Unmanned aerial vehicle data transmission system and method based on big data

Abstract

The invention relates to the field of unmanned aerial vehicle data transmission, in particular to an unmanned aerial vehicle data transmission system and method based on big data, and the system comprises a model construction module, a program presetting module, an inspection positioning module, an accident processing module and a signal compression module, the model construction module is used for constructing an accident model, and the program presetting module is used for selecting a verification angle; the unmanned aerial vehicle inspection system comprises an inspection positioning module, an accident processing module and a signal compression module, an unmanned aerial vehicle inspection program is preset, the inspection positioning module is used for finding an accident area and positioning the accident area, the accident processing module is used for judging the accident type, and the signal compression module is used for outputting shot pictures as point clouds. Therefore, urban management resources are reasonably allocated, rapid response and real-time monitoring of unmanned aerial vehicle management inspection are realized, and the efficiency of unmanned aerial vehicle inspection in urban management is enhanced.

Description

Technical Field

[0001] The present invention relates to the field of unmanned aerial vehicle data transmission, and in particular to a unmanned aerial vehicle data transmission system and method based on big data. Background Art

[0002] In the field of social governance, drone inspections can be used to monitor urban traffic, environmental pollution, natural disasters, building safety, urban management, etc. Through the various sensors and cameras carried by drones, large-scale, high-definition data can be obtained in real time to provide decision support and information feedback. After an incident occurs, drones can also be used to respond quickly, monitor in real time, and make corresponding processing and decisions.

[0003] Drone inspections generally identify abnormal events by taking vertical photos of the inspection route, but they cannot identify the specific type of event, and therefore cannot call relevant personnel in time to handle it, which can easily delay the timing of event handling. The vague handling of events can also cause a high false alarm rate for drones.

[0004] In addition, when some security incidents occur, the communication between the drone and the handling department will be disturbed, making it impossible for the drone to send back pictures of the accident scene in the first time, and the handling personnel will not be able to obtain information from the accident scene in time. Although contact can be established through shortwave communication, the transmission capacity of shortwave communication is small and cannot undertake the task of transmitting high-quality pictures or videos. Summary of the invention

[0005] The purpose of the present invention is to provide a UAV data transmission system and method based on big data to solve the problems raised in the above background technology.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions: a data transmission system for unmanned aerial vehicles based on big data, comprising: a model construction module, a program preset module, an inspection and positioning module, an accident processing module and a signal compression module; The model construction module is used to use a depth of field camera to shoot images of various accidents at different angles, construct a picture set of various accidents, compare the color gamut, depth of field and segmentation area size of each accident image at different angles, and calculate the difference between the image and the image directly above; The program preset module is used to calculate the number of verification angles according to the accident risk on the drone inspection route and the historical false alarm rate of the accident, select representative images of various types of accidents according to the principle of maximum distinction, and use the angles corresponding to the representative images as the verification angles of the accidents. The number of selected representative images is the same as the number of verification angles, and all verification angles corresponding to the accidents are stored in the drone memory, and the drone inspection program is preset; The inspection and positioning module is used to start the UAV inspection program. When the UAV flies at a fixed altitude, it calculates the function of the total color value of the picture taken directly above and the flight distance, and compares it with the standard function of the route. When the residual is higher than the threshold, it is identified as an accident and the accident area is located; The accident processing module is used to take pictures of the accident at various angles in the order of accident risk when an accident occurs, calculate the degree of distinction between each picture and the picture taken directly above, use the picture with the greatest degree of distinction as the positioning image, use the accident type corresponding to the positioning image angle as the suspected accident, verify all verification angles of the suspected accident, and output the suspected accident as the accident type when the total degree of distinction difference is less than a threshold; The signal compression module is used to detect the communication speed between the drone and the department corresponding to the accident. When the communication is not smooth, the picture of the accident scene is divided into sub-areas, and the distinction of each sub-area relative to the corresponding area of ​​the projection directly above is calculated. The rendering radius is determined according to the distinction. According to the rendering radius, a point cloud rendering algorithm is used to output the picture as a point cloud map to the relevant department.

[0007] Further, the model construction module includes: an image training unit and a projection feature unit; The image training unit is used to classify pictures taken at historical accident scenes to form an accident training set; The projection feature unit is used to calculate the degree of distinction between images at each angle and the projection image directly above the accident.

[0008] Further, the program presetting module includes: an adaptation verification unit, an angle addressing unit and a patrol instruction unit; The adaptive verification unit is used to calculate the verification quantity according to the accident risk on the drone inspection route and the historical false alarm rate of accidents; The angle addressing unit is used to take the corresponding number of pictures with the greatest degree of distinction in the accident as representative images, and record the verification angles corresponding to the representative images; The inspection instruction unit is used to store the accident and the verification angle corresponding to the accident into the drone database and set the drone inspection program.

[0009] Further, the inspection and positioning module includes: a drone positioning unit and an abnormality marking unit; The drone positioning unit is used to determine the difference in total color value of the captured image relative to the standard image during drone inspection; The abnormal marking unit is used to enter the accident tracking mode and mark the location of the accident after the drone determines that an accident has occurred.

[0010] Further, the accident handling module includes: a program execution unit and an accident identification unit; The program execution unit is used to control the drone to take pictures at the verification angle corresponding to each accident according to the program preset in the drone; The accident identification unit is used to calculate the degree of distinction of the image relative to the image taken directly above, thereby identifying the type of accident and verifying the identified accident.

[0011] Further, the signal compression module includes: a channel detection unit, a grayscale rendering unit and a point cloud output unit; The channel detection unit is used to detect the speed of the communication connection between the drone and the accident handling center; The grayscale rendering unit is used to segment the accident representative image according to grayscale, and calculate the degree of distinction of each area relative to adjacent areas; The point cloud output unit is used to render each area into feature points of different radius according to the distinctiveness of the area, and output the formed point cloud map to the accident handling center.

[0012] A method for transmitting data from a drone based on big data comprises the following steps: Step S1. Take pictures of various types of accidents at different angles to form an accident picture set, and calculate the difference between each picture in the accident picture set and the projected image directly above according to the picture comparison method; Step S2. Calculate the number of iterations according to the accident risk on the drone inspection route and the historical false alarm rate of accidents, sort the accident pictures in descending order according to the degree of distinction, select the pictures with the highest number of iterations of distinction as verification pictures, and preset the angles corresponding to all verification pictures as verification angles, and use the verification angle with the highest degree of distinction as the representative angle; Step S3. During the drone inspection, the function of the total color value of the projected image directly above with the distance is detected and compared with the function of the previous inspection. When the difference exceeds a threshold, it is determined that an accident has occurred under the drone, and the accident is located; Step S4. The drone flies to the representative angle of each accident to take pictures, calculates the difference between the shooting result and the image projected directly above, takes the accident closest to the preset result as the suspected accident, verifies other verification angles corresponding to the suspected accident according to the accident verification method, and outputs the accident type after verification; Step S5. When the communication rate between the UAV and the processing center is lower than the threshold, the captured image is segmented, the distinction between each segmented area and the adjacent area is calculated, the punctuation drawing radius is determined according to the distinction, and the punctuation of all areas is rendered as a point cloud image and sent to the processing center.

[0013] Further, step S1 includes: Step S11. Pre-take pictures of the accident at all possible angles that the drone can shoot, and store all the pictures taken in the database after marking the angles to form a collection of accident pictures; Step S12. Record the vertically shot accident picture as the top projection image, execute the picture comparison method in step S13, and calculate the difference between each picture in the accident picture set and the top projection image; Step S13. Use the image segmentation algorithm to segment the accident area in the overhead projection image and the accident image, record the area of ​​the segmented area, the color value of the pixel point and the depth of field, and calculate the difference between the overhead projection image and the accident image according to the following formula: ; Wherein, E is the discrimination, D1 represents the segmented area of ​​the accident area in the overhead projection image, D2 represents the segmented area of ​​the accident area in the accident picture, n represents the number of pixel columns in the picture, m represents the number of pixel rows in the picture, UR(i,j) represents the color value of the pixel in the i-th column and j-th row of the overhead projection image, Ur(i,j) represents the color value of the pixel in the i-th column and j-th row of the accident picture, h is the depth of field of the overhead projection image at the pixel point (i,j), σ1 represents the horizontal shooting angle of the accident picture, σ2 represents the vertical shooting angle of the accident picture, and when tanσ1·h and tanσ2·h are not integers, they are taken as integers.

[0014] Further, step S2 includes: Step S21. According to the historical inspection records of the drone, the frequency of the drone reporting accidents and the probability of the drone falsely reporting accidents are obtained from the records, and the number of iterations x is calculated, where x=Set[v·c / (p·a)], where v is the frequency of the drone reporting accidents, c is the probability of the drone falsely reporting accidents, p is the number of accident types, a is the preset descent rate, and Set is an upward value function; Step S22. Arrange all the pictures in the accident picture set in descending order according to the degree of difference from the projected image directly above, select the first picture as the representative picture of the accident, the first x pictures as the verification pictures of the accident, the angle corresponding to the representative picture is the representative angle, and the angle corresponding to the verification picture is the verification angle; Step S23: Store the verification image, representative image, and corresponding verification angles and representative angles in the drone memory.

[0015] Further, step S3 includes: Step S31. Calculate the function F1(d) of the total color value of the overhead projection image captured with the distance according to the video sent back by the drone during the last inspection, where d represents the distance and F(d) represents the sum of the color values ​​of all pixels in the image captured by the drone when the distance is d; Step S32. Start the drone to start the inspection. According to the distance the drone has flown, record the total color value of the pictures taken by the drone, and generate a function F2 (d) of the total color value of the projected image directly above during the inspection as a function of the distance. Step S33. Let the risk coefficient b = F1 (d0) - F2 (d0), where d0 is the distance the drone has flown, and the preset risk coefficient threshold b0. When b ≥ b0, it is determined that there is an abnormality in the current position, and the position of the drone is marked as the location where the accident occurred.

[0016] Further, step S4 includes: Step S41. After taking the overhead projection image from directly above the accident location, control the drone to take pictures from representative angles of each accident, and calculate the difference between the accident picture and the overhead projection image for each picture taken; Step S42. After all pictures are taken, calculate the characteristic coefficient G of the accident picture, where G=|E1-E2|, where E1 represents the degree of distinction of the representative picture stored in the drone memory, and E2 represents the degree of distinction of the accident picture taken by the drone at the same angle; Step S43. Arrange all the captured accident pictures in ascending order according to the characteristic coefficients to form a suspected accident sequence. If there is an element in the suspected accident sequence, mark the first accident in the suspected accident sequence as a suspected accident, and go to step S44. If there is no element in the suspected accident sequence, output the accident type corresponding to the first accident in the suspected accident sequence to the processing center; Step S44. Control the drone to take a picture from the verification angle corresponding to the suspected accident, and calculate the difference between the accident picture taken and the projected image directly above, and compare it with the difference corresponding to the verification picture. If the difference is less than the threshold, it means that the picture verification has passed, otherwise it means that the verification has failed; If the number of images that pass the verification is greater than or equal to the number of images that fail the verification, the accident type corresponding to the suspected accident is output to the processing center, otherwise the current suspected accident is removed from the accident suspicion sequence and the process goes to step S43.

[0017] Further, step S5 includes: Step S51. When the communication rate between the drone and the accident handling center is lower than the threshold, the drone's image transmission is stopped, and the image to be sent is divided into k identical rectangular areas, where k is a preset resolution; Step S52. For each rectangular area, calculate the degree of distinction between the rectangular area and each adjacent area according to the method of step S13, and determine the punctuation radius according to the degree of distinction: ; Where s represents the punctuation radius, L represents the pixel arrangement width of the image, b represents the number of regions adjacent to the rectangular area, st represents the punctuation radius of the t-th adjacent area, bt represents the number of regions adjacent to the t-th adjacent area, and E0 represents the degree of distinction between the rectangular area and the t-th adjacent area. Step S53. According to the punctuation radius of each rectangular area, the area is slightly reduced to the punctuation of the corresponding radius, and the point cloud image composed of all the punctuation is rendered and transmitted to the processing center as an accident image.

[0018] Compared with the prior art, the beneficial effects achieved by the present invention are: 1. The present invention can use historical accident images to build models of various accidents, perform surface projection on the model, calculate the distinction between before and after the accident at the same projection angle, select angles according to route risks and historical false alarm rates, and establish a picture verification model, which is helpful for comprehensive monitoring and investigation of the accident scene and improves the recognition accuracy.

[0019] 2. The present invention can continuously calculate the change in the total color value of pixels in the picture directly above during inspection, and compare it with the picture verification model. If the change law of the distinction conforms to the accident model, the type of accident is output, so as to more accurately understand the nature and degree of harm of the accident, call relevant departments in time, and carry out rescue and treatment work in a targeted manner, which is helpful to reasonably allocate resources, improve the efficiency and effectiveness of urban management assistance work, and reduce the losses caused by accidents.

[0020] 3. The present invention can detect the coordinates of various points on the accident scene when the transmission is not smooth, generate rendering points of different radii according to the grayscale of the coordinates, and transmit the rendering point cloud to the processing department, so as to help the processing personnel to understand the on-site situation in time when the network is not smooth, realize the rapid response and real-time monitoring of drones, and improve the efficiency of drone inspections in urban management. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings: Figure 1 It is a structural schematic diagram of a UAV data transmission system based on big data of the present invention; Figure 2 It is a schematic diagram of the steps of a UAV data transmission method based on big data of the present invention. DETAILED DESCRIPTION

[0022] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0023] See also Figure 1 ,The present invention provides a technical solution: a data transmission system for unmanned aerial vehicles based on big data, comprising: a model construction module, a program preset module, an inspection and positioning module, an accident processing module and a signal compression module; The model construction module is used to use a depth of field camera to shoot images of various accidents at different angles, construct a picture set of various accidents, compare the color gamut, depth of field and segmentation area size of each accident image at different angles, and calculate the difference between the image and the image directly above; The model construction module includes: an image training unit and a projection feature unit; The image training unit is used to classify pictures taken at historical accident scenes to form an accident training set; The projection feature unit is used to calculate the degree of distinction between images at each angle and the projection image directly above the accident.

[0024] The program preset module is used to calculate the number of verification angles according to the accident risk on the drone inspection route and the historical false alarm rate of the accident, select representative images of various types of accidents according to the principle of maximum distinction, and use the angles corresponding to the representative images as the verification angles of the accidents. The number of selected representative images is the same as the number of verification angles, and all verification angles corresponding to the accidents are stored in the drone memory, and the drone inspection program is preset; The program presetting module includes: an adaptation verification unit, an angle addressing unit and a patrol instruction unit; The adaptive verification unit is used to calculate the verification quantity according to the accident risk on the drone inspection route and the historical false alarm rate of accidents; The angle addressing unit is used to take the corresponding number of pictures with the greatest degree of distinction in the accident as representative images, and record the verification angles corresponding to the representative images; The inspection instruction unit is used to store the accident and the verification angle corresponding to the accident into the drone database and set the drone inspection program.

[0025] The inspection and positioning module is used to start the UAV inspection program. When the UAV flies at a fixed altitude, it calculates the function of the total color value of the picture taken directly above and the flight distance, and compares it with the standard function of the route. When the residual is higher than the threshold, it is identified as an accident and the accident area is located; The inspection and positioning module includes: a drone positioning unit and an abnormal marking unit; The drone positioning unit is used to determine the difference in total color value of the captured image relative to the standard image during drone inspection; The abnormal marking unit is used to enter the accident tracking mode and mark the location of the accident after the drone determines that an accident has occurred.

[0026] The accident processing module is used to take pictures of the accident at various angles in the order of accident risk when an accident occurs, calculate the degree of distinction between each picture and the picture taken directly above, use the picture with the greatest degree of distinction as the positioning image, use the accident type corresponding to the positioning image angle as the suspected accident, verify all verification angles of the suspected accident, and output the suspected accident as the accident type when the total degree of distinction difference is less than a threshold; The accident processing module includes: a program execution unit and an accident identification unit; The program execution unit is used to control the drone to take pictures at the verification angle corresponding to each accident according to the program preset in the drone; The accident identification unit is used to calculate the degree of distinction of the image relative to the image taken directly above, thereby identifying the type of accident and verifying the identified accident.

[0027] The signal compression module is used to detect the communication speed between the drone and the department corresponding to the accident. When the communication is not smooth, the picture of the accident scene is divided into sub-areas, and the distinction of each sub-area relative to the corresponding area of ​​the projection directly above is calculated. The rendering radius is determined according to the distinction. According to the rendering radius, a point cloud rendering algorithm is used to output the picture as a point cloud map to the relevant department.

[0028] The signal compression module includes: a channel detection unit, a grayscale rendering unit and a point cloud output unit; The channel detection unit is used to detect the speed of the communication connection between the drone and the accident handling center; The grayscale rendering unit is used to segment the accident representative image according to grayscale, and calculate the degree of distinction of each area relative to adjacent areas; The point cloud output unit is used to render each area into feature points of different radius according to the distinctiveness of the area, and output the formed point cloud map to the accident handling center.

[0029] like Figure 2 As shown, a method for transmitting data from a drone based on big data comprises the following steps: Step S1. Take pictures of various types of accidents at different angles to form an accident picture set, and calculate the difference between each picture in the accident picture set and the projected image directly above according to the picture comparison method; Step S1 includes: Step S11. Pre-take pictures of the accident at all possible angles that the drone can shoot, and store all the pictures taken in the database after marking the angles to form a collection of accident pictures; Step S12. Record the vertically shot accident picture as the top projection image, execute the picture comparison method in step S13, and calculate the difference between each picture in the accident picture set and the top projection image; Step S13. Use the image segmentation algorithm to segment the accident area in the overhead projection image and the accident image, record the area of ​​the segmented area, the color value of the pixel point and the depth of field, and calculate the difference between the overhead projection image and the accident image according to the following formula: ;

[0030] Wherein, E is the discrimination, D1 represents the segmented area of ​​the accident area in the overhead projection image, D2 represents the segmented area of ​​the accident area in the accident picture, n represents the number of pixel columns in the picture, m represents the number of pixel rows in the picture, UR(i,j) represents the color value of the pixel in the i-th column and j-th row of the overhead projection image, Ur(i,j) represents the color value of the pixel in the i-th column and j-th row of the accident picture, h is the depth of field of the overhead projection image at the pixel point (i,j), σ1 represents the horizontal shooting angle of the accident picture, σ2 represents the vertical shooting angle of the accident picture, and when tanσ1·h and tanσ2·h are not integers, they are taken as integers.

[0031] Step S2. Calculate the number of iterations according to the accident risk on the drone inspection route and the historical false alarm rate of accidents, sort the accident pictures in descending order according to the degree of distinction, select the pictures with the highest number of iterations of distinction as verification pictures, and preset the angles corresponding to all verification pictures as verification angles, and use the verification angle with the highest degree of distinction as the representative angle; Step S2 includes: Step S21. According to the historical inspection records of the drone, the frequency of the drone reporting accidents and the probability of the drone falsely reporting accidents are obtained from the records, and the number of iterations x is calculated, where x=Set[v·c / (p·a)], where v is the frequency of the drone reporting accidents, c is the probability of the drone falsely reporting accidents, p is the number of accident types, a is the preset descent rate, and Set is an upward value function; Step S22. Arrange all the pictures in the accident picture set in descending order according to the degree of difference from the projected image directly above, select the first picture as the representative picture of the accident, the first x pictures as the verification pictures of the accident, the angle corresponding to the representative picture is the representative angle, and the angle corresponding to the verification picture is the verification angle; Step S23: Store the verification image, representative image, and corresponding verification angles and representative angles in the drone memory.

[0032] Step S3. During the drone inspection, the function of the total color value of the projected image directly above with the distance is detected and compared with the function of the previous inspection. When the difference exceeds a threshold, it is determined that an accident has occurred under the drone, and the accident is located; Step S3 includes: Step S31. Calculate the function F1(d) of the total color value of the overhead projection image captured with the distance according to the video sent back by the drone during the last inspection, where d represents the distance and F(d) represents the sum of the color values ​​of all pixels in the image captured by the drone when the distance is d; Step S32. Start the drone to start the inspection. According to the distance the drone has flown, record the total color value of the pictures taken by the drone, and generate a function F2 (d) of the total color value of the projected image directly above during the inspection as a function of the distance. Step S33. Let the risk coefficient b = F1 (d0) - F2 (d0), where d0 is the distance the drone has flown, and the preset risk coefficient threshold b0. When b ≥ b0, it is determined that there is an abnormality in the current position, and the position of the drone is marked as the location where the accident occurred.

[0033] Step S4. The drone flies to the representative angle of each accident to take pictures, calculates the difference between the shooting result and the image projected directly above, takes the accident closest to the preset result as the suspected accident, verifies other verification angles corresponding to the suspected accident according to the accident verification method, and outputs the accident type after verification; Step S4 includes: Step S41. After taking the overhead projection image from directly above the accident location, control the drone to take pictures from representative angles of each accident, and calculate the difference between the accident picture and the overhead projection image for each picture taken; Step S42. After all pictures are taken, calculate the characteristic coefficient G of the accident picture, where G=|E1-E2|, where E1 represents the degree of distinction of the representative picture stored in the drone memory, and E2 represents the degree of distinction of the accident picture taken by the drone at the same angle; Step S43. Arrange all the captured accident pictures in ascending order according to the characteristic coefficients to form a suspected accident sequence. If there is an element in the suspected accident sequence, mark the first accident in the suspected accident sequence as a suspected accident, and go to step S44. If there is no element in the suspected accident sequence, output the accident type corresponding to the first accident in the suspected accident sequence to the processing center; Step S44. Control the drone to take a picture from the verification angle corresponding to the suspected accident, and calculate the difference between the accident picture taken and the projected image directly above, and compare it with the difference corresponding to the verification picture. If the difference is less than the threshold, it means that the picture verification has passed, otherwise it means that the verification has failed; If the number of images that pass the verification is greater than or equal to the number of images that fail the verification, the accident type corresponding to the suspected accident is output to the processing center, otherwise the current suspected accident is removed from the accident suspicion sequence and the process goes to step S43.

[0034] Step S5. When the communication rate between the UAV and the processing center is lower than the threshold, the captured image is segmented, the distinction between each segmented area and the adjacent area is calculated, the punctuation drawing radius is determined according to the distinction, and the punctuation of all areas is rendered as a point cloud image and sent to the processing center.

[0035] Step S5 includes: Step S51. When the communication rate between the drone and the accident handling center is lower than the threshold, the drone's image transmission is stopped, and the image to be sent is divided into k identical rectangular areas, where k is a preset resolution; Step S52. For each rectangular area, calculate the degree of distinction between the rectangular area and each adjacent area according to the method of step S13, and determine the punctuation radius according to the degree of distinction: ; Where s represents the punctuation radius, L represents the pixel arrangement width of the image, b represents the number of regions adjacent to the rectangular area, st represents the punctuation radius of the t-th adjacent area, bt represents the number of regions adjacent to the t-th adjacent area, and E0 represents the degree of distinction between the rectangular area and the t-th adjacent area. Step S53. According to the punctuation radius of each rectangular area, the area is slightly reduced to the punctuation of the corresponding radius, and the point cloud image composed of all the punctuation is rendered and transmitted to the processing center as an accident image.

[0036] Example: There are three possible types of accidents on the inspection route of the drone. The number of iterations is determined to be 3 according to the historical records. The representative angles of the accidents are 30°, 60° and 70° respectively. The difference between the representative image of the accident and the image directly above the projection is calculated to be 0.6, 0.4 and 0.8 respectively. During the actual inspection process of the drone, pictures are taken from 30°, 60° and 70° of the accident scene, and the distinctions between each picture and the projected image directly above are calculated to be 0.5, 0.8 and 0.6 respectively. Accident 1 is regarded as a suspected accident, and the two verification angles of accident 1, 15° and 45°, are verified. If both verifications fail, accident 3 is regarded as a suspected accident, and the two verification angles of accident 3, 65° and 85°, are verified. The obtained distinction is close to the predetermined distinction of accident 3, and accident 3 is output as the accident type.

[0037] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0038] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions described in the aforementioned embodiments or replace some of the technical features therein by equivalents. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A UAV data transmission method based on big data, characterized in that: The method comprises the following steps: Step S1. Take pictures of various types of accidents at different angles to form an accident picture set, and calculate the difference between each picture in the accident picture set and the projected image directly above according to the picture comparison method; Step S2. Calculate the number of iterations according to the accident risk on the drone inspection route and the historical false alarm rate of accidents, sort the accident pictures in descending order according to the degree of distinction, select the pictures with the highest number of iterations of distinction as verification pictures, and preset the angles corresponding to all verification pictures as verification angles, and use the verification angle with the highest degree of distinction as the representative angle; Step S3. During the drone inspection, the function of the total color value of the projected image directly above with the distance is detected and compared with the function of the previous inspection. When the difference exceeds a threshold, it is determined that an accident has occurred under the drone, and the accident is located; Step S4. The drone flies to the representative angle of each accident to take pictures, calculates the difference between the shooting result and the image projected directly above, takes the accident closest to the preset result as the suspected accident, verifies other verification angles corresponding to the suspected accident according to the accident verification method, and outputs the accident type after verification; Step S5. When the communication rate between the UAV and the processing center is lower than the threshold, the captured image is segmented, the distinction between each segmented area and the adjacent area is calculated, the punctuation drawing radius is determined according to the distinction, and the punctuation of all areas is rendered as a point cloud image and sent to the processing center.

2. The method for transmitting data from a drone based on big data according to claim 1, characterized in that: Step S1 includes: Step S11. Pre-take pictures of the accident at all possible angles that the drone can shoot, and store all the pictures taken in the database after marking the angles to form a collection of accident pictures; Step S12. Record the vertically shot accident picture as the top projection image, execute the picture comparison method in step S13, and calculate the difference between each picture in the accident picture set and the top projection image; Step S13. Use the image segmentation algorithm to segment the accident area in the overhead projection image and the accident image, record the area of ​​the segmented area, the color value of the pixel point and the depth of field, and calculate the difference between the overhead projection image and the accident image according to the following formula: ； Wherein, E is the discrimination, D1 represents the segmented area of ​​the accident area in the overhead projection image, D2 represents the segmented area of ​​the accident area in the accident picture, n represents the number of pixel columns in the picture, m represents the number of pixel rows in the picture, UR(i,j) represents the color value of the pixel in the i-th column and j-th row of the overhead projection image, Ur(i,j) represents the color value of the pixel in the i-th column and j-th row of the accident picture, h is the depth of field of the overhead projection image at the pixel point (i,j), σ1 represents the horizontal shooting angle of the accident picture, σ2 represents the vertical shooting angle of the accident picture, and when tanσ1·h and tanσ2·h are not integers, they are taken as integers.

3. The method for transmitting data via unmanned aerial vehicle based on big data according to claim 2, characterized in that: Step S2 includes: Step S21. According to the historical inspection records of the drone, the frequency of the drone reporting accidents and the probability of the drone falsely reporting accidents are obtained from the records, and the number of iterations x is calculated, where x=Set[v·c / (p·a)], where v is the frequency of the drone reporting accidents, c is the probability of the drone falsely reporting accidents, p is the number of accident types, a is the preset descent rate, and Set is an upward value function; Step S22. Arrange all the pictures in the accident picture set in descending order according to the degree of difference from the projected image directly above, select the first picture as the representative picture of the accident, the first x pictures as the verification pictures of the accident, the angle corresponding to the representative picture is the representative angle, and the angle corresponding to the verification picture is the verification angle; Step S23. Store the verification image, the representative image, and the corresponding verification angle and representative angle into the drone memory; Step S3 includes: Step S31. Calculate the function F1(d) of the total color value of the overhead projection image captured with the distance according to the video sent back by the drone during the last inspection, where d represents the distance and F(d) represents the sum of the color values ​​of all pixels in the image captured by the drone when the distance is d; Step S32. Start the drone to start the inspection. According to the distance the drone has flown, record the total color value of the pictures taken by the drone, and generate a function F2 (d) of the total color value of the projected image directly above during the inspection as a function of the distance. Step S33. Let the risk coefficient b = F1 (d0) - F2 (d0), where d0 is the distance the drone has flown, and the preset risk coefficient threshold b0. When b ≥ b0, it is determined that there is an abnormality in the current position, and the position of the drone is marked as the location where the accident occurred.

4. The method for transmitting data via unmanned aerial vehicle based on big data according to claim 3, characterized in that: Step S4 includes: Step S41. After taking the overhead projection image from directly above the accident location, control the drone to take pictures from representative angles of each accident, and calculate the difference between the accident picture and the overhead projection image for each picture taken; Step S42. After all pictures are taken, calculate the characteristic coefficient G of the accident picture, where G=|E1-E2|, where E1 represents the degree of distinction of the representative picture stored in the drone memory, and E2 represents the degree of distinction of the accident picture taken by the drone at the same angle; Step S43. Arrange all the captured accident pictures in ascending order according to the characteristic coefficients to form a suspected accident sequence. If there is an element in the suspected accident sequence, mark the first accident in the suspected accident sequence as a suspected accident, and go to step S44. If there is no element in the suspected accident sequence, output the accident type corresponding to the first accident in the suspected accident sequence to the processing center; Step S44. Control the drone to take a picture from the verification angle corresponding to the suspected accident, and calculate the difference between the accident picture taken and the projected image directly above, and compare it with the difference corresponding to the verification picture. If the difference is less than the threshold, it means that the picture verification has passed, otherwise it means that the verification has failed; If the number of images that pass the verification is greater than or equal to the number of images that fail the verification, the accident type corresponding to the suspected accident is output to the processing center, otherwise the current suspected accident is removed from the accident suspicion sequence and the process goes to step S43.

5. The method for transmitting data via unmanned aerial vehicle based on big data according to claim 4, characterized in that: Step S5 includes: Step S51. When the communication rate between the drone and the accident handling center is lower than the threshold, the drone's image transmission is stopped, and the image to be sent is divided into k identical rectangular areas, where k is a preset resolution; Step S52. For each rectangular area, calculate the degree of distinction between the rectangular area and each adjacent area according to the method of step S13, and determine the punctuation radius according to the degree of distinction: ； Where s represents the punctuation radius, L represents the pixel arrangement width of the image, b represents the number of regions adjacent to the rectangular area, st represents the punctuation radius of the t-th adjacent area, bt represents the number of regions adjacent to the t-th adjacent area, and E0 represents the degree of distinction between the rectangular area and the t-th adjacent area. Step S53. According to the punctuation radius of each rectangular area, the area is slightly reduced to the punctuation of the corresponding radius, and the point cloud image composed of all the punctuation is rendered and transmitted to the processing center as an accident image.

6. A UAV data transmission system based on big data, characterized in that: The system includes the following modules: a model construction module, a program preset module, an inspection and positioning module, an accident processing module and a signal compression module; The model construction module is used to use a depth of field camera to shoot images of various accidents at different angles, construct a picture set of various accidents, compare the color gamut, depth of field and segmentation area size of each accident image at different angles, and calculate the difference between the image and the image directly above; The program preset module is used to calculate the number of verification angles according to the accident risk on the drone inspection route and the historical false alarm rate of the accident, select representative images of various types of accidents according to the principle of maximum distinction, and use the angles corresponding to the representative images as the verification angles of the accidents. The number of selected representative images is the same as the number of verification angles, and all verification angles corresponding to the accidents are stored in the drone memory, and the drone inspection program is preset; The inspection and positioning module is used to start the UAV inspection program. When the UAV flies at a fixed altitude, it calculates the function of the total color value of the picture taken directly above and the flight distance, and compares it with the standard function of the route. When the residual is higher than the threshold, it is identified as an accident and the accident area is located; The accident processing module is used to take pictures of the accident at various angles in the order of accident risk when an accident occurs, calculate the degree of distinction between each picture and the picture taken directly above, use the picture with the greatest degree of distinction as the positioning image, use the accident type corresponding to the positioning image angle as the suspected accident, verify all verification angles of the suspected accident, and output the suspected accident as the accident type when the total degree of distinction difference is less than a threshold; The signal compression module is used to detect the communication speed between the drone and the department corresponding to the accident. When the communication is not smooth, the picture of the accident scene is divided into sub-areas, and the distinction of each sub-area relative to the corresponding area of ​​the projection directly above is calculated. The rendering radius is determined according to the distinction. According to the rendering radius, a point cloud rendering algorithm is used to output the picture as a point cloud map to the relevant department.

7. The big data-based drone data transmission system according to claim 6, characterized in that: The model construction module includes: an image training unit and a projection feature unit; The image training unit is used to classify pictures taken at historical accident scenes to form an accident training set; The projection feature unit is used to calculate the degree of distinction between images at each angle and the projection image directly above the accident.

8. The big data-based drone data transmission system according to claim 7, characterized in that: The program presetting module includes: an adaptation verification unit, an angle addressing unit and a patrol instruction unit; The adaptive verification unit is used to calculate the verification quantity according to the accident risk on the drone inspection route and the historical false alarm rate of accidents; The angle addressing unit is used to take the corresponding number of pictures with the greatest degree of distinction in the accident as representative images, and record the verification angles corresponding to the representative images; The inspection instruction unit is used to store the accident and the verification angle corresponding to the accident into the drone database and set the drone inspection program.

9. The big data-based drone data transmission system according to claim 8, characterized in that: The inspection and positioning module includes: a drone positioning unit and an abnormal marking unit; The drone positioning unit is used to determine the difference in total color value of the captured image relative to the standard image during drone inspection; The abnormal marking unit is used to enter the accident tracking mode and mark the location of the accident after the drone determines that an accident has occurred; The accident processing module includes: a program execution unit and an accident identification unit; The program execution unit is used to control the drone to take pictures at the verification angle corresponding to each accident according to the program preset in the drone; The accident identification unit is used to calculate the degree of distinction of the image relative to the image taken directly above, thereby identifying the type of accident and verifying the identified accident.

10. The big data-based drone data transmission system according to claim 9, characterized in that: The signal compression module includes: a channel detection unit, a grayscale rendering unit and a point cloud output unit; The channel detection unit is used to detect the speed of the communication connection between the drone and the accident handling center; The grayscale rendering unit is used to segment the accident representative image according to grayscale, and calculate the degree of distinction of each area relative to adjacent areas; The point cloud output unit is used to render each area into feature points of different radius according to the distinctiveness of the area, and output the formed point cloud map to the accident handling center.

Images