Airport runway lighting anomaly detection and positioning method based on aerial video analysis

Abstract

The present invention discloses a method for detecting and locating airport runway light anomalies based on aerial video analysis, characterized in that the method comprises the following steps: S1, inputting a runway light sample dataset into a YOLOV8 target detector for target detection training; S2, sequentially capturing video images via an unmanned aerial vehicle (UAV) and inputting them into a YOLOV8 target detector for target detection, and using the ByteTrack algorithm to associate target data and obtain a target tracking trajectory set; S3, extracting the coordinates of target trajectory points in the target tracking trajectory set, performing motion linear equation fitting, and obtaining the linear slope and Y-intercept; S4, performing unsupervised clustering of the target trajectory points according to motion characteristics using a Gaussian mixture model to obtain a left sideline light target trajectory set, a middle line light target trajectory set, and a right sideline light target trajectory set; and performing runway light anomaly determination and outputting target and location prompts. The present invention can rapidly detect lighting anomalies of runway centerline lights and sideline lights, providing technical support for airport security.

Description

Technical Field

[0001] The present invention relates to the field of airport runway operation and maintenance, and in particular to a method for detecting and locating anomalies of airport runway lights based on aerial video analysis. Background Art

[0002] In modern aviation, ensuring airport operational safety is crucial. Safety testing of airport runway lighting systems, particularly runway lighting, is crucial for ensuring safe takeoff and landing. Typically, runway lights are located on both sides and in the middle of an airport's straightaways, guiding aircraft for proper takeoff and landing.

[0003] Traditional methods for detecting anomalies in airport runway lighting rely primarily on manual inspections, requiring personnel to walk along the runway at specific intervals to check whether the lights are functioning properly. This approach is not only labor-intensive and time-consuming, but also increases the difficulty and risk of manual inspections at night or in inclement weather. With technological advancements, automated detection methods based on sensors and monitoring devices are gaining adoption. For example, photoelectric sensors are used to detect anomalies in light brightness and color. However, the installation and maintenance costs of these hardware devices, such as sensors, are high, and a single sensor can only monitor a limited area. Comprehensive coverage of all airport runway lights requires a large number of sensors, increasing system complexity and cost.

[0004] With the development of aerial photography technology, it has gradually been applied in the field of civil aviation safety inspection. However, in terms of airport runway lighting detection, the method of using aerial video to analyze lighting anomalies is still in a technical blank. Even if runway light images are obtained through aerial photography, there is no relevant technology in image processing and detection and recognition processing, and it is impossible to accurately identify runway lights (especially under complex lighting conditions and background environments, it is difficult to accurately distinguish normal lights from abnormal lights, and it is even more difficult to accurately locate the position of abnormal lights). Therefore, aerial photography detection of airport runway lights is the current direction of technological development. How to achieve technical issues such as target recognition, target tracking detection, abnormal light detection and abnormal light positioning of airport runway lights is a technical difficulty in current research and development, which also limits the technical expansion of drone aerial photography technology in the field of civil aviation runway inspection. Summary of the Invention

[0005] The purpose of the present invention is to solve the technical problems pointed out by the background technology and provide a method for detecting and locating airport runway light anomalies based on aerial video analysis. Based on the runway light trajectory analysis of the aerial video, the motion trajectory is fitted and the light trajectory with similar motion characteristics is calculated. At the same time, GMM cluster analysis is performed to obtain the light sets of the left line, middle line and right line of the runway respectively. The trajectory information of each set is further analyzed (light count, spacing determination, brightness determination), abnormal runway lights are identified and their precise positions are located.

[0006] The purpose of the present invention is achieved through the following technical solutions:

[0007] A method for detecting and locating anomalies of airport runway lights based on aerial video analysis, the method comprising:

[0008] S1. Build a runway light sample dataset and input the runway light sample dataset into the YOLOV8 target detector for target detection training;

[0009] S2: Use the drone to take video images of the airport runway lighting area in sequence with the runway centerline as the flight reference direction, input them into the YOLOV8 target detector for target detection, and use the ByteTrack algorithm to associate target data and obtain a set of target tracking trajectories;

[0010] S3, constructing a coordinate system based on the horizontal direction of the video image as the X-axis and the vertical direction as the Y-axis, extracting the coordinates of the target trajectory points in the target tracking trajectory set, fitting the motion linear equation and obtaining the slope of the line and the Y-intercept, and using the slope of the line and the Y-intercept as motion features;

[0011] S4. Use a Gaussian mixture model to perform unsupervised clustering of the target trajectory points according to their motion characteristics and divide them into four categories: left sidelights, middle sidelights, right sidelights, and other lights. Eliminate the set of candidate categories of other lights to obtain the left sidelight target trajectory set, the middle sidelight target trajectory set, and the right sidelight target trajectory set;

[0012] S5. Calculate the distance between two adjacent runway lights in the video frame for the left runway light target track set, the middle runway light target track set, and the right runway light target track set. If the distance is greater than a light spacing threshold, then there is a completely unlit runway light between the two adjacent runway lights. Mark the position between the two adjacent runway lights as unlit and output a corresponding value.

[0013] S6. Perform historical brightness average statistics on the left side light target trajectory set, the middle side light target trajectory set, and the right side light target trajectory set according to the runway light target. If the historical brightness average is less than the low brightness average threshold, it is determined that the runway light target brightness is too low and a corresponding low brightness prompt is output; if the historical brightness average is greater than the high brightness average threshold, it is determined that the runway light target brightness is too high and a corresponding high brightness prompt is output.

[0014] In order to better implement the present invention, in step S5, the video frame position where the position mark between two adjacent runway lights is not illuminated is firstly extracted and the position mark is not illuminated output; in step S6, the runway light target where the target brightness is too low and the video frame position where the target brightness is too high is firstly extracted and the brightness is too high prompt output.

[0015] Preferably, the runway light video images of the runway light sample data set are shot by a drone with the runway centerline as the flight reference direction, and the runway light video images are annotated with runway light labels.

[0016] Preferably, the YOLOV8 target detector of the present invention performs runway light target detection and labeling on the video image stream F(t) frame by frame, where t represents the sequential number of the video frame, and the ByteTrack algorithm performs data association and target tracking based on the runway lights in the previous and next frames.

[0017] Preferably, the linear motion equation corresponding to the coordinates (x, y) of the target trajectory point m is fitted as follows: y=a m x+b m , a m is the slope of the straight line, b m is the Y-axis intercept; the sum of squared errors from the trajectory point to the fitting line is set as the optimization objective function E(i), x m represents the x-axis coordinate after fitting the linear equation of motion, y represents the y-axis coordinate before fitting, and M represents the total number of trajectory points associated with the target data.

[0018] Preferably, the least squares method is used to solve the optimal fitting parameter a for the optimization objective function. m , excellent and b m，优 b m，优 =y 均 -a m，优 ×x 均 , where x 均 Indicates the mean value of the horizontal coordinate of the target trajectory point, y 均 Indicates the mean value of the vertical coordinate of the target trajectory point.

[0019] Preferably, the Gaussian mixture model of the present invention includes the following methods in the unsupervised clustering process:

[0020] S41. The motion features of the target trajectory points are used as two-dimensional data points in the clustering process. The Gaussian mixture model determines the probability density function of each data point S(i) as follows:

[0021] Where K is the number of Gaussian components, π k is the weight parameter of the kth Gaussian component, N(S(i)|μ k ,∑ k ) is the k Gaussian distribution probability density function of the data point S(i), μ k is the mean vector parameter of the Gaussian distribution, ∑ k is the covariance matrix parameter of the Gaussian distribution, and θ is the vector parameter composed of all parameters;

[0022] S42, iterative optimization, for each data point S(i), calculate its posterior probability generated by each Gaussian component:

[0023] r ik is the probability that the data point S(i) belongs to the kth Gaussian component, z i is the latent variable corresponding to the data point S(i);

[0024] S43. Re-estimate the mean, covariance matrix, and mixing coefficient of each Gaussian component to maximize the likelihood function of the entire model. The specific update process is:

[0025]

[0026] Where M represents the total number of trajectory points associated with the target data;

[0027] S44, looping through the above steps until the parameters of the Gaussian mixture model converge or the number of loops reaches a preset threshold;

[0028] S45. Cluster the data points based on the optimal parameters of the Gaussian mixture model, and cluster the data points S(i) into corresponding categories.

[0029] Preferably, the parameter converges to the parameter μ k Whether the mean of the changes before and after the update is less than the threshold thr_mu, the preset threshold set by the number of cycles is the threshold thr_loop.

[0030] Preferably, the Gaussian mixture model calculates the mean slope of the data points contained in each cluster set during clustering processing, and takes the category set with the largest absolute value of the mean slope as the center line light set Z_mid, the category set with a mean slope greater than 0 as the left line light set Z_left, and the category set with a mean slope less than 0 as the right line light set Z_right.

[0031] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0032] (1) The present invention analyzes the runway light trajectory based on aerial video, fits the motion trajectory and calculates the light trajectory with similar motion characteristics. At the same time, GMM cluster analysis is performed to obtain the light sets of the left line, middle line, and right line of the runway respectively. The trajectory information of each set (light count, spacing determination, brightness determination) is further analyzed to identify abnormal runway lights and locate their precise positions.

[0033] (2) The present invention realizes the accurate identification and precise positioning of abnormal lights on the left, middle and right lines of the airport runway through the analysis of light trajectories in aerial videos. The present invention relies on a drone to conduct regular runway light inspections, which has the characteristics of low cost, fast deployment, high flexibility and strong scalability, which is conducive to the large-scale promotion of the method and effectively improves the digital level of the airport.

[0034] (3) The present invention can effectively identify abnormal lighting problems (abnormal light brightness, no light) existing in the centerline lights and sideline lights (including left sideline lights and right sideline lights) of the airport runway in aerial videos, effectively saving the cost of manpower inspection and hardware deployment, and providing strong technical support for airport flight safety. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 This is a flow chart of the method for detecting and locating abnormalities in airport runway lighting according to the present invention;

[0036] Figure 2 This is a schematic diagram of runway light recognition at frame 4989 in an embodiment;

[0037] Figure 3 Schematic diagram of runway lighting recognition at frame 5700 in an embodiment. DETAILED DESCRIPTION

[0038] Below in conjunction with embodiment, the present invention is described in further detail:

[0039] Example

[0040] like Figure 1 As shown, a method for detecting and locating anomalies of airport runway lights based on aerial video analysis includes:

[0041] S1. Construct a runway light sample dataset, and input the runway light sample dataset into the YOLOV8 target detector for target detection training. The runway light video images of the runway light sample dataset are taken by a drone (the aerial photography route height is preferably 20 meters) with the runway centerline as the flight reference direction (the flight direction is from the start of the runway to the end of the runway at a constant speed), and the runway light video images are processed by runway light labeling. The runway light video images of the runway light sample dataset need to cover various weather scenes and different light states of all light groups (including too dark and too bright states based on brightness grading). When processing the runway light label, a further preferred technical solution is: all normal runway light areas are marked as light_normal, runway light areas with too low brightness are marked as light_low, and runway light areas with too high brightness are marked as light_high. The brightness is too low and too high respectively. Subsequently, a preliminary identification and judgment of the brightness being too low and too high can be performed first, and then the historical brightness average value is compared with the threshold.

[0042] S2. Use a drone to sequentially capture video images of the airport runway lighting area with the runway centerline as the flight reference direction (the flight direction is from the beginning of the runway to the end of the runway at a constant speed, and the aerial photography route height is preferably 20 meters; the video image is captured at an image frame interval of 1 second). Input the image into the YOLOV8 target detector for target detection and use the ByteTrack algorithm for target data association to obtain a target tracking trajectory set.

[0043] YOLOV8 target detector performs frame-by-frame runway light target detection on the video image stream F(t) (see Figure 2 , for example, the target recognition and target number recognition results of a frame of video image) and labeling, t represents the sequential number of the video frame, the ByteTrack algorithm performs data association and target tracking based on the runway lights in the previous and next frames, the YOLOV8 target detector performs target recognition in the previous and next frames, the ByteTrack algorithm associates the same runway light target, and the YOLOV8 target detector and the ByteTrack algorithm jointly realize target tracking. Figure 3 As shown, the YOLOV8 target detector in this embodiment detects, identifies and numbers runway light targets frame by frame. The YOLOV8 target detector gives a number to a newly appeared target, and the subsequent same target is numbered with the same target. The ByteTrack algorithm associates the same runway light target.

[0044] S3. Construct a coordinate system based on the horizontal X-axis and the vertical Y-axis of the video image. Extract the coordinates of the target trajectory points in the target tracking trajectory set, fit the motion linear equation, and obtain the slope and Y-intercept of the line. Use the slope and Y-intercept of the line as motion features. Runway lights include left-side, middle-side, and right-side lights. In the drone video image, they show historical regularity. Taking the left-side lights as an example, each left-side light is on the left side of the video image. The historical trajectory of the same left-side light target is a straight line located on the left side of the video image. Therefore, the target historical trajectory linear features of the runway light target can be used for feature extraction and runway light type identification. Due to the influence of other light sources (vehicles, warning lights, light sources from distant buildings and towers, other types of navigation lights, etc.) on both sides of the airport runway, these lights will be identified and marked as targets. The historical trajectories of these targets (defined as other lights) are substantially different from the runway light features and need to be excluded first. By analyzing a large number of aerial videos of runway lights, it is creatively discovered that the movement direction (slope) and movement position (offset) of the runway lights located on the left line, middle line, and right line are highly similar in image. Therefore, the present invention creatively uses motion features for identification and other light exclusion processing.

[0045] In some embodiments, the method for fitting the linear equation of motion of the target trajectory point m is as follows: the linear equation of motion corresponding to the coordinates (x, y) of the target trajectory point m is fitted as follows: y=a m x+b m , a m is the slope of the straight line, b m is the Y-axis intercept; the sum of squared errors from the trajectory point to the fitting line is set as the optimization objective function E(i), x m represents the X-axis coordinate after the motion linear equation is fitted (a coordinate system is constructed based on the horizontal direction of the video image as the X-axis and the vertical direction as the Y-axis, and the X-axis coordinate is the X-axis coordinate data of the coordinate system), y is the y-axis coordinate before fitting, and M represents the total number of trajectory points associated with the target data. The least squares method is used to solve the optimal fitting parameter a for the optimization objective function. m , excellent and b m ,excellent, b m，优 =y 均 -a m，优 ×x 均 , where x 均 Indicates the mean value of the horizontal coordinate of the target trajectory point, y 均 It represents the mean value of the vertical coordinate of the target trajectory point. From this, we can get the a of the target at each trajectory point. m 、b m , and the optimal a for the same target after fitting the historical trajectory points m，优 and b m，优, the runway light targets show regularity, and the sum of squared errors is obviously small (preferably, a threshold can be set), and a m，优 and b m，优 (Other lamps that cannot be fitted can be excluded directly), and other lamps can be excluded accordingly.

[0046] S4. Use the Gaussian mixture model to perform unsupervised clustering on the target trajectory points according to the motion characteristics and divide them into four categories, namely left sideline lights, middle sideline lights, right sideline lights, and other lights. Eliminate the set of other light candidate categories to obtain the left sideline light target trajectory set, middle sideline light target trajectory set, and right sideline light target trajectory set.

[0047] In some embodiments, the Gaussian mixture model includes the following methods in the unsupervised clustering process:

[0048] S41. The motion features of the target trajectory points are used as two-dimensional data points in the clustering process. The Gaussian mixture model determines the probability density function of each data point S(i) as follows:

[0049] Where K is the number of Gaussian components (set to 4 in this embodiment), π k is the weight parameter of the kth Gaussian component (satisfying And π k ≥0), N(S(i)|μ k ,∑ k ) is the k Gaussian distribution probability density function of the data point S(i), μ k is the mean vector parameter of the Gaussian distribution, ∑ k is the covariance matrix parameter of the Gaussian distribution, and θ is the vector parameter composed of all parameters. At the beginning of the Gaussian mixture model processing, the above parameters are randomly initialized.

[0050] S42, iterative optimization, for each data point S(i), calculate its posterior probability generated by each Gaussian component:

[0051] r ik is the probability that the data point S(i) belongs to the kth Gaussian component, z i is the latent variable corresponding to the data point S(i);

[0052] S43. Re-estimate the mean, covariance matrix, and mixing coefficient of each Gaussian component to maximize the likelihood function of the entire model. The specific update process is:

[0053]

[0054] Where M represents the total number of trajectory points associated with the target data.

[0055] S44, repeat the above steps until the parameters of the Gaussian mixture model converge or the number of cycles reaches a preset threshold. The parameter convergence is parameter μ k Whether the mean of the changes before and after the update is less than the threshold thr_m (set to 0.01 in this embodiment as an example), the preset threshold set by the number of loops is the threshold thr_loop (set to 500 in this embodiment as an example).

[0056] S45. Cluster the data points based on the optimal parameters of the Gaussian mixture model, and cluster the data points S(i) into corresponding categories. After multiple iterations until convergence, the data points are assigned to the Gaussian component with the largest posterior probability according to the posterior probability of each data point belonging to each Gaussian component, and clustering is completed. The number of lights in the left line, middle line, and right line categories is much larger than the number of lights in other positions (obviously, the number of lights in the left line, middle line, and right line categories is the largest three categories). Therefore, the set containing significantly fewer data points is determined to be the set where the other lights are located, and the set is deleted (the deletion of other lights is further performed here); in this embodiment, after the Gaussian mixture model clustering is completed, all cluster sets are sorted from high to low according to the data points, the top three cluster sets are retained (as category sets), and the remaining cluster sets are all deleted. In some embodiments, the Gaussian mixture model calculates the mean slope of the data points contained in each cluster set during clustering processing, and takes the category set with the largest absolute value of the mean slope as the center line light set Z_mid, the category set with a mean slope greater than 0 as the left line light set Z_left, and the category set with a mean slope less than 0 as the right line light set Z_right.

[0057] S5. Calculate the distance between two adjacent runway lights in the video frame for each of the left runway light target trajectory set, the middle runway light target trajectory set, and the right runway light target trajectory set. If the distance is greater than a light spacing threshold, a completely unlit runway light exists between the two adjacent runway lights. A mark indicating that the position between the two adjacent runway lights is unlit is output accordingly (thereby locating and identifying the target position of the unlit runway light). In some embodiments, the video frame position where the unlit mark first appears between the two adjacent runway lights is extracted and the unlit position is output.

[0058] S6. The historical brightness average values of the left sideline light target trajectory set, the middle sideline light target trajectory set, and the right sideline light target trajectory set are statistically analyzed according to the runway light targets. If the historical brightness average value is less than the low brightness average threshold value thr_low, the runway light target brightness is determined to be too low and a corresponding low brightness prompt is output. If the historical brightness average value is greater than the high brightness average threshold value thr_high, the runway light target brightness is determined to be too high and a corresponding high brightness prompt is output. In some embodiments, the runway light target at the first video frame position where the target brightness is too low is extracted and a low brightness prompt is output (thereby locating and identifying the runway light target position with too low brightness), and the runway light target at the first video frame position where the target brightness is too high is extracted and a high brightness prompt is output (thereby locating and identifying the runway light target position with too high brightness). The present invention can effectively identify the lighting abnormalities (abnormal lighting brightness, no lighting) of the centerline lights and sideline lights (including the left sideline lights and the right sideline lights) of the airport runway in the aerial video, effectively saving the cost of manpower inspection and hardware deployment, and providing strong technical support for airport flight safety.

[0059] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for detecting and locating anomalies in airport runway lighting based on aerial video analysis, characterized by: The methods include: S1. Build a runway light sample dataset and input the runway light sample dataset into the YOLOV8 target detector for target detection training; S2: Use the drone to take video images of the airport runway lighting area in sequence with the runway centerline as the flight reference direction, input them into the YOLOV8 target detector for target detection, and use the ByteTrack algorithm to associate target data and obtain a set of target tracking trajectories; S3, constructing a coordinate system based on the horizontal direction of the video image as the X-axis and the vertical direction as the Y-axis, extracting the coordinates of the target trajectory points in the target tracking trajectory set, fitting the motion linear equation and obtaining the slope of the line and the Y-intercept, and using the slope of the line and the Y-intercept as motion features; S4. Use a Gaussian mixture model to perform unsupervised clustering of the target trajectory points according to their motion characteristics and divide them into four categories: left sidelights, middle sidelights, right sidelights, and other lights. Eliminate the set of candidate categories of other lights to obtain the left sidelight target trajectory set, the middle sidelight target trajectory set, and the right sidelight target trajectory set; S5. Calculate the distance between two adjacent runway lights in the video frame for the left runway light target track set, the middle runway light target track set, and the right runway light target track set. If the distance is greater than a light spacing threshold, then there is a completely unlit runway light between the two adjacent runway lights. Mark the position between the two adjacent runway lights as unlit and output a corresponding value. S6. Perform historical brightness average statistics on the left side light target trajectory set, the middle side light target trajectory set, and the right side light target trajectory set according to the runway light target. If the historical brightness average is less than the low brightness average threshold, it is determined that the runway light target brightness is too low and a corresponding low brightness prompt is output; if the historical brightness average is greater than the high brightness average threshold, it is determined that the runway light target brightness is too high and a corresponding high brightness prompt is output.

2. The method for detecting and locating anomalies of airport runway lighting based on aerial video analysis according to claim 1 is characterized by: In step S5, the video frame position where the position mark between two adjacent runway lights is not illuminated is firstly extracted and the position mark is not illuminated output; in step S6, the runway light target where the target brightness is too low and the video frame position where the target brightness is too high is extracted and the video frame position where the target brightness is too high is output.

3. The method for detecting and locating anomalies of airport runway lighting based on aerial video analysis according to claim 1 is characterized in that: The runway light video images in the runway light sample data set are shot by a drone with the runway centerline as the flight reference direction, and the runway light video images are annotated with runway light labels.

4. The method for detecting and locating anomalies of airport runway lighting based on aerial video analysis according to claim 1 is characterized by: The YOLOV8 target detector detects and labels the runway light targets frame by frame in the video image stream F(t), where t represents the sequential number of the video frame. The ByteTrack algorithm performs data association and target tracking for the runway lights based on the targets in the previous and next frames.

5. The method for detecting and locating anomalies of airport runway lighting based on aerial video analysis according to claim 1 is characterized in that: The linear equation of motion corresponding to the coordinates (x, y) of the target trajectory point m is fitted as: y = a m x+b m , a m is the slope of the straight line, b m is the Y-axis intercept; the sum of squared errors from the trajectory point to the fitting line is set as the optimization objective function E(i), x m represents the x-axis coordinate after fitting the linear equation of motion, y represents the y-axis coordinate before fitting, and M represents the total number of trajectory points associated with the target data.

6. The method for detecting and locating anomalies of airport runway lighting based on aerial video analysis according to claim 5, characterized in that: Use the least squares method to solve the optimal fitting parameter a for the optimization objective function m , excellent and b m ,excellent b m , good = y 均 -a m , Excellent×x 均 , where x 均 Indicates the mean value of the horizontal coordinate of the target trajectory point, y 均 Indicates the mean value of the vertical coordinate of the target trajectory point.

7. The method for detecting and locating anomalies of airport runway lighting based on aerial video analysis according to claim 1 is characterized by: Gaussian mixture models include the following methods in unsupervised clustering: S41. The motion features of the target trajectory points are used as two-dimensional data points in the clustering process. The Gaussian mixture model determines the probability density function of each data point S(i) as follows: Where K is the number of Gaussian components, π k is the weight parameter of the kth Gaussian component, N(S(i)|μ k ,∑ k ) is the k Gaussian distribution probability density function of the data point S(i), μ k is the mean vector parameter of the Gaussian distribution, ∑ k is the covariance matrix parameter of the Gaussian distribution, and θ is the vector parameter composed of all parameters; S42, iterative optimization, for each data point S(i), calculate its posterior probability generated by each Gaussian component: r ik is the probability that the data point S(i) belongs to the kth Gaussian component, z i is the latent variable corresponding to the data point S(i); S43. Re-estimate the mean, covariance matrix, and mixing coefficient of each Gaussian component to maximize the likelihood function of the entire model. The specific update process is: Where M represents the total number of trajectory points associated with the target data; S44, looping through the above steps until the parameters of the Gaussian mixture model converge or the number of loops reaches a preset threshold; S45. Cluster the data points based on the optimal parameters of the Gaussian mixture model, and cluster the data points S(i) into corresponding categories.

8. The method for detecting and locating anomalies of airport runway lighting based on aerial video analysis according to claim 7, characterized in that: The parameter converges to the parameter μ k Whether the mean of the changes before and after the update is less than the threshold thr_mu, the preset threshold set by the number of cycles is the threshold thr_loop.

9. The method for detecting and locating anomalies of airport runway lights based on aerial video analysis according to claim 1, characterized in that: During clustering, the Gaussian mixture model calculates the mean slope of the data points contained in each cluster set, and takes the category set with the largest absolute value of the mean slope as the center line light set Z_mid, the category set with a mean slope greater than 0 as the left line light set Z_left, and the category set with a mean slope less than 0 as the right line light set Z_right.

Images