An aerostructure independent patch extraction method, apparatus, device and medium

Abstract

The application discloses an aviation piece independent face sheet extraction method, device, equipment and medium, and relates to the technical field of independent face sheet extraction of an airplane. The method comprises the following steps: acquiring a plurality of two-dimensional images of the target aviation piece; obtaining a plurality of row pixel point coordinates which are the same as the initial seed point pixel in the plurality of two-dimensional images; reading the plurality of row pixel point coordinates in sequence, so that each row of the pixel point coordinates constructs a background color image; obtaining the clustering center coordinates of each independent face sheet in the two-dimensional image; obtaining a plurality of MASK images; extracting a first extraction result and a second extraction result of the independent face sheet in the two-dimensional image; and obtaining the independent face sheet of the target aviation piece based on the first extraction result and the second extraction result. The application can more accurately extract the independent face sheet of the target aviation piece, so that the target aviation piece can be more accurately programmed, and a target aviation piece which can meet the requirements can be processed.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of extraction of aircraft independent patches, and particularly relates to an aircraft independent patch extraction method, device, equipment and medium. BACKGROUND

[0002] An aircraft part is a general term for functional components such as passenger planes, fighter planes and unmanned aerial vehicles. Taking a passenger plane as an example, a complete passenger plane is composed of thousands of aircraft parts, each of which has a specific function and attribute. However, the processing of aircraft parts is heavily dependent on numerical control machine tools. Only reliable materials, high-precision machine tools and reasonably designed processing programs can produce parts that meet assembly requirements. Among them, the numerical control machine tool needs a program compilation process, which is commonly referred to as programming in the industry. In order to program the aircraft parts, the independent patches of the aircraft parts need to be extracted.

[0003] However, the prior art cannot accurately extract the independent patches of the aircraft parts, so that the aircraft parts cannot be more accurately programmed. SUMMARY

[0004] The main purpose of the present application is to provide an aircraft independent patch extraction method, device, equipment and medium, which aims to solve the technical problem that the prior art cannot accurately extract the independent patches of the aircraft parts, so that the aircraft parts cannot be more accurately programmed.

[0005] To achieve the above purpose, the first aspect of the present application provides an aircraft independent patch extraction method, which comprises:

[0006] Projecting a target aircraft part from multiple perspectives to obtain a plurality of two-dimensional images of the target aircraft part;

[0007] Performing global random initial seed point search on a plurality of two-dimensional images to obtain a plurality of pixel point coordinates in the two-dimensional images that are the same as the initial seed point pixels;

[0008] Reading a plurality of pixel point coordinates in sequence to make each row of pixel point coordinates construct a background color image;

[0009] Based on the background color image, obtaining the cluster center coordinates of each independent patch in the two-dimensional image;

[0010] Based on the cluster center coordinates, determining whether the region corresponding to the two-dimensional image is a background region to obtain a plurality of MASK images;

[0011] extract a first extraction result and a second extraction result of the independent patches in the two-dimensional image based on the plurality of MASK images, wherein a ratio of a foreground area to a background area of the second extraction result is greater than a ratio of a foreground area to a background area of the first extraction result;

[0012] obtain the independent patches of the target aero-component based on the first extraction result and the second extraction result.

[0013] Optionally, the extracting the first extraction result and the second extraction result of the independent patches in the two-dimensional image based on the plurality of MASK images comprises:

[0014] obtaining projection curtains of the plurality of MASK images in X-axis and Y-axis directions;

[0015] obtaining a plurality of extreme value coupled region coordinates of the plurality of MASK images based on the projection curtains of the plurality of MASK images in the X-axis and Y-axis directions;

[0016] sequentially processing the extreme value coupled region coordinates corresponding to the plurality of MASK images in the coordinate axes to obtain projection coordinate values of all groups of the independent patches;

[0017] segmenting the two-dimensional image based on the projection coordinate values to obtain the second extraction result.

[0018] Optionally, the extracting the first extraction result and the second extraction result of the independent patches in the two-dimensional image based on the plurality of MASK images comprises:

[0019] multiplying a plurality of the MASK images with values of a plurality of corresponding pixel positions of the two-dimensional image to extract the first extraction result.

[0020] Optionally, before the step of sequentially reading a plurality of rows of the pixel point coordinates to make each row of the pixel point coordinates construct a background color image, the method further comprises:

[0021] sequentially recording the plurality of rows of the pixel point coordinates in a text document;

[0022] adding the same gray value to the pixel points in each row of the text document;

[0023] initializing the gray value of a boundary region in the two-dimensional image after adding the gray value;

[0024] the sequentially reading a plurality of rows of the pixel point coordinates to make each row of the pixel point coordinates construct a background color image comprises:

[0025] Read the pixel point coordinates of the plurality of rows in sequence after the initialization processing, so that each row of the pixel point coordinates constructs a background color image.

[0026] Optionally, the global random initial seed point search on the plurality of two-dimensional images to obtain a plurality of pixel point coordinates in the plurality of two-dimensional images that are the same as the pixel of the initial seed point comprises:

[0027] If the pixel of the initial seed point is the same as the pixel in the X-axis direction of the two-dimensional image, the initial seed point is moved by one pixel unit in the X-axis direction to obtain an X-axis pixel point coordinate;

[0028] If the pixel of the initial seed point is the same as the pixel in the Y-axis direction of the two-dimensional image, the initial seed point is moved by one pixel unit in the Y-axis direction to obtain a Y-axis pixel point coordinate;

[0029] Based on the X-axis pixel point coordinate and the Y-axis pixel point coordinate, a plurality of pixel point coordinates in the plurality of two-dimensional images that are the same as the pixel of the initial seed point are obtained.

[0030] Optionally, before the global random initial seed point search on the plurality of two-dimensional images to obtain a plurality of pixel point coordinates in the plurality of two-dimensional images that are the same as the pixel of the initial seed point, the method further comprises:

[0031] Based on the constructed noise removal model, the noise of the plurality of two-dimensional images is removed;

[0032] The position of a foreign element in the two-dimensional image after the noise is removed is determined; wherein the foreign element is an object that is different in color from the two-dimensional image;

[0033] The foreign element in the two-dimensional image is removed;

[0034] The global random initial seed point search on the plurality of two-dimensional images to obtain a plurality of pixel point coordinates in the plurality of two-dimensional images that are the same as the pixel of the initial seed point comprises:

[0035] The global random initial seed point search on the plurality of two-dimensional images after the foreign element is removed to obtain a plurality of pixel point coordinates in the plurality of two-dimensional images that are the same as the pixel of the initial seed point.

[0036] Optionally, before the global random initial seed point search on the plurality of two-dimensional images to obtain a plurality of pixel point coordinates in the plurality of two-dimensional images that are the same as the pixel of the initial seed point, the method further comprises:

[0037] A scaling parameter when the target aircraft is projected is obtained; wherein the scaling parameter is a scale parameter in a three-dimensional model represented by each pixel point in a two-dimensional image;

[0038] Project the target aerostructure under multiple perspectives based on the scaling parameter when the target aerostructure is projected, to obtain several two-dimensional images of the target aerostructure.

[0039] In a second aspect, the present application provides an extraction device for an independent patch of an aerostructure, the device comprising:

[0040] An acquisition module is configured to project the target aerostructure under multiple perspectives to obtain several two-dimensional images of the target aerostructure.

[0041] A first obtaining module is configured to perform a global random initial seed point search on the several two-dimensional images to obtain a plurality of row pixel point coordinates in the several two-dimensional images that are the same as the initial seed point pixel.

[0042] A reading module is configured to sequentially read the plurality of row pixel point coordinates, so that each row of the pixel point coordinates constructs a background color image.

[0043] A second obtaining module is configured to obtain a cluster center coordinate of each independent patch in the two-dimensional image based on the background color image.

[0044] A judging module is configured to judge whether a region corresponding to the two-dimensional image is a background region based on the cluster center coordinate, to obtain a plurality of MASK images.

[0045] An extraction module is configured to extract a first extraction result and a second extraction result of an independent patch in the two-dimensional image based on the plurality of MASK images, wherein a ratio of a foreground region area to a background region area of the second extraction result is greater than a ratio of a foreground region area to a background region area of the first extraction result.

[0046] A third obtaining module is configured to obtain an independent patch of the target aerostructure based on the first extraction result and the second extraction result.

[0047] In a third aspect, the present application provides a computer device, which comprises a memory and a processor, the memory stores a computer program, and the processor executes the computer program to implement the method in the embodiments.

[0048] In a fourth aspect, the present application provides a computer readable storage medium, which stores a computer program, and a processor executes the computer program to implement the method in the embodiments.

[0049] Through the above technical solutions, the present application has at least the following beneficial effects:

[0050] The present application provides a method, apparatus, device, and medium for extracting independent facets of aerospace components. The method involves first projecting the target aerospace component from multiple viewpoints to obtain several two-dimensional images of the target aerospace component; then performing a global random initial seed point search on the several two-dimensional images to obtain multiple rows of pixel coordinates that are identical to the initial seed point pixels in the several two-dimensional images; then sequentially reading the multiple rows of pixel coordinates to construct a background color image for each row of pixel coordinates; then obtaining the cluster center coordinates of each independent facet in the two-dimensional images based on the background color image; then determining whether the region corresponding to the two-dimensional image is a background region based on the cluster center coordinates to obtain several mask images; then extracting a first extraction result and a second extraction result of the independent facets in the two-dimensional images based on the several mask images; wherein the ratio of the foreground region area to the background region area in the second extraction result is greater than the ratio of the foreground region area to the background region area in the first extraction result; finally, obtaining the independent facets of the target aerospace component based on the first extraction result and the second extraction result. In other words, when it is necessary to extract individual facets of a target aerospace component, the three-dimensional image of the target aerospace component is first projected into several two-dimensional images from different perspectives. Then, based on the pixels of the initial seed point, all two-dimensional images are searched, and the coordinates of multiple rows of pixels identical to the initial seed point are recorded. Based on the coordinates of these multiple rows of pixels, a background color image is constructed for each row of pixels. Then, the cluster center coordinates of each individual facet in the two-dimensional image are obtained through the background color image. Based on the cluster center coordinates, several mask images are obtained. Based on these several mask images, the first extraction result and the second extraction result of the individual facets are extracted. Finally, the individual facets of the target aerospace component are extracted. That is, because this application traverses the entire two-dimensional image through the initial seed point, all information of the individual facets in the two-dimensional image is extracted, which lays the foundation for accurately extracting the individual facets of the target aerospace component. Based on several mask images, two different types of first extraction results and second extraction results of the individual facets in the two-dimensional image are obtained. The first extraction result and the second extraction result are combined to form the extraction result, i.e., the individual facet. Since the first extraction result and the second extraction result together form independent patches, and the first extraction result and the second extraction result can enrich the different requirements of different algorithms for the feature form of the target aerospace parts, the independent patches of the target aerospace parts can be extracted more accurately, so that the target aerospace parts can be programmed more accurately, and thus the target aerospace parts that can better meet the requirements can be manufactured. Attached Figure Description

[0051] Figure 1 This is a schematic diagram of the computer device structure for the hardware operating environment involved in the embodiments of this application;

[0052] Figure 2 A flow chart of an independent surface patch extraction method of an aviation part according to an embodiment of the present application;

[0053] Figure 3 A three-dimensional image structure diagram of a target aviation part according to an embodiment of the present application;

[0054] Figure 4 A three-dimensional image structure diagram of a target aviation part after parameter design according to an embodiment of the present application;

[0055] Figure 5 A two-dimensional image diagram of a target aviation part under multi-view projection according to an embodiment of the present application;

[0056] Figure 6 A two-dimensional image diagram of pure background single foreground difference based on equal scale construction according to an embodiment of the present application;

[0057] Figure 7 A first extraction result diagram of equal scale independent surface patch extraction of an original input image according to an embodiment of the present application;

[0058] Figure 8 A parallel view projection ray diagram constructed along a horizontal direction of a two-dimensional image according to an embodiment of the present application;

[0059] Figure 9 A parallel view projection ray diagram constructed along a vertical direction of a two-dimensional image according to an embodiment of the present application;

[0060] Figure 10 A coupled region diagram of a pair of projection extreme values obtained along coordinate axes of a two-dimensional image according to an embodiment of the present application;

[0061] Figure 11 A second extraction result diagram of equal scale independent surface patch extraction of an original input image according to an embodiment of the present application;

[0062] Figure 12 An independent surface patch image diagram after adding semantic information in a time margin random number manner according to an embodiment of the present application;

[0063] Figure 13 An image diagram after boundary region semantic information is initialized according to an embodiment of the present application;

[0064] Figure 14 A ray method vector measurement search diagram corresponding to a global random initial seed point according to an embodiment of the present application;

[0065] Figure 15 A schematic diagram of an independent surface patch extraction device of an aviation part according to an embodiment of the present application.

[0066] The implementation, functional features and advantages of the present application will be further described with reference to the embodiments in conjunction with the drawings. DETAILED DESCRIPTION

[0067] It should be understood that the specific embodiments described herein are merely intended to explain the present application and are not intended to limit the present application.

[0068] Aircraft parts are functional components that make up passenger planes, fighter planes, and unmanned aerial vehicles, etc. Taking a passenger plane as an example, a complete passenger plane is composed of thousands of aircraft parts, each of which has its specific function and attributes. However, the processing of aircraft parts is heavily dependent on numerical control machine tools, and only reliable materials, high-precision machine tools, and reasonably designed processing programs can produce parts that meet the assembly requirements. The program preparation process required by the numerical control machine tool is commonly referred to as programming in the industry. The existing programming method in the aviation field is mainly based on the relative position of specific facets in the three-dimensional model of the aircraft part. The specific functional facets can be divided into two categories, namely driving facets and guiding facets. The driving facets play a boundary role in the plane direction to prevent the milling cutter from milling through the part, while the guiding facets determine the range that should be milled. The web can be used as a driving facet, while the side wall and the round corner facets are considered as guiding facets. These functional facets have certain independent characteristics in space, so they can also be called independent facets. However, the existing programming method still heavily relies on manual work worldwide, and lacks effective automatic programming strategies. To achieve automatic programming, the identification and judgment of independent facets in the three-dimensional model of the aircraft part is one of the most core tasks.

[0069] With the development of computer vision and pattern recognition technology, it is possible to realize the identification of independent facets in aircraft parts based on image-based methods. Deep learning has strong feature extraction and abstraction capabilities, which further ensures the accuracy and precision of independent facet identification, and brings new opportunities and possibilities for the realization of automatic programming technology. However, the prerequisite for accurate identification and judgment of independent facets in aircraft parts based on deep learning technology is to first obtain the two-dimensional independent facets for research. However, the independent facets of the aircraft part cannot be accurately extracted at present, so the programming of the aircraft part cannot be more accurate.

[0070] To solve the above technical problems, the present application provides a method, device, equipment and medium for extracting independent facets of aircraft parts. Before introducing the specific technical solutions of the present application, the hardware running environment involved in the embodiment of the present application is introduced.

[0071] Reference Figure 1 , Figure 1 The computer device structure diagram of the hardware running environment involved in the embodiment of the present application is shown in the figure.

[0072] As Figure 1 shown, the computer device can include a processor 1001, such as a central processing unit (CPU), a communication bus 1002, a user interface 1003, a network interface 1004, and a memory 1005. The communication bus 1002 is used to realize the connection and communication between the components. The user interface 1003 can include a display, an input unit such as a keyboard, and can also include a standard wired interface, a wireless interface. The network interface 1004 can optionally include a standard wired interface, a wireless interface (such as a wireless fidelity (WIreless-FIdelity, WI-FI) interface). The memory 1005 can be a high-speed random access memory (RAM) memory, or a stable non-volatile memory (Non-Volatile Memory, NVM), such as a disk memory. The memory 1005 can also be a storage device independent of the aforementioned processor 1001.

[0073] Those skilled in the art can understand that Figure 1 the structure shown in the foregoing embodiments does not constitute a limitation on the computer device, and can include more or fewer components than the illustrated components, or combine certain components, or different component arrangements.

[0074] As Figure 1 shown, the memory 1005 as a storage medium can include an operating system, a data storage module, a network communication module, a user interface module, and an electronic program.

[0075] In Figure 1 the computer device shown, the network interface 1004 is mainly used for data communication with a network server; the user interface 1003 is mainly used for data interaction with a user; the processor 1001 and the memory 1005 in the computer device of the present application can be arranged in the computer device, and the computer device calls the extraction device of the independent face sheet of the aeronautical piece stored in the memory 1005 through the processor 1001, and executes the extraction method of the independent face sheet of the aeronautical piece provided by the present application.

[0076] Referring to Figure 2 , based on the hardware environment of the foregoing embodiments, the embodiments of the present application provide an extraction method of an independent face sheet of an aeronautical piece, which comprises the following steps:

[0077] S10: Projecting the target aeronautical piece under multiple perspectives to obtain a plurality of two-dimensional images of the target aeronautical piece.

[0078] In the implementation process, the target aerospace part refers to the aerospace part that needs to extract the independent face sheet, and the aerospace part refers to the parts that make up the space shuttle, which can be obtained by conventional means. When projecting the three-dimensional target aerospace part image from multiple perspectives into a plurality of two-dimensional images, a corresponding system needs to be set, specifically, the program of the present application is based on Windows 7 system 64 bit, the processor is Intel(R) Xeon(R) W-22233.60Ghz, the running memory size is 32GB, the frequency is 3200Mhz, the graphics card used is Nvidia-P2200 (video memory capacity 5GB), the graphics card driver version is 441.66, the CUDA version is 10.2, and the hard disk capacity is 256GB. The software development platform is based on Visual Studio 2019, the programming language is C++, the corresponding three-dimensional modeling software is CATIA software of Dassault Company corresponding to the version V5.21, and the corresponding image processing library is OpenCV4.5.3 based on Release X64 platform, and the SDK version is 10.0.22000.0. The three-dimensional model of the aircraft structure is designed and drawn using the three-dimensional modeling software CATIA of Dassault Company, and the frame, beam, plate and the like are commonly used in the aircraft structure, but the method proposed in the present application is suitable for the extraction of independent face sheets of all aircraft structure three-dimensional images in the two-dimensional projection plane. The designed and drawn three-dimensional image of the aircraft beam structure is as shown in Figure 3 .

[0079] In order to realize accurate extraction of the face sheet, the contour line gray value, line type, background gray value and face sheet gray value in the three-dimensional model need to be designed. The design method is: in CATIA, the line type of the three-dimensional model is set to 1, the gray value of the contour line is black, the gray value of the corresponding RGB color space 3 channels is the same, and the gray value is 0, the line width of the three-dimensional model is set to 1:0.17mm; the boundary gray value of the curved surface is white, and the line width is 2:0.33mm; the gray value of the background is a solid color, that is, the gray value of any background point is the same hR(x,y)=255, hG(x,y)=255, hB(x,y)=255; the light source is set to no light, and the corresponding scattering parameter=1.00, diffuse reflection parameter=0.59 and reflection parameter=0.86; the RGB 3 channel gray value of all face sheets in the foreground is different and a certain value hR(x,y)=200, hG(x,y)=12, hB(x,y)=109 or other combinations. The advantage of this display setting method is that it can improve the contrast between the background and the foreground in the three-dimensional model, the same face sheet has the same gray value information, and the boundaries of different face sheets are easy to distinguish. The obtained aerospace part image based on the designed three-dimensional model display method is as shown in Figure 4 .

[0080] The method further comprises: obtaining a scaling parameter of the target aircraft when the target aircraft is projected; wherein the scaling parameter is a scale parameter of each pixel point in the two-dimensional image in the three-dimensional model; and projecting the target aircraft at multiple perspectives based on the scaling parameter of the target aircraft when the target aircraft is projected to obtain the several two-dimensional images of the target aircraft.

[0081] Specifically, the scaling parameter of the three-dimensional model image in the two-dimensional projection image, i.e., the scale parameter of each pixel point in the two-dimensional image in the three-dimensional model, is designed. The core role of this step is to ensure that there is no similar gray value at the position points with a distance of 1 pixel unit up, down, left and right in the boundary region of each independent facet in the intercepted two-dimensional image, i.e., each facet is separated by a complete closed contour line. The formula for adaptive adjustment (scaling parameter) of the magnification of the two-dimensional projection image corresponding to the three-dimensional model is:

[0082]

[0083] Any two points Point1(x0, y0, z0) and Point2(x1, y1, z1) in the three-dimensional model can be represented by a vector The angle between the vector and the screen is θ, and the length L(Point1, Point2) of the projection vector u of the vector u in the plane of the two-dimensional projection screen can be represented as:

[0084]

[0085] Size screen (i,j) represents a set area of the displayed image in the screen, and the size is calculated as:

[0086] Size screen (i,j)=(y B -y A )*(x B -x A )

[0087] wherein Size map (x,y,z) represents an area of the corresponding region in the three-dimensional model mapped from point A (the upper left corner coordinate of the rectangular display region of the screen) to point B (the lower right corner coordinate of the rectangular display region of the screen); threshold represents a set projection area size threshold, with a unit of mm / pixel; scale(l) represents a scale of the distance between the three-dimensional model and the screen; and Tmap represents a scaling transformation of the three-dimensional model based on the relationship between the calculated value and the threshold, which can improve the extraction effect of the obtained two-dimensional image. The transformation is an important prerequisite step for extracting facets.

[0088] In order to obtain the projection image information under different viewing angles, a rotation tool based on the centroid of the aircraft part is designed in CATIA software based on the secondary development function. The angle of rotation increases by 1 degree per second, and the centroid is the center of rotation. At the same time, manual adjustment of the three-dimensional model can also be used to obtain images under different viewing angles. The color of the model cursor is set to red, corresponding to the RGB value of 255, 0, 0, and the cursor size is 4 pixels. Small size is used to avoid the discontinuity of independent face regions caused by large cursors.

[0089] Then a video stream capture window tool is designed. The capture area is the display area of the model in CITAI, and there are no irrelevant elements (menu bar, toolbar, etc.) in the area. Video stream is used to obtain two-dimensional projection images under different viewing angles. The frame rate of the video stream is 1 frame per second, and high-fidelity recording is used to ensure that the single-frame image in the obtained video stream has low image feature loss. The saved video stream format is.AVI format, and the saved video stream name is FExtract.avi. Based on the OpenCV4.5.3 image processing library, each frame of image in the video stream is extracted. The data of FExtract.avi is read through the VideoCapture class object, and the class object name is img, that is, VideoCapture img(“FExtract.avi”). Then each frame of image is read in turn using the while loop {Mat fram defines the image storage variable; img>>fram saves the data to the storage variable; imwrite(“save image location.png”,fram). The projection image of the target aircraft part under any viewing angle is shown in FIG. 1, Figure 5 Figure 5 Each small image represents a two-dimensional image. In this way, several two-dimensional images obtained by projecting the three-dimensional target aircraft part under multiple viewing angles can be obtained. This way can more comprehensively and completely obtain the information of the target aircraft part, thereby preparing for the subsequent extraction of more accurate independent face pieces.

[0090] S11: Perform global random initial seed point search on the plurality of two-dimensional images to obtain a plurality of row pixel point coordinates in the plurality of two-dimensional images that are the same as the initial seed point pixel.

[0091] ​In the implementation process, if the pixel of the initial seed point is the same as the pixel in the X-axis direction of the two-dimensional image, the initial seed point is moved one pixel unit in the X-axis direction to obtain an X-axis pixel point coordinate; if the pixel of the initial seed point is the same as the pixel in the Y-axis direction of the two-dimensional image, the initial seed point is moved one pixel unit in the Y-axis direction to obtain a Y-axis pixel point coordinate; and based on the X-axis pixel point coordinate and the Y-axis pixel point coordinate, a plurality of row pixel point coordinates in the two-dimensional image which are the same as the pixel of the initial seed point are obtained. Specifically, based on a designed ray vector measurement search rule, it is determined whether the upper and lower and left and right are connected or not. The ray is the direction of the seed point as the starting point to the right. When the pixel gray value corresponding to the end point of the ray is the same as the gray value of the seed point, the corresponding x value is added by 1 and the search is continued. When the pixel gray value corresponding to the end point of the ray is not the same as the gray value of the seed point, the search in the X-axis direction is stopped, and the pixel point coordinates which satisfy the gray equality and through which the seed point passes are recorded. Then, the pixel point corresponding to the end point position is taken as the starting point, the search is continued in the direction of the end point vector and the Y value increases, if the pixel gray value is the same as the gray value of the original seed point, the value is added by 1, the search is continued downward until the pixel gray value corresponding to the position is not the same as the gray value of the original seed point, the search is stopped, and the pixel point coordinates which satisfy the relationship in the Y-axis positive direction are recorded. Then, the search is carried out in the positive and negative directions of the X-axis perpendicular to the Y-axis, and the pixel points which satisfy the gray relationship compared with the original seed point are recorded, and the positions are recorded. In turn, the search is continued until the gray relationship of the searched pixel point and the direction pixel point is completely determined, and the corresponding coordinate value and the coordinate position of the original seed point are recorded. The traversal is carried out from left to right and from top to bottom, and the pixel points are taken as the original seed points for the search. However, it is worth noting that if the corresponding seed point coordinate is included in the seed point traversed before, the seed point is not independent and the search is not carried out, and the next traversal position is directly turned to, until all the pixel points are processed. In this way, the plurality of row pixel point coordinates in the two-dimensional image which are the same as the pixel of the initial seed point can be more accurately obtained.

[0092] S12: The plurality of row pixel point coordinates are sequentially read, so that each row of the pixel point coordinates constructs a background color image.

[0093] In the implementation process, the pixel point coordinates of each row are sequentially read, a pure background color image (image1…imageN) is newly built for each row of pixel point coordinates, the size, gray scale and other information of the image are consistent with those in the above steps, and the pixel point coordinates of all rows in the text document are processed until the processing is completed. Based on the pure background single foreground difference image constructed in the same scale, as shown in Figure 6 .

[0094] S13: Obtain a cluster center coordinate of each independent patch in the two-dimensional image based on the background color image.

[0095] In the specific implementation, the cluster center of each independent patch is calculated according to the mean value of the x value and the y value of each pixel point with a non-zero gray value in the colored pure background image, and the cluster center is the centroid position of the patch and can be used to represent the accurate position of the independent patch in the intercepted image.

[0096] S14: Determine whether the region corresponding to the two-dimensional image is a background region based on the cluster center coordinate, to obtain a plurality of MASK images.

[0097] In the specific implementation, the MASK is consistent with the role of a mask in the manufacturing of a semiconductor chip, that is, the region with a pixel value other than 0 in the MASK is processed, and the gray value of each channel of the region with a pixel value of 1 remains unchanged. A single-channel image image1-3 with the same size as the pure background single-foreground difference image based on the equal-scale construction and with all gray values being 0 is newly created, and the coordinates of the pixel points with a gray value other than 0 in the pure background single-foreground difference image are obtained. The coordinate values are mapped to the newly created image and the corresponding gray values are changed to 1. The coordinate data of each row in the text document is read in sequence, the cluster center of the image region where the corresponding row is located is calculated based on the data, and whether the corresponding region is a background region in the original extracted video stream image is determined based on the cluster center coordinate. If the corresponding region has a gray value other than 0, 0, 0 or 255, 255, 255, a MASK image is newly created, and all the data in the text document are processed, to obtain N MASK images (no MASK image for the background and the contour).

[0098] S15: Extract a first extraction result and a second extraction result of the independent patch in the two-dimensional image based on the plurality of MASK images, wherein the ratio of the foreground region area to the background region area of the second extraction result is greater than the ratio of the foreground region area to the background region area of the first extraction result.

[0099] In the implementation process, the first extraction result of the independent patches in the two-dimensional image based on the plurality of MASK images is that the plurality of MASK images are multiplied by the values of the corresponding pixel positions of the plurality of two-dimensional images respectively to extract the first extraction result. Specifically, the obtained plurality of MASK images are multiplied by the pixel position numbers of the original image respectively, and the values of the positions other than the pixel points corresponding to the areas with a value of 1 in the MASK image are 0. The gray value of the position pixel point with a value of 1 remains unchanged, that is, the extraction of the independent patches in the same scale as the original input image is completed. Then, a plurality of 3-channel images are newly created for saving the independent patch images obtained based on the MASK images, and the imwrite function is used for saving. The extraction result is the first extraction result. The first extraction result of the independent patches in the same scale as the original input image is shown in Figure 7 Figure 7 Each of the images in the above formula represents a first extraction result.

[0100] The second extraction result of the independent patches in the two-dimensional image based on the plurality of MASK images is that the projection curtain of the plurality of MASK images in the X-axis and Y-axis directions is obtained first; then the plurality of extreme value coupling region coordinates of the morphological difference projection of the plurality of MASK images are obtained based on the projection curtain of the plurality of MASK images in the X-axis and Y-axis directions; then the extreme value coupling region coordinates corresponding to the plurality of MASK images in the coordinate axis are processed in sequence to obtain the projection coordinate values of all groups of independent patches; finally, the two-dimensional image is segmented based on the projection coordinate values to obtain the second extraction result. Specifically, the extracted patches are displayed in the image area with the same size as the original image. Generally, each independent patch occupies a small area of the total image, and if it is used as a training sample of a feature program coding patch attribute category judgment model, the weight of the independent patch is small, and the trained model is difficult to achieve the ideal recognition accuracy. Therefore, the second extraction result needs to be further extracted. More specifically, based on the obtained MASK image, as shown in Figure 8 parallel perspective projection rays are constructed in the horizontal direction and the vertical direction respectively, the width of the rays is the same as the number of rows of the original image, and the image plane is taken as the coordinate system. The horizontal direction, that is, the X-axis direction projection ray starts from the left Y-axis and points to the infinity of the X-axis. At the same time, the projection curtain is constructed in the positive direction of the X-axis, and the position of the projection curtain corresponds to the number of columns of the original image. Similarly, as shown in Figure 9 ​As shown, the projection curtain in the Y-axis direction can be constructed, and the position of the curtain corresponds to the number of rows of the original image. The corresponding ray along the Y-axis direction points to the infinity of the Y-axis from the X-axis, and the width of the parallel ray corresponds to the number of columns of the original image. Since the pair of dark areas formed by the projection in the curtain in the X-axis and Y-axis directions has a coupling relationship with the morphology of the image in the MASK image, the pair of dark areas can be called the independent patch morphology difference projection extreme coupling area in the MASK image. The meaning of the extreme value is that the length of the dark area along a certain direction (X-axis or Y-axis) is determined by the starting extreme value (minimum value) and the end extreme value (maximum value) of the independent patch along the direction coordinate in the MASK image. The pair of projection extreme coupling areas obtained along the coordinate axis are shown as follows. Figure 10 The projection coordinate values of all independent patches are obtained by sequentially processing the corresponding extreme coupling areas of the obtained several MASK images in the coordinate axis. Based on the two polar coordinate values in the X-axis and Y-axis directions, the original image region is segmented, and three segmented regions along the X-axis and Y-axis directions are obtained, and nine segmented regions are obtained by combination. The region where the cluster center position is located is the target region, and the coordinate values of the corresponding regions are calculated respectively. Based on the calculated coordinate values, the original image is segmented, and the second extraction result of the independent patch is obtained, and the corresponding values in all MASK images are sequentially processed, that is, the independent patches in the original image and having the minimum background area feature are obtained, as shown in Figure 11 Figure 11 Each image in the above represents a second extraction result.

[0101] S16: Based on the first extraction result and the second extraction result, the independent patches of the target aviation part are obtained.

[0102] In the specific implementation process, the attribute features of the second extraction result are extracted. The second extraction result is the same as the first extraction result, and both are represented in the form of an image. The difference is that the independent patches in the image of the second extraction result have a large proportion of the total image area, that is, the area of the foreground region is much larger than that of the background region. That is, the first extraction result and the second extraction result are both independent patches of the target aviation part, but the combination of the first extraction result and the second extraction result can improve the recognition accuracy. Specifically, in different cases, the independent patches extracted by the corresponding method can be selected to extract the independent patches of the target aviation part more accurately.

[0103] ​In summary, when the independent patches of the target aircraft part need to be extracted, the three-dimensional image of the target aircraft part is projected into a plurality of two-dimensional images from different perspectives, then the initial seed pixel is taken as the basis to search all the two-dimensional images, and a plurality of pixel point coordinates with the same initial seed pixel are recorded, then a background color image is constructed based on each row of pixel point coordinates, the clustering center coordinates of each independent patch in the two-dimensional image are obtained through the background color image, a plurality of MASK images are obtained based on the clustering center coordinates, the first extraction result and the second extraction result of the independent patch are extracted based on the plurality of MASK images, and finally the independent patch of the target aircraft part is extracted. That is, the information of the independent patch in the two-dimensional image is extracted completely because the initial seed point traverses the entire two-dimensional image, which makes a good preparation for accurately extracting the independent patch of the target aircraft part. The first extraction result and the second extraction result of the two different types of independent patches in the two-dimensional image are obtained based on the plurality of MASK images, and the first extraction result and the second extraction result jointly constitute the extraction result, that is, the final independent patch. Since the first extraction result and the second extraction result jointly constitute the independent patch, and the first extraction result and the second extraction result can meet the differentiated needs of different algorithms for the feature form of the target aircraft part, the independent patch of the target aircraft part can be more accurately extracted, so that the target aircraft part can be more accurately programmed, and a target aircraft part that meets the requirements can be processed.

[0104] In some embodiments, before the step of sequentially reading the plurality of rows of pixel point coordinates to make each row of pixel point coordinates construct a background color image, it further includes: first recording the plurality of rows of pixel point coordinates in a text document; then adding the same gray value to each row of pixel points in the text document; and finally initializing the gray value of the boundary area in the two-dimensional image after adding the gray value.

[0105] The step of sequentially reading the plurality of rows of pixel point coordinates to make each row of pixel point coordinates construct a background color image includes sequentially reading the plurality of rows of pixel point coordinates after the initialization processing to make each row of pixel point coordinates construct a background color image.

[0106] In this embodiment, the pixel points meeting the gray relationship are recorded using a text document, and the pixel points corresponding to the same row represent that they have the same gray value relationship. If there are N independent patches in the image, plus the background and the contour line, there are N+2 rows of recording data. The search data saving result of the text document recording the connected features of the two-dimensional plane gray area is shown in Table 1.

[0107] Table 1 search data saving result of the text document recording the connected features of the two-dimensional plane gray area

[0108]

[0109] The pixel region where the seed point search position is added with semantic information, that is, the same gray value of each line in the corresponding text document is added to the pixel point. In order to make each line represent the independent region gray value different, therefore, the designed time margin random number method is adopted, and the gray value is added to the pixel point of the corresponding region in different channels. The design method of time margin random number is as follows:

[0110] srand(unsigned)time(NULL)

[0111] (1) The time margin random number acquisition corresponding to the red channel:

[0112] numbR=(int)(rand()+21)%255

[0113] (2) The time margin random number acquisition corresponding to the green channel:

[0114] numbG=(int)(rand()+7)%255

[0115] (3) The time margin random number acquisition corresponding to the blue channel:

[0116] numbB=(int)(rand()+3)%255

[0117] Wherein, srand represents the pre-set seed; time represents the time-based variable; (int) represents the rounding of data. The independent facet image after adding semantic information based on the time margin random number method is shown in Figure 12 .

[0118] In order to make the gray value of the boundary region after adding semantic information still 255, 255, 255, the semantic information of the image boundary region needs to be initialized, that is, the four corner points of the image are filled with water. The image after the semantic information of the boundary region is initialized is shown in Figure 13 , which is represented as the original image. Based on the above steps, the non-connected region with N+2 gray difference can be obtained, however, these regions are not separated in the image, and need to be further extracted. The pixel information with the same gray value is recorded in the text document, and the coordinate points of each line in the text document are read in turn, and the corresponding pixel point in the image obtained after the semantic information of the boundary region is initialized is located based on the coordinate point.

[0119] A new image image1-1 with the same size as the intercepted image is created, that is, the number of rows and columns of the new image is the same as that of the intercepted image, and the number of channels and the number of bits corresponding to each channel of each pixel point in the image are the same, that is, 3-channel RGB color space 8bit, and the gray scale of the pure background is 0, 0, 0. Then, an image image1-2 is constructed based on the gray scale difference, that is, the gray scale of the background and the foreground is different. First, the coordinates of the regions with different gray scales in the image initialized by the boundary region semantic information are obtained, and the value of the coordinates is the value corresponding to each row of the text document. The coordinates corresponding to each row value correspond to the coordinates in the new pure background image, and the gray scale of the pixel points in the corresponding position is assigned to 3 channels, and the corresponding value is consistent with the gray scale value in the image initialized by the boundary region semantic information. In this way, the pixel point coordinates of multiple rows can be read more effectively and accurately.

[0120] In order to obtain more information and more accurate multi-row pixel point coordinates, in some embodiments, before the step of searching for a global random initial seed point in a plurality of two-dimensional images to obtain multi-row pixel point coordinates in the plurality of two-dimensional images that are the same as the initial seed point pixel, the method further comprises: first removing noise from the plurality of two-dimensional images based on the constructed noise removal model; then determining the position of foreign elements in the two-dimensional images after removing the noise; wherein the foreign elements are objects with different colors from the two-dimensional images; and then removing the foreign elements in the two-dimensional images.

[0121] The step of searching for a global random initial seed point in a plurality of two-dimensional images to obtain multi-row pixel point coordinates in the plurality of two-dimensional images that are the same as the initial seed point pixel comprises:

[0122] The step of searching for a global random initial seed point in a plurality of two-dimensional images to obtain multi-row pixel point coordinates in the plurality of two-dimensional images that are the same as the initial seed point pixel comprises:

[0123] In this embodiment, since the extracted two-dimensional images are obtained based on a video stream, there is a lot of noise in the images due to the image compression and decompression process. In order to effectively remove the noise, a region threshold difference noise removal method is designed in this embodiment, and the noise removal model corresponding to different regions can be represented as:

[0124] (1) Background noise removal:

[0125]

[0126] where p(i,j) represents any pixel point in the two-dimensional projection image; h R (i,j), h G(i,j), h B (i,j) respectively represent the gray value of the pixel corresponding to the red, green and blue channel at the (i,j) position in the image; thre R-B , thre G-B , thre B-B respectively represent the red, green and blue channel threshold values set for background noise removal.

[0127] (2) Contour noise removal:

[0128]

[0129] thre R-C , thre G-C , thre B-C respectively represent the red, green and blue channel threshold values set for contour noise removal.

[0130] (3) Patch noise removal:

[0131]

[0132] thre R-S , thre G-S , thre B-S respectively represent the red, green and blue channel threshold values set for patch noise removal.

[0133] At the same time, in order to further obtain a more accurate two-dimensional image, there is a step of adjusting the color of the cursor in the above-mentioned two-dimensional image to red. The red cursor (foreign element) is only used to facilitate the acquisition of images under different angles, so it needs to be removed. The removal method is to directly convert the gray value of the pixel with red color, i.e. RGB gray value of 255, 0, 0, in the global:

[0134] (1) When the gray value of the pixel at a distance of 1 pixel around the 4 pixels is 255, 255, 255, the cursor has a higher probability of being in the background, so its gray value needs to be converted to 255, 255, 255. Since the area occupied by the cursor after processing is integrated with the background, it will not affect the extraction of the patch;

[0135] (2) When the gray value of the pixel at a distance of 1 pixel around the 4 pixels is 0, 0, 0, the cursor has a higher probability of being in the contour line area, so its gray value needs to be converted to 0, 0, 0. Since the cursor only occupies 4 pixels, it will not interfere with the contour and boundary with a line width of 1:0.17mm relationship;

[0136] (3) When the gray value of the pixel point around the 4 pixel points is hR(x, y) = 200, hG(x, y) = 12, and hB(x, y) = 109, the cursor is more likely to be in the independent sheet region, so the gray value is hR(x, y) = 200, hG(x, y) = 12, and hB(x, y) = 109. Since the region occupied by the cursor after processing has a two-dimensional plane position containing relationship with the independent sheet, the independent sheet will not be affected.

[0137] Based on the image processed according to the above steps, only 3 kinds of gray difference pixel points can be obtained, which are 0, 0, 0, 255, 255, 255, and 200, 12, 109. Then, the same gray value is given to the pixel points in the 4-connected region by using a global random initial seed point search method. Global means that all pixel points in the image need to judge the gray relationship between adjacent pixels, and the initial seed point means that the pixel point at each position is treated as a seed point. The ray method corresponding to the global random initial seed point search method is shown in FIG. 8. In this way, a two-dimensional image with higher accuracy can be obtained, so that the independent sheet of the target aircraft part can be more accurately extracted. Figure 14

[0138] In summary, the present application can extract the independent sheet of the three-dimensional model of the aircraft part in the two-dimensional plane projection image under any viewing angle. The extraction result includes the first extraction result and the second extraction result, which enriches the need for differentiation of sample features by different algorithms, provides samples for analyzing the attributes of the sheet based on the extraction result, and supports the development of automatic programming work of the target aircraft part. It lays a foundation for realizing the judgment of the sheet attribute based on the image strategy.

[0139] In another embodiment, as shown in FIG. 9, based on the same inventive idea as the foregoing embodiment, the embodiment of the present application also provides an extraction device for an independent sheet of an aircraft part, which comprises: Figure 15 An acquisition module is configured to project a target aircraft part under multiple viewing angles to obtain a plurality of two-dimensional images of the target aircraft part.

[0140] A first obtaining module is configured to perform global random initial seed point search on the plurality of two-dimensional images to obtain a plurality of row pixel point coordinates in the plurality of two-dimensional images which are the same as the initial seed point pixel.

[0141] A reading module is configured to read the plurality of row pixel point coordinates in sequence to make each row of pixel point coordinates construct a background color image.

[0142]

[0143] ​​a second obtaining module, configured to obtain a cluster center coordinate of each independent facet in the two-dimensional image based on the background color image;

[0144] a judging module, configured to judge whether the region corresponding to the two-dimensional image is a background region based on the cluster center coordinate, to obtain a plurality of MASK images;

[0145] an extracting module, configured to extract a first extraction result and a second extraction result of the independent facet in the two-dimensional image based on the plurality of MASK images, wherein a ratio of a foreground region area to a background region area of the second extraction result is greater than a ratio of a foreground region area to a background region area of the first extraction result;

[0146] a third obtaining module, configured to obtain the independent facet of the target aerostructure based on the first extraction result and the second extraction result.

[0147] It should be noted that the modules in the extraction device of the independent facet of the aerostructure in the embodiment correspond to the steps in the extraction method of the independent facet of the aerostructure in the foregoing embodiment one by one, and therefore the specific embodiments and the technical effects achieved in the embodiment can refer to the embodiments of the extraction method of the independent facet of the aerostructure, which will not be described herein again.

[0148] In addition, in an embodiment, the present application further provides a computer device, which comprises a processor, a memory and a computer program stored in the memory, and the computer program realizes the method in the foregoing embodiment when executed by the processor.

[0149] In addition, in an embodiment, the present application further provides a computer storage medium, which stores a computer program, and the computer program realizes the method in the foregoing embodiment when executed by a processor.

[0150] In some embodiments, the computer readable storage medium can be a memory such as FRAM, ROM, PROM, EPROM, EEPROM, flash memory, magnetic surface memory, optical disc or CD-ROM, etc., or various devices comprising one or any combination of the above memories. The computer can be various computing devices including smart terminals and servers.

[0151] In some embodiments, the executable instructions can be in the form of programs, software, software modules, scripts or codes, written in any form of programming language (including compiled or interpreted languages, or declarative or procedural languages), and can be deployed in any form, including being deployed as independent programs or as modules, components, subroutines or other units suitable for use in a computing environment.

[0152] As an example, executable instructions can correspond to a file in a file system, can be stored in a part of a file that is used by the operating system to store application program data, can be stored as an "applet" in a general purpose software application, can be stored as a "plugin" in a web browser, or can be stored as an "app" in a mobile device, to name but a few.

[0153] As an example, executable instructions can be deployed to be executed on one computer, or on multiple computers of a system of computers in one location, or on multiple computers of a system of computers distributed among multiple locations and interconnected by a communication network.

[0154] It has to be noted that, as used herein, the terms "includes" and / or "contains", or any other tautological variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements is not limited to those elements but can include other elements not expressly listed or inherent to such process, method, article, or apparatus. An element proceeded by "comprises... a" does not, without further constraints, exclude the presence of additional identical elements in the process, method, article, or apparatus that comprises the element.

[0155] The above-mentioned sequence numbers of the embodiments of the present application are only for description, and do not represent advantages or disadvantages of the embodiments.

[0156] Those skilled in the art can clearly understand the above-mentioned embodiment method from the description of the embodiments, which can be realized by software and necessary general hardware platform, of course, can also be realized by hardware, but in many cases, the former is a better embodiment. Based on such understanding, the technical solutions of the present application can be embodied in the form of a software product, which is stored in a storage medium (such as a read-only memory / random access memory, a magnetic disk, an optical disk), and includes a plurality of instructions for causing a multimedia terminal device (which can be a mobile phone, a computer, a television receiver, or a network device) to execute the method described in each embodiment of the present application.

[0157] The above is only the preferred embodiment of the present application, and does not limit the patent scope of the present application, and any equivalent structure or equivalent process transformation using the content of the specification and drawings, or direct or indirect application in other related technical fields, are also included in the patent protection scope of the present application.

Claims

1. A method for extracting independent facets of aircraft components, characterized in that, The method includes: Projecting the target aircraft component from multiple perspectives to obtain several two-dimensional images of the target aircraft component; A global random initial seed point search is performed on several two-dimensional images to obtain the coordinates of multiple rows of pixels in the several two-dimensional images that are the same as the initial seed point pixels. Read multiple rows of pixel coordinates sequentially, so that each row of pixel coordinates constructs a background color image; Based on the background color image, obtain the cluster center coordinates of each independent facet in the two-dimensional image; Based on the cluster center coordinates, determine whether the region corresponding to the two-dimensional image is a background region to obtain several mask images; Based on the plurality of mask images, a first extraction result and a second extraction result are extracted from the independent patches in the two-dimensional image; wherein, the ratio of the foreground area to the background area in the second extraction result is greater than the ratio of the foreground area to the background area in the first extraction result; the extraction of the first extraction result and the second extraction result from the independent patches in the two-dimensional image based on the plurality of mask images includes: obtaining the projection screen of the plurality of mask images in the X-axis and Y-axis directions; obtaining the coordinates of a plurality of extreme value coupling regions of the morphological difference projection of the plurality of mask images based on the projection screen of the plurality of mask images in the X-axis and Y-axis directions; sequentially processing the coordinates of the extreme value coupling regions corresponding to the plurality of mask images in the coordinate axes to obtain the projection coordinate values ​​of all independent patches as a group; segmenting the two-dimensional image based on the projection coordinate values ​​to obtain the second extraction result; the extraction of the first extraction result and the second extraction result from the independent patches in the two-dimensional image based on the plurality of mask images includes: multiplying the values ​​of the corresponding pixel positions of the plurality of mask images by the values ​​of the plurality of two-dimensional images to extract the first extraction result; Based on the first extraction result and the second extraction result, the independent surface patches of the target aerospace component are obtained.

2. The method for extracting independent facets of aerospace components as described in claim 1, characterized in that, Before the step of sequentially reading multiple rows of pixel coordinates to construct a background color image from each row of pixel coordinates, the method further includes: Record the coordinates of the multiple rows of pixels sequentially in a text document; Add the same grayscale value to each pixel in each line of the text document; The gray values ​​of the boundary regions in the two-dimensional image after adding gray values ​​are initialized. The step of sequentially reading multiple rows of pixel coordinates to construct a background color image for each row of pixel coordinates includes: The pixel coordinates of the multiple rows after initialization are read sequentially so that each row of pixel coordinates can be used to construct a background color image.

3. The method for extracting independent facets of aerospace components as described in claim 1, characterized in that, The step of performing a global random initial seed point search on several two-dimensional images to obtain the coordinates of multiple rows of pixels in the several two-dimensional images that are the same as the initial seed point pixels includes: If the pixel of the initial seed point is the same as the pixel in the X-axis direction of the two-dimensional image, then the initial seed point is moved one pixel unit towards the X-axis to obtain the X-axis pixel coordinates; If the pixel of the initial seed point is the same as the pixel in the Y-axis direction of the two-dimensional image, then the initial seed point is moved one pixel unit towards the Y-axis to obtain the Y-axis pixel coordinates; Based on the X-axis pixel coordinates and the Y-axis pixel coordinates, obtain the coordinates of multiple rows of pixels in the several two-dimensional images that are the same as the initial seed point pixels.

4. The method for extracting independent facets of aerospace components as described in claim 1, characterized in that, Before the step of performing a global random initial seed point search on several two-dimensional images to obtain the coordinates of multiple rows of pixels in the several two-dimensional images that are the same as the initial seed point pixels, the method further includes: Based on the constructed noise removal model, noise is removed from the several two-dimensional images; Determine the position of foreign object elements in the two-dimensional image after noise removal; wherein, the foreign object element is an object with a different color from the two-dimensional image. Remove foreign elements from the two-dimensional image; The step of performing a global random initial seed point search on several two-dimensional images to obtain the coordinates of multiple rows of pixels in the several two-dimensional images that are the same as the initial seed point pixels includes: A global random initial seed point search is performed on several two-dimensional images after removing foreign elements to obtain the coordinates of multiple rows of pixels in the several two-dimensional images that are the same as the initial seed point pixels.

5. The method for extracting independent facets of aerospace components as described in claim 1, characterized in that, Prior to the step of projecting the target aircraft component from multiple viewpoints to obtain several two-dimensional images of the target aircraft component, the procedure includes: Obtain the scaling parameters when projecting the target aerospace component; wherein, the scaling parameters are the scale parameters in the three-dimensional model represented by each pixel in the two-dimensional image; Based on the scaling parameters when projecting the target aircraft component, the target aircraft component is projected from multiple viewpoints to obtain several two-dimensional images of the target aircraft component.

6. A device for extracting independent faceplates of aircraft parts, characterized in that, The apparatus for performing the method according to any one of claims 1-5 comprises: The acquisition module is used to project the target aircraft part from multiple perspectives to obtain several two-dimensional images of the target aircraft part. The first obtaining module is used to perform a global random initial seed point search on several two-dimensional images to obtain the coordinates of multiple rows of pixels in the several two-dimensional images that are the same as the initial seed point pixels. The reading module is used to sequentially read multiple rows of pixel coordinates so that each row of pixel coordinates constructs a background color image; The second obtaining module is used to obtain the cluster center coordinates of each independent facet in the two-dimensional image based on the background color image; The judgment module is used to determine whether the region corresponding to the two-dimensional image is a background region based on the coordinates of the cluster center, so as to obtain several mask images; An extraction module is used to extract a first extraction result and a second extraction result of independent patches in the two-dimensional image based on the plurality of mask images; wherein the ratio of the foreground area to the background area of ​​the second extraction result is greater than the ratio of the foreground area to the background area of ​​the first extraction result. The third obtaining module is used to obtain independent facets of the target aerospace component based on the first extraction result and the second extraction result.

7. A computer device, characterized in that, The computer device includes a memory and a processor, wherein the memory stores a computer program and the processor executes the computer program to implement the method as described in any one of claims 1-5.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, and the processor executes the computer program to implement the method as described in any one of claims 1-5.

Images