A method for monitoring airport clearance based on intelligent photoelectric technology

Abstract

The present invention discloses an airport clearance monitoring method based on intelligent photoelectric technology, which comprises the following steps: collecting image data in an airport clearance area; preprocessing the collected image data; comparing and analyzing image data of adjacent time series by a deep learning method to determine whether there are changes in buildings in the airport clearance area; if there are changes in buildings in the airport clearance area, determining the location of newly built or under-construction buildings; after determining the location of newly built or under-construction buildings, collecting information on the buildings and calculating the clearance restriction surface and restricted height of the buildings; verifying the information on newly built or under-construction buildings and checking to determine whether their heights meet the restriction requirements of airport clearance influencing factors on the location of the buildings.

Description

Technical Field

[0001] The invention belongs to the technical field of construction engineering and relates to an airport clearance monitoring method based on intelligent photoelectric technology. Background Art

[0002] Airports are an important part of the aviation industry, and are places where aircraft take off, land, park, and maintain. The evolution of airports reflects the development of civil aviation. In recent years, with the development of aviation, the safety of aircraft takeoff and landing has become more and more important to people, and the airport clearance zone directly affects the safety of aircraft takeoff and landing. The airport clearance zone is a spatial area that limits the height of ground obstacles, which is used to ensure that there are no ground obstacles that hinder navigation and flight when the aircraft is flying at low altitudes for takeoff and landing. The clearance conditions of the airport will directly affect the safety of aircraft operations and the flight level of the airport.

[0003] At present, the most advanced technology for air clearance inspection in my country is to use satellite remote sensing images and drone image data for comparative analysis. However, due to the long period of satellite remote sensing data, satellite remote sensing imaging technology cannot detect buildings under construction in a timely manner. At the same time, the image data will be of poor quality due to cloud cover, and thus cannot achieve accurate monitoring of airport air clearance. In addition, although the method of using drone image data to monitor the air clearance around the airport can solve the shortcomings of cloud cover of satellite remote sensing image data, it is costly. In addition, in order to avoid the impact of drone aerial photography missions on the normal operation of the airport, it is necessary to complete communication and coordination with multiple parties such as the airport, air traffic control and the military in advance. The invention and preparation of the workflow is complex and has a long cycle.

[0004] Therefore, it is urgent to design an airport clearance monitoring method based on intelligent photoelectric technology to solve the technical problems existing in the existing technology. Summary of the invention

[0005] The purpose of the present invention is to solve some of the technical problems existing in the prior art to at least a certain extent, and to provide an airport clearance monitoring method based on intelligent photoelectric technology. The method has a reasonable structure and is based on a deep combination of artificial intelligence technology and photoelectric technology to realize airport clearance monitoring. It has the advantages of clear vision, precise positioning, low cost and high efficiency.

[0006] In order to solve the above technical problems, the present invention provides an airport clearance monitoring method based on intelligent photoelectric technology, which includes:

[0007] S1, collects image data within the airport clearance area;

[0008] S2, preprocessing the acquired image data by frame-by-frame extraction and image stitching;

[0009] S3 uses deep learning to compare and analyze image data of adjacent time series to determine whether there are any changes in the buildings within the airport clearance area;

[0010] S4, if there are changes in the buildings within the airport clear area, determine the location of the new buildings or buildings under construction;

[0011] S5, after determining the location of the newly built or under-construction building, collect the building information and calculate the clearance restriction surface and the restricted height of the building;

[0012] S6, verify the information of newly built or under-construction buildings and check whether their heights meet the restrictions on the location of the buildings caused by the factors affecting the airport clearance.

[0013] As a preferred embodiment, in step S1, the image data acquisition position is set according to the required clearance range of the airport, the layout and height of the urban buildings and the detection range of the camera.

[0014] As a preferred embodiment, in step S3, the step of comparing and analyzing image data using a deep learning method specifically includes:

[0015] Label the image data of the building group and establish a sample library that meets the requirements of image data diversity;

[0016] By using the convolutional neural network structure of multiple instance learning and spatial attention model, a neural network that reflects the multiple sensitive elements of the airport, the significant scale characteristics, and the local semantic expression of the buildings is constructed, and a deep learning model suitable for extracting building information based on optoelectronic technology image data is obtained;

[0017] Based on the deep learning model, the temporal features in the effective image data are extracted, and the differences in the spatial structure type information of the building are compared from images of different time series.

[0018] As a preferred embodiment, establishing a sample library that meets the requirements of image data diversity specifically involves performing style transformation according to different regions, seasons, and imaging conditions to obtain images under various conditions, thereby forming a sample library that meets the diversity of actual image data.

[0019] As a preferred embodiment, in step S5, based on the factors affecting the airport clearance, the height restriction requirements of the location of the building are checked, the proposed height information of the building at the location is verified, and it is calculated whether it exceeds the airport clearance restriction requirements.

[0020] As a preferred embodiment, the airport clearance influencing factors include airport clearance restriction surfaces, flight procedures, aircraft performance, visual aids and navigation facilities.

[0021] As a preferred embodiment, in step S5, the limit height h(x, y) is:

[0022] h(x,y)=Min{h 附件十四面 (x,y),h 飞行程序 (x,y),h 飞机性能 (x,y),h 导航设施 (x,y),h 目视灯光 (x,y),…}

[0023] h 附件十四面 (x,y)=Min{h 进近面 (x,y),h 内水平面 (x,y),h 锥形面 (x,y),…}

[0024] h 飞行程序 (x,y)=Min{h 仪表进近程序 (x,y),h 进场程序 (x,y),h 离场程序 (x,y),…}

[0025] h 导航设施 (x,y)=Min{h 航向台 (x,y),h 下滑台 (x,y),h VOR台 (x,y),…}

[0026] h 目视灯光 (x,y)=Min{h 简易灯光 (x,y),h 精密进近灯光 (x,y),h PAPI灯 (x,y),…}

[0027] Among them, (x, y) is the coordinates of the building,

[0028] h 附件十四面 (x,y),h 飞行程序 (x,y),h 飞机性能 (x,y),h 导航设施 (x,y),h 目视灯光 (x, y) are the Annex XIV restricted altitude, flight procedure restricted altitude, aircraft performance restricted altitude, visual aids restricted altitude, and navigation facility restricted altitude at the coordinate position (x, y) respectively.

[0029] As a preferred embodiment, in step S5, the clearance restriction surface is obtained through the total restriction space, wherein the total restriction space L(x, y) is:

[0030] L(x,y)=L(S 附件十四面 (x,y))∪L(S 飞行程序 (x,y))∪L(S 飞机性能 (x,y))∪L(S导航设施 (x,y))∪L(S 目视灯光 (x,y))∪…

[0031] L(S 附件十四面 (x,y))=L(S 进近面 (x,y))∪L(S 内水平面 (x,y))∪L(S 锥形面 (x,y))∪…

[0032] L(S 飞行程序 (x,y))=L(L(S 仪表进近程序 (x,y))∪L(S 进场程序 (x,y))∪L(S 离场程序 (x,y))∪…

[0033] L(S 导航设施 (x,y))=L(S 航向台 (x,y))∪L(S 下滑台 (x,y))∪L(S VOR台 (x,y))∪…

[0034] L(S 目视灯光 (x,y))=L(S 简易灯光 (x,y))∪L(S 精密进近灯光 (x,y))∪L(S PAPI灯 (x,y))∪…

[0035] Among them, (x, y) is the coordinates of the building,

[0036] L 附件十四面 (x,y),L 飞行程序 (x,y),L 飞机性能 (x,y),L 导航设施 (x,y),L 目视灯光 (x, y) are the annex fourteen restricted space, flight procedure restricted space, aircraft performance restricted space, visual aid facility restricted space, and navigation facility restricted space at the coordinate position (x, y).

[0037] Beneficial effects of the present invention:

[0038] The present invention provides an airport clearance monitoring method based on intelligent photoelectric technology. It is based on the deep combination of artificial intelligence technology and photoelectric technology, and integrates interdisciplinary cutting-edge technologies such as analog and digital electronics, big data, image processing, sensors, and optical imaging. It determines whether there are any building changes in the airport clearance area, calculates and determines the airport clearance restriction surface and restricted height of the building, and determines whether its height meets the restriction requirements of the airport clearance influencing factors on the location of the building, thereby realizing airport clearance monitoring, and has the advantages of clear vision, accurate positioning, low cost and high efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] The above advantages of the present invention will become clearer and easier to understand through the detailed description made in conjunction with the following drawings, which are only exemplary and do not limit the present invention, wherein:

[0040] Attached Figure 1 It is a flow chart of an airport clearance monitoring method based on intelligent photoelectric technology according to the present invention;

[0041] Attached Figure 2 This is a step of comparing and analyzing image data using a deep learning approach in the present invention;

[0042] Attached Figure 3 It is a deep learning model diagram for extracting building information based on optoelectronic technology image data according to the present invention;

[0043] Attached Figure 4 It is a schematic diagram of the monitoring process of determining building elevation change in the present invention;

[0044] Attached Figure 5 It is a partial schematic diagram of the fourteenth surface of the appendix in the present invention;

[0045] Attached Figure 6 yes Figure 4 The corresponding AA cross-sectional view;

[0046] Attached Figure 7 yes Figure 4 Corresponding to the cross-sectional view of BB. DETAILED DESCRIPTION

[0047] Attached Figure 1 To Attachment Figure 7 Detailed description of the invention The present invention is related to an airport clearance monitoring method based on intelligent photoelectric technology. The present invention is described in detail below in conjunction with specific embodiments and drawings.

[0048] The embodiments described herein are specific embodiments of the present invention, which are used to illustrate the concept of the present invention, are illustrative and exemplary, and should not be interpreted as limiting the embodiments of the present invention and the scope of the present invention. In addition to the embodiments described herein, those skilled in the art can also adopt other obvious technical solutions based on the contents disclosed in the claims and the specification of the present invention, including technical solutions that adopt any obvious replacement and modification of the embodiments described herein.

[0049] The drawings of this specification are schematic diagrams, which assist in explaining the concept of the present invention and schematically show the shapes of various parts and their mutual relationships. Please note that in order to clearly show the structures of various components of the embodiments of the present invention, the drawings are not drawn according to the same scale. The same reference numerals are used to represent the same parts.

[0050] The flowchart of the airport clearance monitoring method based on intelligent photoelectric technology described in the present invention is as shown in the attached figure. Figure 1 The airport clearance monitoring method based on intelligent photoelectric technology includes the following steps:

[0051] S1, within a period of time, using a high-definition camera installed on the highest building in the airport clear area to batch collect video data in the airport clear area, and extract image data within the period of time from the video data;

[0052] In the above step S1 of this embodiment, the image data collection position is set according to the clearance requirement range of the airport, the layout and height of the urban buildings, and the detection range of the high-definition camera. Specifically, according to the clearance requirement range of the airport, the layout and height of the urban buildings, and the detection range of the camera, a reasonable image data collection position is set, and the camera is usually set at the highest point of the urban building, so that the camera data will not be blocked by other buildings during collection, and high-quality image data can be collected.

[0053] For areas around airports, cameras can be set up at the highest point of the airport, usually the top of the control tower or terminal building, to conduct targeted video surveillance and collection of surrounding village buildings, high-voltage lines, signal towers, cranes and other facilities. For areas with dense urban construction, high-definition cameras can be set up on the top of existing tall buildings to promptly detect new buildings and cranes in the city.

[0054] In step S1 of the present invention, the video data collected by optoelectronic technology has richer ground object detail information, improved time, space, and spectral resolution, and an increase in data sources, which makes it possible to dynamically monitor and promptly discover abnormally sensitive ground objects within the airport clearance range.

[0055] S2, preprocessing the image data extracted in the above step S1 by frame-by-frame extraction and image stitching;

[0056] Since the image data collected by high-definition cameras within the airport's airspace is video data, it is necessary to extract the video data frame by frame and pre-process the collected image data through image stitching.

[0057] The specific implementation process of preprocessing the collected image data by image stitching is as follows:

[0058] Step S201: performing geometric distortion correction and noise point suppression on the above image data;

[0059] The geometric distortion correction is specifically to use the distortion matrix for correction. The specific correction method is as follows:

[0060] (I) Assume that the coordinates of the four vertices before correction are (x 1 ,y 1 ),(x 2 ,y 2 ),(x 3 ,y 3 ),(x 4 ,y 4 ), the coordinates of the four vertices of the corrected image are (x 1 ',y 1 '),(x 2 ',y 2 '),(x 3 ',y 3 '),(x 4 ',y 4 '), and the four vertices before correction are called input points, and the four vertices after correction are called reference points;

[0061] (ii) Let the distortion correction matrix H be And introduce the vector h, h = (h 11 ,h 12 ,h 13 ,h 21 ,h 22 ,h 23 ,h 31 ,h 32 ,h 33 ) T , enter the coordinates of the input point and the reference point, and use the formula Calculate h and obtain the distortion correction matrix H;

[0062] (3) The original image can be corrected using the distortion correction matrix.

[0063] The reason why the above-mentioned image is geometrically distorted is that the size and shape of the image undergo certain changes during the acquisition of the camera image data. The corrected image becomes rectangular and the original aspect ratio does not change.

[0064] S202: extracting feature points to perform image matching on two sets of image data in adjacent time series;

[0065] The specific implementation method of the above image matching is as follows:

[0066] (I) First, create an image Gaussian pyramid and use the Gaussian kernel function to filter so that the original image retains the most detail features. After Gaussian filtering, the detail features are gradually reduced to simulate the feature representation in large-scale conditions;

[0067] Among them: the number of layers of the image pyramid is O = log 2 (min(M,N))-3, M is the row height of the original image, N is the column width of the original image; O is the number of groups of the image Gaussian pyramid. The number of layers in each group of the image Gaussian pyramid is: S=n+3; where n is the number of images to be extracted. Each image in the pyramid is represented by L(x,y,σ), then Where I(x,y) represents the image, G(x,y,σ) is the Gaussian function, represents convolution, and σ is the image scale parameter.

[0068] (ii) confirming the extreme points in the image in the Gaussian pyramid, and establishing a feature description vector from the extreme points; the specific steps are as follows:

[0069] (1) Accurately locate the extreme point: First, use the formula Threshold the image;

[0070] Wherein T=0.04, which can be set manually; n is the number of images to be extracted; abs(val) is the pixel threshold of the image, and the pixel threshold is set to remove unstable pixels in the image.

[0071] (2) Determine whether each pixel point is an extreme point in the x, y and σ directions in turn, and confirm the final extreme point;

[0072] When determining whether a pixel is an extreme point, it is necessary to compare it with other pixels around the pixel to confirm several final extreme points, which are the feature points of the image.

[0073] (3) Using the feature description vector established by the above-mentioned final extreme point to perform similarity matching, feature points extracted from two adjacent time series images are compared pairwise to find feature points of several pairs of images that match each other.

[0074] S203: Image stitching: After completing the matching of the two images in the adjacent time sequence, stitching the images at different positions at the same time.

[0075] The above image stitching is to directly add each RGB value of the overlapping area of ​​the two images and calculate the average value as the RGB value of the point after the merged image. The specific stitching formula is:

[0076] Where (x, y) is the pixel coordinate after stitching, F(x, y) is the stitched image, H(x, y) and L(x, y) are the two images before stitching, and H∩L represents the overlapping area of ​​H(x, y) and L(x, y).

[0077] The image is digitized through RGB values ​​and image stitching is performed through digital operations, with the aim of stitching images from different orientations.

[0078] The purpose of stitching images at different locations at the same time is to stitch images at different locations within the airport clearance range collected at the same time into one image before proceeding to the subsequent step S3, so as to more quickly detect changes in buildings in step S3.

[0079] In addition, in the entire step S2, the preprocessing of the above-mentioned image data by frame-by-frame extraction and image stitching reduces the noise of the image data, thereby reducing the problem of model overfitting in the convolutional neural operation of the image data in the subsequent step S3.

[0080] S3 uses deep learning to compare and analyze image data of adjacent time series to determine whether there are any changes in the buildings within the airport clearance area;

[0081] See attached Figure 3 As shown, attached Figure 3 It is a deep learning model diagram for extracting building information based on optoelectronic technology image data as described in the present invention. This model targets the feature points and extraction requirements of buildings in image data. On the basis of adaptive analysis of the main deep learning neural network models, a convolutional neural network structure based on multi-instance learning and spatial attention models is obtained, and a network is constructed that reflects the multi-scale significance features of airport sensitive elements and highlights the local semantic expression of buildings, thereby realizing a deep learning model suitable for extracting buildings from image data.

[0082] Wherein, step S3 specifically includes:

[0083] Step S301: inputting the pre-processed adjacent time-series image data containing feature point information into the convolutional neural network respectively;

[0084] Step S302: Then, through machine deep learning, the image data is subjected to a dot multiplication operation using a convolutional neural network module, a multi-scale dilated convolution feature fusion module, and an attention model;

[0085] Among them, the attention model is used to simulate the attention model of the human brain. The role of the attention model in the embodiment of the present invention is to select the feature point information that is more critical to the current task from the multiple feature points in the preprocessed image data obtained in step S2, that is, the building feature point information required in the embodiment of the present invention.

[0086] The purpose of performing the dot multiplication operation on the above-mentioned image data is to input the image data obtained by the multi-scale dilated convolution feature fusion module into the attention model, and obtain the building feature point information required in the embodiment of the present invention after performing the dot multiplication operation in the attention model.

[0087] As the present invention Figure 3 As shown, multi-scale dilated convolution is used in the embodiment of the present invention to expand the receptive field of the convolutional neural network and control the size of the output feature map without increasing the model parameters. The convolution features from the shallow, middle and deep layers are respectively fed into the dilated convolution kernels of different window sizes to obtain feature maps of the same size and stronger separability, and obtain richer spatial structural feature information such as the edge and shape of the building.

[0088] Step S303: performing multi-instance learning pooling on the image data after the dot multiplication operation is completed, and highlighting the building feature point information obtained in the attention model;

[0089] Step S304: extracting building feature points from different time series image data after multi-example learning pooling;

[0090] Step S305: Compare the changes of the building feature points of the above-mentioned different time-series image data.

[0091] Among them, the convolutional neural network is a deep learning model designed for feature extraction of two-dimensional image data, see the attached Figure 3 As shown in the figure, complex functions are fitted through multiple convolutional layers. The features of each convolutional layer are obtained by sharing weights from the local features of the previous layer, which effectively reduces the number of network parameters and alleviates the overfitting problem of the model. It is a deep learning model widely used in image processing fields such as target detection and recognition.

[0092] The AlexNet model of convolutional neural network is adopted in the present invention. The AlexNet model has 8 layers of structure, wherein the first 5 layers are convolutional layers and the last 3 layers are fully connected layers. After the convolution operation of the first 2 convolutional layers and the 5th convolutional layer, there will be a pooling layer, so that the pixel value of the feature map after the feature extraction is greatly reduced, which facilitates the operation and makes the feature more obvious, which is a feature that other convolutional layers do not have. The 3rd convolutional layer merges the channels, and merges the data of the previous two channels again, which is a kind of tandem operation.

[0093] The embodiment of the present invention simplifies the classic three-dimensional Alexnet to effectively extract temporal features, and adds a channel attention model. The channel attention model uses multiple 1×1 convolutional layers to refine the convolutional features generated from the three-dimensional convolutional neural network backbone, optimizes the extracted building features, emphasizes areas where the buildings change significantly, and finally generates fused optimized features.

[0094] As an embodiment of the present invention, before step S3, a step of comparing and analyzing image data by deep learning is performed, as shown in the attached Figure 3 and attached Figure 4 As shown, it must also include:

[0095] The image data of the building complex is labeled to establish a sample library that meets the requirements of image data diversity; that is, style transformation is performed according to different regions, seasons, and imaging conditions to obtain image data under various conditions and form a sample library that meets the diversity of actual image data.

[0096] As an embodiment of the present invention, before step S301, a step of comparing and analyzing image data by deep learning is performed, as shown in the attached Figure 3 and attached Figure 4 As shown, it must also include:

[0097] By using a convolutional neural network structure with multi-instance learning pooling and an attention model, a neural network is constructed that reflects the multiple sensitive elements of airports, has significant scale features, and highlights the local semantic expression of buildings. This results in a deep learning model suitable for extracting building information based on optoelectronic technology image data.

[0098] Based on the above deep learning model, the temporal features in the effective image data can be extracted, and the differences in the spatial structure type information of the building can be compared from images of different temporal sequences.

[0099] Attached Figure 4 is a schematic diagram of the monitoring process for determining building elevation changes in the present invention; Figure 4 As shown,

[0100] In step S3, the specific steps of determining whether the building elevation within the airport clearance area has changed are as follows:

[0101] (1) Input a pair of image data of different time sequences after preprocessing into the main body of the three-dimensional convolutional network;

[0102] The three-dimensional convolution is defined as the depth of the filter being less than the depth of the input layer, so the three-dimensional filter needs to slide in three dimensions (the length, width, and height of the input layer). A convolution operation is performed at each position where the filter slides to obtain a value. When the filter slides through the entire three-dimensional space, the output result is also three-dimensional.

[0103] (2) Using the Rectified Linear Unit of the main body of the three-dimensional convolutional network to increase the nonlinear relationship between the layers of the neural network;

[0104] This step is used to implement the sparse model through a linear rectification function to better mine relevant feature points to fit the image data input in the training step (1);

[0105] According to the attached Figure 4 As shown in the figure, the Rectified Linear Unit (RELU) function is a commonly used activation function in artificial neural networks, usually referring to nonlinear functions represented by ramp functions and their variants. The RELU function formula is

[0106] (3) Using the spatiotemporal attention module, the image data emphasizing the significant changes of the building and the building features optimized by the linear rectification function are extracted and input into the fully connected layer;

[0107] According to the attached Figure 4 As shown, the fully connected layer in the embodiment of the present invention is used to connect the building feature points after the linear rectification function and output them to the softmax classifier for classification.

[0108] According to the attached Figure 4 As shown, the spatiotemporal attention module draws on the mechanism of human visual attention, obtains the target area that needs to be focused on, i.e., the focus of attention, by quickly scanning the global image, and then invests more attention resources in this area to obtain more detailed information of the target that needs to be focused on, while suppressing other useless information. In essence, it selects the information that is more critical to the current task goal from a large amount of information. The spatiotemporal attention module of the present invention is used to obtain key information of image data in two dimensions of time and space.

[0109] According to the attached Figure 4 As shown, global average pooling is used to compress the amount of data and parameters to reduce overfitting.

[0110] (4) Use the Softmax classifier to map the image data output by the fully connected layer to the interval [0,1] in order to classify buildings with and without changes;

[0111] According to the attached Figure 4As shown in the figure, the Softmax classifier is a generalized induction of the logistic regression classifier facing multiple classifications. The softmax function is a commonly used multi-classifier in machine learning, especially in convolutional neural networks, where the last layer often uses the softmax classifier for multi-class classification tasks. Its softmax function is: softmax(z j ) is the softmax function, z j Represents a single original output value, and j represents the output value of the current operation. Softmax is used in the multi-classification process to map multiple inputs to the interval [0,1]. The softmax function uses a natural base e to first increase the difference between the input values, and then normalizes it to a probability distribution, so that the data can be classified.

[0112] (5) Output the building elevation change results of a pair of image data at different time series.

[0113] S4, if there are changes in the buildings within the airport clear area, determine the location of the new buildings or buildings under construction;

[0114] S5, after determining the location of the newly built or under-construction building, collect the building information and calculate the clearance restriction surface and the restricted height of the building;

[0115] As a preferred embodiment, in step S5, according to the airport clearance influencing factors, the height restriction requirements of the location of the building are checked, the proposed height information of the building at the location is verified, and whether the clearance restriction requirements are exceeded is calculated. Specifically, the airport clearance influencing factors include the airport clearance restriction surface, flight procedures, aircraft performance, visual aids and navigation facilities.

[0116] As a preferred embodiment, in step S5, the limit height h(x, y) is:

[0117] h(x,y)=Min{h 附件十四面 (x,y),h 飞行程序 (x,y),h 飞机性能 (x,y),h 导航设施 (x,y),h 目视灯光 (x,y),…}

[0118] h 附件十四面 (x,y)=Min{h 进近面 (x,y),h 内水平面 (x,y),h 锥形面 (x,y),…}

[0119] h 飞行程序 (x,y)=Min{h 仪表进近程序(x,y),h 进场程序 (x,y),h 离场程序 (x,y),…}

[0120] h 导航设施 (x,y)=Min{h 航向台 (x,y),h 下滑台 (x,y),h VOR台 (x,y),…}

[0121] h 目视灯光 (x,y)=Min{h 简易灯光 (x,y),h 精密进近灯光 (x,y),h PAPI灯 (x,y),…}

[0122] Among them, (x, y) is the coordinates of the building,

[0123] h 附件十四面 (x,y),h 飞行程序 (x,y),h 飞机性能 (x,y),h 导航设施 (x,y),h 目视灯光 (x, y) are the restricted altitude of Annex XIV at the coordinate position (x, y), the restricted altitude of the flight procedure, the restricted altitude of the aircraft performance, the restricted altitude of the visual aids, and the restricted altitude of the navigation facilities. Figure 5 To Attachment Figure 7 The corresponding schematic diagram and position relationship are provided.

[0124] The space directly above any surface is the restricted space, and the space directly below the surface is the allowed space. For any point on the surface, there is a corresponding allowed space and restricted space. For a point S1(x,y) on surface S1, the allowed space is C(S1(x,y)) and the restricted space is L(S1(x,y)); similarly for surface S2, a point S2(x,y) on S2, the allowed space is C(S2(x,y)) and the restricted space is L(S2(x,y)). The restricted space is the unallowed space. The principle of clearance restriction is that all points take the lowest point, that is, the restricted space L(x,y) for a certain point is the sum of the corresponding restricted spaces of all faces, that is:

[0125] L(x,y)=L(S 1 (x,y))∪L(S 2 (x,y)

[0126] Similarly, the clearance restriction spaces corresponding to multiple clearance restriction surfaces can be combined to obtain the total restriction space.

[0127] In step S5, the clearance restriction surface is obtained through the total restriction space, wherein the total restriction space L(x, y) is:

[0128] L(x,y)=L(S 附件十四面 (x,y))∪L(S 飞行程序 (x,y))∪L(S 飞机性能 (x,y))∪L(S 导航设施 (x,y))∪L(S 目视灯光 (x,y))∪…

[0129] L(S 附件十四面 (x,y))=L(S 进近面 (x,y))∪L(S 内水平面 (x,y))∪L(S 锥形面 (x,y))∪…

[0130] L(S 飞行程序 (x,y))=L(L(S 仪表进近程序 (x,y))∪L(S 进场程序 (x,y))∪L(S 离场程序 (x,y))∪…

[0131] L(S 导航设施 (x,y))=L(S 航向台 (x,y))∪L(S 下滑台 (x,y))∪L(S VOR台 (x,y))∪…

[0132] L(S 目视灯光 (x,y))=L(S 简易灯光 (x,y))∪L(S 精密进近灯光 (x,y))∪L(S PAPI灯 (x,y))∪…

[0133] Among them, (x, y) is the coordinates of the building,

[0134] L 附件十四面 (x,y),L 飞行程序 (x,y),L 飞机性能 (x,y),L 导航设施 (x,y),L 目视灯光 (x, y) are the annex fourteen restricted space, flight procedure restricted space, aircraft performance restricted space, visual aid facility restricted space, and navigation facility restricted space at the coordinate position (x, y).

[0135] The Annex 14 restricted space, flight procedure restricted space, visual aids restricted space and navigation facilities restricted space each contain several components.

[0136] Among them, Annex 14 surfaces include approach surface, inner horizontal surface, conical surface, transition surface, lift strip, take-off climb surface, etc.;

[0137] The flight procedures include the departure procedures, arrival procedures, approach procedures, holding areas, sectors, etc. for each runway;

[0138] Visual lighting includes simple approach lighting, Category I / II precision approach lighting, and PAPI lighting;

[0139] Navigation facilities include localizer, glide path, VOR, NDB, etc.

[0140] By calculating the minimum value of each restricted space, the total restricted space L(x,y) can be obtained.

[0141] The surface is composed of points, that is, the restricted space corresponding to the surface is:

[0142] S={(x 1 ,y 1 ),(x 2 ,y 2 ),…,(x n ,y n )}

[0143] L(S) = {L(x 1 ,y 1 ),L(x 2 ,y 2 ),…,L(x n ,y n )}

[0144] The space directly above any surface is restricted space, and the space directly below the surface is allowed space. After obtaining the restricted space corresponding to all surfaces, the clearance restriction surface can be obtained through the restricted space, and then the restricted height of the building can be obtained.

[0145] S6, verify the information of newly built or under-construction buildings and check whether their heights meet the restrictions on the location of the buildings caused by the factors affecting the airport clearance.

[0146] An airport clearance monitoring method based on intelligent photoelectric technology is provided in an embodiment of the present invention. Based on the deep combination of artificial intelligence technology and photoelectric technology, the embodiment of the present invention improves the efficiency of judging building changes through preprocessing methods such as image stitching and image noise reduction; further, the embodiment of the present invention determines whether there are building changes in the airport clearance area through deep learning technology, calculates and determines the clearance restriction surface and restricted height of the building, and determines whether its height meets the restriction requirements of the airport clearance influencing factors on the location of the building. The method is simple and efficient, realizes airport clearance monitoring, and has the advantages of clear vision, accurate positioning, low cost and high efficiency.

[0147] In addition, the present invention also provides an airport clearance monitoring system based on intelligent photoelectric technology, which implements detection according to the steps of the airport clearance monitoring method described above, so as to improve the efficiency of airport clearance inspections, monitor buildings within the airport clearance range in real time, and improve the airport's operational safety.

[0148] Compared with the shortcomings and deficiencies of the prior art, the airport clearance monitoring method based on intelligent photoelectric technology provided by the present invention is based on the deep combination of artificial intelligence technology and photoelectric technology, and integrates interdisciplinary cutting-edge technologies such as analog and digital electronics, big data, image processing, sensors, optical imaging, etc., to determine whether there are any building changes in the airport clearance area, calculate and determine the clearance restriction surface and restricted height of the building, and determine whether its height meets the restriction requirements of the airport clearance influencing factors on the location of the building, so as to realize airport clearance monitoring, and has the advantages of clear vision, accurate positioning, low cost and high efficiency; in addition, the method can effectively replace manual inspections of airport clearance, greatly reduce the manpower and material costs of airport clearance inspections, and has good promotion value.

[0149] The present invention is not limited to the above-mentioned embodiments. Anyone can derive other various forms of products under the inspiration of the present invention. However, no matter what changes are made in the shape or structure, all those having the same or similar technical solutions as the present invention fall within the protection scope of the present invention.

Claims

1. An airport clearance monitoring method based on intelligent photoelectric technology, It is characterized in that The following steps are involved: S1, collects image data within the airport clearance area; S2, preprocessing the acquired image data by frame-by-frame extraction, image stitching, and image correction; S3 uses deep learning to compare and analyze image data of adjacent time series to determine whether there are any changes in the buildings within the airport clearance area; S4, if there are changes in the buildings within the airport clear area, determine the location of the new buildings or buildings under construction; S5, after determining the location of the newly built or under-construction building, collect the building information and calculate the clearance restriction surface and the restricted height of the building; S6, verify the information of newly built or under-construction buildings and check whether their heights meet the restrictions on the location of the buildings by factors affecting the airport clearance; Among them, in step S3, comparing and analyzing the image data by using deep learning specifically includes: Label the image data of the building group and establish a sample library that meets the requirements of image data diversity; By using the convolutional neural network structure of multiple instance learning and spatial attention model, a neural network that reflects the multiple sensitive elements of the airport, the significant scale characteristics, and the local semantic expression of the buildings is constructed, and a deep learning model suitable for extracting building information based on optoelectronic technology image data is obtained; Based on the deep learning model, the time series features in the effective image data are extracted, and the differences in the spatial structure type information of the building are compared from images of different time series; The deep learning model specifically includes: a convolutional neural network module, a multi-scale dilated convolution feature fusion module, an attention model, and a multi-instance learning pooling module; The convolution application network module is connected to a multi-scale dilated convolution feature fusion module, the multi-scale dilated convolution feature fusion module is connected to an attention model, and the attention model is connected to a multi-instance learning pooling module; The convolutional neural network module is used to perform dot multiplication on the image data and input it into the attention model. The multi-scale dilated convolution feature fusion module is used to expand the perception threshold of the convolutional neural network and control the size of the output feature map without increasing the parameters; The attention model is used to select required building feature point information from multiple feature points in the image data; The multi-instance learning pooling module is used to highlight the building feature point information obtained in the attention model.

2. The airport clearance monitoring method according to claim 1, It is characterized in that The establishment of a sample library that meets the requirements of image data diversity is to perform style transformation according to different regions, seasons, and imaging conditions, obtain images under various conditions, and form a sample library that meets the diversity of actual image data.

3. The airport clearance monitoring method according to claim 1, It is characterized in that In step S5, the height restriction requirements of the location of the building are checked according to the factors affecting the airport clearance, the proposed height information of the building at the location is verified, and it is calculated whether it exceeds the airport clearance restriction requirements.

4. The airport clearance monitoring method according to claim 3, It is characterized in that The airport clearance influencing factors include airport clearance restriction surfaces, flight procedures, aircraft performance, visual aids and navigation facilities.

5. The airport clearance monitoring method according to claim 1, It is characterized in that In step S5, the limit height h(x,y) is: h(x,y)=Min{h 附件十四面 (x,y),h 飞行程序 (x,y),h 飞机性能 (x,y),h 导航设施 (x,y),h 目视灯光 (x,y),…} h 附件十四面 (x,y)=Min{h 进近面 (x,y),h 内水平面 (x,y),h 锥形面 (x,y),…} h 飞行程序 (x,y)=Min{h 仪表进近程序 (x,y),h 进场程序 (x,y),h 离场程序 (x,y),…} h 导航设施 (x,y)=Min{h 航向台 (x,y),h 下滑台 (x,y),h VOR台 (x,y),…} h 目视灯光 (x,y)=Min{h 简易灯光 (x,y),h 精密进近灯光 (x,y),h PAPI灯 (x,y),…} Among them, (x, y) is the coordinates of the building, h 附件十四面 (x,y),h 飞行程序 (x,y),h 飞机性能 (x,y), h 航设施 (x,y),h 目视灯光 (x, y) are the Annex XIV restricted altitude, flight procedure restricted altitude, aircraft performance restricted altitude, visual aids restricted altitude, and navigation facility restricted altitude at the coordinate position (x, y) respectively.

6. The airport clearance monitoring method according to claim 1, It is characterized in that In step S5, the clearance restriction surface is obtained through the total restriction space, wherein the total restriction space L(x, y) is: L(x,y)=L(S 附件十四面 (x,y))UL(S 飞行程序 (x,y))UL(S 飞机性能 (x,y))UL(S 导航设施 (x,y))UL(S 目视灯光 (x,y)) And. L(S 附件十四面 (x,y))=L(S 进近面 (x,y))UL(S 内水平面 (x,y))UL(S 锥形面 (x,y)) And. L(S 飞行程序 (x,y))=L(L(S 仪表进近程序 (x,y))UL(S 进场程序 (x,y))UL(S 离场程序 (x,y)) And. L(S 导航设施 (x,y))=L(S 航向台 (x,y))UL(S 下滑台 (x,y))UL(S VOR台 (x,y)) And. L(S 目视灯光 (x,y))=L(S 简易灯光 (x,y))UL(S 精密进近灯光 (x,y))UL(S PAPI灯 (x,y)) And. Among them, (x, y) is the coordinates of the building, L 附件十四面 (x,y), L 飞行程序 (x,y), L 飞机性能 (x,y), L guide 航设施 (x,y), L 目视灯光 (x, y) are the annex fourteen restricted space, flight procedure restricted space, aircraft performance restricted space, visual aid facility restricted space, and navigation facility restricted space at the coordinate position (x, y).

Images