Unmanned aircraft obstacle avoidance monitoring method and system of cloud computer, and medium

Abstract

The embodiment of the invention provides an unmanned aircraft obstacle avoidance monitoring method and system based on a cloud computer and a medium. The method comprises the steps that an image of the advancing direction of an unmanned aircraft is collected, an initial image is formed, the initial image is preprocessed, and feature points of the initial image are extracted; the average visual expansion rate between continuous image frames is calculated through the initial image feature points; performing region segmentation on the initial image to obtain a plurality of sub-regions, and obtaining gray values of the sub-regions; generating an obstacle region according to the gray values of the sub-regions, and extracting obstacle features; matching the obstacle features with preset features; if matching succeeds, obstacle distribution information is calculated; if the matching is unsuccessful, carrying out region segmentation adjustment on the initial image; the initial image is acquired according to the advancing direction, the obstacle distribution information is analyzed according to the average visual expansion rate, the obstacle is accurately avoided through the obstacle distribution information in the flight process, and the flight safety is improved.

Description

Technical Field

[0001] This application relates to the field of obstacle avoidance, and more particularly, to an obstacle avoidance monitoring method, system and medium for an unmanned aerial vehicle of a cloud computer. Background Art

[0002] A cloud computer is an overall service solution that includes cloud resources, a transmission protocol, and a cloud terminal. Using an open cloud terminal and Tianting's unique CHP transmission protocol, resources such as desktops, applications, and hardware are provided to users in a service mode of on-demand service and elastic allocation. Users only need a small terminal device, access the network anywhere with a network, and connect a monitor and keyboard and mouse to access their personal desktops, data, and applications. Through the cloud computer, flight control and interface manipulation of unmanned aerial vehicles can be achieved. During the control process of an unmanned aerial vehicle by an existing cloud computer, it is impossible to obtain an initial image by analyzing the image of the traveling direction of the unmanned aerial vehicle, so it is difficult to accurately judge the distribution of obstacles during the flight of the unmanned aerial vehicle through image analysis technology, and the obstacle avoidance effect during the flight of the unmanned aerial vehicle is poor. In view of the above problems, an effective technical solution is urgently needed at present. Summary of the Invention

[0003] The purpose of the embodiments of this application is to provide an obstacle avoidance monitoring method, system and medium for an unmanned aerial vehicle of a cloud computer, obtain an initial image according to the traveling direction of the unmanned aerial vehicle, analyze the obstacle distribution information according to the average visual expansion rate, and accurately avoid obstacles during the flight of the unmanned aerial vehicle through the obstacle distribution information, improving the flight safety of the unmanned aerial vehicle.

[0004] The embodiments of this application also provide an obstacle avoidance monitoring method for an unmanned aerial vehicle of a cloud computer, including:

[0005] Collect an image of the traveling direction of the unmanned aerial vehicle to form an initial image, preprocess the initial image, and extract the feature points of the initial image;

[0006] Calculate the average visual expansion rate between consecutive image frames through the feature points of the initial image;

[0007] Perform region segmentation on the initial image to obtain a number of sub-regions, and obtain the gray values of the sub-regions;

[0008] Generate an obstacle region according to the gray values of the sub-regions, and extract the obstacle features;

[0009] Match the obstacle features with the preset features;

[0010] If the match is successful, calculate the obstacle distribution information according to the average visual expansion rate;

[0011] If the matching is unsuccessful, the segmentation parameters are adjusted to perform regional segmentation adjustments on the initial image.

[0012] Optionally, in the unmanned aerial vehicle obstacle avoidance monitoring method of the cloud computer described in the embodiment of the present application, before collecting the unmanned aerial vehicle moving direction image and forming the initial image, it also includes:

[0013] Obtain the initial position and destination position of the unmanned aerial vehicle;

[0014] generating an initial movement trajectory according to the initial position and the destination position of the unmanned aerial vehicle;

[0015] The unmanned aerial vehicle moves according to the initial moving trajectory, and the video of the moving direction of the unmanned aerial vehicle is collected in real time;

[0016] Perform frame difference processing on the video of the traveling direction of the unmanned aerial vehicle to obtain a number of single-frame images;

[0017] The single-frame images are combined according to a predetermined number of frames to generate an image of the travel direction of the unmanned aerial vehicle.

[0018] Optionally, in the unmanned aerial vehicle obstacle avoidance monitoring method of the cloud computer described in the embodiment of the present application, an image of the traveling direction of the unmanned aerial vehicle is collected to form an initial image, and the initial image is preprocessed to extract feature points of the initial image, specifically:

[0019] Obtain the initial image and extract image features;

[0020] Compare the image features with the preset features to obtain the feature deviation rate;

[0021] Determining whether the characteristic deviation rate is greater than or equal to a preset characteristic deviation rate threshold;

[0022] If it is greater than or equal to, the corresponding image features will be eliminated;

[0023] If it is less than, the image feature points are established and the visual expansion information of different image frames is calculated.

[0024] Optionally, in the unmanned aerial vehicle obstacle avoidance monitoring method of the cloud computer described in the embodiment of the present application, the average visual expansion rate between consecutive image frames is calculated by initial image feature points, specifically:

[0025] Obtain visual information of different image frames, compare visual information of adjacent image frames, and obtain visual deviation;

[0026] Calculate the visual dilation rate of adjacent image frames according to the visual deviation;

[0027] Group the image frames, with two adjacent image frames grouped together to obtain multiple groups of image frames;

[0028] Obtain the visual dilation rates of different groups respectively to get multiple groups of visual dilation rates;

[0029] Process the multiple groups of visual dilation rates, excluding the maximum and minimum visual dilation rates, to obtain the visual dilation rates of the remaining groups;

[0030] Calculate the mean of the visual dilation rates of the remaining groups to obtain the average visual dilation rate.

[0031] Optionally, in the obstacle avoidance monitoring method for an unmanned aerial vehicle of the cloud computer described in the embodiments of the present application, generating an obstacle area according to the sub-region gray values and extracting obstacle features specifically includes:

[0032] Obtain the sub-region gray values, compare the sub-region gray values with a preset gray value to obtain a gray deviation rate;

[0033] Determine whether the gray deviation rate is greater than or equal to a preset gray deviation rate threshold;

[0034] If it is greater than or equal to the preset gray deviation rate threshold, determine that the corresponding sub-region is an obstacle area;

[0035] If it is less than the preset gray deviation rate threshold, determine that the corresponding sub-region is a background area;

[0036] Extract the obstacle area features, generate obstacle features, and obtain obstacle distribution information.

[0037] Optionally, after extracting the obstacle area features, generating obstacle features, and obtaining obstacle distribution information in the obstacle avoidance monitoring method for an unmanned aerial vehicle of the cloud computer described in the embodiments of the present application, it further includes:

[0038] Obtain the obstacle features and generate an obstacle edge line according to the obstacle features;

[0039] Generate an unmanned aerial vehicle movement trajectory according to the initial position and destination position of the unmanned aerial vehicle;

[0040] Calculate the distance between the unmanned aerial vehicle movement trajectory and the obstacle edge line at the corresponding position to obtain distance information;

[0041] Determine whether the distance information is greater than or equal to a preset distance threshold;

[0042] If it is greater than or equal to, determine that the unmanned aerial vehicle can avoid obstacles;

[0043] If it is less than, correction information is generated, and the initial movement trajectory of the unmanned aerial vehicle is adjusted according to the correction information.

[0044] In a second aspect, an embodiment of the present application provides an obstacle avoidance monitoring system for an unmanned aerial vehicle of a cloud computer. The system includes: a memory and a processor. The memory includes a program for the obstacle avoidance monitoring method of the unmanned aerial vehicle of the cloud computer. When the program for the obstacle avoidance monitoring method of the unmanned aerial vehicle of the cloud computer is executed by the processor, the following steps are implemented:

[0045] Collect an image of the traveling direction of the unmanned aerial vehicle to form an initial image, preprocess the initial image, and extract the feature points of the initial image;

[0046] Calculate the average visual expansion rate between consecutive image frames through the feature points of the initial image;

[0047] Perform region segmentation on the initial image to obtain a number of sub-regions, and obtain the gray values of the sub-regions;

[0048] Generate an obstacle region according to the gray values of the sub-regions, and extract the obstacle features;

[0049] Match the obstacle features with the preset features;

[0050] If the matching is successful, calculate the obstacle distribution information according to the average visual expansion rate;

[0051] If the matching is unsuccessful, adjust the segmentation parameters and perform region segmentation adjustment on the initial image.

[0052] Optionally, in the obstacle avoidance monitoring system for an unmanned aerial vehicle of the cloud computer described in the embodiment of the present application, before collecting an image of the traveling direction of the unmanned aerial vehicle to form an initial image, it further includes:

[0053] Obtain the initial position and destination position of the unmanned aerial vehicle;

[0054] Generate an initial movement trajectory according to the initial position and destination position of the unmanned aerial vehicle;

[0055] The unmanned aerial vehicle moves according to the initial movement trajectory and continuously collects a video of the traveling direction of the unmanned aerial vehicle;

[0056] Perform frame difference processing on the video of the traveling direction of the unmanned aerial vehicle to obtain a number of single-frame images;

[0057] Combine the single-frame images according to a predetermined number of frames to generate an image of the traveling direction of the unmanned aerial vehicle.

[0058] Optionally, in the obstacle avoidance monitoring system of an unmanned aerial vehicle for a cloud computer described in the embodiments of the present application, an image in the traveling direction of the unmanned aerial vehicle is collected to form an initial image, and the initial image is preprocessed to extract feature points of the initial image. Specifically:

[0059] Obtain the initial image and extract image features;

[0060] Compare the image features with preset features to obtain a feature deviation rate;

[0061] Determine whether the feature deviation rate is greater than or equal to a preset feature deviation rate threshold;

[0062] If it is greater than or equal to, then eliminate the corresponding image features;

[0063] If it is less than, then establish image feature points and calculate the visual expansion information of different image frames.

[0064] In a third aspect, the embodiments of the present application also provide a computer-readable storage medium, which includes a program for the obstacle avoidance monitoring method of an unmanned aerial vehicle for a cloud computer. When the program for the obstacle avoidance monitoring method of an unmanned aerial vehicle for a cloud computer is executed by a processor, the steps of the obstacle avoidance monitoring method of an unmanned aerial vehicle for a cloud computer as described in any one of the above are implemented.

[0065] As can be seen from the above, an obstacle avoidance monitoring method, system and medium for an unmanned aerial vehicle of a cloud computer provided by the embodiments of the present application collect an image in the traveling direction of the unmanned aerial vehicle to form an initial image, preprocess the initial image, and extract feature points of the initial image; calculate the average visual expansion rate between consecutive image frames through the feature points of the initial image; perform region segmentation on the initial image to obtain several sub-regions, and obtain the gray values of the sub-regions; generate an obstacle region according to the gray values of the sub-regions and extract obstacle features; match the obstacle features with preset features; if the match is successful, calculate the obstacle distribution information according to the average visual expansion rate; if the match is unsuccessful, adjust the segmentation parameters and perform region segmentation adjustment on the initial image; obtain the initial image according to the traveling direction of the unmanned aerial vehicle, and analyze the obstacle distribution information according to the average visual expansion rate. During the flight of the unmanned aerial vehicle, it precisely avoids obstacles through the obstacle distribution information, improving the flight safety of the unmanned aerial vehicle.

[0066] Other features and advantages of the present application will be described in the subsequent specification. The objectives and advantages of the present application are achieved and obtained through the structures specifically pointed out in the written specification, claims, and drawings. Description of the Drawings

[0067] To more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the accompanying drawings required for use in the embodiments of the present application. It should be understood that the following drawings only show some embodiments of the present application, and thus should not be regarded as limiting the scope. For those of ordinary skill in the art, without creative efforts, other related drawings can also be obtained based on these drawings.

[0068] Figure 1 It is a flowchart of the obstacle avoidance monitoring method for an unmanned aerial vehicle of a cloud computer provided by an embodiment of the present application;

[0069] Figure 2 It is a flowchart for obtaining an image of the traveling direction of an unmanned aerial vehicle in the obstacle avoidance monitoring method for an unmanned aerial vehicle of a cloud computer provided by an embodiment of the present application;

[0070] Figure 3 It is a flowchart of the method for calculating visual expansion information in the obstacle avoidance monitoring method for an unmanned aerial vehicle of a cloud computer provided by an embodiment of the present application;

[0071] Figure 4 It is a schematic structural diagram of the obstacle avoidance monitoring system for an unmanned aerial vehicle of a cloud computer provided by an embodiment of the present application. Detailed implementation manners

[0072] The following will clearly and completely describe the technical solutions in the embodiments of the present application in conjunction with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only some embodiments of the present application, rather than all embodiments. Usually, the components of the embodiments of the present application described and shown in the accompanying drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present application provided in the accompanying drawings is not intended to limit the scope of the present application to be protected, but only represents the selected embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without creative efforts belong to the scope of protection of the present application.

[0073] It should be noted that similar reference numerals and letters indicate similar items in the following drawings. Therefore, once an item is defined in one drawing, it does not need to be further defined and explained in subsequent drawings. At the same time, in the description of the present application, terms such as "first" and "second" are only used for distinguishing descriptions and cannot be understood as indicating or implying relative importance.

[0074] Please refer to Figure 1 , Figure 1 which is a flowchart of an obstacle avoidance monitoring method for an unmanned aerial vehicle of a cloud computer in some embodiments of the present application. The obstacle avoidance monitoring method for an unmanned aerial vehicle of the cloud computer is used in a terminal device. The obstacle avoidance monitoring method for an unmanned aerial vehicle of the cloud computer includes the following steps:

[0075] S101, Collect an image of the traveling direction of the unmanned aerial vehicle (i.e., the forward direction of the unmanned aerial vehicle), form an initial image, preprocess the initial image, and extract the feature points of the initial image;

[0076] S102, Calculate the average visual expansion rate between consecutive image frames through the initial image feature points;

[0077] S103, Perform region segmentation on the initial image to obtain a number of sub-regions, and obtain the gray values of the sub-regions;

[0078] S104, Generate an obstacle region according to the gray values of the sub-regions, extract the obstacle features; match the obstacle features with the preset features;

[0079] S105, If the matching is successful, calculate the obstacle distribution information according to the average visual expansion rate; if the matching is unsuccessful, adjust the segmentation parameters and perform region segmentation adjustment on the initial image.

[0080] It should be noted that a preset image recognition algorithm is used to preprocess the image of the forward direction of the unmanned aerial vehicle, and the local feature points of the image are extracted to represent the key structures such as edges and corners in the image. Match the feature points adjacent in degree, calculate the displacement vector of the matching points, and calculate the average value of the displacement to calculate the visual expansion rate; wherein, the visual expansion rate is the displacement change rate of the feature points between consecutive frames, reflecting the size change or relative motion speed of the feature points in the image. The image is divided into multiple sub-regions according to a preset region segmentation algorithm; wherein, the segmentation algorithm includes but is not limited to the watershed algorithm or superpixel segmentation. Then, the average gray value of each sub-region is extracted to generate a gray distribution map. After extracting the obstacle shape features of the gray map based on a pre-trained obstacle feature recognition model, match them with a preset obstacle feature library. If the matching is successful, mark it as an obstacle and calculate the distribution area of the obstacle; if the matching fails, adjust the segmentation parameters, perform region segmentation adjustment on the initial image, and perform obstacle matching again; thus accurately identifying the obstacle distribution.

[0081] As an implementation manner, the original images collected by the on-board monocular camera of the unmanned aerial vehicle are first preprocessed. The images are cropped into 416*416 pixel size according to a preset cropping method, and then the target detection network is used to detect the suspected obstacle areas in the images. At the same time, the SURF algorithm is used to extract the feature points of the whole image, and the extracted feature points are used to calculate the average visual expansion rate between consecutive image frames. The suspected obstacle areas are matched between consecutive image frames. Feature points are extracted and matched between the successfully matched pairs of suspected obstacles, and the obstacles are determined according to the average visual expansion rate. For the feature points of the whole image, they are matched between consecutive image frames, and then the feature points without visual expansion are filtered out. The remaining feature points are dynamically clustered to divide the distribution ranges of different obstacles, so as to realize the detection of obstacles in the traveling direction of the unmanned aerial vehicle.

[0082] Please refer to Figure 2 , Figure 2 Figure 2 is a flowchart for obtaining an image in the traveling direction of an unmanned aerial vehicle for an obstacle avoidance monitoring method of a cloud computer in some embodiments of the present application. According to the embodiments of the present invention, before collecting the image in the traveling direction of the unmanned aerial vehicle to form an initial image, it further includes:

[0083] S201, obtaining the initial position and the destination position of the unmanned aerial vehicle;

[0084] S202, generating an initial movement trajectory according to the initial position and the destination position of the unmanned aerial vehicle;

[0085] S203, the unmanned aerial vehicle moves according to the initial movement trajectory and collects the video in the traveling direction of the unmanned aerial vehicle in real time;

[0086] S204, performing frame difference processing on the video in the traveling direction of the unmanned aerial vehicle to obtain a plurality of single-frame images;

[0087] S205, combining the single-frame images according to a predetermined number of frames to generate an image in the traveling direction of the unmanned aerial vehicle.

[0088] It should be noted that, first, the current position coordinates and destination coordinates of the unmanned aerial vehicle are obtained through the GPS module, where the coordinate values are expressed in longitude and latitude. Secondly, based on the preset optimal path search algorithm, known static obstacles are avoided, and the initial movement trajectory is generated according to the current coordinates and destination coordinates of the unmanned aerial vehicle; wherein the movement trajectory is composed of multiple waypoints, and each waypoint includes information such as position, altitude, and speed. Then, the image acquisition device carried by the unmanned aerial vehicle is used to capture and transmit the video information in front to the server background in real time. Finally, the background processes the video information captured by the unmanned aerial vehicle through the OpenCV image processing library, which includes frame difference processing of the video to obtain a single-frame image, and then synthesizes continuous single-frame images according to a preset number of frames to obtain an image of the unmanned aerial vehicle's forward direction; thereby enhancing the continuity detection of obstacles.

[0089] Please refer to Figure 3 , Figure 3 This is a flow chart of a method for calculating visual expansion information of a cloud computer unmanned aerial vehicle obstacle avoidance monitoring method in some embodiments of the present application. According to an embodiment of the present invention, an image of the traveling direction of the unmanned aerial vehicle is collected to form an initial image, the initial image is preprocessed, and feature points of the initial image are extracted, specifically:

[0090] S301, obtaining an initial image and extracting image features;

[0091] S302, comparing the image features with preset features to obtain a feature deviation rate;

[0092] S303, determining whether the characteristic deviation rate is greater than or equal to a preset characteristic deviation rate threshold;

[0093] S304, if it is greater than or equal to, the corresponding image feature is eliminated;

[0094] S305: If it is less than , establish image feature points and calculate visual expansion information of different image frames.

[0095] It should be noted that the image features of the initial image are extracted based on the preset image recognition algorithm. The current image features are matched with the preset template library, and the feature deviation rate with the obstacles in the template library is calculated in turn. If the feature deviation rate is greater than or equal to the preset feature deviation rate threshold, it means that the current obstacle does not match the obstacle in the image, the current obstacle feature is removed, and the next obstacle in the template library is selected to calculate the feature deviation rate for feature comparison. If the feature deviation rate is less than the preset feature deviation rate threshold, it means that the current image matches the obstacle, then the current feature point is retained, and the visual expansion information of different image frames is calculated to improve the accuracy of obstacle judgment.

[0096] According to an embodiment of the present invention, the average visual expansion rate between consecutive image frames is calculated through initial image feature points, specifically as follows:

[0097] Obtain the visual information of different image frames, compare the visual information of adjacent image frames, and obtain the visual deviation;

[0098] Calculate the visual expansion rate of adjacent image frames according to the visual deviation;

[0099] Group the image frames, group two adjacent image frames into one group, and obtain multiple groups of image frames;

[0100] Obtain the visual expansion rates of different groups respectively, and obtain multiple groups of visual expansion rates;

[0101] Process the multiple groups of visual expansion rates, remove the maximum visual expansion rate and the minimum visual expansion rate, and obtain the visual expansion rates of the remaining groups;

[0102] Calculate the mean value of the visual expansion rates of the remaining groups to obtain the average visual expansion rate.

[0103] It should be noted that the visual expansion rate is used to reflect the size change rate of obstacles in the image and is positively correlated with the approaching speed of the unmanned aerial vehicle. Match the feature points of consecutive frames, calculate the displacement vector of each pair of matching points to obtain the visual deviation; then calculate the visual expansion rate of adjacent image frames according to the displacement vector of the matching points. Calculate the visual expansion rates of multiple groups according to multiple consecutive frames. Then, based on a preset outlier rejection mechanism, remove some of the maximum visual expansion rates and some of the minimum visual expansion rates. Finally, calculate the mean value of the visual expansion rates of the remaining groups after outlier rejection as the final average visual expansion rate; it is used to improve the accuracy of obstacle distribution information, so that the unmanned aerial vehicle can accurately avoid obstacles.

[0104] According to an embodiment of the present invention, an obstacle area is generated according to the sub-region gray value, and the obstacle features are extracted, specifically as follows:

[0105] Obtain the sub-region gray value, compare the sub-region gray value with a preset gray value, and obtain the gray deviation rate;

[0106] Judge whether the gray deviation rate is greater than or equal to a preset gray deviation rate threshold;

[0107] If it is greater than or equal to the preset gray deviation rate threshold, determine that the corresponding sub-region is an obstacle area;

[0108] If it is less than the preset gray deviation rate threshold, determine that the corresponding sub-region is a background area;

[0109] Extract the features of the obstacle area, generate obstacle features, and obtain the obstacle distribution information.

[0110] It should be noted that the gray deviation rate is the difference ratio between the gray value of the sub-region and the preset background gray value. As an implementation, by obtaining the gray value of the sub-region, comparing it with the preset gray value, the gray deviation rate is calculated. If the gray deviation rate is greater than or equal to the preset gray deviation rate threshold, the corresponding sub-region is determined as the obstacle area. If the gray deviation rate is less than the preset gray deviation rate threshold, it is marked as the background area, that is, the safe area, and does not participate in the subsequent obstacle calculation. Finally, the contour of the obstacle area is extracted, and features such as area, length, width, and coordinates are calculated to obtain the obstacle distribution information; by analyzing the gray value, the accurate segmentation of the obstacle area and the background area is realized.

[0111] According to the embodiment of the present invention, after extracting the features of the obstacle area, generating the obstacle features, and obtaining the obstacle distribution information, it further includes:

[0112] Obtain the obstacle features and generate the obstacle edge line according to the obstacle features;

[0113] Generate the movement trajectory of the unmanned aerial vehicle according to the initial position and the destination position of the unmanned aerial vehicle;

[0114] Calculate the distance between the movement trajectory of the unmanned aerial vehicle and the obstacle edge line at the corresponding position to obtain the distance information;

[0115] Judge whether the distance information is greater than or equal to the preset distance threshold;

[0116] If it is greater than or equal to, it is determined that the unmanned aerial vehicle can avoid obstacles;

[0117] If it is less than, generate correction information and adjust the initial movement trajectory of the unmanned aerial vehicle according to the correction information.

[0118] It should be noted that in this embodiment, first, an edge detection algorithm, such as the Canny edge detection algorithm, is used for the obstacle area to generate a binary edge map of the obstacle; then, the Hough transform is used to fit a straight line to obtain the obstacle edge line. Then, according to the movement trajectory determined by the initial position and the destination position of the unmanned aerial vehicle, the distance from the obstacle projected onto the movement trajectory line is calculated to obtain the shortest distance between the obstacle and the movement trajectory. If the shortest distance is greater than or equal to the preset distance threshold, it means that the unmanned aerial vehicle can avoid the current obstacle, that is, it is determined as a low-risk area. If the shortest distance is less than the preset distance threshold, it is determined as a high-risk collision area, and the movement trajectory of the unmanned aerial vehicle is adjusted based on the shortest distance to ensure that the unmanned aerial vehicle can effectively avoid obstacles during flight.

[0119] It is worth mentioning that it also includes:

[0120] Set the acquisition time node and obtain the obstacle position information of adjacent time nodes;

[0121] Determine whether there is a position difference in the obstacle positions of adjacent time nodes;

[0122] If there is a position difference, determine that the obstacle is a dynamic obstacle, analyze the positions of the dynamic obstacles at adjacent time nodes to obtain the position difference;

[0123] Generate the moving speed and moving direction of the dynamic obstacle according to the position difference to obtain the moving trajectory of the dynamic obstacle;

[0124] Analyze the moving trajectory of the dynamic obstacle and the moving trajectory of the unmanned aerial vehicle to obtain the Euclidean distance;

[0125] Determine whether the Euclidean distance is greater than a preset Euclidean distance value. If it is greater, determine that the unmanned aerial vehicle and the dynamic obstacle will not collide. If it is less, the unmanned aerial vehicle and the dynamic obstacle will collide. Calculate the time of collision and dynamically adjust the flight parameters of the unmanned aerial vehicle according to the time of collision;

[0126] If there is no position difference, determine that the obstacle is a static obstacle.

[0127] It should be noted that this embodiment provides a processing mechanism for dynamic / static obstacles, specifically: set the acquisition time node, and determine whether there is a position difference in the obstacle positions of adjacent time nodes; if so, determine it as a dynamic obstacle; if not, determine it as a static obstacle; when it is a dynamic obstacle, determine the moving trajectory of the obstacle according to the position difference; calculate the Euclidean distance between the moving trajectory of the unmanned aerial vehicle and the moving trajectory of the obstacle; adjust the flight parameters of the unmanned aerial vehicle according to the Euclidean distance. By analyzing the obstacle positions at different time nodes to analyze the obstacle types, if the obstacle is a dynamic obstacle, analyze the distance change between the dynamic obstacle and the unmanned aerial vehicle during movement to determine whether the two will collide, and flexibly adjust the flight direction or flight speed of the unmanned aerial vehicle according to the analysis results to ensure that the unmanned aerial vehicle can accurately avoid dynamic and static obstacles during flight and improve the flight safety of the unmanned aerial vehicle.

[0128] It is worth mentioning that it also includes:

[0129] Input the obstacle type, threat level, and obstacle avoidance strategy score into a preset first neural network model for training;

[0130] When a dynamic obstacle is detected, the emergency avoidance mechanism is triggered first, and then the obstacle avoidance strategy is output according to the first neural network model;

[0131] When a static obstacle is detected, a local fine-tuning is performed along the original path. It should be noted that in this embodiment, the accuracy of the obstacle avoidance strategy is improved by continuously training the first neural network model. When encountering a dynamic obstacle, the emergency avoidance operation is triggered first, including but not limited to vertical climbing, vertical descending, horizontal left shifting, horizontal right shifting, etc. When the threat level (used to represent the distance between the unmanned aerial vehicle and the obstacle) is lower than the preset threshold, the first neural network model outputs the obstacle avoidance strategy according to the obstacle type and movement trajectory. When encountering a static obstacle, a local fine-tuning method is used to avoid it.

[0132] Please refer to Figure 4 , Figure 4 FIG. is a schematic structural diagram of an obstacle avoidance monitoring system for an unmanned aerial vehicle of a cloud computer in some embodiments of the present application. In a second aspect, an embodiment of the present application provides an obstacle avoidance monitoring system 4 for an unmanned aerial vehicle of a cloud computer, the system includes: a memory 41 and a processor 42. The memory 41 includes a program of an obstacle avoidance monitoring method for an unmanned aerial vehicle of a cloud computer. When the program of the obstacle avoidance monitoring method for an unmanned aerial vehicle of a cloud computer is executed by the processor, the following steps are implemented:

[0133] Collect images of the traveling direction of the unmanned aerial vehicle (i.e., the forward direction of the unmanned aerial vehicle) to form an initial image, preprocess the initial image, and extract the feature points of the initial image;

[0134] Calculate the average visual expansion rate between consecutive image frames through the feature points of the initial image;

[0135] Perform region segmentation on the initial image to obtain a number of sub-regions, and obtain the gray values of the sub-regions;

[0136] Generate an obstacle region according to the gray values of the sub-regions, extract the obstacle features; match the obstacle features with the preset features;

[0137] If the matching is successful, calculate the obstacle distribution information according to the average visual expansion rate. If the matching is unsuccessful, adjust the segmentation parameters and perform region segmentation adjustment on the initial image.

[0138] It should be noted that the image in the forward direction of the unmanned aerial vehicle is pre - processed using a preset image recognition algorithm to extract local feature points of the image, which are used to represent key structures such as edges and corner points in the image. Feature points that are adjacent in degree are matched, the displacement vector of the matching points is calculated, and the average value of the displacement is calculated to calculate the visual expansion rate; wherein, the visual expansion rate is the displacement change rate of feature points between consecutive frames, reflecting the size change or relative movement speed of feature points in the image. The image is divided into multiple sub - regions according to a preset region segmentation algorithm; wherein, the segmentation algorithm includes, but is not limited to, the watershed algorithm or superpixel segmentation. Then, the average gray value of each sub - region is extracted to generate a gray - scale distribution map. Based on a pre - trained obstacle feature recognition model, after extracting the obstacle shape features of the gray - scale map, they are matched with a preset obstacle feature library. If the match is successful, it is marked as an obstacle, and the distribution area of the obstacle is calculated; if the match fails, the segmentation parameters are adjusted, the region segmentation of the initial image is adjusted, and the obstacle matching is performed again; thus, the distribution of obstacles is accurately identified.

[0139] As an implementation manner, the original image collected by the on - board monocular camera of the unmanned aerial vehicle is first pre - processed. The image is cropped to a size of 416 * 416 pixels according to a preset cropping method, and then a target detection network is used to detect the suspected obstacle area in the image. At the same time, the SURF algorithm is used to extract feature points from the entire image, and the extracted feature points are used to calculate the average visual expansion rate between consecutive image frames. The suspected obstacle areas are matched between consecutive image frames. Feature points are extracted and matched between pairs of successfully matched suspected obstacles, and obstacles are determined according to the average visual expansion rate. For the feature points of the entire image, they are matched between consecutive image frames, and then the feature points without visual expansion are filtered out, and the remaining feature points are dynamically clustered to divide the distribution ranges of different obstacles, thereby realizing the detection of obstacles in the forward direction of the unmanned aerial vehicle.

[0140] According to the embodiment of the present invention, before collecting the image in the forward direction of the unmanned aerial vehicle to form an initial image, it further includes:

[0141] Obtain the initial position and destination position of the unmanned aerial vehicle;

[0142] Generate an initial movement trajectory according to the initial position and destination position of the unmanned aerial vehicle;

[0143] The unmanned aerial vehicle moves according to the initial movement trajectory and collects the video in the forward direction of the unmanned aerial vehicle in real time;

[0144] Perform frame - difference processing on the video in the forward direction of the unmanned aerial vehicle to obtain a number of single - frame images;

[0145] Combine single-frame images according to a predetermined number of frames to generate an image of the traveling direction of an unmanned aerial vehicle.

[0146] It should be noted that, first, the current position coordinates and destination coordinates of the unmanned aerial vehicle are obtained through the GPS module, where the coordinate values are represented by longitude and latitude. Secondly, based on a preset optimal path search algorithm, known static obstacles are avoided, and an initial movement trajectory is generated according to the current coordinates and destination coordinates of the unmanned aerial vehicle; among them, the movement trajectory consists of multiple waypoints, and each waypoint includes information such as position, altitude, and speed. Then, through the image acquisition device carried by the unmanned aerial vehicle, video information in the front is captured and transmitted to the server background in real time. Finally, the background processes the video information captured by the unmanned aerial vehicle through the OpenCV image processing library, including performing frame difference processing on the video to obtain single-frame images, and then synthesizing consecutive single-frame images according to the preset number of frames to obtain an image of the forward direction of the unmanned aerial vehicle; thereby enhancing the continuity detection of obstacles.

[0147] According to an embodiment of the present invention, an image of the traveling direction of an unmanned aerial vehicle is collected to form an initial image, and the initial image is preprocessed to extract feature points of the initial image, specifically:

[0148] Obtain the initial image and extract image features;

[0149] Compare the image features with preset features to obtain a feature deviation rate;

[0150] Judge whether the feature deviation rate is greater than or equal to a preset feature deviation rate threshold;

[0151] If it is greater than or equal to, the corresponding image features are removed;

[0152] If it is less than, establish image feature points and calculate the visual expansion information of different image frames.

[0153] It should be noted that based on a preset image recognition algorithm, the image features of the initial image are extracted. The current image features are matched with a preset template library, and the feature deviation rate from the obstacles in the template library is calculated in turn. If the feature deviation rate is greater than or equal to the preset feature deviation rate threshold, it means that the current obstacle does not match the obstacle in the image, the current obstacle feature is removed, and the next obstacle in the template library is selected to calculate the feature deviation rate for feature comparison. If the feature deviation rate is less than the preset feature deviation rate threshold, it means that the current image matches the obstacle, then the current feature points are retained, and the visual expansion information of different image frames is calculated; improving the accuracy of obstacle judgment.

[0154] According to an embodiment of the present invention, the average visual expansion rate between consecutive image frames is calculated through the initial image feature points, specifically:

[0155] Obtain the visual information of different image frames, compare the visual information of adjacent image frames, and obtain the visual deviation;

[0156] Calculate the visual expansion rate of adjacent image frames according to the visual deviation;

[0157] Group the image frames, group two adjacent image frames into one group, and obtain multiple groups of image frames;

[0158] Obtain the visual expansion rates of different groups respectively, and obtain multiple groups of visual expansion rates;

[0159] Process the multiple groups of visual expansion rates, eliminate the maximum visual expansion rate and the minimum visual expansion rate, and obtain the visual expansion rates of the remaining groups;

[0160] Calculate the mean value of the visual expansion rates of the remaining groups to obtain the average visual expansion rate.

[0161] It should be noted that the visual expansion rate is used to reflect the size change rate of the obstacle in the image, and is positively correlated with the approaching speed of the unmanned aerial vehicle. Match the feature points of consecutive frames, calculate the displacement vector of each pair of matching points to obtain the visual deviation; then, according to the displacement vector of the matching points, calculate the visual expansion rate of adjacent image frames. Then, calculate the visual expansion rate based on multiple consecutive frames to obtain multiple groups of visual expansion rates. Then, based on a preset outlier rejection mechanism, eliminate some of the maximum visual expansion rates and some of the minimum visual expansion rates. Finally, calculate the mean value of the visual expansion rates of the remaining groups after outlier rejection as the final average visual expansion rate; it is used to improve the accuracy of the obstacle distribution information, so that the unmanned aerial vehicle can accurately avoid obstacles.

[0162] According to the embodiment of the present invention, an obstacle area is generated according to the sub-region gray value, and the obstacle features are extracted, specifically:

[0163] Obtain the sub-region gray value, compare the sub-region gray value with a preset gray value, and obtain the gray deviation rate;

[0164] Judge whether the gray deviation rate is greater than or equal to a preset gray deviation rate threshold;

[0165] If it is greater than or equal to the preset gray deviation rate threshold, determine that the corresponding sub-region is an obstacle area;

[0166] If it is less than the preset gray deviation rate threshold, determine that the corresponding sub-region is a background area;

[0167] Extract the obstacle area features, generate obstacle features, and obtain the obstacle distribution information.

[0168] It should be noted that the gray deviation rate is the difference ratio between the gray value of the sub-region and the preset background gray value. As an implementation manner, by obtaining the gray value of the sub-region, comparing it with the preset gray value, the gray deviation rate is calculated. If the gray deviation rate is greater than or equal to the preset gray deviation rate threshold, it is determined that the corresponding sub-region is an obstacle region. If the gray deviation rate is less than the preset gray deviation rate threshold, it is marked as the background region, that is, the safe region, and does not participate in the subsequent obstacle calculation. Finally, the contour of the obstacle region is extracted, and features such as area, length, width, and coordinates are calculated to obtain the obstacle distribution information; by analyzing the gray value, the accurate segmentation of the obstacle region and the background region is realized.

[0169] According to the embodiment of the present invention, after extracting the obstacle region features, generating the obstacle features, and obtaining the obstacle distribution information, it further includes:

[0170] Obtain the obstacle features and generate the obstacle edge line according to the obstacle features;

[0171] Generate the movement trajectory of the unmanned aerial vehicle according to the initial position and the destination position of the unmanned aerial vehicle;

[0172] Calculate the distance between the movement trajectory of the unmanned aerial vehicle and the obstacle edge line at the corresponding position to obtain the distance information;

[0173] Judge whether the distance information is greater than or equal to the preset distance threshold;

[0174] If it is greater than or equal to, it is determined that the unmanned aerial vehicle can avoid the obstacle;

[0175] If it is less than, generate the correction information and adjust the initial movement trajectory of the unmanned aerial vehicle according to the correction information.

[0176] It should be noted that in this embodiment, first, an edge detection algorithm, such as the Canny edge detection algorithm, is used for the obstacle region to generate a binary edge map of the obstacle; then, the Hough transform is used to fit a straight line to obtain the obstacle edge line. Then, according to the movement trajectory determined by the initial position and the destination position of the unmanned aerial vehicle, the distance from the obstacle projected onto the movement trajectory line is calculated to obtain the shortest distance between the obstacle and the movement trajectory. If the shortest distance is greater than or equal to the preset distance threshold, it means that the unmanned aerial vehicle can avoid the current obstacle, that is, it is determined as a low-risk region. If the shortest distance is less than the preset distance threshold, it is determined as a high-risk collision region, and the movement trajectory of the unmanned aerial vehicle is adjusted based on the shortest distance to ensure that the unmanned aerial vehicle can effectively avoid obstacles during flight.

[0177] It is worth mentioning that it further includes:

[0178] Set the acquisition time node and obtain the obstacle position information at adjacent time nodes;

[0179] Determine whether there is a position difference in the obstacle positions at adjacent time nodes;

[0180] If there is a position difference, determine the obstacle as a dynamic obstacle, analyze the positions of the dynamic obstacles at adjacent time nodes to obtain the position difference;

[0181] Generate the moving speed and moving direction of the dynamic obstacle based on the position difference to obtain the moving trajectory of the dynamic obstacle;

[0182] Analyze the moving trajectory of the dynamic obstacle and the moving trajectory of the unmanned aerial vehicle to obtain the Euclidean distance;

[0183] Determine whether the Euclidean distance is greater than a preset Euclidean distance value. If it is greater, determine that the unmanned aerial vehicle and the dynamic obstacle will not collide. If it is less, the unmanned aerial vehicle and the dynamic obstacle will collide. Calculate the time of collision and dynamically adjust the flight parameters of the unmanned aerial vehicle according to the time of collision;

[0184] If there is no position difference, determine the obstacle as a static obstacle.

[0185] It should be noted that this embodiment provides a processing mechanism for dynamic / static obstacles, specifically: setting the acquisition time node, determining whether there is a position difference in the obstacle positions at adjacent time nodes; if so, determining it as a dynamic obstacle; if not, determining it as a static obstacle; when it is a dynamic obstacle, determining the moving trajectory of the obstacle according to the position difference; calculating the Euclidean distance between the moving trajectory of the unmanned aerial vehicle and the moving trajectory of the obstacle; adjusting the flight parameters of the unmanned aerial vehicle according to the Euclidean distance. By analyzing the obstacle positions at different time nodes to analyze the obstacle types, if the obstacle is a dynamic obstacle, analyze the distance change between the dynamic obstacle and the unmanned aerial vehicle during movement to determine whether the two will collide, and flexibly adjust the flight direction or flight speed of the unmanned aerial vehicle according to the analysis results to ensure that the unmanned aerial vehicle can accurately avoid dynamic and static obstacles during flight and improve the flight safety of the unmanned aerial vehicle.

[0186] It is worth mentioning that it also includes:

[0187] Input the obstacle type, threat level, and obstacle avoidance strategy score into a preset first neural network model for training;

[0188] When a dynamic obstacle is detected, first trigger the emergency avoidance mechanism, and then output the obstacle avoidance strategy according to the first neural network model;

[0189] When a static obstacle is detected, local fine-tuning is performed along the original path. It should be noted that in this embodiment, the accuracy of the obstacle avoidance strategy is improved by continuously training the first neural network model. When a dynamic obstacle is encountered, an emergency avoidance operation is first triggered, including but not limited to vertical climb, vertical descent, horizontal left shift, horizontal right shift, etc. When the threat level (used to represent the distance between the unmanned aerial vehicle and the obstacle) is lower than the preset threshold, the first neural network model outputs an obstacle avoidance strategy according to the obstacle type and movement trajectory. When a static obstacle is encountered, it is avoided by means of local fine-tuning.

[0190] The third aspect of the present invention provides a computer-readable storage medium, which includes a program for the obstacle avoidance monitoring method of an unmanned aerial vehicle of a cloud computer. When the program for the obstacle avoidance monitoring method of an unmanned aerial vehicle of a cloud computer is executed by a processor, the steps of the obstacle avoidance monitoring method of an unmanned aerial vehicle of a cloud computer as described in any one of the above are implemented.

[0191] An obstacle avoidance monitoring method, system and medium for an unmanned aerial vehicle of a cloud computer disclosed in the present invention form an initial image by collecting an image in the traveling direction of the unmanned aerial vehicle, preprocesses the initial image, and extracts feature points of the initial image; calculates the average visual expansion rate between consecutive image frames through the feature points of the initial image; divides the initial image into regions to obtain several sub-regions, and obtains the gray values of the sub-regions; generates an obstacle region according to the gray values of the sub-regions, and extracts obstacle features; matches the obstacle features with preset features; if the match is successful, calculates the obstacle distribution information according to the average visual expansion rate; if the match is unsuccessful, adjusts the segmentation parameters and adjusts the region segmentation of the initial image; obtains the initial image according to the traveling direction of the unmanned aerial vehicle, and analyzes the obstacle distribution information according to the average visual expansion rate. During the flight of the unmanned aerial vehicle, the obstacle distribution information is used to accurately avoid obstacles, thereby improving the flight safety of the unmanned aerial vehicle.

[0192] In several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are only illustrative. For example, the division of units is only a logical function division, and there may be other division methods in actual implementation. For example, multiple units or components can be combined, or can be integrated into another system, or some features can be ignored, or not executed. In addition, the couplings, direct couplings, or communication connections shown or discussed between the various components can be through some interfaces, and the indirect couplings or communication connections of devices or units can be electrical, mechanical, or other forms.

[0193] The units described above as separate components may or may not be physically separated, and the components shown as units may or may not be physical units; they may be located in one place or distributed across multiple network units; some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0194] In addition, each functional unit in the embodiments of the present invention may be all integrated in a processing unit, or each unit may be separately regarded as a unit, or two or more units may be integrated in one unit; the above integrated units may be implemented in the form of hardware, or in the form of a combination of hardware and software functional units.

[0195] Those of ordinary skill in the art can understand that all or part of the steps to implement the above method embodiments can be completed by hardware related to program instructions. The foregoing program can be stored in a readable storage medium. When the program is executed, it performs the steps including the above method embodiments; and the foregoing storage medium includes: removable storage devices, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical discs and other various media that can store program codes.

[0196] Alternatively, if the above integrated units of the present invention are implemented in the form of software functional modules and sold or used as independent products, they can also be stored in a readable storage medium. Based on this understanding, the technical solution of the embodiments of the present invention essentially or the part that contributes to the prior art can be embodied in the form of a software product. The software product is stored in a storage medium and includes several instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the methods described in the embodiments of the present invention. And the foregoing storage medium includes: removable storage devices, ROM, RAM, magnetic disks, or optical discs and other various media that can store program codes.

Claims

1. A method for obstacle avoidance monitoring of an unmanned aerial vehicle of a cloud computer, characterized in that, Including: Collect images in the traveling direction of the unmanned aerial vehicle to form initial images, preprocess the initial images, and extract feature points of the initial images; Calculate the average visual expansion rate between consecutive image frames through the feature points of the initial images; Perform region segmentation on the initial images to obtain several sub-regions, and acquire the gray values of the sub-regions; Generate obstacle regions according to the gray values of the sub-regions, and extract obstacle features; Match the obstacle features with preset features; If the matching is successful, calculate the obstacle distribution information according to the average visual expansion rate; If the matching is unsuccessful, adjust the segmentation parameters and perform region segmentation adjustment on the initial images.

2. The obstacle avoidance monitoring method for an unmanned aerial vehicle of a cloud computer according to claim 1, wherein Before collecting images in the traveling direction of the unmanned aerial vehicle to form initial images, it further includes: Obtain the initial position and destination position of the unmanned aerial vehicle; Generate an initial movement trajectory according to the initial position and destination position of the unmanned aerial vehicle; The unmanned aerial vehicle moves according to the initial movement trajectory and continuously collects videos in the traveling direction of the unmanned aerial vehicle; Perform frame difference processing on the videos in the traveling direction of the unmanned aerial vehicle to obtain several single-frame images; Combine the single-frame images according to a predetermined number of frames to generate images in the traveling direction of the unmanned aerial vehicle.

3. The obstacle avoidance monitoring method for an unmanned aerial vehicle of a cloud computer according to claim 2, wherein Collect images in the traveling direction of the unmanned aerial vehicle to form initial images, preprocess the initial images, and extract feature points of the initial images. Specifically: Obtain the initial images and extract image features; Compare the image features with preset features to obtain a feature deviation rate; Judge whether the feature deviation rate is greater than or equal to a preset feature deviation rate threshold; If it is greater than or equal to, eliminate the corresponding image features; If it is less than, establish image feature points and calculate the visual expansion information of different image frames.

4. The obstacle avoidance monitoring method for an unmanned aerial vehicle of a cloud computer according to claim 3, wherein Calculate the average visual expansion rate between consecutive image frames through the feature points of the initial images. Specifically: Obtain the visual information of different image frames, compare the visual information of adjacent image frames to obtain a visual deviation; Calculate the visual expansion rate of adjacent image frames according to the visual deviation; Group the image frames, group two adjacent image frames into one group to obtain multiple groups of image frames; Respectively obtain the visual expansion rates of different groups to obtain multiple groups of visual expansion rates; Process the multiple groups of visual expansion rates, eliminate the maximum visual expansion rate and the minimum visual expansion rate, and obtain the visual expansion rates of the remaining groups; Calculate the mean value of the visual expansion rates of the remaining groups to obtain the average visual expansion rate.

5. The method for obstacle avoidance monitoring of an unmanned aerial vehicle of a cloud computer according to claim 4, wherein, Generate obstacle regions according to the gray values of the sub-regions, and extract obstacle features. Specifically: Obtain the gray values of the sub-regions, compare the gray values of the sub-regions with preset gray values to obtain a gray deviation rate; Judge whether the gray deviation rate is greater than or equal to a preset gray deviation rate threshold; If it is greater than or equal to the preset gray deviation rate threshold, determine the corresponding sub-region as an obstacle region; If it is less than the preset gray deviation rate threshold, determine the corresponding sub-region as a background region; Extract the features of the obstacle regions, generate obstacle features, and obtain the obstacle distribution information.

6. The obstacle avoidance monitoring method for an unmanned aerial vehicle of a cloud computer according to claim 5, characterized in that, After extracting the features of the obstacle regions, generating obstacle features, and obtaining the obstacle distribution information, it further includes: Obtain the obstacle features and generate obstacle edge lines according to the obstacle features; generating a movement trajectory of the unmanned aerial vehicle according to the initial position and the destination position of the unmanned aerial vehicle; Calculate the distance between the moving trajectory of the unmanned aerial vehicle and the obstacle edge line at the corresponding position to obtain distance information; Determining whether the distance information is greater than or equal to a preset distance threshold; If it is greater than or equal to, it is determined that the unmanned aerial vehicle can avoid obstacles; If it is less than, correction information is generated and the initial moving trajectory of the unmanned aerial vehicle is adjusted according to the correction information.

7. An obstacle avoidance monitoring system for an unmanned aerial vehicle of a cloud computer, characterized in that, The system includes: a memory and a processor, wherein the memory includes a program of an unmanned aerial vehicle obstacle avoidance monitoring method of a cloud computer, and when the program of the unmanned aerial vehicle obstacle avoidance monitoring method of the cloud computer is executed by the processor, the following steps are implemented: Collecting images of the traveling direction of the unmanned aerial vehicle to form an initial image, preprocessing the initial image, and extracting feature points of the initial image; Calculate the average visual expansion rate between consecutive image frames through the initial image feature points; Performing region segmentation on the initial image to obtain a number of sub-regions, and obtaining the grayscale values of the sub-regions; Generate obstacle regions according to the grayscale values of sub-regions and extract obstacle features; Match obstacle features with preset features; If the match is successful, the obstacle distribution information is calculated based on the average visual expansion rate; If the matching is unsuccessful, the segmentation parameters are adjusted to perform regional segmentation adjustments on the initial image.

8. The obstacle avoidance monitoring system for an unmanned aerial vehicle of a cloud computer according to claim 7, characterized in that, Before collecting the image of the direction of travel of the unmanned aerial vehicle and forming the initial image, it also includes: Obtain the initial position and destination position of the unmanned aerial vehicle; generating an initial movement trajectory according to the initial position and the destination position of the unmanned aerial vehicle; The unmanned aerial vehicle moves according to the initial moving trajectory, and the video of the moving direction of the unmanned aerial vehicle is collected in real time; Perform frame difference processing on the video of the traveling direction of the unmanned aerial vehicle to obtain a number of single-frame images; The single-frame images are combined according to a predetermined number of frames to generate an image of the travel direction of the unmanned aerial vehicle.

9. The obstacle avoidance monitoring system for an unmanned aerial vehicle of a cloud computer according to claim 8, characterized in that, Collect the image of the unmanned aerial vehicle in the direction of travel to form an initial image, pre-process the initial image, and extract the feature points of the initial image, specifically: Obtain the initial image and extract image features; Compare the image features with the preset features to obtain the feature deviation rate; Determining whether the characteristic deviation rate is greater than or equal to a preset characteristic deviation rate threshold; If it is greater than or equal to, the corresponding image features will be eliminated; If it is less than, the image feature points are established and the visual expansion information of different image frames is calculated.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a cloud computer's unmanned aerial vehicle obstacle avoidance monitoring method program. When the cloud computer's unmanned aerial vehicle obstacle avoidance monitoring method program is executed by a processor, the steps of the cloud computer's unmanned aerial vehicle obstacle avoidance monitoring method as described in any one of claims 1 to 6 are implemented.

Images