A low-code data transmission method and system for unmanned aerial vehicles

Abstract

The present invention relates to the field of low-code transmission technology, and specifically discloses a low-code data transmission method and system for unmanned aerial vehicles. The method includes regularly acquiring remote sensing data, identifying the remote sensing data, determining regional segmentation information and the stability of each region; sending the stability of each region to the unmanned aerial vehicle, receiving images with timestamps fed back by the unmanned aerial vehicle, and calculating the actual transmission frequency of the unmanned aerial vehicle based on the timestamps; calculating the standard transmission frequency based on the stability, comparing the standard transmission frequency with the actual transmission frequency, and correcting the stability of each region; when the unmanned aerial vehicle returns, determining the image acquisition ratio of each region based on the stability, and reading the image from the database of the unmanned aerial vehicle based on the image acquisition ratio; the present invention has a built-in image recognition algorithm on the unmanned aerial vehicle, and further introduces an upload judgment process on the basis of intermittent transmission, thereby reducing the amount of image upload and reducing the working pressure of the image receiving end.

Description

Technical Field

[0001] The present invention relates to the field of low-code transmission technology, and specifically to a low-code data transmission method and system for unmanned aerial vehicles. Background Art

[0002] In the process of field data collection, drones are a commonly used collection equipment. They are highly flexible and can obtain close-up images and obtain very detailed field data.

[0003] However, the transmission architecture of existing drones is relatively simple, either a global transmission mode or an intermittent transmission mode at a fixed frequency. The intermittent transmission mode is a type of low-code transmission mode. When the number of drones is large, the working pressure of the image receiving end will be very high. How to provide a simpler low-code transmission mode and further reduce the working pressure of the image receiving end is the technical problem that the technical solution of the present invention wants to solve. Summary of the Invention

[0004] The purpose of the present invention is to provide a low-code data transmission method and system for unmanned aerial vehicles to solve the problems raised in the above background technology.

[0005] To achieve the above object, the present invention provides the following technical solutions:

[0006] A low-code data transmission method and system for unmanned aerial vehicles, the method comprising:

[0007] acquiring remote sensing data at regular intervals, identifying the remote sensing data, and determining regional segmentation information and the stability of each region; the stability is represented by a stability value including a time period;

[0008] The stability of each area is sent to the drone, and the drone receives a time-stamped image feedback. The actual transmission frequency of the drone is calculated based on the time stamp. The drone queries the stability of the area where it is located in real time, obtains an image based on the stability, and recognizes the obtained image. The decision on whether to upload the image is made based on the stability and the image recognition result. The process of recognizing the obtained image is the image comparison process. When the drone obtains an image, the image acquisition location is recorded.

[0009] Calculate the standard transmission frequency based on the stability, compare the standard transmission frequency with the actual transmission frequency, and correct the stability of each area;

[0010] When the UAV returns, the image acquisition ratio of each area is determined according to the degree of stability, and the image is read from the UAV's database according to the image acquisition ratio.

[0011] As a further solution of the present invention, the steps of regularly acquiring remote sensing data, identifying the remote sensing data, and determining regional segmentation information and the stability of each region include:

[0012] Acquire remote sensing data of the drone monitoring area; the remote sensing data is a remote sensing image containing bands;

[0013] Perform contour recognition on the remote sensing image of each band and mark the contour points;

[0014] Determine region segmentation information and each region based on the marked contour points; the region segmentation information is a collection of contours determined by the contour points;

[0015] Remote sensing images containing bands in each region are identified and stability values ​​are calculated; the stability values ​​include spatial stability values ​​and temporal stability values, the spatial stability value is determined by the difference between the region and adjacent regions, and the temporal stability value is determined by the difference between the region and the same region at adjacent moments.

[0016] As a further solution of the present invention, the steps of identifying the remote sensing images containing bands in each region and calculating the stability value include:

[0017] Classify remote sensing images based on bands;

[0018] Perform frequency domain conversion on each type of remote sensing image to obtain a frequency domain map containing time for each region;

[0019] For a certain area at a certain moment, determine the areas that are spatially adjacent at the same moment, compare the frequency domain graphs of the two areas, and calculate the spatial stability value;

[0020] For a certain area at a certain moment, determine the adjacent areas in the time domain at the same location, compare the frequency domain graphs of the two areas, and calculate the time stability value;

[0021] The final stability value is determined based on the spatial stability value and the temporal stability value.

[0022] As a further solution of the present invention, the steps of sending the stability level of each area to the drone, receiving the image with a timestamp fed back by the drone, and calculating the actual transmission frequency of the drone based on the timestamp include:

[0023] Send the stability value of each area to the drone;

[0024] Receive time-stamped images from the drone;

[0025] Identify the validity of the image, read the timestamp of the valid image, and generate a time series;

[0026] The actual transmission frequency of the UAV in each time period is determined based on the difference of the time series, wherein the time period is determined by the elements in the time series.

[0027] As a further solution of the present invention, the steps of calculating the standard transmission frequency according to the stability, comparing the standard transmission frequency with the actual transmission frequency, and correcting the stability of each area include:

[0028] Read the stability of the area where the drone is located, and query the standard transmission frequency corresponding to the stability in the preset frequency table;

[0029] Comparing the standard transmission frequency and the actual transmission frequency, calculating the frequency ratio of each time period; the frequency ratio is the actual transmission frequency divided by the standard transmission frequency;

[0030] A correction coefficient is determined according to the frequency ratio of each time period, and the stability of the area is corrected based on the correction coefficient; the correction coefficient is inversely proportional to the frequency ratio of each time period.

[0031] As a further solution of the present invention: when the drone returns, the image acquisition ratio of each area is determined according to the stability, and the step of reading images from the drone database according to the image acquisition ratio includes:

[0032] When the drone returns, the stability of each area determined during the collection process is counted;

[0033] Calculate the reciprocal of the stability level, calculate the proportional relationship between the reciprocals, and determine the image acquisition ratio of each area;

[0034] Based on the image acquisition ratio of each area, the corresponding image of each area is read from the UAV database;

[0035] The process of reading the image includes a position determination process, which is used to determine whether the position of the image belongs to the area.

[0036] As a further embodiment of the present invention, the method further comprises:

[0037] On the drone side, the stability of the area where it is located is checked in real time;

[0038] Searching a preset frequency table for an image acquisition frequency corresponding to the degree of stability;

[0039] Acquire an image based on the image acquisition frequency, input the acquired image into a trained convolutional recognition model, and identify abnormal objects and their abnormality levels in the image;

[0040] When the abnormality levels of all abnormal objects in the image meet the preset abnormality conditions, the image is determined to be uploaded.

[0041] The technical solution of the present invention also provides a low-code data transmission system for a drone, the system comprising:

[0042] A stability determination module is used to regularly acquire remote sensing data, identify the remote sensing data, and determine regional segmentation information and the stability of each region; the stability is represented by a stability value containing a time period;

[0043] The frequency inversion module is used to send the stability level of each area to the drone, receive the time-stamped image feedback from the drone, and calculate the actual transmission frequency of the drone based on the time stamp. The drone queries the stability level of the area where it is located in real time, obtains images based on the stability level, and recognizes the obtained images. The decision on whether to upload the images is made based on the stability level and the image recognition results. The process of recognizing the obtained images is the image comparison process. When the drone obtains images, the image acquisition location is recorded.

[0044] The stability determination module is used to calculate the standard transmission frequency according to the stability level, compare the standard transmission frequency with the actual transmission frequency, and correct the stability level of each area;

[0045] The image reading module is used to determine the image acquisition ratio of each area according to the stability when the drone returns, and read the image from the drone's database according to the image acquisition ratio.

[0046] As a further solution of the present invention: the stability determination module includes:

[0047] A remote sensing data acquisition unit, configured to acquire remote sensing data of the drone monitoring area; the remote sensing data is a remote sensing image containing bands;

[0048] Contour recognition unit, used to perform contour recognition on the remote sensing image of each band and mark contour points;

[0049] A region segmentation unit, configured to determine region segmentation information and each region based on the marked contour points; the region segmentation information is a collection of contours determined by the contour points;

[0050] The stability value calculation unit is used to identify the remote sensing images containing bands in each area and calculate the stability value; the stability value includes a spatial stability value and a temporal stability value. The spatial stability value is determined by the difference between the area and the adjacent areas, and the temporal stability value is determined by the difference between the area and the same area at adjacent moments.

[0051] As a further solution of the present invention: the frequency inversion module includes:

[0052] A stability value sending unit, used to send the stability value of each area to the drone;

[0053] An image receiving unit, used to receive images with timestamps fed back by the drone;

[0054] A time series generation unit is used to identify the validity of images, read the timestamps of valid images, and generate time series;

[0055] A difference application unit is used to determine the actual transmission frequency of the drone in each time period based on the difference of the time series; wherein the time period is determined by the elements in the time series.

[0056] Compared with the prior art, the present invention has the following beneficial effects:

[0057] The present invention incorporates an image recognition algorithm into the drone. Based on intermittent transmission, it further introduces an upload determination process, which reduces the amount of image uploads and alleviates the workload of the image receiving end. At the same time, the image receiving end can also infer which areas are at risk based on the specific amount of image uploads. When the drone returns, it reads the corresponding images, thus largely avoiding data omissions. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, 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 invention.

[0059] Figure 1 The overall flow chart of the low-code data transmission method for drones is shown.

[0060] Figure 2 A structural diagram of a low-code data transmission system for drones is shown. DETAILED DESCRIPTION

[0061] In order to make the technical problems, technical solutions and beneficial effects to be solved by the present invention more clearly understood, the present invention is further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0062] Figure 1 The following is a general flow chart of a low-code data transmission method and system for drones. In an embodiment of the present invention, a low-code data transmission method for drones includes:

[0063] Step S100: acquiring remote sensing data at regular intervals, identifying the remote sensing data, and determining regional segmentation information and the stability of each region; the stability is represented by a stability value including a time period;

[0064] The application scenario of this application is the data collection scenario in the wild area, and remote sensing data of the wild area is obtained at regular intervals. The remote sensing data is images of different bands, and the data obtained in different bands have different focuses. For example, the visible light band can directly reflect the color of surface objects, so it can provide intuitive images; the near-infrared band is invisible to the human eye, but can be detected by remote sensing equipment, which can highlight the difference between vegetation and non-vegetation, especially in vegetation growth and health status analysis. It has high sensitivity; the mid-infrared band has strong penetrating ability and good perception of thermal energy, and is widely used in fire monitoring, urban heat island effect and other aspects.

[0065] The acquired remote sensing data can be identified and the entire area to be monitored can be segmented into multiple regions. After the multiple regions are obtained, the stability of each region is calculated based on the remote sensing data of each region. The stability is used to characterize the stability of the region. There are intervals in the acquisition process of remote sensing data. For example, if it is acquired every two hours, the stability calculated at each moment is actually the stability for the next two hours. Therefore, it is also called the stability containing the time period and is represented by the stability value containing the time period.

[0066] Step S200: Send the stability level of each area to the drone, receive the image with a time stamp fed back by the drone, and calculate the actual transmission frequency of the drone based on the time stamp; wherein, on the drone side, the stability level of the area where it is located is queried in real time, an image is acquired based on the stability level, and the acquired image is recognized, and whether to upload the image is determined based on the stability level and the image recognition result; the process of recognizing the acquired image is an image comparison process; when the drone acquires an image, the image acquisition location is recorded;

[0067] The stability of each area is sent to the drone. After receiving the stability of each area sent by the main end, the drone obtains its own position in real time, determines which area it is in, reads the stability of the corresponding area, and determines the image acquisition frequency based on the stability. The higher the stability, the lower the image acquisition frequency. After acquiring the image based on the image acquisition frequency, the drone locally recognizes the image and determines whether to upload the image. If the judgment result is to upload the image, then the image will be uploaded immediately. If the judgment result is not to upload the image, then the image will not be uploaded. In addition, when acquiring images, the drone needs to record the acquisition time and location of each image.

[0068] Regarding the working process of the drone, in summary, it has two judgment processes. First, the image acquisition frequency is determined based on the degree of stability, and then the acquired image is judged for the second time to determine whether to upload it. Therefore, in fact, the frequency of the drone uploading images is the frequency of the images uploaded after the second judgment. From the perspective of the execution subject (main end) of this method, the image with a timestamp fed back by the drone is received, and the actual transmission frequency of the drone is calculated according to the timestamp. The actual transmission frequency is the frequency of the image uploaded by the drone after the second judgment.

[0069] Step S300: Calculate the standard transmission frequency according to the stability, compare the standard transmission frequency with the actual transmission frequency, and correct the stability of each area;

[0070] The relationship between stability and standard transmission frequency is fixed. The standard transmission frequency is calculated at the main end based on the stability. This process is actually equivalent to the first judgment process of the drone (the image acquisition frequency is also determined based on the stability). The standard transmission frequency and the image acquisition frequency are numerically the same, but the calculation subjects are different. The standard transmission frequency is calculated by the main end, and the image acquisition frequency is calculated by the drone end. The calculation processes of the two are independent. At the main end, by comparing the calculated actual transmission frequency and the standard transmission frequency, the judgment process of the drone can be roughly inferred. Only two values ​​need to be compared, and there is no need to identify the images uploaded by the drone, which is extremely efficient. Especially when the number of drones is large, there are also many images fed back. If the images are to be identified, a lot of work is required. If only two values ​​are compared for each drone, a lot of work can be saved.

[0071] Since the images fed back by the drone contain location information, when comparing the standard transmission frequency and the actual transmission frequency, what is actually being compared is the standard transmission frequency and the actual transmission frequency of the same area. The stability of each area is corrected based on the comparison results; at this time, the corrected stability is more in line with the actual situation.

[0072] To facilitate understanding of the above content, an example is provided as follows: the stability of each area is determined based on remote sensing data every three hours. Then, during these three hours, the drone will collect regional information multiple times. After each collection is completed, the stability of each area will be updated. The drone can be used to increase the update frequency of the stability of each area.

[0073] Step S400: When the drone returns, the image acquisition ratio of each area is determined according to the stability, and the image is read from the drone's database according to the image acquisition ratio;

[0074] When the drone returns, the execution subject of this method has already obtained the stability level of each area, and can read images from the drone's database based on the stability level. The specific reading method is to determine the image acquisition ratio of each area based on the stability level, and read images from the drone's database based on the image acquisition ratio. This is a sampling method. For areas with high stability, the corresponding image acquisition ratio will be lower, and the number of images read will be smaller.

[0075] In addition, after the drone returns, it is very simple to read the image in the drone's database. The image can be read directly through a wired connection, such as a data cable and a card reader, which is extremely convenient.

[0076] In the technical solution of the present invention, the execution subject of this method actually only needs to identify the remote sensing data at regular intervals. Within a period of time, it only needs to compare some values ​​to update the stability of each area. In addition, when reading the image, it is also a wired reading method, and the workload of the execution subject of this method is extremely small.

[0077] Regarding step S100, the steps of regularly acquiring remote sensing data, identifying the remote sensing data, and determining region segmentation information and the stability of each region include:

[0078] Acquire remote sensing data of the drone monitoring area; the remote sensing data is a remote sensing image containing bands;

[0079] Perform contour recognition on the remote sensing image of each band and mark the contour points;

[0080] Determine region segmentation information and each region based on the marked contour points; the region segmentation information is a collection of contours determined by the contour points;

[0081] Remote sensing images containing bands in each region are identified and stability values ​​are calculated; the stability values ​​include spatial stability values ​​and temporal stability values, the spatial stability value is determined by the difference between the region and adjacent regions, and the temporal stability value is determined by the difference between the region and the same region at adjacent moments.

[0082] The above content describes the recognition process of remote sensing data. In this application, the area to be detected is called the drone monitoring area, and remote sensing data of the drone monitoring area is obtained. The remote sensing data is a remote sensing image containing bands; contour recognition is performed on the remote sensing image of each band, and contour points are marked; the contour recognition scheme adopts the existing contour recognition scheme applied to remote sensing data. The existing contour recognition schemes are basically based on the recognition scheme of the color value difference of pixel points; by connecting the identified contour points, the contours of each area (region segmentation information and each area) can be obtained. Finally, the remote sensing image containing the bands of each area is identified, and the stability value is calculated.

[0083] Specifically, the steps of identifying the remote sensing images containing bands in each region and calculating the stability value include:

[0084] Classify remote sensing images based on bands;

[0085] Perform frequency domain conversion on each type of remote sensing image to obtain a frequency domain map containing time for each region;

[0086] For a certain area at a certain moment, determine the areas that are spatially adjacent at the same moment, compare the frequency domain graphs of the two areas, and calculate the spatial stability value;

[0087] For a certain area at a certain moment, determine the adjacent areas in the time domain at the same location, compare the frequency domain graphs of the two areas, and calculate the time stability value;

[0088] The final stability value is determined based on the spatial stability value and the temporal stability value;

[0089] The conversion process of the frequency domain diagram is: Where F(u, v) represents the value at point (u, v) in the frequency domain image, f(x, y) represents the value at point (x, y) in the spatial domain image, N and M are the maximum length and width of the spatial domain image, respectively. When there is no pixel at point (x, y) in the spatial domain image, f(x, y) is set to zero.

[0090] The comparison process of the frequency domain graph is:

[0091] The information radius is determined with the origin of the frequency domain graph as the center. The information radius is determined as follows: Where R is the final information radius determined, Indicates finding the maximum value of parameter S, ∈ is the preset threshold, Indicates the conditions that parameter S satisfies.

[0092] Regarding the information radius, the low-frequency information in the frequency domain graph is concentrated near the origin. An increasing radius is determined, and a continuously expanding circle is determined with the origin as the center. The data sum of each point in the circle is calculated (usually the amplitude sum is used). Under the premise that it does not exceed the preset threshold, the maximum radius that can be achieved is the information radius.

[0093] Comparison information radius, as the comparison result of frequency domain graph;

[0094] The calculation process of spatial stability value is: Wherein, R(i, j) represents the area to be calculated, R(i-1, j), R(i+1, j), R(i, j-1) and R(i, j+1) represent adjacent areas, respectively, where i represents the i-th area in the horizontal direction (generally from left to right) in the spatial domain, and j represents the j-th area in the vertical direction (generally from top to bottom) in the spatial domain.

[0095] The calculation process of the time stable value is: Where R t Indicates the information radius of the area at the current moment, R t+1 and R t-1 They represent the information radius of the same area at the next moment and the previous moment respectively.

[0096] In the above content, the frequency map is simplified into the information radius, which simplifies the comparison process. The comparison process includes two parts, one is the spatial domain comparison, and the other is the time domain comparison. For the spatial domain comparison process, a four-neighborhood comparison scheme is adopted. For the time domain comparison process, the current moment is compared with the two adjacent moments before and after. It is worth mentioning that for the boundaries in the spatial domain and the endpoints in the time domain, there will be some missing data in the above calculation process. At this time, adaptive adjustment can be made to calculate the calculable values ​​and then calculate the mean. For example, when there are only three spatially adjacent areas, the coefficient of W1 will become 1 / 3, and there are only three differences that need to be calculated.

[0097] Regarding step S200, the steps of sending the stability level of each area to the drone, receiving the image with a timestamp fed back by the drone, and calculating the actual transmission frequency of the drone based on the timestamp include:

[0098] Send the stability value of each area to the drone;

[0099] Receive time-stamped images from the drone;

[0100] Identify the validity of the image, read the timestamp of the valid image, and generate a time series;

[0101] The actual transmission frequency of the UAV in each time period is determined based on the difference of the time series, wherein the time period is determined by the elements in the time series.

[0102] The above content explains the calculation process of the actual transmission frequency. Since the stability level is limited to a stable value, the stable value of each area is sent to the drone. After receiving the stable value, the drone will perform independent operations and feedback an image with a timestamp; the image is identified for validity. The process of validity identification is very simple, which is to detect whether there are abnormal problems such as large-area occlusion in the image, which generally do not exist. The timestamp of the valid image is read to generate a time series. The time series is the time at which the drone uploaded the image when it was in the area; the difference of the time series is calculated to obtain the interval between the drone uploading images when it was in the area, thereby calculating the actual transmission frequency of each time period.

[0103] Specifically, the calculation process is generally to subtract the previous time from the next time. The last time is not calculated and is placed in the same time period as the previous time. For example:

[0104] Assume that two images are uploaded at 1:00 and 1:05, and the time period is from 1:00 to 1:05. The actual transmission frequency is 1 divided by 5 minutes. In reality, the interval between image uploads is in seconds. The above example is just for ease of understanding and expands the time span.

[0105] Regarding step S300, the steps of calculating the standard transmission frequency based on the stability, comparing the standard transmission frequency with the actual transmission frequency, and correcting the stability of each area include:

[0106] Read the stability of the area where the drone is located, and query the standard transmission frequency corresponding to the stability in the preset frequency table;

[0107] Comparing the standard transmission frequency and the actual transmission frequency, calculating the frequency ratio of each time period; the frequency ratio is the actual transmission frequency divided by the standard transmission frequency;

[0108] A correction coefficient is determined according to the frequency ratio of each time period, and the stability of the area is corrected based on the correction coefficient; the correction coefficient is inversely proportional to the frequency ratio of each time period.

[0109] In one example of the technical solution of the present invention, the stability of the area where the drone is located is read, and the standard transmission frequency corresponding to the stability is queried in a preset frequency table. This process actually has another hidden meaning, that is, the way the drone side determines the image acquisition frequency is also the same. On the drone side, its own position is obtained in real time, the area it is located in is determined, and then the image acquisition frequency corresponding to the stability is queried in the preset frequency table.

[0110] The actual transmission frequency corresponds to the second judgment process. It is not greater than the standard transmission frequency. The ratio of the actual transmission frequency to the standard transmission frequency is calculated. The larger the actual transmission frequency is, the more the drone believes that all acquired images need to be uploaded during the second judgment process. At this time, it means that the actual situation is more complicated and the stability needs to be reduced. In other words, the larger the result of dividing the actual transmission frequency by the standard transmission frequency, the more unstable it is, and the smaller the stability after correction. When a correction coefficient is used in the correction process, the correction coefficient is inversely proportional to the frequency ratio of each time period.

[0111] Regarding step S400, when the drone returns, determining the image acquisition ratio of each area according to the stability, and reading images from the drone's database according to the image acquisition ratio includes:

[0112] When the drone returns, the stability of each area determined during the collection process is counted;

[0113] Calculate the reciprocal of the stability level, calculate the proportional relationship between the reciprocals, and determine the image acquisition ratio of each area;

[0114] Based on the image acquisition ratio of each area, the corresponding image of each area is read from the UAV database;

[0115] The process of reading the image includes a position determination process, which is used to determine whether the position of the image belongs to the area.

[0116] In an example of the technical solution of the present invention, the image reading process is described in detail. When the drone returns, the stability of each area determined during the acquisition process is counted, the inverse of the stability is calculated, and then the proportional relationship between the inverses is calculated as the image acquisition ratio; the higher the stability of the area, the smaller the corresponding image acquisition ratio; on this basis, the total number of image reads pre-set by the management personnel is queried, and the number of image reads is allocated to each area based on the image acquisition ratio. Then, the image set corresponding to each area is queried in the database, and the image reading number of images is randomly read.

[0117] As an example of the technical solution of the present invention, the method further includes:

[0118] On the drone side, the stability of the area where it is located is checked in real time;

[0119] Searching a preset frequency table for an image acquisition frequency corresponding to the degree of stability;

[0120] Acquire an image based on the image acquisition frequency, input the acquired image into a trained convolutional recognition model, and identify abnormal objects and their abnormality levels in the image;

[0121] When the abnormality levels of all abnormal objects in the image meet the preset abnormality conditions, the image is determined to be uploaded.

[0122] The above content specifically defines the working process of the drone. The process of image acquisition and determining whether to upload has been explained before. The process of identifying the acquired images is as follows: the trained convolutional recognition model is used to identify the image. The training process of the convolutional recognition model is not complicated. First, some abnormal images of abnormal samples are obtained in advance, and the management personnel determine their abnormality levels. That is, the abnormal samples correspond to the abnormality levels one by one and the corresponding relationship is pre-set; the image features of the abnormal samples are extracted from the abnormal images, which are called convolution kernels. When there are more abnormal images of abnormal samples, the number of convolution kernels extracted by the convolutional recognition model increases, and its recognition function becomes more and more perfect.

[0123] In practical applications, the convolutional recognition model can be used to quickly identify abnormal objects in images. After identifying the abnormal objects, the abnormality level is queried based on the preset correspondence. Finally, the abnormality levels of all abnormal objects are accumulated to obtain the total abnormality level. When the total abnormality level meets the preset level threshold, it is determined to be an uploaded image.

[0124] It should be noted that the smaller the level threshold is set, the easier it is to upload images and the more resources are consumed. Therefore, its specific value needs to be determined by the staff based on the actual situation. For example, when there are fewer drones, the level threshold can be set smaller, and the central end will receive more images. When there are more drones, the level threshold can be set larger, and the same drone will upload fewer images, and the working pressure of the central end will be much smaller. This is also the meaning of low-code transmission in this application.

[0125] Figure 2 A structural diagram of a low-code data transmission system for a drone is shown. In a preferred embodiment of the technical solution of the present invention, a low-code data transmission system for a drone is also provided. The system 10 includes:

[0126] The stability determination module 11 is used to regularly acquire remote sensing data, identify the remote sensing data, and determine regional segmentation information and the stability of each region; the stability is represented by a stability value containing a time period;

[0127] The frequency inversion module 12 is used to send the stability level of each area to the drone, receive the image with a time stamp fed back by the drone, and calculate the actual transmission frequency of the drone based on the time stamp. The drone queries the stability level of the area where it is located in real time, obtains an image based on the stability level, and recognizes the obtained image. The decision on whether to upload the image is made based on the stability level and the image recognition result. The process of recognizing the obtained image is the image comparison process. When the drone obtains an image, the image acquisition location is recorded.

[0128] The stability determination module 13 is used to calculate the standard transmission frequency according to the stability level, compare the standard transmission frequency with the actual transmission frequency, and correct the stability level of each area;

[0129] The image reading module 14 is used to determine the image acquisition ratio of each area according to the stability when the drone returns, and read the image from the drone's database according to the image acquisition ratio.

[0130] Furthermore, the stability determination module 11 includes:

[0131] A remote sensing data acquisition unit, configured to acquire remote sensing data of the drone monitoring area; the remote sensing data is a remote sensing image containing bands;

[0132] Contour recognition unit, used to perform contour recognition on the remote sensing image of each band and mark contour points;

[0133] A region segmentation unit, configured to determine region segmentation information and each region based on the marked contour points; the region segmentation information is a collection of contours determined by the contour points;

[0134] The stability value calculation unit is used to identify the remote sensing images containing bands in each area and calculate the stability value; the stability value includes a spatial stability value and a temporal stability value. The spatial stability value is determined by the difference between the area and the adjacent areas, and the temporal stability value is determined by the difference between the area and the same area at adjacent moments.

[0135] Specifically, the frequency inversion module 12 includes:

[0136] A stability value sending unit, used to send the stability value of each area to the drone;

[0137] An image receiving unit, used to receive images with timestamps fed back by the drone;

[0138] A time series generation unit is used to identify the validity of images, read the timestamps of valid images, and generate time series;

[0139] A difference application unit is used to determine the actual transmission frequency of the drone in each time period based on the difference of the time series; wherein the time period is determined by the elements in the time series.

[0140] The above are only preferred embodiments of the present invention and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention description and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A low-code data transmission method for drones, characterized in that: The method comprises: acquiring remote sensing data at regular intervals, identifying the remote sensing data, and determining regional segmentation information and the stability of each region; the stability is represented by a stability value including a time period; The stability of each area is sent to the drone, and the drone receives a time-stamped image feedback. The actual transmission frequency of the drone is calculated based on the time stamp. The drone queries the stability of the area where it is located in real time, obtains an image based on the stability, and recognizes the obtained image. The decision on whether to upload the image is made based on the stability and the image recognition result. The process of recognizing the obtained image is the image comparison process. When the drone obtains an image, the image acquisition location is recorded. Calculate the standard transmission frequency based on the stability, compare the standard transmission frequency with the actual transmission frequency, and correct the stability of each area; When the drone returns, the image acquisition ratio of each area is determined according to the stability, and the image is read from the drone's database according to the image acquisition ratio; The steps of regularly acquiring remote sensing data, identifying the remote sensing data, and determining regional segmentation information and the stability of each region include: Acquire remote sensing data of the drone monitoring area; the remote sensing data is a remote sensing image containing bands; Perform contour recognition on the remote sensing image of each band and mark the contour points; Determine region segmentation information and each region based on the marked contour points; the region segmentation information is a collection of contours determined by the contour points; Remote sensing images containing bands in each region are identified and stability values ​​are calculated; the stability values ​​include spatial stability values ​​and temporal stability values, the spatial stability value is determined by the difference between the region and adjacent regions, and the temporal stability value is determined by the difference between the region and the same region at adjacent moments.

2. The low-code data transmission method for drones according to claim 1, characterized in that: The steps of identifying the remote sensing images containing bands in each region and calculating the stability value include: Classify remote sensing images based on bands; Perform frequency domain conversion on each type of remote sensing image to obtain a frequency domain map containing time for each region; For a certain area at a certain moment, determine the areas that are spatially adjacent at the same moment, compare the frequency domain graphs of the two areas, and calculate the spatial stability value; For a certain area at a certain moment, determine the adjacent areas in the time domain at the same location, compare the frequency domain graphs of the two areas, and calculate the time stability value; The final stability value is determined based on the spatial stability value and the temporal stability value.

3. The low-code data transmission method for drones according to claim 1, characterized in that: The steps of sending the stability level of each area to the drone, receiving an image with a timestamp fed back by the drone, and calculating the actual transmission frequency of the drone based on the timestamp include: Send the stability value of each area to the drone; Receive time-stamped images from the drone; Identify the validity of the image, read the timestamp of the valid image, and generate a time series; The actual transmission frequency of the UAV in each time period is determined based on the difference of the time series, wherein the time period is determined by the elements in the time series.

4. The low-code data transmission method for drones according to claim 1, characterized in that: The steps of calculating the standard transmission frequency according to the stability, comparing the standard transmission frequency with the actual transmission frequency, and correcting the stability of each area include: Read the stability of the area where the drone is located, and query the standard transmission frequency corresponding to the stability in the preset frequency table; Comparing the standard transmission frequency and the actual transmission frequency, calculating the frequency ratio of each time period; the frequency ratio is the actual transmission frequency divided by the standard transmission frequency; A correction coefficient is determined according to the frequency ratio of each time period, and the stability of the area is corrected based on the correction coefficient; the correction coefficient is inversely proportional to the frequency ratio of each time period.

5. The low-code data transmission method for drones according to claim 1, characterized in that: When the drone returns, the image acquisition ratio of each area is determined according to the stability, and the step of reading images from the drone database according to the image acquisition ratio includes: When the drone returns, the stability of each area determined during the collection process is counted; Calculate the reciprocal of the stability level, calculate the proportional relationship between the reciprocals, and determine the image acquisition ratio of each area; Based on the image acquisition ratio of each area, the corresponding image of each area is read from the UAV database; The process of reading the image includes a position determination process, which is used to determine whether the position of the image belongs to the area.

6. The low-code data transmission method for drones according to claim 1, characterized in that: The method further comprises: On the drone side, the stability of the area where it is located is checked in real time; Searching a preset frequency table for an image acquisition frequency corresponding to the degree of stability; Acquire an image based on the image acquisition frequency, input the acquired image into a trained convolutional recognition model, and identify abnormal objects and their abnormality levels in the image; When the abnormality levels of all abnormal objects in the image meet the preset abnormality conditions, the image is determined to be uploaded.

7. A low-code data transmission system for drones, characterized in that: The system comprises: A stability determination module is used to regularly acquire remote sensing data, identify the remote sensing data, and determine regional segmentation information and the stability of each region; the stability is represented by a stability value containing a time period; The frequency inversion module is used to send the stability level of each area to the drone, receive the time-stamped image feedback from the drone, and calculate the actual transmission frequency of the drone based on the time stamp. The drone queries the stability level of the area where it is located in real time, obtains images based on the stability level, and recognizes the obtained images. The decision on whether to upload the images is made based on the stability level and the image recognition results. The process of recognizing the obtained images is the image comparison process. When the drone obtains images, the image acquisition location is recorded. The stability determination module is used to calculate the standard transmission frequency according to the stability level, compare the standard transmission frequency with the actual transmission frequency, and correct the stability level of each area; The image reading module is used to determine the image acquisition ratio of each area according to the stability when the drone returns, and read the image from the drone's database according to the image acquisition ratio; The stability determination module includes: A remote sensing data acquisition unit, configured to acquire remote sensing data of the drone monitoring area; the remote sensing data is a remote sensing image containing bands; Contour recognition unit, used to perform contour recognition on the remote sensing image of each band and mark contour points; A region segmentation unit, configured to determine region segmentation information and each region based on the marked contour points; the region segmentation information is a collection of contours determined by the contour points; The stability value calculation unit is used to identify the remote sensing images containing bands in each area and calculate the stability value; the stability value includes a spatial stability value and a temporal stability value. The spatial stability value is determined by the difference between the area and the adjacent areas, and the temporal stability value is determined by the difference between the area and the same area at adjacent moments.

8. The low-code data transmission system for drones according to claim 7, characterized in that: The frequency inversion module includes: A stability value sending unit, used to send the stability value of each area to the drone; An image receiving unit, used to receive images with timestamps fed back by the drone; A time series generation unit is used to identify the validity of images, read the timestamps of valid images, and generate time series; A difference application unit is used to determine the actual transmission frequency of the drone in each time period based on the difference of the time series; wherein the time period is determined by the elements in the time series.

Images