A multi-source information fusion polar ice and snow runway quality assessment method and system

Abstract

The application provides a polar ice and snow runway quality evaluation method and system based on multi-source information fusion, and belongs to the technical field of aviation infrastructure detection and polar engineering. The evaluation method specifically comprises the following steps: generating a panoramic orthographic image based on collected sequence images, defining a runway physical boundary and dividing the runway into an edge reference area and a center load evaluation area; extracting a healthy runway frequency domain fingerprint by using the edge reference area, performing frequency domain scanning on the center load evaluation area, and locking a key quality evaluation area set; extracting a local point cloud based on the coordinate index of the key quality evaluation area set, constructing a local reference plane, calculating the maximum physical concave depth, and calculating a comprehensive hazard index at a microscopic level; quantifying a bending risk and a transverse snow abnormal index based on a macroscopic settlement field; mapping the comprehensive hazard index, the bending risk and the transverse snow abnormal index to a unified runway geographic coordinate system, generating a risk weight map and outputting a maintenance strategy.

Description

Technical Field

[0001] This application belongs to the field of aviation infrastructure inspection and polar engineering technology, specifically involving a method and system for assessing the quality of polar ice and snow runways through multi-source information fusion. Background Technology

[0002] Polar air transport is a vital lifeline for polar scientific research, resupply, and personnel rotation. Unlike the concrete or asphalt pavements widely used in low-latitude regions, polar airports are typically constructed based on blue ice, gravel, sea ice, and snow. Among these, ice and snow runways are the most common type of runway in the polar regions. These runways are created by repeatedly crushing, compacting, sintering, and leveling natural snow using heavy machinery, and then artificially grooving the surface to create texture, thereby meeting the load-bearing capacity and friction coefficient requirements for wheeled aircraft takeoffs and landings. Due to the extreme nature of the polar environment and the unique rheological properties of ice and snow materials, the operation and maintenance of these runways face challenges that are drastically different from those of conventional airports.

[0003] Polar compacted snow runways exist in a dynamically changing low-temperature environment, making their physical properties highly susceptible to the combined effects of weather conditions and loads. On one hand, frequent aircraft takeoffs and landings, along with tire friction, cause the grooved structure of the runway surface to gradually wear down and smooth out. Frictional heat can even lead to surface micro-melting followed by re-icing, significantly reducing the coefficient of friction and increasing the risk of aircraft overrunning the runway. On the other hand, the unique viscoelasticity and thermal sensitivity of snow and ice make the runway prone to uneven settlement, erosion pits, or long-wavelength undulations under the influence of solar radiation and the movement of underlying glaciers. If these geometric changes are not detected in time, they can cause irreversible damage to the aircraft landing gear structure and even lead to serious aviation safety accidents.

[0004] Existing methods for assessing the quality of polar runways primarily rely on manual inspections and contact measurements. Traditional methods typically involve professionals driving specialized vehicles equipped with friction coefficient testers to conduct sampling tests on the runway. However, this approach has significant limitations. The polar climate is harsh, with frequent blizzards, resulting in short window periods for manual on-site operations, high labor intensity, and significant personal safety risks. Secondly, contact measurements are usually based on discrete sampling data, making it difficult to achieve continuous, full-coverage monitoring of the runway and easily overlooking hidden cracks. Furthermore, the movement of heavy inspection vehicles can cause secondary disturbances to the fragile snow surface structure.

[0005] With the development of remote sensing technology, non-contact detection based on optical images has been gradually introduced into airport pavement assessment. However, in polar environments, single visual sensors face severe optical challenges. Polar ice and snow surfaces have extremely high reflectivity and low contrast characteristics, and are prone to whitening under diffuse light conditions, resulting in images lacking texture and making it difficult to distinguish pavement undulations from surface stains using conventional image processing algorithms. In addition, the low solar altitude angle unique to the polar regions casts elongated shadows over even small surface irregularities, and this high-frequency shadow noise can easily be misjudged as cracks or foreign objects.

[0006] Meanwhile, although 3D laser scanning technology can acquire high-precision geometric point cloud data, it also faces technical bottlenecks when applied to ice and snow. The smooth ice surface produces strong specular reflections of the laser, leading to the loss of echo signals and creating data gaps. Furthermore, relying on a single data source makes it difficult to accurately decouple the two failure modes of runway geometric settlement and texture degradation, failing to provide precise quantitative basis for maintenance decisions at polar airports. Therefore, there is an urgent need for a runway quality assessment technology that can adapt to the unique optical environment of polar regions and integrate microscopic textures with macroscopic geometric features. Summary of the Invention

[0007] This invention provides a method for assessing the quality of polar ice and snow runways by fusing multi-source information. The method includes: generating panoramic orthophotos based on acquired sequential images, defining the physical boundaries of the runway and dividing it into an edge reference zone and a central load-bearing assessment zone. By extracting the frequency domain fingerprint of the healthy running track using the edge benchmark area, the central load assessment area is scanned in the frequency domain to identify the set of key quality assessment areas. Local point clouds are extracted based on the coordinate index of the key quality assessment area set, a local reference plane is constructed and the maximum physical depression depth is calculated. Logical verification is performed by combining the texture degradation index and the comprehensive hazard index at the micro level is calculated. The theoretical baseline of the transverse slice is reconstructed using the edge benchmark area, the macroscopic settlement field of the central bearing assessment area is calculated, and the bending risk and transverse snow accumulation anomaly index are quantified based on the macroscopic settlement field. The comprehensive hazard index, bending risk, and lateral snow accumulation anomaly index are mapped to a unified runway geographic coordinate system to generate a risk weight map and output maintenance strategies.

[0008] Generate panoramic orthophotos based on the acquired sequence of images, including: extracting the frequency domain image of the overlapping region between two adjacent corrected frames. and Calculate the normalized cross-power spectrum : in, For frequency domain coordinates, express The complex conjugate; Performing an inverse discrete Fourier transform on the normalized cross-power spectrum yields the phase correlation function matrix. And select the peak coordinates of the phase correlation function matrix. A quadratic surface is fitted to the region and its neighboring pixels to obtain an accurate translation vector. : The requirements section is repeated. in, This represents the probability distribution of overlap between two images under different relative displacements.

[0009] Generating panoramic orthophotos also includes: calculating the first Global coordinate offset of frame image in panoramic coordinate system : in, For the first Frame and the Precise translation vectors between frames; Constructing panoramic orthophotos based on global coordinate offsets : in, For the coordinates of the panoramic image, For the set of image frames covering this coordinate, For the weight function, For the corrected first Frame image, for The amount.

[0010] Defining the physical boundaries of the runway and dividing it into an edge reference zone and a center load-bearing assessment zone includes: defining the lateral gradient operator. Calculate panoramic orthophotos column-to-column cumulative gradient energy spectrum : in, The length of the runway panoramic view. Image coordinates; Based on the column-accumulated gradient energy spectrum, the coordinates of the left and right physical boundaries of the runway are identified. and : in, An adaptive threshold; based on and The runway space is divided into the left edge reference zone. Right edge reference area and the central carrying capacity assessment area .

[0011] A frequency domain scan of the central carrying capacity assessment area was performed, including: Extracting the first using a sliding window line, number Local sub-images of a column And introduce a two-dimensional Hanning window function. Weighted processing yields image patches: in, Step size, Local coordinates; Performing a two-dimensional discrete Fourier transform on the image patch yields the complex spectrum. And calculate the logarithmic power spectral density. : in, These are frequency domain coordinates.

[0012] Extract the frequency domain fingerprint of the healthy running track and identify key quality assessment areas, including: Calculate the average power spectrum of the edge reference region as the reference spectrum. Define the set of groove signal frequencies As a frequency domain fingerprint of a healthy running track: in, The relative intensity threshold coefficient is used to calculate the texture degradation index of each sub-block in the central load-bearing evaluation area. : in, For small stability constants; select those that satisfy The regions form a set of key quality assessment areas, among which This is the threshold for damage assessment.

[0013] Constructing a local reference plane and calculating the maximum physical indentation depth includes: extracting the center coordinates of the key quality assessment area. Centered on, with radius Local point cloud collection ;right The equation of the reference plane is obtained by performing least-squares plane fitting. Calculation points relative depth residual : Extracting the maximum physical depression depth from a local point cloud set : .

[0014] By combining the texture degradation index with logical cross-verification, a comprehensive hazard index at the microscopic level is calculated, including: Mapping the maximum physical indentation depth to a geometric risk probability using the Sigmoid function and the texture degradation index Mapped to texture risk probability : in, This represents the critical threshold for the depth of the depression. This is the steepness coefficient. Set a saturation threshold; construct a state conflict discrimination function. To identify State II thermal fusion settlement and calculate the comprehensive hazard index. : in, The geometric weight bias coefficient. As an additional penalty factor for thermal fusion settlement, This is an indicator function.

[0015] The theoretical baseline of the transverse slice is reconstructed using the edge benchmark region. The macroscopic subsidence field is calculated, and the bending risk and transverse snow accumulation anomaly index are quantified, including using the Huber loss function. For point sets belonging to the edge reference region Perform regression to obtain the first... Theoretical baseline parameters for each slice : Constructing a macroscopic settling field And calculate the transverse curvature based on its second-order differential. and bending risk : Define local gradient and the potential energy density of windblown snow accumulation Calculate the transverse snow cover anomaly index : in, This is an indicator function.

[0016] This invention also provides a multi-source information fusion system for assessing the quality of polar ice and snow runways, the system comprising: Region division module: Generate panoramic orthophotos based on the acquired sequence of images, define the physical boundaries of the runway and divide it into an edge reference area and a central load-bearing assessment area; use the edge reference area to extract the frequency domain fingerprint of the healthy runway, perform frequency domain scanning on the central load-bearing assessment area, and lock in the set of key quality assessment areas; Local hazard identification module: Extract local point clouds based on the coordinate index of the key quality assessment area set, construct a local reference plane and calculate the maximum physical depression depth, combine with the texture degradation index for logical verification, and calculate the comprehensive hazard index at the micro level; Global Hazard Identification Module: Reconstructs the theoretical baseline of the transverse slice using the edge benchmark area, calculates the macroscopic settlement field of the central bearing capacity assessment area, and quantifies the bending risk and transverse snow accumulation anomaly index based on the macroscopic settlement field; Maintenance module: Maps the comprehensive hazard index, bending risk and lateral snow accumulation anomaly index to a unified runway geographic coordinate system, generates a risk weight map and outputs maintenance strategies.

[0017] This invention provides a multi-source information fusion method and system for evaluating the quality of polar ice and snow runways, offering significant advantages. This application solves the problem of constructing high-precision orthophoto panoramic data in polar environments with weak texture. Addressing the highly homogeneous surface of compacted snow runways and the repetitive texture features caused by artificial grooving, this application abandons traditional stitching algorithms that rely on corner extraction and adopts a frequency-domain phase correlation registration strategy under inertial navigation attitude constraints. This strategy utilizes the overall spectral energy distribution of the image for alignment, effectively avoiding geometric misalignment and ghosting phenomena caused by missing or mismatched feature points on ice and snow surfaces without significant ground features, ensuring a unified and high-precision spatial benchmark for subsequent quality assessment. Simultaneously, this application automatically defines the physical boundaries of the runway using the cumulative characteristics of the grooved texture gradient, establishing a relatively static internal reference system on polar ice sheets lacking absolute geographical control points, providing a reliable physical benchmark for subsequent differential assessment. This application introduces a frequency-domain analysis mechanism, significantly improving the sensitivity to micro-texture degradation. This method extracts the frequency domain fingerprint of a healthy running track using edge reference areas and transforms the detection dimension from spatial domain grayscale, which is highly susceptible to light interference, to frequency domain energy distribution. Leveraging the inherent periodicity of artificial grooves, this method can accurately detect periodic signal attenuation caused by mechanical wear or ice deposits. Therefore, even under the diffused light or low solar altitude conditions unique to polar regions, it can still accurately identify early pavement damage that is difficult to detect with the naked eye, effectively solving the technical challenge of visual detection failure in low-contrast scenarios.

[0018] This application constructs a deep logic verification system based on graph-cloud mutual verification, significantly reducing the false alarm rate and false negative rate of the assessment. By mapping the texture degradation index extracted visually and the physical concavity depth extracted from the point cloud to a unified risk probability space, this application constructs a state conflict discrimination logic. This logic can not only eliminate visual false alarms caused by projected shadows or surface oil stains through geometric flatness, but more importantly, it can accurately decouple the two distinct failure modes of "surface wear" and "deep thermal melting". In particular, for hidden disasters where the surface grooves are intact but the foundation has undergone thermal melting and collapse, this method can accurately capture them through the logical mutual exclusion relationship between texture and depth, filling the gap in existing technology for the detection of hidden defects. This application adopts a "coarse-to-fine" detection strategy, using visual frequency domain analysis to quickly locate key areas, and then guiding laser point cloud for local fine scanning, effectively solving the problem of limited computing power and energy for equipment in polar fields. The resulting visualized weighted graph and hierarchical maintenance strategy, which integrates multi-dimensional risk characteristics, can directly guide maintenance personnel to take targeted engineering measures for different types of defects, realizing a closed loop from data perception to decision execution, and significantly improving the safety and maintenance efficiency of polar airport operations. Attached Figure Description

[0019] Figure 1 This is a flowchart of the quality assessment of polar ice and snow runways based on multi-source information fusion, as described in this invention. Figure 2 This is the orthophoto panoramic image of the runway obtained after stitching together according to the present invention; Figure 3 This invention relates to the lateral cumulative gradient energy spectrum of orthophoto panoramic images; Figure 4 This is a comparison chart of the frequency domain characteristics of healthy groove textures and texture degradation regions. The top row is divided into healthy groove textures with clear regularity within the edge reference area, and texture degradation regions due to wear within the central bearing area; the bottom row shows the spectral results for the two types of regions respectively. Figure 5 This is a calculation map of the local point cloud depression depth in the key quality assessment area. Detailed Implementation

[0020] The following specific examples illustrate the implementation of this application. Those skilled in the art can easily understand other advantages and effects of this application from the content disclosed in this specification. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. This application can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of this application. It should be noted that, in the absence of conflict, the following embodiments and features in the embodiments can be combined with each other. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0021] It should be noted that various aspects of embodiments within the scope of the appended claims are described below. It will be apparent that the aspects described herein can be embodied in a wide variety of forms, and any particular structure and / or function described herein is merely illustrative. Based on this application, those skilled in the art will understand that one aspect described herein can be implemented independently of any other aspect, and two or more of these aspects can be combined in various ways. For example, any number and aspects set forth herein can be used to implement the device and / or practice the method. Additionally, this device and / or method can be implemented using structures and / or functionalities other than one or more of the aspects set forth herein.

[0022] Additionally, specific details are provided in the following description to facilitate a thorough understanding of the examples. However, those skilled in the art will understand that practice can be carried out without these specific details.

[0023] This embodiment provides a multi-source information fusion method for assessing the quality of polar ice and snow runways. The scenario is a compacted snow runway constructed in a high-latitude polar ice cap region. The runway is located in an extremely cold climate, with an average annual temperature below -20 degrees Celsius. The foundation consists of deep permafrost and a layer of granular snow that has undergone multiple compaction processes. The runway surface exhibits extremely high optical reflectivity and low contrast under diffused light conditions. The physical environment of the runway is affected by the low solar altitude angle in the polar region, resulting in a persistently tilted angle of sunlight incidence. Furthermore, it is susceptible to surface wind and snowfall, causing the runway surface microtexture to include not only static structural features but also dynamic meteorological and environmental depositional characteristics.

[0024] The evaluation object in this embodiment is an ice and snow runway serving the take-off and landing of heavy fixed-wing transport aircraft. The runway structure is physically divided into a central load-bearing zone and an edge protection zone. The runway surface is pre-grooved with periodic artificial grooves perpendicular to the runway centerline through heat treatment and mechanical etching processes. These grooves are regularly arranged along the entire length of the runway to increase tire grip and drain trace amounts of meltwater generated by friction. The edge protection zone is located beyond the outermost groove boundaries on both sides of the runway. Although this area has undergone foundation compaction, it is not paved with artificial grooves and rarely bears direct rolling loads from aircraft tires during daily operation. Therefore, this area retains the original elevation benchmark and surface texture from the initial runway construction, serving as the physical reference standard for quality evaluation in this embodiment.

[0025] In this application scenario, the main factors leading to runway quality degradation include mechanical wear and thermal melt settlement. During takeoff and landing, heavy wheeled aircraft experience frequent contact between their landing gear tires and the runway's center load-bearing area. The high-intensity frictional shear forces cause the serrated edges of the artificial grooves to gradually become rounded or filled with ice debris. Simultaneously, the concentrated load generated by the aircraft's weight and the heat conduction effect of the dark tires cause viscoelastic creep or localized micro-melting of the ice beneath the pavement, resulting in geometric deformations of varying scales in the central area of ​​the pavement. This deformation manifests as microscopic irregular depressions (deep pits) and macroscopic disc-shaped settlement grooves. These physical damages directly impact the safety of subsequent flight takeoffs and landings.

[0026] To conduct non-contact assessment of the aforementioned scenario, this embodiment employs a multi-source information acquisition system. This system is mounted on an industrial-grade drone platform capable of operating in low-temperature environments, integrating a high dynamic range visible light camera and a high-precision LiDAR (Light Detection and Ranging) system. The drone is configured to perform low-altitude, bow-shaped or strip-shaped flight missions along the runway, with the operating altitude controlled within a range where the artificial groove texture can be clearly distinguished. The visible light camera is configured to acquire a sequence of images with a high overlap rate to generate orthorectified panoramic images reflecting the details of the runway texture. The LiDAR is configured to synchronously emit high-frequency laser pulses and acquire 3D point cloud data reflecting the geometric undulations of the runway by receiving echo signals. The acquisition process covers the runway's central bearing area and the edge protection zones on both sides, ensuring spatial alignment and logical correlation between texture and geometric features in subsequent data processing.

[0027] On compacted snow runways in polar regions, the surface cover exhibits high homogeneity, characterized by a highly reflective white background and periodic artificial groove textures. This optical characteristic of both weak and repetitive textures makes it difficult for traditional feature-point-based image stitching algorithms to extract robust corner points, and they are prone to mismatches between adjacent similar grooves, leading to geometric misalignment or ghosting in the stitched panoramic image. This embodiment addresses two problems: utilizing UAV pose information and frequency domain analysis technology to achieve high-precision sequential image stitching on snow and ice surfaces lacking significant feature points; and automatically defining the physical boundaries of the runway based on the statistical characteristics of the texture gradient in the stitched image.

[0028] S1-1: Image preprocessing and registration based on inertial navigation attitude constraints; During data collection, drones experience attitude jitter due to airflow. By utilizing the high-precision inertial measurement unit and RTK-GPS data onboard the drone, the acquired sequence of images is corrected for planar orientation, projecting them onto a unified horizontal reference plane to eliminate perspective distortion caused by camera tilt.

[0029] Let the first The original image of the frame is The corresponding camera rotation matrix is and the camera intrinsic parameter matrix are Define the corrected image. The projection transformation is as follows: in Let be the rotational homography matrix.

[0030] Calculate adjacent frames using GPS location information and coarse translation vector between This is used to determine the search window ROI of the overlapping area of ​​the image, avoiding getting stuck in local optima in the repeated groove texture, that is, preventing matching to the wrong adjacent groove.

[0031] S1-2: Frequency domain registration based on phase correlation; To address the lack of feature points on the surface of ice and snow, Fourier transform is used to convert image data from the spatial domain to the frequency domain. Frequency domain matching utilizes the global information of the image and has strong robustness to local whitening or noise.

[0032] Extract the grayscale matrix of the overlapping region between two adjacent corrected images. and Perform a two-dimensional discrete Fourier transform on it to obtain the frequency domain image. and : in Given frequency domain coordinates, calculate the normalized cross-power spectrum of the two frequency domain images. : In the formula, express The complex conjugate of . According to the Fourier translation theorem, if two images have a translation relationship, the inverse Fourier transform of their cross power spectrum will produce a Dirac impulse function in the spatial domain.

[0033] Calculate the phase correlation function : On the polar runway Peak coordinates This corresponds to the precise translation amount between the two images. A quadratic surface is fitted to the peak neighborhood, and then... and its neighboring areas Pixel region, to obtain accurate translation vector .

[0034] The phase correlation function matrix is ​​a two-dimensional real matrix in the spatial domain, representing the probability distribution of overlap between two images under different relative displacements. This is achieved by analyzing the normalized cross-power spectrum. Obtained by the inverse discrete Fourier transform. For a size of The overlapping region, any coordinate in the matrix The element value at a given position is defined by the following formula: in, The base of the natural logarithm, The imaginary unit is denoted by . Under ideal, noise-free translation conditions, the matrix... Only in the corresponding actual translation amount The location is represented by a Dirac pulse, where the value at that point is 1 and the values ​​at all other points are 0; in actual polar images, this matrix represents an energy peak distribution centered on the actual translation.

[0035] based on All image sequences are translated and fused to generate a high-resolution panoramic orthophoto. .

[0036] After obtaining the translation vectors between the images in the sequence, a unified panoramic coordinate system needs to be constructed. Let the first frame image... The top left corner is the origin of the panoramic coordinate system. . No. Global coordinate offset of frame image in panoramic coordinate system This is the sum of the relative displacements between all preceding frames: Constructing panoramic orthophotos : in For the coordinates of the panoramic image, For the set of image frames covering this coordinate, for The amount, It is a fade-in / fade-out weight function based on the distance from the image center to eliminate stitching gaps.

[0037] S1-3: Physical boundary delineation based on lateral gradient accumulation.

[0038] In generating panoramic images Next, the physical edges of the runway need to be identified. This is achieved by utilizing the textural difference between the densely packed artificial grooves in the runway area and the smooth, naturally accumulated snow in the edge areas. Since the artificial grooves are perpendicular to the runway direction (i.e., the image...),... (axial direction), which is in the image The grayscale changes drastically along the axial direction (lateral). Define the lateral gradient operator. The gradient magnitude of each row in the panoramic image is calculated. To eliminate the influence of individual noise points on the ice surface, a column-oriented cumulative gradient energy spectrum is constructed. : in This represents the length of the panoramic view of the runway.

[0039] This energy spectrum The value oscillates at high amplitude in the runway groove area, but rapidly decays to a low value in the edge snow accumulation area. An adaptive threshold is set. The preferred setting is the average value of the energy spectrum. times, Identify the coordinates of the left and right physical boundaries of the runway and : Based on the coordinates of the left and right physical boundaries of the runway and The panoramic space is divided into three independent areas: the left edge reference area. Right edge reference area Central carrying capacity assessment area .

[0040] in, The preset sampling width for the edge reference area; This serves as a safe buffer distance for the central area to contract inwards.

[0041] This invention addresses the unique scenario of polar ice and snow runways, which lack significant feature points and contain numerous repetitive groove textures. It employs a combined strategy of inertial navigation constraints and frequency domain phase correlation matching. This method avoids the failure problem of traditional feature point extraction algorithms on homogenized surfaces, utilizing the overall spectral characteristics of the grooves for alignment, effectively preventing geometric misalignment during image stitching. Furthermore, by leveraging the significant features of the groove texture in gradient energy, the physical boundaries of the runway are automatically segmented. This process not only acquires complete panoramic runway data but, more importantly, establishes... (Base zone) and The spatial logical relationship of the (assessment area).

[0042] For compacted snow runways in polar regions, in the visual spatial domain, due to the diffuse reflection of the ice and snow medium and the high dynamic range of illumination, worn areas and intact areas often exhibit extremely similar grayscale characteristics, making it difficult for conventional algorithms based on pixel difference or edge detection to distinguish between grooves being worn smooth and changes in illumination. However, in the frequency domain, the pre-fabricated artificial grooves on the runway surface constitute a strong periodic signal, and wear or filling signifies this periodic damage.

[0043] Therefore, this invention utilizes an established edge reference region as a reference source for an ideal physical state to extract the inherent spectral characteristics of the groove; subsequently, a frequency domain scan is performed on the central load-bearing evaluation region, and by quantifying the degree of degradation of local regions from ordered periodic texture to disordered chaotic texture, the texture anomaly region is quickly located across the entire field. This process can effectively avoid polar light interference.

[0044] S2-1: Sliding window segmentation; Given the highly uneven wear distribution of polar runways, typically concentrated in the ruts, it is necessary to analyze panoramic orthophotos. Perform local slice analysis.

[0045] set up These are the index counts of the sliding window in the row and column directions of the panorama, respectively. Extract the... line, number Local sub-images of a column : in, This represents the number of pixels the sliding window moves in. These are local pixel coordinate variables within the sub-image, with a value range of [value range missing]. ; The size is the number of pixels on the side of the local sliding window. This size should be chosen to cover at least 3-5 complete groove cycles to meet the statistical significance of the frequency domain analysis.

[0046] Because rectangular blocks are forcibly cropped from continuous runway images, abrupt changes in grayscale at the image edges produce high-frequency artifacts in the spectrum, which weakens the subtle ice wear signal. To address this, a two-dimensional Hanning window function is introduced. The sub-images are weighted to simulate the human eye's focusing mechanism on the center of the field of vision, thus suppressing edge noise. in, It is a two-dimensional Hanning window function matrix; The local sub-image matrix after windowing has its edge grayscale smoothly transitioned to zero, ensuring that the subsequently extracted spectral features only reflect the texture structure of the runway itself.

[0047] S2-2: Construct the logarithmic power spectral density of the perturbation; For image patches A two-dimensional discrete Fourier transform is performed. This invention transforms image information from a "position-grayscale" space to a "frequency-energy" space, thereby utilizing the periodicity of the grooves to combat the uneven illumination on the polar surface. The transformed complex spectrum is defined as follows: : in, For the frequency coordinate variables in the frequency domain space of the local sub-image; The imaginary unit; This is the complex spectrum of the two-dimensional discrete Fourier transform of a local image.

[0048] Because of the specular reflection on the surface of ice and snow, its DC component energy is extremely high, often masking the high-frequency signals representing the groove texture. Therefore, calculating the logarithmic power spectral density... Perform dynamic range compression: in, is the logarithmic power spectral density matrix of the local sub-image. The groove texture of a normal track will appear as significant bright spots, i.e., concentrated energy, while low-frequency interferences such as lighting and shadows are compressed near the origin.

[0049] S2-3: Frequency domain fingerprint extraction of a healthy running track based on a physical reference system; The grooved texture features in polar environments vary slightly with the drone's flight altitude and shooting angle. To obtain the most accurate standard for the current moment, a locked physical reference area is used. .from Randomly selected For each sample block, its average power spectrum is calculated as the reference spectrum. This spectrum represents the ideal runway texture characteristics under current lighting and flight conditions, without aircraft tire wear.

[0050] Based on this, the characteristic frequencies of the grooves are identified. Since the grooves are physically made, their direction and spacing are fixed, and the energy is necessarily concentrated in a specific frequency band. The set of groove signal frequencies is defined. : in, The average reference power spectrum is calculated based on the reference area samples; This is the relative intensity threshold coefficient used to extract feature frequencies; This is a set of frequency coordinates representing the periodic characteristics of a well-formed groove. This set represents the frequency domain fingerprint of the healthy runway in this evaluation task.

[0051] S2-4: Obtain a set of key quality assessment areas.

[0052] In the central carrying area Frequent takeoffs and landings of aircraft lead to two types of physical damage: first, the rounding of grooved edges due to friction (attenuation of periodic signals); and second, surface smoothing due to ice buildup (loss of high-frequency components). Both phenomena manifest in the frequency domain as energy shifting from... The region diverges towards the incoherent noise region.

[0053] Therefore, a texture degradation index is defined. Quantifying this energy transfer process: in, To characterize the first A quantitative index of the wear and tear of individual blocks; This is a small stability constant used in numerical calculations.

[0054] In the texture degradation index formula, the denominator corresponds to the structural integrity of the groove, while the numerator corresponds to the random roughness or impurities on the surface. When the groove is ground down or filled, the denominator decreases sharply, while the numerator relatively increases. The value increases exponentially, thus indicating physical damage.

[0055] Set damage assessment threshold Filter out those that meet the requirements The abnormal blocks form a set of key quality assessment areas. : in, The threshold used to determine whether a texture has undergone abnormal degradation; This is the set of spatial coordinates for the key quality assessment areas in the final output.

[0056] This invention solves the detection challenge caused by the low visual contrast of polar ice and snow running tracks by introducing a frequency domain analysis mechanism. Dynamic extraction of edge reference areas is also included. The frequency domain fingerprinting strategy allows the evaluation process to adapt to the varying lighting and flight conditions in the polar regions, eliminating the need for pre-setting fixed texture parameters. While frequency domain analysis outputs visually key quality assessment areas, a single data source has severe physical limitations under polar ice and snow runways. The extremely low solar altitude angle in the polar regions projects long, dark shadows behind tiny snow ridges, leading to a significant decrease in the visual texture degradation index. The runway surface may show a significant drop in depth, but this is not structural damage. The unique polar ice melt can cause subsidence craters, but the surface grooves may remain intact, which visual inspection might miss. To address this issue, this step employs microscopic depth extraction based on a local reference surface, nonlinear normalization of heterogeneous data, and logical verification based on a conflict matrix.

[0057] S3-1: Calculation of relative depth residuals based on spatial fitting of local point clouds; Set up key quality assessment areas center coordinates As an index, local microenvironments are extracted from the lidar point cloud library.

[0058] Because polar runways have a natural slope due to the movement of ice shelves, direct elevation measurements cannot reflect local potholes; therefore, a reference datum adapted to the local terrain must be constructed. The central coordinates are then extracted. , radius is Point cloud collection inside a cylinder : in, For the set of The coordinates of the laser points are determined using the least squares method. By performing plane fitting, the equation of the local reference plane is obtained. Calculate the vertical distance from each point to the reference surface, i.e., the relative depth residual. : in, The coefficients of the fitted plane represent the macroscopic topographic trend of the local area; To remove the micro-level undulations after removing the macro-level topography, It indicates a depression.

[0059] Extract the maximum physical depression depth of the region. : in, The physical depth (in meters) of the deepest point in the local area.

[0060] S3-2: Nonlinear normalization of heterogeneous data.

[0061] For physical depth The Sigmoid function is used for mapping. There is a safety threshold for runway dents; exceeding this threshold indicates saturation of risk. In other words, dents deeper than this threshold, regardless of their actual depth, are fatal to tires. The geometric risk probability is defined. : in, This represents the critical threshold for the depth of the depression. This is the steepness coefficient, used to control the sensitivity of the transition from safe to dangerous risk. .

[0062] For texture index Perform maximum value normalization and define texture risk probability. : in, The saturation threshold at which the texture is completely destroyed; .

[0063] S3-3: Construct the logic conditions for conflict detection; This invention constructs a state conflict discrimination function. This function is based on and The high and low combinations output a clear physical state of the disease.

[0064] in, The preset probability grading threshold.

[0065] The meaning and processing strategy of physical mutual verification for each state are as follows: State I indicates a severe visual anomaly, such as a blackboard, but the point cloud display surface is flat. Visual sensors affected by low-angle polar projections or surface oil contamination are identified as having false positive defects.

[0066] State II indicates that the visual texture is normal and the groove is intact, but the point cloud depth exceeds the standard. This is the most dangerous hidden hazard in the polar regions. It indicates that the permafrost or ice base beneath the runway has thawed and collapsed, but the surface ice sheet has not yet broken.

[0067] State III is characterized by visual texture wear and geometric indentation. Typical mechanical damage includes tire punctures or severe wear.

[0068] S3-4: Generate a quality assessment map; calculate the comprehensive hazard index only for the confidence areas of State II and State III. : in, The bias coefficient for geometric weights; It is an indicator function; This is an additional penalty factor for thermal settling.

[0069] Based on the assessment of microscopic defects, polar compacted snow runways face the more macroscopic risk of long-wave disk-shaped deformation. Because the runways are constructed on viscoelastic glaciers or granular snow layers, irreversible plastic creep will occur in the central load-bearing zone under the long-term cumulative effect of heavy transport aircraft wheel loads. This settlement is not only manifested as a decrease in elevation, but more fundamentally as an abnormal curvature of the ice sheet.

[0070] Therefore, this invention establishes a full-field transverse slice model. The theoretical zero plane is reconstructed using the established edge reference region, and the curvature modulus characterizing the bending stress of the ice sheet is calculated.

[0071] S4-1: Horizontal slice reconstruction.

[0072] The processor retrieves the entire field's laser point cloud data and performs differential slicing processing along the runway's orientation (Y-axis). The slice index is defined as... The slice step size is Extract the first A subset of point clouds in a horizontal slice Using the set of reference points belonging to the defined edge region in the slice. This reconstructs the theoretical baseline of the location under unloaded conditions.

[0073] Considering the possibility of windblown snow accumulation at the edges, the baseline parameters for noise reduction are first obtained: in, For the first Optimal estimates of the slope and intercept of the theoretical baseline for each slice. The horizontal and vertical coordinates of the point cloud within the edge reference area; The Huber loss function is used to suppress the drag of edge snow ridge noise on the reference surface.

[0074] Based on the regression parameters, construct the first... Macroscopic settling field of slices : in, Horizontal coordinate Macroscopic settlement at the location (positive for downward); The smooth elevation surface measured in the central bearing area.

[0075] S4-2: Bending risk calculation.

[0076] Subsidence depth alone is insufficient to reflect the risk of ice sheet fracturing; the degree of ice sheet curvature directly corresponds to its internal tensile stress. Regarding the subsidence field... Perform second-order differential analysis to calculate the transverse curvature. : in, The local curvature modulus represents the bending tendency of the macroscopic cross-section.

[0077] Step-by-step definition of slice Bending risk Characterizes the overall bending deformation energy of the cross section: in, This represents the risk of bending. The higher the value, the more severe the disc-shaped bending of the runway section, and the higher the potential risk of through-cracks in the ice sheet.

[0078] S4-3: Calculation of Snowdrift Resistance and Snow Accumulation Potential Energy: Another major hazard of macroscopic settlement is the alteration of lateral slope, leading to abnormal accumulation of snowdrifts in settlement troughs during windy weather. Defining the local gradient... The negative first derivative of the settlement field: when When the sign of the y-value is opposite to the runway design slope, or when the absolute value approaches 0, the surface wind speed drops sharply, and the ice and snow particles it carries, i.e., blown snow, will form deposits at that location. The potential energy density of blown snow is defined as follows: : in, For the indicator function, when the horizontal position The slope at a point points towards the center of the runway, meaning it obstructs the lateral migration of windblown snow; otherwise, it is set to 0. It represents the volume of windblown snow that may be deposited per unit area.

[0079] The transverse snow cover anomaly index of the slice is obtained by integrating the density. : in, This represents the physical potential of anomalous snow accumulation caused by macroscopic deformation in this slice.

[0080] Thus, this invention has quantified the comprehensive hazard index at the microscopic level for potholes and cracks. At the macro level, the bending risk and the lateral snow accumulation anomaly index were quantified for ice sheet fracturing. However, isolated numerical values ​​are insufficient to directly guide operations for on-site maintenance personnel. Maintenance windows at polar airports are extremely short, necessitating priority for handling complex, high-risk areas. Therefore, this embodiment maps the aforementioned heterogeneous risk indicators to a unified runway geographic coordinate system, generating a layered, overlaid risk weight map. Based on this, targeted maintenance strategy instructions are output, achieving a closed loop from data detection to engineering decision-making.

[0081] S5-1: Spatial mapping of multi-dimensional risk layers; To visually represent the health status of the runway, a two-dimensional grid map based on the actual physical dimensions of the runway was created. The grid cell size was defined as follows: .

[0082] Micro-risk points and its corresponding Projected onto a mesh, generating a micro-defect layer. : in, Indicates the first The grid cell in row col; if there are no defects in the cell, assign a value of 0.

[0083] The macroscopic slice data is extended along the runway width direction to generate a macroscopic curvature layer. and snow accumulation anomaly layer : in, This refers to the slice index corresponding to the grid column coordinate *col*; through normalization, the physical quantity is mapped to... The relative risk value.

[0084] Next, based on the polar ice and snow runway quality assessment method disclosed in the above embodiments, this embodiment further specifies the discretization strategy for data acquisition and the fusion assessment algorithm for multi-source parameters. This embodiment uses a combination of UAV scanning and ground vehicle contact sampling to achieve quantitative assessment of runway quality.

[0085] In this preferred embodiment, the evaluation method is specifically implemented through the following steps: Gridded mapping of runway assessment units: Considering the spatial variability of polar runway surface characteristics, this system establishes a runway coordinate system. The total length of the runway to be evaluated is set as... (meters), to achieve a balance between accuracy and efficiency, the runway is divided longitudinally into... There are 1 standard evaluation segment, and the length of each evaluation segment is set to 1. Rice, that is .

[0086] For any i One evaluation segment, in which Four key parameters were collected using a combined air-ground approach: non-contact full-coverage sampling of flatness and contaminants, and contact discrete sampling of intensity and friction coefficient. Specifically, drones equipped with lidar and multispectral cameras are used to obtain the first... Imagery and digital elevation model (DEM) of each assessment segment. Based on the imagery, pollutants within the segment are identified, namely blown snow accumulation and oil spills, to obtain pollutant distribution data. Based on the DEM, microscopic undulations of the runway surface are extracted to obtain smoothness data.

[0087] Furthermore, regarding the contact-based discrete sampling intensity and friction coefficient: using a ground-based detection vehicle in the first... Contact measurements were conducted within each assessment section. Based on a sampling standard of 5 points per 100 meters, measurements were taken in the [number]th [section / section]. Five feature points are selected within each evaluation segment.

[0088] The structural strength of the ice / snow layer at these five points was measured using a vehicle-mounted penetrator. The surface friction coefficient at these five points was measured using a vehicle-mounted friction test wheel.

[0089] Preprocessing methods for collected data: To eliminate the inconsistencies in scale among different physical quantities, the raw data collected needs to be converted into a normalized mass index. In this embodiment, for the first... Each evaluation segment defines an evaluation function with four dimensions: Smoothness Quality Index The smoothness of polar ice and snow runways is mainly affected by uneven ice melting or snow accumulation. This embodiment uses elevation data collected by a drone to calculate the root mean square height deviation. Let the first... The total number of valid elevation sampling points within each assessment segment is , No. The elevation value of each sampling point is The height of the reference fitting surface for this segment is .

[0090] Calculate the root mean square deviation of the flatness of this section. : Define the flatness design tolerance threshold as Then the first Smoothness quality index of each assessment section The structure is as follows: The closer to 1, the smoother the surface; if the deviation exceeds the threshold, the exponent returns to zero.

[0091] Pollutant Impact Index : Calculation based on UAV imagery Pollutant coverage area within each assessment segment The total area of ​​this section of the track is known to be .

[0092] Calculate pollutant coverage : Define the maximum allowable pollutant coverage threshold as follows: Pollutant Impact Index The structure is as follows: Runway strength quality index model The strength data comes from five discrete measuring points on a ground vehicle. Let the first... Within the evaluation segment, the first One measuring point ( The measured strength value is (Unit: MPa)

[0093] To ensure safe takeoff and landing and avoid accidents caused by localized weak layers, this embodiment uses a weighted method to calculate the representative strength of the segment. The minimum value is given a higher weight. The minimum value weight coefficient is set to... : The required rated strength standard for this runway service aircraft model is defined as follows: Strength-mass index : Friction coefficient mass index model The friction coefficient also originates from 5 discrete measurement points. Let the first... Within the evaluation segment, the first The measured friction coefficient at each measuring point is: Calculate the average friction coefficient for this section. : The lower limit of the safe coefficient of friction for ice and snow running tracks is defined as follows: Friction coefficient and mass index The structure is as follows: in, For an excellent buffer zone width, when the measured value is higher than A score of 100 is considered full marks.

[0094] S5-2 Multidimensional Factors Runway Assessment: This step uses a logical threshold determination algorithm for the first... Each evaluation segment has an input variable consisting of the calculated set of indices. .

[0095] These four indices are divided into two categories: rigid safety index ( ) and Flexible Service Index ( Based on this, the following three-level decision logic is constructed: Level 1: Airworthiness Interruption Judgment Based on Rigid Safety Index: The rigidity index indicates whether the runway meets the most basic physical support and braking requirements. When or This means that the measured physical values ​​did not meet the design standards. or safety bottom line .

[0096] Therefore, a runway blocking state determination function is defined. : like If TRUE, then directly determine the first... Runway condition level This constitutes a serious defect in the quality inspection, and subsequent assessments will be terminated.

[0097] Level 2: Operational Restriction Detection If the first level of judgment is passed (i.e.) A result of FALSE indicates that the runway is safe. At this point, the index needs to be checked. and These two indices are defined as follows: arrive The value is a continuous value, and the lower the value, the further away from the ideal state.

[0098] Set service maintenance threshold Define a restricted decision function : like If TRUE, then determine This refers to the defective part. If... If the level is too low, mechanical leveling is recommended; if The reading is low, indicating a need to clean up contaminants.

[0099] Normal operation determination: The runway is in ideal condition only when both the initial operation restriction determination and the airworthiness disruption determination are higher than the service maintenance threshold.

[0100] Next, in a further preferred embodiment of the present invention, the profound impact of harsh polar weather and complex ice and snow physical properties on runway quality is considered. Based on the above evaluation model, this embodiment further introduces dynamic wind field parameters, snow temperature parameters, and snow layer structure parameters to perform multi-dimensional and in-depth calculations on the evaluation logic.

[0101] In polar environments, the hazards of runway lateral slope are not static but highly correlated with strong crosswinds. Furthermore, the load-bearing capacity of compacted snow runways is extremely sensitive to temperature and is prone to forming a dangerous layered structure with a hard top and soft bottom under frequent compaction. Therefore, this embodiment constructs an enhanced assessment mechanism combining meteorological and physical properties, specifically implemented as follows: Step 1: Correction of the lateral snow retention anomaly index based on dynamic wind field: Considering the strong vector characteristics of windblown snow migration during strong winds, a single lateral geometric slope is insufficient to determine the degree of snow accumulation. This embodiment incorporates wind field vectors acquired in real-time from external weather stations, including horizontal wind speed. Angle with wind direction ,definition Wind direction and runway lateral The counterclockwise angle of the axis.

[0102] Calculate the effective crosswind component of the wind field across the runway. : When crosswinds sweep across the runway surface, they accelerate on the windward slope and, on the leeward slope (or reverse slope), cause airflow separation, resulting in the deposition of large amounts of ice and snow particles. Based on this, the wind-blown snow retention potential energy density considering the dynamic wind field is redefined. : Among them, indicator function For dynamic determination of the windward and leeward sides. When the effective crosswind... With local lateral gradient When the direction is opposite, it indicates that the airflow is obstructed, and it is identified as a high-risk snow accumulation area. This indicates that the stronger the crosswind and the steeper the blocking slope, the faster the instantaneous snow accumulation rate.

[0103] The modified enhanced transverse snow retention anomaly index was obtained through spatial integration. and use this to replace the original Participate in the generation of subsequent risk space mapping layers.

[0104] Step 2: Reconstructing the effective structural bearing capacity index based on snow temperature and snow layer structure variance: Polar ice and snow are not rigid bodies; the strength of the ice crystal bonds within them decreases exponentially with increasing temperature. Meanwhile, the impact of heavy aircraft wheels easily compacts the runway surface, while the subsurface becomes porous due to the decrease in stress, forming a highly concealed, weak underlying layer. Once an aircraft breaches this hard outer shell, it will trigger catastrophic impact and shear failure.

[0105] In this step, obtain the first The measured average snow temperature within each assessment segment The structural critical phase transition temperature of the polar ice and snow runway is set as follows: Preferably -5℃, defining the snow softening coefficient. : in, This is the temperature sensitivity gain parameter for the ice and snow medium. When the snow temperature is much lower than the critical temperature... Approaching 1 (no intensity decay); when snow temperature approaches or exceeds the critical temperature, A rapid decline indicates the risk of loss of load-bearing capacity due to thermal melting.

[0106] Furthermore, regarding the vehicle-mounted penetration instrument in the first... The depth structure data obtained from each measuring point no longer only takes the lowest intensity, but extracts its intensity along the depth. Structural strength profile Define the surface reference depth. With deep reference depth Calculate the structural stratification gradient at the measuring point. : when This indicates that the surface layer is stronger than the deeper layers, demonstrating a mechanical reversal phenomenon of hard surface and soft underlying structure. Based on this, the most dangerous structural abrupt change features within the assessment section are extracted, and a weak underlying layer penalty factor is constructed. : The original rigid safety index: runway strength quality index After correction, the effective structural bearing capacity index is obtained. : in, This is the preset structural abrupt change attenuation coefficient. The corrected... It combines the snow temperature weakening effect under the current meteorological environment with the mechanical attenuation induced by cracks / loosening in the internal cover layer.

[0107] Step 3: Update the adaptive decision logic for multi-dimensional fusion: The revised effective structural load-bearing index Replace the original three-level decision logic Participating in Level 1: Airworthiness disruption determination based on rigid safety index.

[0108] When encountering extreme warming weather or when deep snowbeds undergo dark melting and degradation, even the initial intensity of surface contact sampling points... Barely meets the standard, due to the snow softening coefficient. Intervention and lower-level punishment factor The exponential amplification effect, effective structural load-bearing capacity index It will quickly fall below the threshold of 1, thereby immediately triggering the blocking state determination function. This significantly advances the warning time for the hidden thermal structural degradation of polar runways, effectively preventing safety accidents caused by the sudden collapse of the deep bearing capacity of the runway for heavy transport aircraft, and truly achieving a leap from the perception of superficial defects to the prediction of deep mechanisms.

[0109] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the above-described method for assessing the quality of polar ice and snow runways by multi-source information fusion.

[0110] The present invention also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the above-described method for assessing the quality of polar ice and snow runways through multi-source information fusion.

[0111] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0112] In this specification, the same or similar parts between the various embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the descriptions of the embodiments described later are relatively simple, and relevant parts can be referred to the descriptions of the foregoing embodiments.

[0113] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for quality assessment of polar ice and snow runways through multi-source information fusion, characterized in that, The evaluation method includes: generating panoramic orthophotos based on the acquired sequence of images, defining the physical boundaries of the runway and dividing it into an edge reference zone and a central load-bearing evaluation zone; By extracting the frequency domain fingerprint of the healthy running track using the edge benchmark area, the central load assessment area is scanned in the frequency domain to identify the set of key quality assessment areas. Local point clouds are extracted based on the coordinate index of the key quality assessment area set, a local reference plane is constructed and the maximum physical depression depth is calculated. Logical verification is performed by combining the texture degradation index and the comprehensive hazard index at the micro level is calculated. The theoretical baseline of the transverse slice is reconstructed using the edge benchmark area, the macroscopic settlement field of the central bearing assessment area is calculated, and the bending risk and transverse snow accumulation anomaly index are quantified based on the macroscopic settlement field. The comprehensive hazard index, bending risk, and lateral snow accumulation anomaly index are mapped to a unified runway geographic coordinate system to generate a risk weight map and output maintenance strategies.

2. The method for assessing the quality of polar ice and snow runways by multi-source information fusion according to claim 1, characterized in that, Generate panoramic orthophotos based on the acquired sequence of images, including: extracting the frequency domain image of the overlapping region between two adjacent corrected frames. and Calculate the normalized cross-power spectrum : in, For frequency domain coordinates, express The complex conjugate; Performing an inverse discrete Fourier transform on the normalized cross-power spectrum yields the phase correlation function matrix. And select the peak coordinates of the phase correlation function matrix. A quadratic surface is fitted to the region and its neighboring pixels to obtain an accurate translation vector. : in, This represents the probability distribution of overlap between two images under different relative displacements.

3. The method for assessing the quality of polar ice and snow runways by multi-source information fusion according to claim 2, characterized in that, Generating panoramic orthophotos also includes: calculating the first Global coordinate offset of frame image in panoramic coordinate system : in, For the first Frame and the Precise translation vectors between frames; Constructing panoramic orthophotos based on global coordinate offsets : in, For the coordinates of the panoramic image, For the set of image frames covering this coordinate, For the weight function, For the corrected first Frame image, for The amount.

4. The method for assessing the quality of polar ice and snow runways by multi-source information fusion according to claim 1, characterized in that, Defining the physical boundaries of the runway and dividing it into an edge reference zone and a center load-bearing assessment zone includes: defining the lateral gradient operator. Calculate panoramic orthophotos column-to-column cumulative gradient energy spectrum : in, The length of the runway panoramic view. Image coordinates; Based on the column-accumulated gradient energy spectrum, the coordinates of the left and right physical boundaries of the runway are identified. and : in, An adaptive threshold; based on and The runway space is divided into the left edge reference zone. Right edge reference area and the central carrying capacity assessment area .

5. The method for assessing the quality of polar ice and snow runways by multi-source information fusion according to claim 1, characterized in that, A frequency domain scan of the central carrying capacity assessment area was performed, including: Extracting the first using a sliding window line, number Local sub-images of a column And introduce a two-dimensional Hanning window function. Weighted processing is performed to obtain image patches: in, Step size, Local coordinates; Performing a two-dimensional discrete Fourier transform on the image patch yields the complex spectrum. And calculate the logarithmic power spectral density. : in, These are frequency domain coordinates.

6. The method for assessing the quality of polar ice and snow runways by multi-source information fusion according to claim 5, characterized in that, Extract the frequency domain fingerprint of the healthy running track and identify key quality assessment areas, including: Calculate the average power spectrum of the edge reference region as the reference spectrum. Define the set of groove signal frequencies As a frequency domain fingerprint of a healthy running track: in, The relative intensity threshold coefficient is used to calculate the texture degradation index of each sub-block in the central load-bearing evaluation area. : in, For small stability constants; select those that satisfy The regions form a set of key quality assessment areas, among which This is the threshold for damage assessment.

7. The method for assessing the quality of polar ice and snow runways by multi-source information fusion according to claim 1, characterized in that, Constructing a local reference plane and calculating the maximum physical indentation depth includes: extracting the center coordinates of the key quality assessment area. Centered on, with radius Local point cloud collection ;right The equation of the reference plane is obtained by performing least-squares plane fitting. Calculation points relative depth residual : Extracting the maximum physical depression depth from a local point cloud set : .

8. The method for assessing the quality of polar ice and snow runways by multi-source information fusion according to claim 7, characterized in that, By combining the texture degradation index with logical cross-verification, a comprehensive hazard index at the microscopic level is calculated, including: Mapping the maximum physical indentation depth to a geometric risk probability using the Sigmoid function and the texture degradation index Mapped to texture risk probability : in, This represents the critical threshold for the depth of the depression. This is the steepness coefficient. Set a saturation threshold; construct a state conflict discrimination function. To identify State II thermal fusion settlement and calculate the comprehensive hazard index. : in, The geometric weight bias coefficient. As an additional penalty factor for thermal fusion settlement, This is an indicator function.

9. A method for assessing the quality of polar ice and snow runways based on multi-source information fusion according to claim 1, characterized in that, The theoretical baseline of the transverse slice is reconstructed using the edge benchmark region. The macroscopic subsidence field is calculated, and the bending risk and transverse snow accumulation anomaly index are quantified, including using the Huber loss function. For point sets belonging to the edge reference region Perform regression to obtain the first... Theoretical baseline parameters for each slice : Constructing a macroscopic settling field And calculate the transverse curvature based on its second-order differential. and bending risk : Define local gradient and the potential energy density of windblown snow accumulation Calculate the transverse snow cover anomaly index : in, This is an indicator function.

10. A multi-source information fusion system for assessing the quality of polar ice and snow runways, characterized in that, The system includes: Region division module: Generate panoramic orthophotos based on the acquired sequence of images, define the physical boundaries of the runway and divide it into an edge reference area and a central load-bearing assessment area; use the edge reference area to extract the frequency domain fingerprint of the healthy runway, perform frequency domain scanning on the central load-bearing assessment area, and lock in the set of key quality assessment areas; Local hazard identification module: Extract local point cloud based on coordinate index of key quality assessment area set, construct local reference plane and calculate maximum physical depression depth, combine with texture degradation index for logical verification, and calculate comprehensive hazard index at the micro level; Global Hazard Identification Module: Reconstructs the theoretical baseline of the transverse slice using the edge benchmark area, calculates the macroscopic settlement field of the central bearing capacity assessment area, and quantifies the bending risk and transverse snow accumulation anomaly index based on the macroscopic settlement field; Maintenance module: Maps the comprehensive hazard index, bending risk and lateral snow accumulation anomaly index to a unified runway geographic coordinate system, generates a risk weight map and outputs maintenance strategies.

Images