles of the four vertices of the minimum circumscribed rectangle relative to the center; based on the edge point set, calculate the circularity index of each image subset, and eliminate the area that does not meet the circular feature to obtain the area that meets the circular feature; randomly select an edge point subset in the retained area for circle fitting, adopt an optimized sampling strategy to reduce the computational complexity, and generate a candidate circle including the center coordinates and radius parameters; verify each candidate circle, and if a sufficient number of edge points fall within the preset radius tolerance range of each candidate circle, it is determined to be a valid circle, and the circular structure area image is output.

[0100] In some embodiments, the two-dimensional slice data preprocessing module 100 also includes a filtering and denoising submodule 120; the filtering and denoising submodule 120 is used to perform local area division on the circular structure area image to obtain a circular area image after division; use an adaptive median filter to filter the circular area image after division to remove pulse noise and obtain a preliminary denoised circular structure area image; calculate the gradient modulus of the preliminary denoised circular structure area image, replace the gradient modulus of the original image with the gradient modulus, and dynamically adjust the diffusion coefficient in the anisotropic diffusion model; based on the adjusted diffusion coefficient, perform anisotropic diffusion denoising on the circular structure area image after preliminary denoising to retain image edge details and obtain the circular structure area image after multiple denoising.

[0101] In some embodiments, the two-dimensional slice data preprocessing module 100 also includes an adaptive Canny submodule 130; the adaptive Canny submodule 130 is used to grayscale the selected area in the circular structure area image after multiple denoising to obtain a grayscale image; construct a Gaussian filter to perform two-dimensional convolution on the grayscale image to reduce the fluctuation of local details of the image to obtain a convolved image; use the finite difference of the first-order partial derivative to calculate the gradient amplitude and direction of the convolved image; perform non-maximum suppression on the gradient amplitude and retain the local gradient maximum point; adopt a dual threshold algorithm, use a high threshold to segment the background and the target, and a low threshold to connect the broken edges; perform fine processing on the edge area, extract the defect contour and identify the potential defect characteristics to obtain the high-precision edge image.

[0102] In some embodiments, the model post-processing and analysis module 300 includes a pore ellipsoid model building module 310, a pore shape parameter extraction module 320, a three-dimensional display module 330 and a volume calculation module 340; the pore ellipsoid model building module 310 is used to simulate the deformed pore into two upper and lower three-axis ellipsoids, and to construct a three-dimensional ellipsoid model by determining the three-dimensional coordinates of the ellipsoid center, the three semi-axis lengths and the three-axis rotation angles, a total of 9 parameters, and the origin of the ellipsoid model is taken at the center of the solid engine base; the pore shape parameter extraction module 320 is used to pre-process the pore image to obtain the two-dimensional coordinates of the pore edge. The three-dimensional display module 330 is used to visualize the upper and lower half ellipsoids of the pore according to the ellipsoid parameters based on the constructed three-dimensional regular volume data field; the volume calculation module 340 is used to perform three-dimensional reconstruction of the pore edge binary image, calculate the pore volume using the volume calculation formula, and calculate the average value of the pore volume in multiple images.

[0103] In some embodiments, the system further includes a centralized management platform 400; the centralized management platform 400 includes a front-end interface module 410, a confidentiality management module 420, a data security management module 430, and a flaw detection report management module 440;

[0104] The front-end interface module 410 is used to create a cross-platform operation interface based on the PyQt library and QtDesigner. The interface includes a left-side interactive bar and a right-side preview box. The left-side interactive bar is used to receive user operation instructions, and the right-side preview box is used to load CT images and 3D reconstructed images. The front-end interface module supports users to import CT slice images, generate 3D images, analyze and interact with 3D images, and output reconstruction results and defect statistics.

[0105] The confidentiality management module 420 is used to encrypt sensitive data using the AES-GCM algorithm of the Crypto.js library, store encryption keys on the back end, and ensure data transmission security through HTTPS; based on the RBAC model, access control is implemented by storing user access control lists in the database and recording sensitive data access logs;

[0106] The data security management module 430 includes a data encryption submodule and a data backup and recovery submodule. The data encryption submodule uses a database engine that supports transparent data encryption to encrypt database and file system data and implements a key rotation strategy. The data backup and recovery submodule uses Veeam Backup & Replication to perform full and incremental backups and manage the storage and access rights of backup data.

[0107] The flaw detection report management module 440 is used to store document template information based on the Django backend framework and MySQL / PostgreSQL database, and combines the React / Vue.js front-end framework and Ant Design / Bootstrap for interface design; uses Python's ReportLab library to generate PDF format reports, and processes report data logic through Django; uses the RESTful API provided by Django to process report viewing and modification operations, and realizes report interaction through state management; stores the flaw detection report file in the storage server, uses the MySQL / PostgreSQL database to store report metadata, and combines Elasticsearch to realize report retrieval.

[0108] An embodiment of the present invention provides a method for intelligent detection and visual characterization of solid motor grain defects, which is applied to a system for intelligent detection and visual characterization of solid motor grain defects, including the following steps:

[0109] Step S210 , based on a preset circular structural component target automatic recognition and detection algorithm, structural component recognition and positioning are performed on the acquired two-dimensional ICT image in the solid motor ICT file to determine the circular structural region image of the solid motor.

[0110] In some embodiments, a solid motor ICT file refers to a tomographic image data file generated by an industrial computed tomography (ICT) scan of the solid motor's internal structure. This file uses X-rays or gamma rays to penetrate an object, creating a grayscale distribution that reflects material density and structural differences. This file contains two-dimensional tomographic image information of the solid motor's internal structure, including the casing, insulation, and lining, as well as potential defects.

[0111] Step S220 , based on the improved anisotropic diffusion denoising algorithm of adaptive median filtering, the pulse interference in the circular structure region image is eliminated multiple times to obtain the circular structure region image after multiple denoising.

[0112] In this invention, the improved algorithm introduces the results of adaptive median filtering, replacing the gradient modulus of the original image with the gradient modulus after filtering, and then dynamically adjusts the diffusion coefficient in the anisotropic diffusion model. The principle is to calculate the gradient of each pixel in the image to determine whether the point is in the edge region or the smooth region, reduce diffusion in the edge region to preserve edge details, and enhance diffusion in the smooth region to remove noise. The improved anisotropic diffusion denoising algorithm based on adaptive median filtering can more effectively suppress pulse noise such as gunpowder slag that is prevalent in solid rocket engine ICT images, while preserving the edge detail structure in the image and enhancing the ability to express target boundary information.

[0113] Step S230 , based on the robust adaptive Canny algorithm of defect edges, circular structure boundary detail extraction processing and noise interference elimination processing are sequentially performed on the circular structure area image after multiple denoising to obtain a high-precision edge image.

[0114] In this invention, the robust adaptive Canny algorithm for defect edges first uses a Gaussian filter to smooth the image, reducing the fluctuation of local details in the image and highlighting the main structural outlines. It then uses non-maximum suppression technology to retain the local maximum of the edge response, refining the edge to a single pixel width to ensure the continuity and accuracy of the edge detection results. Finally, a dual-threshold method is used to determine the final edge image. A high threshold is used to segment the image content to be detected from the background, finding each valid line segment. A low threshold is then used to extend these line segments in both directions to find edge breaks and connect the discontinuous edges of the image. This algorithm can accurately extract subtle changes in the boundaries of circular structures, assist in discovering potential material defects or structural anomalies in the boundary area, and improve the sensitivity and reliability of detection.

[0115] In step S240 , based on a preset deep learning ICT slice image interpolation algorithm, image registration, normalization, and ROI extraction are performed on the original ICT slice image to obtain an integrated continuous slice sequence.

[0116] In this paper, the interpolation algorithm aims to improve the continuity and integrity of the model by inserting new intermediate slices between the original slices. This is primarily achieved by leveraging deep learning models such as 3D CNN and GAN. 3D CNN is used to extract deep features from adjacent slices and capture the spatial context between them. The U-Net in the GAN architecture is responsible for generating intermediate interpolated slices from two adjacent frames. PatchGAN is used to evaluate the realism of the generated images. Through multi-objective optimization such as L1 / L2 loss, adversarial loss, and perceptual loss, the generated images achieve high levels of pixel-level similarity, visual quality, and semantic consistency.

[0117] In step S250, a 3D convolutional neural network is used to advance the deep features of adjacent slices, and the features are aligned by optical flow estimation or deformable convolution.

[0118] In step S260, a generative adversarial network architecture is used to generate an intermediate interpolated slice from two adjacent frame slices through a U-Net generator, and the authenticity of the generated image is evaluated using a PatchGAN discriminator, and optimized through a multi-objective loss function.

[0119] In step S270 , anisotropic diffusion filtering is used to smooth the edges of the interpolated slices, and then the interpolated data and the original data are integrated using voxel fusion technology to obtain processed slice data.

[0120] Step S280: import the processed slice data into a 3D modeling module, construct a 3D voxel grid and reconstruct the model; perform edge smoothing and voxel fusion optimization on the reconstructed 3D model to generate a high-precision interactive 3D visualization model.

[0121] Step S290 , performing three-dimensional reconstruction of the pore image in the two-dimensional ICT image to obtain a three-dimensional reconstruction model of the deformed pore and a volume estimation result; and determining a fault diagnosis result of the solid motor based on the volume estimation result.

[0122] The following describes an exemplary application of an embodiment of the present invention in a practical application scenario.

[0123] The solid rocket motor grain defect intelligent detection and visualization characterization system constructed by this invention adopts a modular architecture design and realizes accurate detection and quantitative evaluation of internal defects of grains through a multi-level data processing chain. The system consists of four core units: a 2D slice data preprocessing module, a 3D visualization modeling module, a model post-processing and analysis module, and a centralized management platform (system architecture as shown in the figure). Figure 1 The technical process is as follows:

[0124] In the system's operation and management interface, encrypted and integrity-verified 2D ICT scan slices are imported into the system through a secure channel and then transferred to the defect analysis and visualization platform for standardized preprocessing. The 2D processing stage utilizes a multi-scale feature extraction algorithm to automatically identify the boundaries of multi-layer structures, such as the shell, insulation layer, and lining, from the sequential tomographic images. Using an improved Canny operator and morphological gradient fusion technology, defect contours are delineated with sub-pixel accuracy, laying the data foundation for 3D modeling. The preprocessed 2D dataset undergoes spatial registration and is then input into the 3D visualization modeling module. This data is first used to construct a 3D regular volume data field, and then, combined with surface topology optimization, a high-fidelity 3D solid model is generated. The 3D volume data is then imported into the model post-processing module for precise defect location and measurement. A structured flaw detection report, including a defect distribution map, quantitative parameter tables, and safety level assessment, is generated, completing the closed-loop processing process. Each module seamlessly interacts via a data bus, forming a closed-loop processing chain: "data input - feature analysis - 3D reconstruction - engineering output."

[0125] 1) 2D slice data preprocessing module

[0126] In the slice data preprocessing module, incoming ICT files undergo detailed 2D preprocessing. This step involves extracting key information from the ICT files and converting it into a matrix format to accurately identify the shell, insulation, lining, and any defects. These preprocessed 2D images and associated ICT data are then fed into the 3D visualization modeling module, laying the foundation for subsequent steps.

[0127] 1.1) Circular region extraction

[0128] The present invention proposes an automatic recognition and detection algorithm based on circular structural parts targets, which is used to realize the automatic recognition and precise positioning of structural parts such as shells, insulation layers, and linings in two-dimensional ICT images. This method combines actual engineering needs and, based on the circular geometric features of the target, sequentially performs steps such as image binarization background segmentation, morphological processing, feature region extraction, and improved circular Hough Transform detection. In engineering applications, due to factors such as camera lens distortion, illumination changes, and noise interference, the originally regular circular structures in the image often appear blurred, incomplete, or deformed at the edges, and their contours may be further distorted after binary segmentation processing. In response to the above problems, this method first uses connected domain analysis to eliminate obvious non-circular areas, marks connected domains that meet certain geometric constraints, and divides the original image into several sub-image sets. Circular target detection operations are then performed in these candidate sub-images. The overall process is as follows:

[0129] (1) Image preprocessing and region segmentation: First, the input two-dimensional ICT image (including the shell, insulation layer and lining) is subjected to binary background segmentation and morphological processing. The image is divided into several candidate regions using connected domain analysis, and the area, boundary and geometric features of each region are extracted.

[0130] (2) Region screening and image subset generation: Non-target areas with an area smaller than a set threshold are eliminated to exclude noise interference and areas of no interest. The retained areas are numbered and divided into independent image subsets for subsequent processing.

[0131] (3) Edge detection and coordinate system establishment: Apply the Sobel operator to each image subset to extract edges and obtain edge point sets. Establish a rectangular coordinate system within each region with its center of gravity as the origin, and calculate the polar angles of the four vertices of the minimum bounding box relative to the center of gravity to further assist in circularity determination.

[0132] (4) Circularity feature extraction and non-circularity elimination: Based on the extracted edge point set, the circularity index of the region (such as roundness, boundary symmetry, curvature consistency, etc.) is calculated, and the areas that do not meet the circular characteristics are eliminated, thereby improving the accuracy and robustness of circle fitting.

[0133] (5) Circle detection based on improved random Hough transform: A subset of edge points is randomly selected in the retained area for circle fitting, candidate circles are generated, and parameters such as their center coordinates and radius are recorded. This process uses an optimized sampling strategy to reduce computational complexity.

[0134] (6) Candidate circle verification and true circle screening: Each candidate circle is verified. If a sufficient number of edge points fall within its radius tolerance, it is considered a valid circle; otherwise, false circles are excluded. Finally, the target circle information for engineering judgment is output as the basis for identifying components such as shells, insulation layers, and linings.

[0135] 1.2) Improved anisotropic diffusion denoising algorithm based on adaptive median filtering

[0136] After circular target extraction, to further improve recognition accuracy, this paper addresses the widespread presence of gunpowder slag noise in solid rocket rocket ICT images by designing an improved anisotropic diffusion denoising algorithm based on adaptive median filtering. Specifically, after extracting circular structural regions such as the shell, insulation layer, and lining, local filtering is performed only within these candidate circular regions, avoiding redundant calculations of non-target areas outside the shell and effectively improving processing efficiency. By introducing an adaptive median filter, this method significantly enhances the ability to suppress non-Gaussian gunpowder slag noise. Specifically, an adaptive median filter is first used to accurately remove pulse-like interference (such as gunpowder slag). The gradient modulus of the filtered image is then used to replace the gradient modulus of the original image. This allows for dynamic adjustment of the diffusion coefficient in the anisotropic diffusion model, achieving refined control of the diffusion process. Compared to traditional anisotropic diffusion models that only smooth Gaussian noise, this method maintains the Gaussian noise smoothing effect while providing a more targeted and efficient suppression of pulse interference such as gunpowder slag. The anisotropic adaptive median diffusion filter finally constructed can not only effectively remove the mixed noise in the circular area, but also better preserve the edge detail structure in the image and enhance the expression ability of the target boundary information. The mathematical definition of the filter is as follows:

[0137] ;

[0138] Where, Indicates the location and diffusion time The image intensity at time , the initial image is . Representing an image The gradient of is a two-dimensional vector defined as: . is a divergence operator used to calculate the degree of divergence of a vector field. The diffusion process is controlled by partial differential equations. Its core idea is to guide the image to be selectively smoothed in a non-uniform space by the direction and magnitude of the image gradient. Compared with the traditional anisotropic diffusion method, this model introduces adaptive median filtering (AMF) to pre-process the image and converts the filtered gradient model into a By replacing the original gradient modulus and dynamically adjusting the diffusion coefficient, this improved strategy can more effectively suppress impulse noise such as gunpowder slag in solid rocket ICT images, while preserving key edge structures and details in the image, enhancing boundary expression capabilities, and providing a more reliable data foundation for subsequent target recognition.

[0139] After extracting circular structures such as shells, insulation layers, and linings and suppressing gunpowder slag noise, this paper proposes an efficient and robust edge detection method, namely a robust adaptive Canny algorithm, for detecting defect edges. First, the extracted circular structure regions in the denoised image are processed to ensure highly targeted and efficient edge detection. Subsequently, the selected regions are grayscaled to reduce data dimensionality while retaining key structural information, providing a clear image foundation for subsequent edge extraction.

[0140] Based on the grayscale image, this method uses the Canny edge detection operator to perform high-precision image processing, extracting fine pixel-level boundaries. The Canny operator, with its excellent noise suppression capabilities, high edge location accuracy, and excellent continuity performance, is particularly suitable for analyzing solid rocket engine ICT images with complex structures and high detail requirements. Using this operator, the system can accurately capture subtle changes in the boundaries of circular structures, thereby assisting in the detection of potential material defects or structural anomalies in the boundary area, improving detection sensitivity and reliability.

[0141] It's important to note that the extracted edge information not only serves defect detection but also serves as an important input for subsequent 3D modeling and visualization analysis. This high-precision edge data provides more accurate geometric constraints and shape contours in 3D space, supporting the 3D reconstruction of complex internal engine structures. This process significantly improves the accuracy and completeness of 3D models recovered from 2D image data, providing solid technical support and data assurance for digital inspection, structural visualization, and subsequent performance evaluation of solid rocket engines.

[0142] The basic idea of ​​this method is to first use a Gaussian filter to smooth the image to reduce the fluctuation of local details in the image and highlight the main structural outlines; then, use non-maximum suppression technology to accurately retain the local maximum of the edge response to ensure the continuity and accuracy of the edge detection results; finally, use the double threshold method to determine the final edge image to further enhance the detection effect of the real boundary. In the actual solid engine ICT image analysis, this preliminary smoothing process of Gaussian filtering not only helps to improve the accuracy of edge positioning, but also lays a good foundation for subsequent structural modeling and defect identification. Its Gaussian function :

[0143] ;

[0144] in, Represented in two-dimensional space coordinates The Gaussian function value at is the standard deviation, which controls the spread of the Gaussian distribution; and Represents the horizontal and vertical offsets relative to the center pixel of the filter, respectively. It is a natural exponential function, and the entire function presents a centrally symmetric bell-shaped distribution.

[0145] After constructing the Gaussian function filter, use Gaussian filtering For images Perform a two-dimensional convolution operation:

[0146] ;

[0147] Where, is the number of images, i.e. the number of images in the grayscale process ,image Is the filtered image. The magnitude and direction of the gradient are calculated using the finite difference of the first-order partial derivative, where the first-order difference convolution is:

[0148] .

[0149] After that, the two differential convolution templates are convolved with the image respectively to obtain its horizontal and vertical gradients, and then the gradient calculation is used to obtain the amplitude and direction. The calculation method is as follows:

[0150] ;

[0151] ;

[0152] ;

[0153] ;

[0154] Where, and are the horizontal and vertical gradients, is the gradient magnitude, is the gradient direction. To determine the image edge, it is necessary to perform non-maximum suppression on the obtained gradient magnitude. The points with the maximum local gradient value are retained instead of cleared to zero, thus achieving the effect of thinning the edge.

[0155] Finally, the double threshold algorithm is used to detect edges. Select two thresholds and ( < ), using a high threshold Segment the image content and background to be detected, find each valid line segment; use a low threshold Extend these line segments in both directions to find the edge breaks and connect the discontinuous edges of the image. Use two thresholds and Get two threshold edge images and . It is obtained by using a high threshold, which can effectively separate the background from the image content and contain fewer false edges, but the edges obtained are discontinuous due to the high threshold. The algorithm connects the edges into contours and when it reaches the endpoint of the contour, Find the edge that can be connected to the contour in the A neighborhood points of Continue to collect edges and connect them until Until it becomes uninterrupted.

[0156] After edge detection, the method further refines the edge region, extracting flaw contours and identifying potential defect features. This process not only accurately locates possible material defects or structural anomalies but also provides critical geometric information for subsequent 3D modeling and visualization analysis.

[0157] To ensure accurate 3D modeling, the system specifically retains key edge contours as mesh boundaries for geometric reconstruction. This edge-preserving strategy effectively reduces the interference of redundant data and ensures the integrity and accuracy of the reconstructed model at the detailed level. This strategy provides more constrained boundary information in 3D space, providing an accurate geometric basis for high-precision 3D reconstruction of the complex internal structure of solid rocket engines.

[0158] This series of processing not only significantly improves the reliability of subsequent digital analysis and diagnosis, but also lays a solid technical foundation for digital inspection and evaluation of solid rocket engines. Ultimately, it provides more accurate and efficient support for engine performance evaluation, structural safety analysis, and defect location.

[0159] 2) 3D visualization modeling module

[0160] After 2D processing, the processed CT files are fed into the 3D visualization modeling module for 3D modeling, enabling accurate visualization of defects. This step is crucial for in-depth exploration of defects within solid motor fuel.

[0161] 2.1) Construction of 3D regular volume data field

[0162] In order to perform three-dimensional reconstruction of two-dimensional CT slices of a solid rocket engine, the present invention needs to utilize the information provided by two-dimensional serial tomographic images, analyze three-dimensional volume data, and construct a three-dimensional regular volume data field.

[0163] Common methods for constructing 3D regular volume data fields include: (1) Surface reconstruction: Extract surface points from 2D slices. Use triangulation or other algorithms to connect the surface points into a mesh. Advantages: Can generate high-resolution models. Disadvantages: May contain holes or topological errors. (2) Volume-based reconstruction treats 2D slices as projections of 3D volumes. Use back-projection algorithms to merge projections into 3D volumes. Advantages: Can generate smooth, continuous data. Disadvantages: High computational effort and possible artifacts. (3) Deep learning-based reconstruction uses neural networks to learn 3D representations from 2D slices. Generate 3D data through techniques such as Generative Adversarial Networks (GANs) or autoencoders. Advantages: Can generate realistic models. Disadvantages: Requires a large amount of training data and may have biases.

[0164] The characteristic of three-dimensional volume data sets is that they are multidimensional arrays composed of scalar or vector data. These data are usually defined on a grid structure and represent values ​​sampled in three-dimensional space. There are two basic types of three-dimensional volume data: (1) scalar three-dimensional volume data, each point of which contains one value. (2) vector three-dimensional volume data, each point of which contains two or three values, which define the components of a vector. In view of the characteristics of solid motors, the present invention chooses to use scalar three-dimensional volume data to represent solid motors. The specific steps are:

[0165] (1) First, for each 2D serial tomographic image, a regularized grid partitioning method is used to discretize the image at the pixel or voxel level, and the image pixels are mapped to 2D grid points to form a data representation based on the grid structure. This operation provides a uniform and regular spatial sampling basis for the subsequent 3D data construction, ensuring the spatial consistency and operability of the data distribution.

[0166] (2) Subsequently, the actual spatial coordinate information of each 2D tomographic image is extracted and integrated into the corresponding 2D grid point data, completing the mapping from the pixel plane coordinates to the actual physical space coordinate system. This step is crucial for constructing a high-precision 3D volume dataset, ensuring the accurate positioning and registration of each tomographic image in 3D space.

[0167] (3) By stacking all the 2D grid image data with spatial coordinate calibration in an orderly manner layer by layer according to the scanning order or tomographic slice sequence, a 3D array with voxel consistency and spatial topological structure is formed. This 3D array fully records the geometric structure and physical property distribution of the tomographic image in 3D space, forming the core data framework for subsequent 3D modeling and analysis.

[0168] (4) For the above three-dimensional discrete data field, in order to overcome the limitations of spatial sampling and achieve continuous reconstruction of the data field, the cubic interpolation method is used to perform numerical interpolation calculations on the three-dimensional array. Specifically, the cubic interpolation method uses the position of the query point in three-dimensional space as the reference, selects the values ​​of four adjacent grid nodes in each dimension, and uses a cubic convolution kernel or a cubic polynomial basis function for weighted fitting. This interpolation strategy can significantly improve the overall reconstruction accuracy of the volume data while ensuring the smoothness of the local data. It is particularly suitable for the construction of three-dimensional volume data fields with rich details and high spatial continuity requirements.

[0169] (5) Finally, the continuous 3D data field generated after cubic interpolation is further organized and packaged to form a high-resolution, regularized 3D volume data set that is suitable for subsequent 3D visualization, geometric modeling, and detailed analysis. This data set not only has good spatial consistency and continuity, but also provides high-quality geometric and physical data support for 3D morphology reconstruction, defect detection, and digital evaluation of complex structures such as solid rocket engines, ensuring the reliability and accuracy of digital detection and structural analysis results.

[0170] 2.2) 3D Visual Modeling

[0171] This paper addresses the practical needs of solid rocket motors, which face complex internal structures and subtle defect distribution. It proposes a 3D visualization modeling method based on ICT images. By combining various image processing and 3D modeling optimization techniques, it achieves high-precision modeling and defect visualization for structural components such as the shell, insulation layer, and liner. 3D modeling not only accurately reproduces the shape and size of internal defects in the motor grain, but also provides a reliable basis for defect identification, performance evaluation, and maintenance decision-making.

[0172] The entire 3D visualization process is based on ICT images. First, the ICT slice image is processed in two dimensions to extract the areas that need to be visualized, including: shell edge, shell, insulation layer, lining and potential defect contours; then the processed slice data is imported into the 3D modeling module to construct a 3D voxel grid and reconstruct the model. During the modeling process, the actual size differences corresponding to the ICT slices in space are fully taken into account, and new data layers are inserted between the original slices through interpolation technology to improve the continuity and integrity of the model. In order to enhance modeling accuracy and visual quality, the present invention proposes a number of optimization strategies, including an ICT slice image interpolation algorithm based on deep learning, edge smoothing processing and voxel fusion algorithm.

[0173] 2.2.1) ICT Slice Image Interpolation Algorithm Based on Deep Learning

[0174] In the 3D visualization pipeline, ICT slice images often suffer from insufficient inter-layer resolution, resulting in staircase artifacts in the reconstructed 3D model. To address this, we propose a deep learning-based ICT slice image interpolation algorithm. This algorithm first preprocesses the raw data, including image registration, normalization, and ROI extraction, to ensure input data accuracy and consistency. A 3D convolutional neural network (3D CNN) is then used to extract deep features from adjacent slices. Features are aligned through optical flow estimation or deformable convolutions to ensure that the interpolated intermediate layers are spatially consistent with the ground-truth data. For the interpolation generation stage, we employ a GAN architecture. The generator (U-Net) generates intermediate interpolated slices from two adjacent frames, while the discriminator (PatchGAN) evaluates the realism of the generated images. Multi-objective optimization, including L1 / L2 loss, adversarial loss, and perceptual loss, is used to ensure pixel-level similarity, visual quality, and semantic consistency of the generated images. In the post-processing stage, the algorithm further uses anisotropic diffusion filtering for edge smoothing and seamlessly integrates the interpolated data with the original data through voxel fusion technology to improve the continuity of 3D reconstruction.

[0175] 2.2.2) Edge smoothing and voxel fusion algorithm

[0176] In the 3D visualization process, edge smoothing and voxel fusion algorithms are key steps to improve model quality. Edge smoothing uses anisotropic diffusion filtering technology to dynamically adjust the smoothing intensity to eliminate noise interference while preserving key structural features such as shell edges, insulation boundaries, and defect contours. The algorithm first calculates the 3D gradient of each voxel ( ), identifying high-gradient regions (such as edges) and low-gradient regions (such as the interior of a homogeneous material). It then performs iterative diffusion based on the Perona-Malik equation, preserving edge sharpness in high-gradient regions and applying uniform smoothing in low-gradient regions, effectively addressing the edge blurring caused by traditional Gaussian filtering. For interfaces between different materials (such as a metal shell and a rubber insulation layer), the algorithm introduces material property constraints and uses a pre-segmented semantic label map to guide the diffusion process, ensuring a smooth transition between material boundaries. During the voxel fusion phase, the algorithm first builds a multi-resolution voxel pyramid and performs global registration at the coarse-grained level to eliminate minor misalignments between interpolated slices and the original data. Then, at the fine-grained level, a weighted fusion strategy based on local similarity is employed. By calculating structural similarity and grayscale histogram matching within a neighborhood window, the fusion weights are dynamically adjusted to ensure geometric coherence and artifact-free transitions in transition regions. For regions containing potential defects, the algorithm preprocesses them using morphological opening and closing operations to remove isolated noise points and repair discontinuous defect contours. The algorithm then generates smooth isosurfaces using the MarchingCubes algorithm. Finally, the fused voxel data is rendered via GPU-accelerated 3D texture mapping, supporting real-time interactive visualization.

[0177] Figure 2 After the ICT file processing in the slice data preprocessing module, Figure 2 (a) is the original image; Figure 2 (b) is circular region extraction, which retains the main structure and eliminates background redundancy; Figure 2 (c) Shadow removal, suppressing image interference factors such as gunpowder slag and improving image uniformity; Figure 2 (d) is multi-layer structure identification, highlighting the boundaries between material layers such as shell, insulation layer, lining, etc. Figure 2 (e) Edge detection, accurately extracting structural contours and potential defect edges; Figure 2 (f) is noise removal, which further removes residual irrelevant content and retains effective boundary information, providing a clear image basis for subsequent modeling and recognition.

[0178] In terms of visualization module implementation, the present invention developed a 3D modeling and display system based on the open-source Mayavi library. Built on the Visualization Toolkit (VTK), Mayavi supports 3D interactive visualization of scalar fields, vector fields, and tensor data. Its powerful rendering capabilities and flexible view control functions allow the generated 3D models to be rotated, scaled, and sliced, while also enabling clear visualization of structural details and defect distribution through transparency adjustment. In practical applications, this visualization system can help domain experts intuitively identify defect types and locations, allowing for model analysis from multiple angles, thereby improving the accuracy and depth of defect analysis. The resulting reconstructed 3D model not only meticulously displays the spatial hierarchical relationships between the engine casing, insulation layer, and lining, but also accurately reflects the distribution characteristics of local defects. The graphical interface allows for magnified observation of any area, enabling panoramic scanning and local analysis of defects, greatly enhancing the efficiency of human-computer interaction and the reliability of judgment. Figure 3 is a schematic diagram of the three-dimensional visualization results, where Figure 3 (a) is the 3D visualization result view from the X-axis direction; Figure 3 (b) is a three-dimensional visualization view from the Y-axis direction; Figure 3 (c) is the 3D visualization view from the Z-axis direction. Figure 4 It is a detailed display of defects; among them, Figure 4 (a) is a detailed display of the defect below; Figure 4 (b) is a detailed display of side defects; Figure 4 (c) is a detailed display of the defect above. Figure 5 It is the hierarchical structure and internal mold diagram provided by the present invention; wherein, Figure 5 (a) is a hierarchical structure diagram; Figure 5 (b) is the internal mold diagram. Figure 5 The visualization effects of some reconstructed models and typical defects are demonstrated in the paper, which verifies the applicability and effectiveness of the method in complex structure three-dimensional modeling and defect presentation.

[0179] 3) Model post-processing and analysis module

[0180] During solid rocket engine fault diagnosis, measuring the distances between different types of defects, and thus assessing their impact on the engine, is a crucial aspect of solid rocket engine nondestructive testing research. Detecting deformed pores is a crucial fault diagnosis tool for solid rocket engines. This approach exploits the fact that pores, when formed, are solids of revolution. Deformed pores can generally be considered to consist of two upper and lower hemi-ellipsoids, each a triaxial ellipsoid (Jacobian ellipsoid). Using pore images obtained from solid rocket engine ICT image detection, 3D reconstruction of the pores is performed based on this 3D spatial representation, allowing a rough estimate of the volume of the rising pores.

[0181] 3.1) Establishment of the ellipsoidal model of stomata

[0182] Figure 6 (a) is a projection diagram of the deformed bubble model provided by the present invention; Figure 6 (b) is the ellipsoid model diagram in spherical coordinates provided by the present invention, as shown in the attached figure. Figure 6 As shown, the pore can be simulated by two upper and lower half ellipsoids. The semi-major axis length is , the semi-medial axis lengths are and , the semi-minor axis length is Attached Figure 6 (a) is the projection of the ellipsoid model on the projection plane. This parameter allows us to represent various pore shapes, such as the hemispherical cap with concave patterns ( ), hemispherical hat shape ( ), deformed spherical ( ),spherical( ). Figure 6 In the spherical coordinate system shown in (b), the parametric equation of the upper semi-ellipse is:

[0183] .

[0184] Where, represents the polar angle; represents the azimuth angle; r represents the radial distance.

[0185] Figure 7 (a) is a schematic diagram of a three-dimensional ellipsoid in space; Figure 7 (b) Schematic diagram of spatial three-dimensional ellipsoid parameters, as shown in the attached figure. Figure 7 As shown, to uniquely determine a three-dimensional ellipsoid in space, at least 9 parameters are required: the three-dimensional coordinates of the ellipsoid center ( ), the three semi-axis lengths of the ellipsoid 、 、 , and the attitude angles of its three main axes in space ,in, express Axis and The angle between the axes, express Axis and The angle between the axes, express Axis and The angle between the axes.

[0186] Based on the previously constructed 3D regular volume data field, the specific location of the pores is determined directly by establishing a 3D Cartesian coordinate system. The origin of the 3D coordinate system is located at the exact center of the solid rocket motor base. Based on the numerical results of the 3D modeling, the edge coordinates of the pores in the newly established coordinate system can be directly determined.

[0187] 3.2) Extracting pore shape parameters using Hough transform

[0188] Usually, the projection of the pores in the ICT image can be simulated as two upper and lower semi-ellipses, where the semi-major axis is , and the semi-minor axes are and . Variation parameters , can represent pores of different shapes. In solid rocket engines, pores are mostly deformed spherical ( Taking the upper part of the pore as an example, let the angle between the major axis of the elliptical model of the pore and the horizontal axis of the image coordinate system be , the semi-major axis length of the elliptical model is , the semi-minor axis length is , the center of the ellipse model is , then the equation of this elliptical model is:

[0189] .

[0190] The shape parameters of the elliptical model of the pore can be extracted from the pore image using the edge image of the pore and the Hough transform method. From the above formula, we can see that at least five parameters need to be identified from the image to determine an ellipse: the horizontal and vertical coordinates of the ellipse center , the angle between the major axis of the ellipse and the horizontal axis of the image coordinate system , the length of the semi-major axis of the ellipse , semi-minor axis length .

[0191] If Hough transform is used to detect ellipse with 5-dimensional parameters, it will consume too much resources. Therefore, it is necessary to try to reduce the dimension and reduce the computational complexity of Hough transform. In the present invention, the ellipse center of the pore is first directly extracted. , semi-major axis length , the angle between the major axis of the ellipse and the horizontal axis of the image coordinate system Then use Hough transform to detect the length of the ellipse semi-minor axis The specific detection steps are as follows:

[0192] (1) Perform image preprocessing on the image to obtain a binary image of the pore edge;

[0193] (2) Find the distance between two edge points in the image. The longest distance is the major axis, and the two points are the two endpoints of the major axis. From these two endpoints, we can find the angle between the major axis and the horizontal axis.

[0194] (3) In order to make the center of the ellipse fall on the long axis, calculate the geometric center of the pore pattern as the center of the ellipse. Scan the closed pore edge line by line, and take the two boundary points with the farthest distance in the i-th row. and , take the horizontal symmetric point ; Scan the closed pore edges row by row, Take the two farthest boundary points in the column and , take the vertical symmetry point ;when When , take the coordinates of the center of the ellipse ;

[0195] (4) Using the major axis as the dividing line, the pore edge image is divided into two parts, the upper and lower parts. Hough transform is used to detect ellipses in the upper and lower parts respectively, and the upper and lower minor axes of the ellipse are extracted. and For the parameter and exist Within the range, according to the model equation established above, the center of the ellipse , semi-major axis and angle Draw an ellipse on the image plane and count the number of edge points that fall on the ellipse Finally, find the largest Parameter corresponding to the value and The value of corresponds to the length of the semi-minor axis of the ellipse. When extracting the semi-minor axis parameters of an ellipse using the Hough transform, multiple sets of solutions may be obtained (with the same number of edge points falling on the ellipse). In this case, the area minimum error method is used to find the optimal parameters from these multiple sets of solutions.

[0196] (5) The parameters are obtained through the above calculation: , we can reconstruct the elliptical model of the pore, as shown in the attached figure. Figure 8 In order to clearly observe the reconstruction effect, the pore edge binary image is inverted and displayed. Figure 8It can be seen that the ellipse detected by the Hough transform method can reconstruct the true shape of the pore more accurately.

[0197] 3.3) Pore Display Based on 3D Regular Volume Data Field

[0198] Since the three-dimensional ellipsoid model of the bubble is composed of two different half ellipsoids, the reconstruction method of the upper and lower half ellipsoids is the same. According to the obtained bubble characteristic parameters, the above method is used to visualize the upper and lower half ellipsoids. Figure 9 Taking the bubble image in the figure as an example, the visualization result of the ellipsoid is shown in the attached figure. Figure 9 shown.

[0199] 3.4) Pore volume calculation

[0200] The solid rocket motor ICT image sequence is reconstructed in three dimensions in sequence, and the volume of the pores can be calculated using the following volume calculation formula: , where is the semi-major axis length, is the semi-minor axis length, and c is the shortest major axis length.

[0201] After determining the pore volume in multiple images, the average pore volume is calculated. The accuracy of the pore volume calculation method based on the ellipsoid reconstruction algorithm is related to many factors, including the shape and size of the defect bubble itself, the accuracy of the 2D sequence tomographic images, the accuracy of the automatic identification of defect contour features, and the effectiveness of the 3D reconstruction. Although the accuracy of this 3D pore reconstruction requires further verification, the pore volume calculation method based on the ellipsoid reconstruction algorithm has significant advantages in computational efficiency and is expected to be applied in the fault diagnosis of solid rocket engines to further evaluate the impact of specific defects on the engine.

[0202] 4) Centralized management platform

[0203] 4.1) Operating System Management Interface

[0204] This paper uses the Qt framework to develop front-end software interfaces. PyQt is a Python library for creating desktop applications. It provides a rich set of functions and tools, making the development of front-end software interfaces simpler and more efficient. PyQt can run on multiple platforms, including Windows, macOS, and Linux, so applications for different operating systems can be created using the same code base. PyQt provides a rich library of widgets, including buttons, text boxes, list boxes, menus, toolbars, etc., and QtDesigner can be used to quickly create various types of interface elements.

[0205] The overall interface of system operation. Figure 10As shown in the figure, the left side is the interaction bar, where users can interact with the software through the buttons they set. The right side is the preview box, which is used to load CT images and 3D reconstructed images. After opening the software, click the Select Import File button to select a saved CT slice image to import. Clicking the Import button will jump to the file management system for selection.

[0206] After the import is complete, click the 3D Generation button. The system will generate 3D images based on the imported 2D CT images, and then display the effect on the right interface, as shown in the attached figure. Figure 11 . After the generation is completed, you can preview it on the right and interact with the mouse. After clicking to enter the analysis mode, you can slice and analyze the 3D image by controlling the analysis angle and height with the mouse. After selecting the defect point, the defect parameter display module below will output the defect parameters (position, volume, etc.), which can also be selectively displayed by checking. You can zoom, rotate, and translate the 3D image by clicking the buttons on the left. Finally, click the Output Reconstruction Result button to output the 3D reconstruction image and defect statistics list. In addition, for defect details, the system supports displaying defects separately to achieve visualization of individual defects, as shown in the attached figure. Figure 12 、 13 14. The user can zoom, rotate and translate the defect, and can output detailed information such as the defect volume, location and spatial parameters.

[0207] For multiple slices, the system supports direct import of 3D engine models and supports rotating some slices for 3D reconstruction. In addition, the system also supports marking defect locations on the 3D engine model and directly exporting the 3D engine model with defect marks. Users can scale, rotate and translate the 3D engine model, as shown in the attached figure. Figure 15 shown.

[0208] 4.2) Confidentiality Management Interface

[0209] In order to achieve large-scale data confidentiality management, the present invention uses the Crypto.js library to implement AES front-end encryption of sensitive data. The encryption key is stored in the back end, and the security of the data during transmission is guaranteed by HTTPS. Access control: Store the user access control list (ACL) in the database to limit user access to sensitive data. Develop access audit policies and record access logs for sensitive data. Technical solution: Encryption and decryption: Use the AES-GCM algorithm in the Crypto.js library to encrypt and decrypt data. Update the encryption key regularly and distribute the key in a secure manner. Access control: Design a user role table in the database to define the relationship between roles and resources. Use the RBAC model for permission management to ensure that users can only access their authorized data. The interface is as shown in the attached figure. Figure 16 shown.

[0210] 4.3) Data Security Management Submodule

[0211] Data security is a crucial task in software design. The data security management submodule of this invention includes two important submodules: data encryption and backup and recovery. We will explore the technical routes, implementation plans and possible best practices of these submodules. The architecture of the entire data security management submodule is shown in the attached figure. Figure 17 shown.

[0212] 4.3.1) Data Encryption

[0213] Data encryption is one of the key measures to maintain data security. By encrypting data in databases and file systems, unauthorized access and data leakage can be effectively prevented. To achieve this goal, the present invention uses a database engine that supports Transparent Data Encryption (TDE), such as MySQL or Microsoft SQL Server. These database engines have built-in powerful encryption capabilities, allowing data to be encrypted without modifying application code. When TDE is enabled in the database, an appropriate key management strategy must be set, including key generation, storage, and access control. Enable TDE on the database server and configure the database instance to support transparent data encryption to ensure that data is encrypted when stored on disk, thereby improving the overall security of the data. Implement a key rotation strategy regularly to prevent potential impacts on data after the key is cracked or leaked.

[0214] 4.3.2) Data backup and recovery

[0215] Backup and recovery are key measures to ensure data integrity and recoverability. By regularly backing up data and designing a disaster recovery plan, you can minimize the risk of data loss and business interruption.

[0216] When choosing a backup tool, I used Veeam Backup & Replication, a professional backup tool that supports both full and incremental backups. Develop a regularly scheduled backup plan to ensure data integrity and recoverability. When configuring Veeam Backup & Replication, set up a schedule for both full and incremental backups to ensure regular data backups and minimize the risk of data loss.

[0217] Backup data storage and access rights can be managed through the use of backup repositories, including storage quotas and access controls. This allows for efficient management of backup data and ensures its security.

[0218] 4.4) NDT report management submodule

[0219] The entire flaw detection report management submodule is as follows Figure 18 As shown. During the solid rocket defect detection process, it is normal to generate a large number of flaw detection reports. In order to standardize and simplify the report generation process, the present invention first designs a document template management system.

[0220] 4.4.1) Document template management

[0221] This invention uses Django as the backend framework, combined with a MySQL or PostgreSQL database to store document template information. On the front end, React or Vue.js is used as the front-end framework, paired with Ant Design or Bootstrap for interface design. By building a Django-based back-end system, template information for flaw detection reports can be effectively managed and stored. On the front end, React or Vue.js is used to build the user interface, providing an intuitive and user-friendly experience.

[0222] 4.4.2) Report Generation

[0223] Report generation is a core component of the entire process. To efficiently generate reports while ensuring accuracy, the present invention has designed a report generation system. This system uses Python libraries, such as ReportLab, to generate PDF reports, while leveraging Django to handle the data logic involved in report generation. The present invention develops a backend service that utilizes Python's ReportLab library to generate PDF reports. Furthermore, the Django framework handles the data logic involved in report generation, ensuring accurate and complete report content.

[0224] 4.4.3) View and modify

[0225] The generated flaw detection report needs to be viewed and modified at any time to meet the needs of users. The present invention uses the RESTful API provided by Django to handle viewing and modification operations, and processes the data logic of the report. React or Vue.js framework is used in conjunction with Redux or Vuex for state management to realize the viewing and modification functions of the report. The present invention establishes a RESTful API, processes user viewing and modification requests through the Django framework, and ensures the consistency of the report data. On the front end, the present invention uses React or Vue.js to build the user interface, and combines Redux or Vuex for state management to realize the viewing and modification functions of the report.

[0226] 4.4.4) Storage and Retrieval

[0227] To efficiently manage NDT report data, a suitable storage and retrieval solution is required. This invention proposes storing NDT report files and related data on a storage server, such as AWS S3, to ensure data security and reliability. Furthermore, this invention utilizes a MySQL or PostgreSQL database to store report metadata and integrates Elasticsearch to achieve more efficient report retrieval, improving the user experience. This storage and retrieval solution allows the invention to effectively manage large amounts of report data, ensuring its security and reliability.

[0228] It should be understood that "one embodiment" or "an embodiment" mentioned throughout the specification means that the specific features, structures or characteristics related to the embodiment are included in at least one embodiment of the present invention. Therefore, "in one embodiment" or "in an embodiment" appearing throughout the specification does not necessarily refer to the same embodiment. In addition, these specific features, structures or characteristics can be combined in one or more embodiments in any suitable manner. It should be understood that in various embodiments of the present invention, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiment of the present invention. The serial numbers of the above-mentioned embodiments of the present invention are for description only and do not represent the advantages and disadvantages of the embodiments.

[0229] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any modifications or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in the present invention should be included in the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be based on the scope of protection of the claims.

Claims

1. A solid rocket grain defect intelligent detection and visualization characterization system, characterized by: The system includes: a two-dimensional slice data preprocessing module, a three-dimensional visual modeling module, and a model post-processing and analysis module; The two-dimensional slice data preprocessing module is used to perform structural part identification and positioning on the two-dimensional ICT image in the acquired solid motor ICT file based on a preset circular structural part target automatic identification and detection algorithm, and determine the circular structure area image of the solid motor; An improved anisotropic diffusion denoising algorithm based on adaptive median filtering is used to remove pulse interference from the circular structure area image multiple times to obtain a circular structure area image after multiple denoising; Based on the robust adaptive Canny algorithm of defect edges, circular structure boundary detail extraction and noise interference elimination processing are sequentially performed on the circular structure area image after multiple denoising to obtain a high-precision edge image; The three-dimensional visualization modeling module is used to perform image registration, normalization, and ROI extraction on the original ICT slice image based on a preset deep learning ICT slice image interpolation algorithm to obtain an integrated continuous slice sequence; Use 3D convolutional neural networks to advance deep features of adjacent slices and align features through optical flow estimation or deformable convolution; Using the generative adversarial network architecture, the U-Net generator generates intermediate interpolated slices from two adjacent frame slices, and the PatchGAN discriminator is used to evaluate the authenticity of the generated image and optimize it through a multi-objective loss function; Anisotropic diffusion filtering is used to smooth the edges of the interpolated slices, and then the interpolated data and the original data are integrated through voxel fusion technology to obtain the processed slice data; Importing the processed slice data into a three-dimensional modeling module to construct a three-dimensional voxel grid and perform model reconstruction; Perform edge smoothing and voxel fusion optimization on the reconstructed 3D model to generate a high-precision interactive 3D visualization model; The model post-processing and analysis module is used to perform three-dimensional reconstruction of the pores on the pore image in the two-dimensional ICT image to obtain a three-dimensional reconstruction model of the deformed pores and a volume estimation result; Based on the volume estimation result, a fault diagnosis result of the solid motor is determined.

2. The system according to claim 1, wherein: The two-dimensional slice data preprocessing module also includes a circular area extraction submodule; The circular region extraction submodule is used to perform binary background segmentation, morphological processing and connected domain analysis on the two-dimensional ICT image to obtain multiple candidate regions; each candidate region carries the basic features of the solid motor; the basic features include area features, boundary features and geometric features; Eliminate non-target areas with an area smaller than a preset area threshold from the plurality of candidate areas to obtain a retained area; Numbering and segmenting the reserved area to obtain multiple image subsets; Performing Sobel operator edge extraction on each of the image subsets to obtain an edge point set; establishing a rectangular coordinate system based on the center of gravity of each of the edge point sets as the origin, and calculating the polar angles of the four vertices of the minimum circumscribed rectangle relative to the center; Based on the edge point set, calculating the circularity index of each of the image subsets, and eliminating areas that do not meet the circular feature to obtain areas that meet the circular feature; Randomly selecting a subset of edge points in the reserved area for circle fitting, using an optimized sampling strategy to reduce computational complexity, and generating candidate circles containing center coordinates and radius parameters; Each candidate circle is verified. If a sufficient number of edge points fall within a preset radius tolerance range of each candidate circle, it is determined to be a valid circle and the circular structure area image is output.

3. The system according to claim 1, wherein: The two-dimensional slice data preprocessing module also includes a filtering and denoising submodule; The filtering and denoising submodule is used to divide the circular structure area image into local areas to obtain the circular area image after division; Adopting an adaptive median filter to filter the divided circular region image, remove impulse noise, and obtain a preliminary denoised circular structure region image; Calculating the gradient modulus of the circular structure region image after preliminary denoising, replacing the gradient modulus of the original image with the gradient modulus, and dynamically adjusting the diffusion coefficient in the anisotropic diffusion model; Based on the adjusted diffusion coefficient, anisotropic diffusion denoising is performed on the circular structure region image after the preliminary denoising, retaining image edge details to obtain the circular structure region image after multiple denoising.

4. The system according to claim 1, wherein: The two-dimensional slice data preprocessing module also includes an adaptive Canny submodule; The adaptive Canny submodule is used to perform grayscale processing on the selected area in the circular structure area image after multiple denoising to obtain a grayscale processed image; Constructing a Gaussian filter to perform two-dimensional convolution on the grayscale processed image to reduce fluctuations in local details of the image, thereby obtaining a convolved image; Calculating the gradient magnitude and direction of the convolved image using the finite difference of the first-order partial derivative; performing non-maximum suppression on the gradient magnitude and retaining the local gradient maximum point; A dual threshold algorithm is used, with a high threshold used to segment the background and the target, and a low threshold used to connect the broken edges; The edge area is refined, the defect outline is extracted and the potential defect features are identified to obtain the high-precision edge image.

5. The system according to claim 1, wherein: The model post-processing and analysis module includes a pore ellipsoid model building module, a pore shape parameter extraction module, a three-dimensional display module and a volume calculation module; The pore ellipsoid model building module is used to simulate the deformed pore into two upper and lower triaxial ellipsoids. By determining the three-dimensional coordinates of the ellipsoid center, the three semi-axis lengths and the three-axis rotation angles, a total of nine parameters, a three-dimensional ellipsoid model is constructed. The origin of the ellipsoid model is the center of the solid rocket engine base. The pore shape parameter extraction module is used to preprocess the pore image to obtain a pore edge binary image, first directly calculate the ellipse center coordinates, the semi-major axis length, and the angle between the major axis and the horizontal axis, then divide the pore edge image according to the major axis, use Hough transform to detect the semi-minor axis lengths of the upper and lower halves of the ellipse respectively, and determine the optimal semi-minor axis parameter from multiple solutions using the area minimum error method; The three-dimensional display module is used to visualize the upper and lower half ellipsoids of the pore based on the constructed three-dimensional regular volume data field and the ellipsoid parameters; The volume calculation module is used to perform three-dimensional reconstruction on the pore edge binary image, calculate the pore volume using a volume calculation formula, and calculate the average value of the pore volume in multiple images.

6. The system according to claim 1, wherein: The system also includes a centralized management platform; the centralized management platform includes a front-end interface module, a confidentiality management module, a data security management module and a flaw detection report management module; The front-end interface module is used to create a cross-platform operation interface based on the PyQt library and QtDesigner. The interface includes an interactive bar on the left and a preview box on the right. The interactive bar on the left is used to receive user operation instructions, and the preview box on the right is used to load CT images and 3D reconstructed images. The front-end interface module supports users to import CT slice images, generate 3D images, analyze and interact with 3D images, and output reconstruction results and defect statistics. The confidentiality management module is used to encrypt sensitive data using the AES-GCM algorithm of the Crypto.js library, store encryption keys on the back end, and ensure data transmission security through HTTPS; based on the RBAC model, access control is implemented by storing user access control lists in the database and recording sensitive data access logs; The data security management module includes a data encryption submodule and a data backup and recovery submodule. The data encryption submodule uses a database engine that supports transparent data encryption to encrypt database and file system data and implements a key rotation strategy. The data backup and recovery submodule uses Veeam Backup & Replication to perform full and incremental backups and manage the storage and access rights of backup data. The flaw detection report management module is used to store document template information based on the Django backend framework and MySQL / PostgreSQL database, and combines the React / Vue.js front-end framework and Ant Design / Bootstrap for interface design; uses Python's ReportLab library to generate PDF format reports, and uses Django to process report data logic; uses the RESTful API provided by Django to handle report viewing and modification operations, and realizes report interaction through state management; stores flaw detection report files on a storage server, uses the MySQL / PostgreSQL database to store report metadata, and combines with Elasticsearch to realize report retrieval.

7. A method for intelligent detection and visualization of defects in solid motor grains, applied to a system for intelligent detection and visualization of defects in solid motor grains, characterized in that: The method comprises: Based on a preset circular structural component target automatic recognition and detection algorithm, the structure component recognition and positioning are performed on the two-dimensional ICT image in the acquired solid motor ICT file to determine the circular structure area image of the solid motor; An improved anisotropic diffusion denoising algorithm based on adaptive median filtering is used to remove pulse interference from the circular structure area image multiple times to obtain a circular structure area image after multiple denoising; Based on the robust adaptive Canny algorithm of defect edges, circular structure boundary detail extraction and noise interference elimination processing are sequentially performed on the circular structure area image after multiple denoising to obtain a high-precision edge image; Based on the preset deep learning ICT slice image interpolation algorithm, the original ICT slice images are processed with image registration, normalization, and ROI extraction to obtain an integrated continuous slice sequence; Use 3D convolutional neural networks to advance deep features of adjacent slices and align features through optical flow estimation or deformable convolution; Using the generative adversarial network architecture, the U-Net generator generates intermediate interpolated slices from two adjacent frame slices, and the PatchGAN discriminator is used to evaluate the authenticity of the generated image and optimize it through a multi-objective loss function; Anisotropic diffusion filtering is used to smooth the edges of the interpolated slices, and then the interpolated data and the original data are integrated through voxel fusion technology to obtain the processed slice data; Importing the processed slice data into a 3D modeling module, constructing a 3D voxel grid and reconstructing the model; performing edge smoothing and voxel fusion optimization on the reconstructed 3D model to generate a high-precision interactive 3D visualization model; Performing three-dimensional reconstruction of the pores on the pore image in the two-dimensional ICT image to obtain a three-dimensional reconstruction model of the deformed pores and a volume estimation result; and determining a fault diagnosis result of the solid motor based on the volume estimation result.

Images