An image-based aircraft shape parameter inversion and analytical reconstruction method
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-04-01
- Publication Date
- 2026-08-11
AI Technical Summary
1、人工测量依赖度高：传统的参数提取方式主要依赖设计人员在CAD或制图软件中导入图片，通过手工描点、拉线测量像素距离并进行比例换算
[0018]有益效果：与现有技术相比，本发明的显著效果是：本发明仅需输入包含飞行目标的图片数据和飞行器参考长度即可输出完整的几何参数报告。在进行结构分割时引入Active Contour算法，相比传统边缘检测方法，主动轮廓模型能有效抑制背景噪声干扰，保证了轮廓提取的连续性与封闭性。同时，本发明实现了从图像数据到设计参数的反演优化：采用理论参考数据与图像识别数据对比的策略，通过反演优化算法在参数空间内进行迭代搜索，寻找能使数据库生成的理论模型与图像真实坐标误差最小的最优设计参数，实现非合作目标图像数据到高置信度模型的获取，显著提升了目标特性分析的效率与准确度。
Smart Images
Figure CN122548855A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the interdisciplinary field of aircraft design and computer vision, specifically to an image-based method for inverting and reconstructing aircraft shape parameters. Background Technology
[0002] In applications such as the early stages of aircraft conceptual design, aerodynamic layout evaluation, and non-cooperative target characteristic analysis, obtaining accurate geometric parameters of the target is a prerequisite for computational fluid dynamics (CFD) simulation and aircraft performance calculation. However, in practical engineering applications, it is often difficult to directly obtain detailed CAD models or design drawings of the target; only image data containing the target (such as publicly available photographs or design sketches) is available. Therefore, researchers need to extract the geometric parameters of the target aircraft from images.
[0003] Currently, methods for obtaining aircraft geometric parameters from images mainly face the following technical bottlenecks: 1. High reliance on manual measurement: Traditional parameter extraction methods mainly rely on designers importing images into CAD or drafting software, manually plotting points, drawing lines to measure pixel distances, and performing proportional conversions. This method is not only labor-intensive, but the obtained geometric parameters are also affected by the designer's subjective judgment, resulting in poor data consistency and repeatability, and cannot meet the needs of large-scale data analysis.
[0004] 2. Traditional image segmentation algorithms lack specificity and robustness: Existing edge extraction algorithms typically use general gradient operators directly; however, aircraft images often face complex background interference (such as clouds, ground textures) or imaging noise. Ordinary edge detection cannot distinguish between target edges and background textures, easily generating a large number of false edges or causing key contours to break, resulting in inaccurate contour extraction results.
[0005] 3. Lack of a Parameter Inversion Mechanism Based on a Physical Model: Most existing extraction techniques fall under the category of forward fitting, which involves first extracting discrete pixel edge points and then directly using polynomials or spline curves for mathematical fitting. The drawbacks of this method are the lack of constraints from prior knowledge of geometric configuration: Firstly, the physical meaning is missing; simple fitting may be affected by image noise, generating wavy contours that do not conform to fluid dynamics and cannot correspond to normal aircraft design parameters. Secondly, errors generated in the preprocessing stage cannot be corrected; traditional fitting methods are open-loop, and when contour deviations occur during image preprocessing, subsequent processing steps can only accept this error. Finally, there is a lack of a closed-loop optimization mechanism; there is no optimization step that compares the virtual image generated by the theoretical model with the real image, making it impossible to use feedback loops to correct parameter errors.
[0006] Therefore, there is an urgent need for a method that can extract contours from complex backgrounds, combine them with an aircraft shape design database, and use an inversion optimization algorithm to find the optimal solution in the parameter space, so as to achieve intelligent reconstruction from image pixels to high-confidence physical parameters. Summary of the Invention
[0007] Purpose of the invention: The purpose of this invention is to provide a method for inverting and reconstructing the shape parameters of an aircraft based on image extraction of the aircraft contour and optimization of the aircraft parameters.
[0008] Technical solution: A method for image-based inversion and analytical reconstruction of aircraft shape parameters, comprising the following steps: S1. Acquire images of the aircraft and perform preprocessing; S2. Perform structural segmentation on the preprocessed aircraft image and extract the binarized outline mask of the aircraft target; S3. Perform geometric screening on the extracted binary outline mask to obtain the aircraft target, perform attitude normalization on the aircraft target, and correct the aircraft target to a horizontal attitude. S4. Construct a coordinate system based on the corrected binarized outline mask and extract the pixel coordinate dataset of the aircraft target outline; S5. Based on the true value of the aircraft feature length, scale transformation is performed on the pixel coordinates of the aircraft target outline to obtain the physical space coordinate dataset of the true outline of the aircraft target. S6. Based on the aircraft shape database, define the characteristic parameters of the aircraft to be inverted, construct a parameterized analytical model driven by the characteristic parameters, extract the physical space coordinate dataset of the theoretical outline of the aircraft target, calculate the geometric deviation between the physical space coordinate dataset of the actual outline and the physical space coordinate dataset of the theoretical outline, use the geometric deviation as the inversion objective function, and use numerical optimization algorithms to iteratively optimize the characteristic parameters until the geometric deviation is minimized to obtain the optimal aircraft characteristic parameters. Based on the optimal aircraft characteristic parameters, construct the aircraft geometric model.
[0009] Specifically, the preprocessing includes: converting the acquired aircraft image from a color image to a single-channel grayscale image, smoothing it using a median filter, and then using a morphological gradient method to enhance the contrast between the target and the background in the aircraft image.
[0010] Specifically, the morphological gradient method includes: performing dilation and erosion operations on the aircraft image respectively, using the image difference between the dilation and erosion operations as a gradient map, and using an image adjustment function to stretch the gradient map to enhance the edge features in the aircraft image.
[0011] Specifically, step S2 includes: using the Canny operator to extract edge features from the aircraft image, performing morphological closing operations, connecting the broken contours in the aircraft image, generating an initial mask by filling holes, and then applying the ActiveContour algorithm to iteratively evolve the initial mask to obtain a binary outline mask of the aircraft target.
[0012] Specifically, the geometric screening includes: using region attribute functions to extract the area, eccentricity, and orientation of all connected regions in the binary outline mask; setting the product of the area of the connected region and the square of the eccentricity as the target score; and selecting the connected region with the largest target score as the aircraft target.
[0013] Specifically, attitude normalization processing includes: constructing a rotation matrix based on the orientation of the aircraft target, and rotating the aircraft target to a horizontal position through affine transformation.
[0014] Specifically, step S4 includes: integrating and summing the pixels along the vertical direction on the corrected binarized outline mask to obtain the thickness distribution curve; identifying the pixel index of the aircraft target shape feature through the gradient abrupt change feature of the thickness distribution curve; and constructing a pixel coordinate dataset of the aircraft target outline.
[0015] Specifically, step S5 includes: calculating the scale between the target aircraft and the actual aircraft based on the input true value of the aircraft feature length, and using the scale to perform a scale transformation on the pixel coordinates of the target aircraft outline.
[0016] Specifically, geometric deviation is the root mean square error between the physical spatial coordinates of the actual outline and the physical spatial coordinates of the theoretical outline.
[0017] Specifically, the numerical optimization algorithm is a nonlinear optimization algorithm.
[0018] Beneficial Effects: Compared with existing technologies, the significant advantages of this invention are: it only requires input of image data containing the flying target and the aircraft reference length to output a complete geometric parameter report. When performing structural segmentation, the Active Contour algorithm is introduced. Compared with traditional edge detection methods, the active contour model can effectively suppress background noise interference, ensuring the continuity and closure of contour extraction. Simultaneously, this invention achieves inversion optimization from image data to design parameters: employing a strategy of comparing theoretical reference data with image recognition data, the inversion optimization algorithm iteratively searches within the parameter space to find the optimal design parameters that minimize the error between the theoretical model generated by the database and the actual coordinates of the image. This enables the acquisition of a high-confidence model from non-cooperative target image data, significantly improving the efficiency and accuracy of target characteristic analysis. Attached Figure Description
[0019] Figure 1This is a flowchart of the method of the present invention.
[0020] Figure 2 This is a schematic diagram illustrating the effect of the image processing process of the present invention.
[0021] Figure 3 This is a schematic diagram of the thickness distribution curve of the present invention.
[0022] Figure 4 This is a schematic diagram comparing the initial parameters and inversion optimization parameters of the aircraft of this invention. Detailed Implementation
[0023] A preferred embodiment of the present invention will be further described below with reference to the accompanying drawings.
[0024] Example
[0025] Please see Figure 1 As shown, this embodiment provides an image-based method for inverting and analytically reconstructing the external parameters of an aircraft, including the following steps: S1. Acquire images of the aircraft and perform preprocessing.
[0026] In this embodiment, after reading the aircraft image, the image is first converted from color to single-channel grayscale. To eliminate salt-and-pepper noise during the imaging process, a median filter with a window size of 3x3 is used for smoothing. Then, a morphological gradient method is used to enhance the contrast between the target and the background in the aircraft image: dilation and erosion operations are performed on the aircraft image respectively, and the difference between the dilation and erosion operations is used as a gradient map. The gradient map is then stretched in grayscale using an image adjustment function to enhance the edge features in the aircraft image and provide clear gradient information for subsequent segmentation.
[0027] S2. Perform structural segmentation on the preprocessed aircraft image and extract the binarized outline mask of the aircraft target.
[0028] Traditional Canny operators are prone to producing discontinuous lines or false edges in complex backgrounds. Therefore, this embodiment adopts a hybrid segmentation strategy. First, the edge features in the aircraft image are extracted using the Canny operator, and morphological closing operations are performed to connect the broken contours in the aircraft image. A solid initial mask is generated by filling holes. Then, the ActiveContour algorithm is applied to iteratively evolve the initial mask to obtain a binary outline mask of the aircraft target. This algorithm constructs a region-based energy function and minimizes the energy to drive the contour lines to approach the maximum gray-level gradient in the image, thereby stripping away complex backgrounds and obtaining a smooth, accurate, and closed binary outline mask.
[0029] Please refer to Figure 2 As shown, Figure 2 (a) is the original image, obtained through structural segmentation. Figure 2 (b) shows the binarized image.
[0030] S3. Perform geometric screening on the extracted binary outline mask to obtain the aircraft target, perform attitude normalization on the aircraft target, and correct the aircraft target to a horizontal attitude.
[0031] In this embodiment, the area, eccentricity, and orientation of all connected components in the binary outline mask are first extracted using a region attribute function. Since aircraft are typically elongated structures, the product of the area of the connected component and the square of the eccentricity is set as the target score. The connected component with the largest target score is selected as the aircraft target. This scoring and filtering effectively removes interference noise such as clouds and ground clutter. After selecting the aircraft target, a rotation matrix is constructed based on the direction of the aircraft target, and the aircraft target is rotated to a horizontal position through an affine transformation. Then, the total pixel mass of the left and right halves of the rotated aircraft image is calculated. Based on the prior information that the aerodynamic layout of aircraft is usually nose-light and tail-heavy (the tail usually has a tail fin with a large pixel area), if the mass of the left side is greater than that of the right side, it is determined that the nose is facing right, and a 180-degree flip operation is performed on the image to ensure that all output results are consistent with the standard attitude of nose facing left and tail facing right.
[0032] S4. Construct a coordinate system based on the corrected binary outline mask and extract the pixel coordinate dataset of the aircraft target outline.
[0033] Please refer to Figure 2 (c) and Figure 3 As shown, in this embodiment, the pixels of the corrected binarized outline mask are integrated and summed along the vertical direction to obtain a thickness distribution curve that varies along the axial direction. The gradient abrupt change features of the thickness distribution curve are used to identify the main shape features of the aircraft target, such as the nose, wing root, and wingtip. Figure 2 (c) shows the identified feature points on the wingtips and tail, and obtains their pixel indices to construct a pixel coordinate dataset of the aircraft target outline.
[0034] S5. Based on the true value of the aircraft feature length, scale transformation is performed on the pixel coordinates of the aircraft target outline to obtain the physical space coordinate dataset of the true outline of the aircraft target.
[0035] Specifically, based on the true value of the input aircraft feature length, such as the actual total length of the aircraft, the scale between the aircraft target and the actual aircraft is calculated. The scale is then used to perform a scale transformation on the pixel coordinates of the aircraft target's outline, thereby obtaining the physical space coordinate dataset of the true outline.
[0036] S6. Based on the aircraft shape database, define the characteristic parameters of the aircraft to be inverted, construct a parameterized analytical model driven by the characteristic parameters, extract the physical space coordinate dataset of the theoretical outline of the aircraft target, calculate the geometric deviation between the physical space coordinate dataset of the actual outline and the physical space coordinate dataset of the theoretical outline, use the geometric deviation as the inversion objective function, and use numerical optimization algorithms to iteratively optimize the characteristic parameters until the geometric deviation is minimized to obtain the optimal aircraft characteristic parameters. Based on the optimal aircraft characteristic parameters, construct the aircraft geometric model.
[0037] In this embodiment, the inversion objective function is defined. Physical space coordinates of the actual outline Physical spatial coordinates of the theoretical outline Root mean square error between:
[0038] In the formula: This represents the number of coordinate points in the coordinate dataset.
[0039] In this embodiment, a nonlinear optimization algorithm is used. Optimization stops when the inversion objective function converges below a preset threshold or reaches the maximum number of iterations. The parameter vector at this point represents the optimal aircraft characteristic parameters obtained through inversion, and the system then outputs the final aircraft geometric model based on these parameters.
[0040] Please refer to Figure 4 As shown, the inverted parameters correspond to a geometric model of the aircraft that is closer to the actual physical structure of the aircraft than the initial parameter conditions.
Claims
1. An image-based aircraft shape parameter inversion and analytical reconstruction method, characterized in that, Includes the following steps: S1. Acquire images of the aircraft and perform preprocessing; S2. Perform structural segmentation on the preprocessed aircraft image and extract the binarized outline mask of the aircraft target; S3. Perform geometric screening on the extracted binary outline mask to obtain the aircraft target, perform attitude normalization on the aircraft target, and correct the aircraft target to a horizontal attitude. S4. Construct a coordinate system based on the corrected binarized outline mask and extract the pixel coordinate dataset of the aircraft target outline; S5. Based on the true value of the aircraft feature length, scale transformation is performed on the pixel coordinates of the aircraft target outline to obtain the physical space coordinate dataset of the true outline of the aircraft target. S6. Based on the aircraft shape database, define the characteristic parameters of the aircraft to be inverted, construct a parameterized analytical model driven by the characteristic parameters, extract the physical space coordinate dataset of the theoretical outline of the aircraft target, calculate the geometric deviation between the physical space coordinate dataset of the actual outline and the physical space coordinate dataset of the theoretical outline, use the geometric deviation as the inversion objective function, and use numerical optimization algorithms to iteratively optimize the characteristic parameters until the geometric deviation is minimized to obtain the optimal aircraft characteristic parameters. Based on the optimal aircraft characteristic parameters, construct the aircraft geometric model.
2. The image-based aircraft shape parameter inversion and analytical reconstruction method of claim 1, wherein, The preprocessing includes: converting the acquired aircraft image from a color image to a single-channel grayscale image, smoothing it using a median filter, and then using a morphological gradient method to enhance the contrast between the target and the background in the aircraft image.
3. The image-based aircraft configuration parameter inversion and analytical reconstruction method of claim 2, wherein, The morphological gradient method includes: performing dilation and erosion operations on the aircraft image respectively, using the image difference between the dilation and erosion operations as a gradient map, and using an image adjustment function to stretch the gradient map to enhance the edge features in the aircraft image.
4. The image-based aircraft configuration parameter inversion and analytical reconstruction method of claim 1, wherein, Step S2 includes: extracting edge features from the aircraft image using the Canny operator, performing morphological closing operations, connecting broken contours in the aircraft image, generating an initial mask by filling holes, and then applying the Active Contour algorithm to iteratively evolve the initial mask to obtain a binary outline mask of the aircraft target.
5. The image-based aircraft configuration parameter inversion and analytical reconstruction method of claim 1, wherein, The geometric screening includes: extracting the area, eccentricity, and orientation of all connected regions in the binary outline mask using a region attribute function; setting the product of the area of the connected region and the square of the eccentricity as the target score; and selecting the connected region with the largest target score as the aircraft target.
6. The image-based method for inverting and analytically reconstructing aircraft shape parameters according to claim 5, characterized in that, The attitude normalization process includes: constructing a rotation matrix based on the orientation of the aircraft target, and rotating the aircraft target to a horizontal position through affine transformation.
7. The image-based aircraft configuration parameter inversion and analytical reconstruction method of claim 1, wherein, Step S4 includes: integrating and summing the pixels along the vertical direction on the corrected binarized outline mask to obtain the thickness distribution curve; identifying the pixel index of the aircraft target shape feature through the gradient abrupt change feature of the thickness distribution curve; and constructing a pixel coordinate dataset of the aircraft target outline.
8. The image-based aircraft configuration parameter inversion and analytical reconstruction method of claim 1, wherein, Step S5 includes: calculating the scale between the target aircraft and the actual aircraft based on the input true value of the aircraft feature length, and using the scale to perform a scale transformation on the pixel coordinates of the target aircraft outline.
9. The image-based aircraft configuration parameter inversion and analytical reconstruction method of claim 1, wherein, The geometric deviation is the root mean square error between the physical spatial coordinates of the actual outline and the physical spatial coordinates of the theoretical outline.
10. The image-based aircraft configuration parameter inversion and analytical reconstruction method of claim 9, wherein, The numerical optimization algorithm is a nonlinear optimization algorithm.