Image analysis system based on UAV intelligent cruise

Abstract

The present application discloses an image analysis system based on an intelligent cruise of a drone, which relates to the field of intelligent cruise technology and solves the technical problem that the existing drone cruise system has poor ability to cope with the complex situation of the human settlement environment when collecting images of the human settlement environment, resulting in low comprehensiveness of the collected data and low accuracy of the analysis of the human settlement environment. The system comprises: selecting a cruise route according to the route data corresponding to each initial route in the planning data and the basic data, generating a plurality of image collection points according to the route data corresponding to the cruise route; controlling the drone to cruise according to the cruise route in the route data; acquiring the flight data of the drone in real time and collecting images to obtain cruise images; updating the image collection point status according to the cruise image and uploading the cruise image; generating a cruise report according to the plurality of cruise images; and being able to completely collect images of the cruise area to increase the accuracy of the human settlement environment analysis.

Description

Technical Field

[0001] This application belongs to the field of intelligent cruise technology, and relates to intelligent cruise technology, specifically an intelligent cruise image analysis system based on a drone. Background Art

[0002] Drone intelligent cruise is an automated aerial inspection solution that integrates advanced flight control, precise positioning, sensor integration and intelligent data processing technologies. It can perform inspection tasks efficiently, accurately and safely in multiple fields such as urban traffic management, environmental monitoring, public security and emergency response, and transmit images and data in real time to provide timely and accurate information support for decision-making, showing strong application value and broad development prospects.

[0003] When current drones detect human settlements, they generally control the drone to fly to a specific location to take pictures of the residential area, or set a route, set fixed collection points on the route, and automatically control the drone to take pictures of the residential area on the route, and then analyze the photos to obtain the detection results of the human settlement environment. Regardless of which method is used, the pilot who controls the drone must have sufficient understanding of the detection environment to ensure that the collected images can fully reflect the residential area. In rural environments, since people's residential areas are relatively scattered and the floor heights of each household are different, the above method is difficult to adapt to the complex rural environment for image collection of the human settlement environment, resulting in less comprehensiveness of the collected data, and thus lower accuracy of the analysis of the human settlement environment. Therefore, a system for analyzing images based on drone intelligent cruise is needed. Among them, the image collection of the human settlement environment complies with relevant regulations and is legal and compliant. Summary of the Invention

[0004] The present application aims to solve at least one of the technical problems existing in the prior art; to this end, the present application proposes an image analysis system based on an intelligent drone cruise, which is used to solve the technical problem that the existing drone cruise system has poor ability to cope with the complex conditions of the human settlement environment when collecting images of the human settlement environment, resulting in low comprehensiveness of the collected data and low accuracy of the analysis of the human settlement environment.

[0005] To achieve the above-mentioned purpose, the first aspect of the present application provides an image analysis system based on intelligent cruise of unmanned aerial vehicles, comprising: an interaction module, an acquisition planning module, a cruise control module, an image acquisition module, an image analysis module and a database;

[0006] The interactive module is used to obtain basic data and planning data of the cruise area and display the cruise report;

[0007] The acquisition planning module extracts route data corresponding to each initial route in the planning data, selects a cruise route based on each route data and basic data, and generates a number of image acquisition points based on the route data corresponding to the cruise route;

[0008] The cruise control module controls the UAV to cruise according to the cruise route in the route data;

[0009] The image acquisition module: acquires the flight data of the UAV in real time, acquires images based on the flight data and the corresponding image acquisition points to obtain cruise images; updates the image acquisition point status based on the cruise images; and uploads the cruise images based on the acquisition point status;

[0010] The image analysis module is configured to obtain a number of cruise images and generate a cruise report based on the cruise images.

[0011] The present application obtains basic data and planning data of the cruise area, selects the most suitable route as the cruise route according to the route data and basic data corresponding to each initial route in the planning data, and reviews the image acquisition points according to the adaptability of the route data and the basic data of the cruise area, and controls the drone to cruise according to the cruise route in the route data; obtains the flight data of the drone in real time, and obtains cruise images by acquiring images based on the flight data and the corresponding image acquisition points; updates the status of the image acquisition points according to the cruise images; uploads the cruise images according to the status of the acquisition points; obtains several cruise images, and generates a cruise report based on the several cruise images; selects the optimal cruise route according to the characteristics of the cruise area, and generates image acquisition points that meet the requirements of the cruise area, so that the image of the cruise area can be acquired as completely as possible, so as to increase the accuracy of the analysis of the human living environment in the cruise area.

[0012] Preferably, the selecting of a cruise route according to each route data and basic data includes:

[0013] The cruise area is rasterized to obtain a number of cruise grids, and the available airspace height within each cruise grid in the basic data is extracted;

[0014] Extract the initial route from the route data and number each cruise grid according to the initial route; extract the flight altitude within each cruise area from the route data;

[0015] Generate a route score for the corresponding initial route based on the flight altitude and available airspace altitude within each cruise grid;

[0016] An initial route with a high route score is selected as the cruise route, and the route data corresponding to the initial route is used as the cruise route data.

[0017] Preferably, generating a route score corresponding to the initial route according to the flight altitude and the available airspace altitude in each cruise grid includes:

[0018] Get the number i of each cruise grid; set the available airspace altitude corresponding to the cruise grid to KHi; set the flight altitude corresponding to the cruise grid to FHin; n is the number of the initial route;

[0019] By formula Calculate the route score HPn of the initial route numbered n; where, and is the proportional coefficient; DH is the unit height; is the optimal height adjustment coefficient corresponding to the grid area numbered i, FH0=0.

[0020] Preferably, the generating of a plurality of image acquisition points according to the route data corresponding to the cruise route includes:

[0021] Obtain the flight altitude FHi and flight track length FCi in each cruise grid in the cruise route data corresponding to the cruise route;

[0022] By formula The number of acquisition points TS is calculated; where BH is the standard acquisition height, and BC is the distance between the corresponding image acquisition points at the standard acquisition height;

[0023] An equal number of image acquisition points are set on the cruise route within the corresponding cruise grid; the number of acquisition points corresponding to each cruise grid is calculated in turn, and the image acquisition points of each cruise grid are set according to the number of acquisition points.

[0024] Preferably, updating the image acquisition point status according to the cruise image includes:

[0025] Acquire a cruise image, and input the cruise image into a clarity detection model to obtain a clarity score of the cruise image;

[0026] Determine whether the clarity score is greater than a set clarity score threshold; if so, set the image acquisition point status of the image acquisition point corresponding to the cruise image to the acquired state; if not, set the image acquisition point status of the image acquisition point corresponding to the cruise image to the unqualified state;

[0027] The collection point status includes a waiting state, a collected state, and an unqualified state.

[0028] Preferably, the clarity detection model is obtained by training an artificial intelligence model, including:

[0029] Obtain a number of cruise images and their corresponding clarity scores from a database; integrate the cruise images and clarity scores into a number of training data and test data;

[0030] The artificial intelligence model is trained using training data; the trained artificial intelligence model is tested using test data; and finally a clarity detection model is obtained, which takes the cruise image as input and outputs the corresponding clarity score; wherein the artificial intelligence model includes a BP neural network model and an RBF neural network model.

[0031] Preferably, the uploading of the cruise image according to the state of the collection point includes:

[0032] When the state of the acquisition point is the acquired state, the flight altitude of the image acquisition point is obtained and marked as the acquisition altitude, and the cruise image and acquisition altitude corresponding to the image acquisition point are packaged and uploaded to the image analysis module.

[0033] Preferably, generating a cruise report based on a plurality of cruise images includes:

[0034] Obtain a cruise image, pre-process the cruise image to obtain a test image; input the test image into a debris recognition model to obtain a number of debris labels and corresponding debris coordinates;

[0035] An environmental score will be generated based on the debris label and debris coordinates;

[0036] The environmental score and the corresponding cruise image are integrated into a collection point analysis result data packet of the image collection point; the collection point analysis result data packet is integrated into a cruise report according to the order of the cruise route; and the debris recognition model is obtained through artificial intelligence model training.

[0037] Preferably, the preprocessing of the cruise image to obtain the inspection image includes:

[0038] Obtain the cruise image and its corresponding acquisition altitude; set the magnification ratio according to the acquisition altitude, and the higher the flight altitude, the larger the magnification ratio; the cropping size is the size of the image input into the debris recognition model; enlarge the cruise image according to the magnification ratio, and then crop it to obtain the inspection image.

[0039] Preferably, the debris recognition model is obtained by training an artificial intelligence model, including:

[0040] Obtain a number of test images, and their corresponding number of debris labels and debris coordinates from a database; integrate the test images, and their corresponding number of debris labels and debris coordinates into a number of training data and test data;

[0041] The artificial intelligence model is trained using training data, and the trained artificial intelligence model is tested using test data; ultimately, a debris recognition model is obtained, whose input is a test image and outputs a number of debris labels and debris coordinates; wherein, the artificial intelligence model includes a BP neural network model and an RBF neural network model.

[0042] Preferably, the generating of an environmental score based on the debris labels and debris coordinates includes:

[0043] For the debris coordinates corresponding to several debris labels identified in the test image, cluster analysis is performed on each debris coordinate to obtain several clusters, and the cluster density PDMm of each cluster is obtained; m is the cluster number, and the number of debris labels and the number of types of debris labels in each cluster are obtained, marked as ZWSm and ZZSm respectively;

[0044] By formula The environmental score JP of the inspection image is calculated, wherein MP is the full score of the set environmental score; ZDM is the set standard clustering density; α1 and α2 are proportional coefficients.

[0045] Compared with the prior art, the present invention has the following advantages:

[0046] 1. This application obtains basic data and planning data of the cruise area, selects the most suitable route as the cruise route according to the route data and basic data corresponding to each initial route in the planning data, and reviews the image acquisition points according to the adaptability of the route data and the basic data of the cruise area, and controls the drone to cruise according to the cruise route in the route data; obtains the flight data of the drone in real time, and acquires cruise images according to the flight data and the corresponding image acquisition points; updates the image acquisition point status according to the cruise image; uploads the cruise image according to the acquisition point status; acquires several cruise images, and generates a cruise report based on the several cruise images; selects the optimal cruise route according to the characteristics of the cruise area, and generates image acquisition points that meet the requirements of the cruise area, so that the image of the cruise area can be acquired as completely as possible to increase the accuracy of the analysis of the human living environment in the cruise area.

[0047] 2. This embodiment selects the most appropriate route as the current cruise route by considering the matching between each set route and the specific cruise area, so that the final cruise effect is better and can better reflect the human living environment characteristics of the cruise area. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0049] Figure 1 This is a module diagram of the intelligent cruise image analysis system in this application;

[0050] Figure 2 Schematic diagram of the execution flow of the intelligent cruise image analysis system in this application. DETAILED DESCRIPTION

[0051] The following will clearly and completely describe the technical solutions of this application in conjunction with the embodiments. Obviously, the embodiments described are only a part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0052] See also Figure 1-Figure 2 , the first embodiment of the present application provides an image analysis system based on an intelligent cruise of a UAV, comprising: an interaction module, an acquisition planning module, a cruise control module, an image acquisition module, an image analysis module and a database;

[0053] Interaction module: The interaction module includes a display unit and an interaction input unit; the interaction input unit is used to obtain basic data and planning data of the cruise area, and the display unit is used to display the cruise report; the cruise area is the area where the drone will cruise, the basic data is the height of several buildings or non-buildings in the cruise area, the planning data is several routes set by the staff for cruising the cruise area, and route-related data, including several initial routes and their corresponding route data, the initial route is the route set by the staff; the route data is the flight speed and flight altitude of each point on the route corresponding to the initial route;

[0054] Acquisition planning module: extracts the route data corresponding to each initial route in the planning data, selects a cruise route based on each route data and basic data, and generates several image acquisition points based on the route data corresponding to the cruise route; the cruise route is the final flight route executed by the UAV, and the image acquisition point is the position where the UAV needs to hover for image acquisition;

[0055] Cruise control module: controls the UAV to cruise according to the cruise route in the route data;

[0056] Image acquisition module: real-time acquisition of drone flight data, including the real-time position, flight speed, angle and other data of the drone; image acquisition is performed according to the flight data and the corresponding image acquisition point to obtain a cruise image; the image acquisition point status is updated according to the cruise image; the acquisition point status includes a pending acquisition state, an acquired state and an unqualified state; the pending acquisition state means that the drone has not passed the image acquisition point and has not performed image acquisition at the image acquisition point, which is the initial state of the image acquisition point; the acquired state means that the image acquisition point has performed image acquisition and the acquired image meets the requirements; the unqualified state means that the image acquisition point has performed image acquisition and the acquired image does not meet the requirements; the cruise image is uploaded according to the acquisition point status; specifically, when the real-time position of the drone is the same as the position of the image acquisition point, image acquisition is performed to determine whether the image clarity of the acquired image is qualified. If qualified, the acquisition point status corresponding to the image acquisition point is modified to the acquired state; otherwise, the acquisition point status corresponding to the image acquisition point is modified to the unqualified state; if the acquisition point status of the image acquisition point is qualified, the acquired image is uploaded, otherwise, it is not uploaded;

[0057] In this embodiment, through the above process, unqualified image points are marked during the cruise and re-collected during the next cruise, thereby ensuring the speed of the cruise.

[0058] The image analysis module acquires a number of cruise images and generates a cruise report based on the cruise images; the cruise report includes images collected at each image collection point and corresponding environmental scores.

[0059] This embodiment obtains basic data and planning data of the cruise area, selects the most suitable route as the cruise route according to the route data and basic data corresponding to each initial route in the planning data, and reviews the image acquisition points according to the adaptability of the route data and the basic data of the cruise area, and controls the drone to cruise according to the cruise route in the route data; obtains the flight data of the drone in real time, and obtains cruise images based on the flight data and the corresponding image acquisition points; updates the image acquisition point status according to the cruise image; uploads the cruise image according to the acquisition point status; obtains several cruise images, and generates a cruise report based on the several cruise images; selects the optimal cruise route according to the characteristics of the cruise area, and generates image acquisition points that meet the requirements of the cruise area, so that the image of the cruise area can be collected as completely as possible to increase the accuracy of the analysis of the human living environment in the cruise area.

[0060] The cruise route is selected based on the route data and basic data, including: rasterizing the cruise area to obtain a number of cruise grids. There are many rasterization methods. In this embodiment, the length and width of the grid are set in advance by the staff, and then the cruise area is rasterized; extracting the available airspace height within each cruise grid in the basic data; the available airspace height is the altitude at which the drone can fly within the corresponding cruise grid; specifically, if the height of the corresponding residential building in the grid area is 8 meters, then the available airspace height of the drone is 8 meters, and the drone's flight altitude is at least greater than 8 meters;

[0061] Extract the initial route from the route data and number each cruise grid according to the initial route. That is, number each cruise grid according to the order of the grids that the UAV passes through in the initial route. The first cruise grid passed is numbered 1; the second cruise grid is numbered 2. Extract the flight altitude within each cruise area in the route data.

[0062] Generate a route score for the corresponding initial route based on the flight altitude and available airspace altitude within each cruise grid;

[0063] An initial route with a high route score is selected as the cruise route, and the route data corresponding to the initial route is used as the cruise route data.

[0064] This embodiment selects the most suitable route as the current cruise route by considering the matching between each set route and the specific cruise area, so that the final cruise effect is better and the specific environmental characteristics of the cruise area are better reflected.

[0065] Generate a route score for the corresponding initial route based on the flight altitude and available airspace altitude within each cruise grid, including: obtaining the number i of each cruise grid; setting the available airspace altitude corresponding to the cruise grid to KHi; setting the flight altitude corresponding to the cruise grid to FHin; n is the number of the initial route;

[0066] By formula Calculate the route score HPn of the initial route numbered n; where, and is the proportional coefficient; it is used to adjust the flight altitude within the cruise grid, as well as the ratio of the flight altitude difference between adjacent cruise grids to the route score of the initial route. The specific value is set based on experience; DH is the unit height, which is used to remove the parameter unit; The optimal height adjustment coefficient corresponding to the grid area numbered i is used to adjust the available airspace altitude to the optimal acquisition altitude. It is set according to the height distribution of buildings in the cruising grid. When the height of buildings in the cruising grid is concentrated close to the available airspace altitude, the higher the altitude required to capture the ground, the larger the optimal height adjustment coefficient is set. The specific value is set according to expert experience; FH0=0.

[0067] In this embodiment, the route score of each initial route is calculated using the above formula. The closer the flight altitude of the initial route in each cruise grid is to the optimal acquisition altitude, the better the image effect acquired in the corresponding cruise grid using the route is, and the corresponding route score is set higher. When the flight altitudes within each cruise grid vary greatly, the proportion of objects in the image will vary too much during subsequent image analysis and processing, and the reference between images acquired in adjacent cruise grids will be weak. This indicates that the more unreasonable the initial route planning is, the lower the corresponding route score is.

[0068] Generating a number of image acquisition points according to the route data corresponding to the cruise route, including: obtaining the flight altitude FHi and the flight track length FCi in each cruise grid in the cruise route data corresponding to the cruise route;

[0069] By formula The number of acquisition points TS is calculated; where BH is the standard acquisition height, and BC is the distance between the corresponding image acquisition points at the standard acquisition height;

[0070] An equal number of image acquisition points are set on the cruise route within the corresponding cruise grid; it can be understood that the intercepted flight trajectory within each cruise grid is approximately a straight line; the number of acquisition points corresponding to each cruise grid is calculated in turn, and the image acquisition points of each cruise grid are set according to the number of acquisition points.

[0071] In this embodiment, the number of collection points within each cruise grid is calculated using the above formula. When the flight altitude within the cruise grid is higher, the range of the ground area included in the image captured by the drone is larger. By increasing the distance between collection points, the content duplication of images collected by adjacent collection points is avoided.

[0072] Updating the image acquisition point state according to the cruise image includes: acquiring the cruise image, inputting the cruise image into a clarity detection model to obtain a clarity score of the cruise image;

[0073] Determine whether the clarity score is greater than a set clarity score threshold; if so, set the image acquisition point status of the image acquisition point corresponding to the cruise image to the acquired state; if not, set the image acquisition point status of the image acquisition point corresponding to the cruise image to the unqualified state; the acquisition point status includes the to-be-acquired state, the acquired state and the unqualified state.

[0074] The clarity detection model is obtained through artificial intelligence model training, including: obtaining a number of cruise images and their corresponding clarity scores from a database; the clarity scores are based on the clarity of the cruise images by experts; the cruise images and clarity scores are integrated into a number of training data and test data;

[0075] The artificial intelligence model is trained using training data; the trained artificial intelligence model is tested using test data; specifically, the cruise image in the test data is input into the trained artificial intelligence model to obtain an output clarity score, and it is determined whether the difference between the output clarity score and the clarity score recorded in the test data is within an acceptable range; if so, it means that the test data has passed the test, and the next set of test data is tested; if not, it means that the set of data has failed the test, and the relevant parameters of the artificial intelligence are adjusted, and the test data is continued to be used for testing; until a set proportion of the test data passes the test, a clarity detection model is finally obtained with the cruise image as input and the corresponding clarity score as output; wherein the artificial intelligence model includes a BP neural network model and an RBF neural network model.

[0076] This embodiment updates the state of the collection points in the above manner to ensure that no collection points are missed or collected repeatedly.

[0077] The cruise image is uploaded according to the status of the collection point, including: when the status of the collection point is the collected state, the flight altitude of the image collection point is obtained and marked as the collection altitude, and the cruise image and collection altitude corresponding to the image collection point are packaged and uploaded to the image analysis module; that is, when the collected image meets the upload requirements, it is uploaded.

[0078] Generating a cruise report based on a number of cruise images includes: acquiring cruise images, and pre-processing the cruise images to obtain inspection images; specifically: pre-processing the cruise images to obtain inspection images includes: acquiring cruise images and their corresponding acquisition altitudes; setting a magnification ratio based on the acquisition altitude, with the magnification ratio being set larger as the flight altitude increases; cropping the image to the size of the image input into the debris recognition model; magnifying the cruise image based on the magnification ratio, and then cropping the image to obtain the inspection image; cropping, amplifying, and other operations on the image based on the acquisition data, and unifying the sizes and ratios of images acquired at different altitudes to facilitate model analysis; and inputting the inspection image into the debris recognition model to obtain a number of debris labels and corresponding debris coordinates.

[0079] An environmental score will be generated based on the debris label and debris coordinates;

[0080] The environmental score and the corresponding cruise image are integrated into a collection point analysis result data packet of the image collection point; the collection point analysis result data packet is integrated into a cruise report according to the order of the cruise route; and the debris recognition model is obtained through artificial intelligence model training.

[0081] The debris recognition model is obtained through artificial intelligence model training, including: obtaining a number of test images and their corresponding debris labels and debris coordinates from a database. The debris labels are the type labels of the debris in the test images, such as garbage and firewood piles, and the debris coordinates are the center coordinates of the recognition box corresponding to the debris label; integrating the test images, their corresponding debris labels and debris coordinates into a number of training data and test data;

[0082] The artificial intelligence model is trained using training data, and the trained artificial intelligence model is tested using test data. Specifically, the test image in the test data is input into the trained artificial intelligence model to obtain output debris labels and debris coordinates. It is determined whether the number and type of the output debris labels are the same as the number and type of the debris labels in the test data, and whether the Euclidean distance between each output debris label and the corresponding debris coordinate in the test data is within an acceptable range. If so, it means that the group of test data has passed the test and the next group of test data is tested. If not, the relevant parameters of the artificial intelligence model are adjusted and the test is continued using the current group of test data. Until a set proportion of test data passes the test, a debris recognition model is finally obtained with the test image as input and a number of debris labels and debris coordinates as output. The artificial intelligence model includes a BP neural network model and an RBF neural network model.

[0083] An environmental score is generated based on the debris labels and debris coordinates, including: performing cluster analysis on the debris coordinates corresponding to several debris labels identified in the inspection image to obtain several clusters, and obtaining the cluster density PDMm of each cluster. In this embodiment, the cluster density can be the average value of the point density of each cluster point in the corresponding cluster; m is the cluster number; the number of debris labels and the number of types of debris labels in each cluster are obtained, marked as ZWSm and ZZSm respectively;

[0084] By formula The environmental score JP of the inspection image is calculated, where MP is the set full score of the environmental score; ZDM is the set standard clustering density; α1 and α2 are proportional coefficients used to adjust the proportion of the impact of the number of debris labels and the number of types of debris labels on the environmental score. The specific values are set according to expert experience.

[0085] In this embodiment, the environmental score corresponding to the inspection image is calculated by the above formula. When more types and numbers of debris identified in the inspection image are identified, it means that the environment of the area corresponding to the inspection image is poor, and the corresponding environmental score is set lower; when the number of clusters is greater and the density of each cluster is smaller, it means that the distribution of debris in the inspection image is more dispersed, which means that the environment of the area corresponding to the inspection image is more cluttered, and the corresponding environmental score is set lower.

[0086] Some of the data in the above formula are calculated by removing the dimensions and taking their numerical values. The formula is a formula that is closest to the actual situation obtained by software simulation of a large amount of collected data; the preset parameters and preset thresholds in the formula are set by technical personnel in this field according to actual conditions or obtained through simulation of a large amount of data.

[0087] How this application works:

[0088] The present application obtains basic data and planning data of the cruise area, selects the most suitable route as the cruise route according to the route data and basic data corresponding to each initial route in the planning data, and reviews the image acquisition points according to the adaptability of the route data and the basic data of the cruise area, and controls the drone to cruise according to the cruise route in the route data; obtains the flight data of the drone in real time, and obtains cruise images by acquiring images based on the flight data and the corresponding image acquisition points; updates the status of the image acquisition points according to the cruise images; uploads the cruise images according to the status of the acquisition points; obtains several cruise images, and generates a cruise report based on the several cruise images; selects the optimal cruise route according to the characteristics of the cruise area, and generates image acquisition points that meet the requirements of the cruise area, so that the image of the cruise area can be acquired as completely as possible, so as to increase the accuracy of the analysis of the human living environment in the cruise area.

[0089] The above embodiments are only used to illustrate the technical method of the present application and are not intended to limit it. Although the present application has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present application can be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present application.

Claims

1. Based on the UAV intelligent cruise image analysis system, including: Interaction module, acquisition planning module, cruise control module, image acquisition module, image analysis module and database; characterized in that, The interactive module is used to obtain basic data and planning data of the cruise area and display the cruise report; The acquisition planning module extracts route data corresponding to each initial route in the planning data, and selects a cruise route based on each route data and the basic data, including: rasterizing the cruise area to obtain a plurality of cruise grids, and extracting the available airspace altitude within each cruise grid in the basic data; Extract the initial route from the route data and number each cruise grid according to the initial route; extract the flight altitude within each cruise area from the route data; Generate a route score corresponding to the initial route according to the flight altitude and available airspace altitude in each cruise grid; including: obtaining the number i of each cruise grid; setting the available airspace altitude corresponding to the cruise grid to KHi; setting the flight altitude corresponding to the cruise grid to FHin; n is the number of the initial route; By formula Calculate the route score HPn of the initial route numbered n; where, and is the proportional coefficient; DH is the unit height; is the optimal height adjustment coefficient corresponding to the grid area numbered i; FH0=0; Selecting an initial route with a high route score as a cruise route, and using the route data corresponding to the initial route as cruise route data; Generate a number of image acquisition points according to the route data corresponding to the cruise route; including: obtaining the flight altitude FHi and the flight track length FCi in each cruise grid in the cruise route data corresponding to the cruise route; By formula The number of acquisition points TS is calculated; where BH is the standard acquisition height, and BC is the distance between the corresponding image acquisition points at the standard acquisition height; An equal number of image acquisition points are set on the cruise route within the corresponding cruise grid; the number of acquisition points corresponding to each cruise grid is calculated in sequence, and the image acquisition points of each cruise grid are set according to the number of acquisition points; The cruise control module controls the UAV to cruise according to the cruise route in the route data; The image acquisition module is used to acquire the flight data of the UAV in real time, and to acquire the cruise image based on the flight data and the corresponding image acquisition points; Update the image acquisition point status according to the cruise image; upload the cruise image according to the acquisition point status; The image analysis module is configured to obtain a number of cruise images and generate a cruise report based on the cruise images.

2. The UAV-based intelligent cruise image analysis system according to claim 1 is characterized in that: The updating of the image acquisition point status according to the cruise image includes: Acquire a cruise image, and input the cruise image into a clarity detection model to obtain a clarity score of the cruise image; Determine whether the clarity score is greater than a set clarity score threshold; if so, set the image acquisition point status of the image acquisition point corresponding to the cruise image to the acquired state; if not, set the image acquisition point status of the image acquisition point corresponding to the cruise image to the unqualified state; The collection point status includes a waiting state, a collected state, and an unqualified state.

3. The UAV-based intelligent cruise image analysis system according to claim 2 is characterized in that: The clarity detection model is obtained through artificial intelligence model training, including: Obtain a number of cruise images and their corresponding clarity scores from a database; integrate the cruise images and clarity scores into a number of training data and test data; The artificial intelligence model is trained using training data; the trained artificial intelligence model is tested using test data; and finally a clarity detection model is obtained whose input is the cruise image and whose output is the corresponding clarity score.

4. The UAV-based intelligent cruise image analysis system according to claim 1 is characterized in that: The process of uploading the cruise image according to the state of the acquisition point includes: When the state of the acquisition point is the acquired state, the flight altitude of the image acquisition point is obtained and marked as the acquisition altitude, and the cruise image and acquisition altitude corresponding to the image acquisition point are packaged and uploaded to the image analysis module.

5. The UAV-based intelligent cruise image analysis system according to claim 1 is characterized in that: The generating of the cruise report according to the plurality of cruise images includes: Obtain a cruise image, pre-process the cruise image to obtain a test image; input the test image into a debris recognition model to obtain a number of debris labels and corresponding debris coordinates; An environmental score will be generated based on the debris label and debris coordinates; The environmental score and the corresponding cruise image are integrated into a collection point analysis result data packet of the image collection point; the collection point analysis result data packet is integrated into a cruise report according to the order of the cruise route; and the debris recognition model is obtained through artificial intelligence model training.

6. The UAV-based intelligent cruise image analysis system according to claim 5 is characterized in that: The debris recognition model is obtained through artificial intelligence model training, including: Obtain a number of test images, and their corresponding number of debris labels and debris coordinates from a database; integrate the test images, and their corresponding number of debris labels and debris coordinates into a number of training data and test data; The artificial intelligence model is trained using training data, and the trained artificial intelligence model is tested using test data; ultimately, a debris recognition model is obtained whose input is a test image and whose output is a number of debris labels and debris coordinates.

7. The UAV-based intelligent cruise image analysis system according to claim 5 is characterized in that: The environment score is generated based on the debris labels and debris coordinates, including: For the debris coordinates corresponding to several debris labels identified in the test image, cluster analysis is performed on each debris coordinate to obtain several clusters, and the cluster density PDMm of each cluster is obtained; m is the cluster number, and the number of debris labels and the number of types of debris labels in each cluster are obtained, marked as ZWSm and ZZSm respectively; By formula The environmental score JP of the inspection image is calculated, wherein MP is the full score of the set environmental score; ZDM is the set standard clustering density; α1 and α2 are proportional coefficients.

Images