A target positioning method and system combining vision and satellite navigation

Abstract

The present invention relates to the technical field of target positioning, and specifically discloses a target positioning method combining vision and satellite navigation. The method includes obtaining satellite images of an application area at different precisions, extracting image features in the target area of the satellite images at different precisions as target features; statistically analyzing the target features and their corresponding precisions to obtain a target feature table; in the application stage, obtaining satellite images based on a preset acquisition frequency, positioning the target in the satellite images based on the target feature table, determining the operation path of the unmanned aerial vehicle based on the target position, sending the operation path to the unmanned aerial vehicle, and the unmanned aerial vehicle obtaining supplementary video data of the target; the present invention first preliminarily obtains the target position by satellite, and then the unmanned aerial vehicle performs data-assisted acquisition. The satellite has low precision requirements and low cost. The unmanned aerial vehicle only needs to be controlled on the ground and does not involve high-altitude operations, providing a target positioning solution with low cost and extremely high safety.

Description

Technical Field

[0001] The present invention relates to the technical field of target positioning, and in particular, to a target positioning method and system combining vision and satellite navigation. Background Art

[0002] There is a need for target positioning in many existing areas, mainly in outdoor areas. For example, in a breeding area, the simplest target positioning solution is to use a watchtower. Workers can perform manual positioning at a high place with the help of a telescope. This solution is relatively convenient and highly flexible. However, the outdoor environment is relatively harsh, and the positioning work is at a high altitude. Even with safety facilities such as air conditioners or safety ropes, there are still certain risks. Based on this, how to provide a target positioning solution without high-altitude operation is the technical problem that the technical solution of the present invention wants to solve. Summary of the Invention

[0003] The purpose of the present invention is to provide a target positioning method and system combining vision and satellite navigation to solve the problems raised in the above background art.

[0004] To achieve the above purpose, the present invention provides the following technical solutions:

[0005] A target positioning method combining vision and satellite navigation, the method comprising:

[0006] Receiving the application area selected by the requester, obtaining and displaying the satellite image with the highest precision of the application area, and receiving the target area calibrated by the requester;

[0007] Obtaining satellite images of the application area at different precisions, extracting the image features of the satellite images at different precisions in the target area as target features;

[0008] Statistical target features and their corresponding precisions to obtain a target feature table;

[0009] In the application stage, obtaining satellite images based on a preset acquisition frequency, positioning the target in the satellite images based on the target feature table, and obtaining the target position; wherein, each time a satellite image is obtained, the precision is randomly determined;

[0010] Determining the flight path of the UAV based on the target position, sending the flight path to the UAV, and obtaining supplementary video data of the target by the UAV; wherein, during the operation of the UAV, determining the coverage rate of the target feature table and feeding it back to the requester.

[0011] As a further solution of the present invention: the step of obtaining satellite images of the application area at different precisions, extracting the image features of the satellite images at different precisions in the target area as target features includes:

[0012] Obtain satellite images of the application area at different precisions;

[0013] For the satellite images at each precision, read the target area calibrated by the requester, and intercept the sub-image corresponding to the target area in the satellite image;

[0014] Perform Gaussian convolution on the sub-image at different scales, and synchronously calculate the texture complexity;

[0015] Select the scale corresponding to the maximum texture complexity, and use this scale and the maximum texture complexity as the target features at the corresponding precision.

[0016] As a further solution of the present invention: the content of performing Gaussian convolution on the sub-image at different scales and synchronously calculating the texture complexity includes:

[0017] L(x, y, σ) = G(x, y, σ) * I(x, y);

[0018]

[0019] In the formula, L(x, y, σ) is the value of the point (x, y) in the sub-image after Gaussian convolution, I(x, y) is the value of the point (x, y) in the original image; G(x, y, σ) is the Gaussian convolution kernel, σ is the preset scale, and m and n are both the sizes of the preset Gaussian convolution kernel;

[0020] In the formula, μ is the texture complexity, z i is the number of the i-th gray level in the gray histogram corresponding to the sub-image after Gaussian convolution, p(z i ) is the proportion of z i in the gray histogram, L is the total number of gray levels; m is the mean value of z i ;

[0021] As a further solution of the present invention: in the application stage, obtaining satellite images based on a preset acquisition frequency, and positioning the target in the satellite image based on the target feature table and obtaining the target position includes:

[0022] In the application stage, randomly select a precision from all precisions as the acquisition precision;

[0023] Obtain satellite images based on a preset acquisition frequency;

[0024] Traverse the satellite image in sequence based on a preset target size, and read the images to be compared;

[0025] Read the target features from the target feature table based on the acquisition precision, perform Gaussian convolution on the images to be compared according to the scale in the target features, and synchronously calculate the texture complexity;

[0026] Compare the texture complexity of the target feature and the texture complexity of the image to be compared, and calculate the difference rate;

[0027] When the difference rate is less than the preset difference threshold, obtain the position of the image to be compared as the target position.

[0028] As a further solution of the present invention: The step of determining the flight path of the drone based on the target position, sending the flight path to the drone, and having the drone obtain supplementary video data of the target includes:

[0029] Obtain the acquisition altitude and acquisition wide angle of the drone, and calculate the acquisition radius according to the acquisition altitude and acquisition wide angle;

[0030] Select 90% of the acquisition radius as the equal division length, and equally divide the length of each target position in the preset direction based on the equal division length to obtain equal division line segments;

[0031] Obtain the perpendicular bisector of each equal division line segment, and use the obtained perpendicular bisector as the flight path;

[0032] Send the flight path to the drone, and have the drone obtain supplementary video data of the target.

[0033] As a further solution of the present invention: The step of sending the flight path to the drone and having the drone obtain supplementary video data of the target includes:

[0034] Send the flight path to the drone and receive the video data real-time feedback by the drone;

[0035] Convert the video data into image frames, and perform target recognition on the image frames;

[0036] When the target recognition is successful, determine the target position according to the image position of the target and the position of the drone at the corresponding moment of the image frame;

[0037] Compare the target position determined by the drone and the target position determined by the satellite image, and calculate the coverage rate.

[0038] The technical solution of the present invention also provides a target positioning system combining vision and satellite navigation, and the system includes:

[0039] A target calibration module, configured to receive the application area selected by the requester, obtain and display the satellite image with the highest accuracy of the application area, and receive the target area calibrated by the requester;

[0040] A feature extraction module, configured to obtain satellite images of the application area at different accuracies, extract image features of the satellite images at different accuracies in the target area as target features;

[0041] A feature statistics module, configured to statistically analyze target features and their corresponding accuracies to obtain a target feature table;

[0042] A position acquisition module, configured to, in the application stage, acquire satellite images based on a preset acquisition frequency, locate a target in the satellite images based on the target feature table, and obtain the target position; wherein, each time a satellite image is acquired, the accuracy is randomly determined;

[0043] A supplementary data acquisition module, configured to determine the operation path of a drone based on the target position, send the operation path to the drone, and have the drone acquire supplementary video data of the target; wherein, during the operation of the drone, determine the coverage rate of the target feature table and feedback it to the requester.

[0044] As a further solution of the present invention: the feature extraction module includes:

[0045] A satellite image acquisition unit, configured to acquire satellite images of the application area at different accuracies;

[0046] An image cropping unit, configured to, for each satellite image at a certain accuracy, read the target area calibrated by the requester and crop the sub-image corresponding to the target area in the satellite image;

[0047] An image processing unit, configured to perform Gaussian convolution on the sub-image at different scales and synchronously calculate the texture complexity;

[0048] A data selection unit, configured to select the scale corresponding to the maximum texture complexity and use this scale and the maximum texture complexity as the target features at the corresponding accuracy.

[0049] As a further solution of the present invention: the position acquisition module includes:

[0050] An accuracy determination unit, configured to, in the application stage, randomly select an accuracy from all accuracies as the acquisition accuracy;

[0051] An image acquisition unit, configured to acquire satellite images based on a preset acquisition frequency;

[0052] An image traversal unit, configured to traverse the satellite image in sequence based on a preset target size and read the image to be compared;

[0053] A transcoding comparison unit, configured to read the target features from the target feature table based on the acquisition accuracy, perform Gaussian convolution on the image to be compared according to the scale in the target features, and synchronously calculate the texture complexity;

[0054] A difference calculation unit, configured to compare the texture complexity of the target features and the texture complexity of the image to be compared and calculate the difference rate;

[0055] A position determination unit, configured to obtain the position of the image to be compared as the target position when the difference rate is less than a preset difference threshold.

[0056] As a further solution of the present invention: The supplementary data acquisition module includes:

[0057] An acquisition parameter acquisition unit, configured to obtain the acquisition altitude and acquisition wide angle of the drone, and calculate the acquisition radius according to the acquisition altitude and acquisition wide angle;

[0058] A position equal division unit, configured to select 90% of the acquisition radius as the equal division length, and equally divide the length of each target position in a preset direction based on the equal division length to obtain equal division line segments;

[0059] A path generation unit, configured to obtain the perpendicular bisector of each equal division line segment, and use the obtained perpendicular bisector as the operation path;

[0060] A path application unit, configured to send the operation path to the drone, and the drone obtains the supplementary video data of the target.

[0061] Compared with the prior art, the beneficial effects of the present invention are as follows: The present invention first preliminarily obtains the target position through a satellite, and then the drone performs data-assisted acquisition. The accuracy requirement for the satellite is not high and the cost is low. The drone only needs to be controlled on the ground without involving high-altitude operations, providing a target positioning solution with low cost and extremely high safety. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings in the following description are only some embodiments of the present invention.

[0063] Figure 1 It is a flowchart of a target positioning method combining vision and satellite navigation.

[0064] Figure 2 It is a structural block diagram of a target positioning system combining vision and satellite navigation. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0065] In order to make the technical problems, technical solutions and beneficial effects to be solved by the present invention clearer, the following further details the present invention with reference to the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0066] Figure 1 It is a general flowchart of a target positioning method combining vision and satellite navigation. In an embodiment of the present invention, a target positioning method combining vision and satellite navigation, the method includes:

[0067] Step S100: Receive the application area selected by the requester, obtain and display the satellite image with the highest precision of the application area, and receive the target area calibrated by the requester;

[0068] Step S200: Obtain satellite images of the application area at different precisions, extract the image features of the satellite images at different precisions in the target area as target features;

[0069] Step S300: Statistically analyze the target features and their corresponding precisions to obtain a target feature table;

[0070] Step S400: In the application stage, obtain satellite images based on a preset acquisition frequency, locate the target in the satellite images based on the target feature table, and obtain the target position; wherein, each time a satellite image is obtained, the precision is randomly determined;

[0071] Step S500: Determine the operation path of the drone based on the target position, send the operation path to the drone, and have the drone obtain supplementary video data of the target; wherein, during the operation of the drone, determine the coverage rate of the target feature table and feedback it to the requester.

[0072] In an example of the technical solution of the present invention, a target positioning solution based on satellites and drones is provided. The acquisition area of the satellite is extremely large, but the acquisition area of the drone is limited, generally limited to a small area. Therefore, in this application, the requester needs to upload an application area first, and all subsequent target positioning processes occur within this application area, and the drone only moves within the application area.

[0073] After receiving the application area selected by the requester, obtain the satellite image with the highest precision (the highest precision that the provider of this method can obtain) of the application area by means of the satellite, display the satellite image, and receive the target area calibrated by the requester. This process is a preprocessing process used to characterize which are the targets that the requester wants to locate. This calibration process needs to be carried out multiple times and is calibrated by multiple requesters. The requester calibration process is actually to select a small area in the satellite image, called the target area, and the target area is a range.

[0074] Obtain satellite images of the application area at different precisions, intercept them from different satellite images by the target area, identify the intercepted areas, extract the target features. Each satellite image at each precision can extract a target feature, so as to obtain the corresponding relationship between the precision and the target feature, and statistically analyze all precisions and target features to obtain a target feature table.

[0075] In the application stage, satellite images are acquired based on a preset acquisition frequency, and traversal matching is performed in the satellite images based on the target feature table. If the matching is successful, the target can be located. Specifically, in this application, the accuracies of satellite images are different. The acquisition resource requirements for low-accuracy satellite images are less. By adopting this differential accuracy, while ensuring the amount of information, the data processing cost can be reduced. Therefore, when identifying each satellite image, it is necessary to query the target features corresponding to the accuracy in the target feature table according to its accuracy, and then perform traversal matching.

[0076] Finally, when the target is matched, the target location is obtained, the movement path of the drone is determined based on the target location, and the drone further obtains the data at the target location. Under normal circumstances, ordinary service providers do not have high satellite permissions, and the highest accuracy of the acquired satellite images is extremely limited, and the image quality is far inferior to that of the images obtained by the drone. Therefore, the video obtained by the drone is more refined data at the target location.

[0077] In addition, during the movement of the drone, an identification algorithm can be introduced to evaluate the target feature table, thereby updating the calibration process, that is, step S100.

[0078] As a preferred embodiment of the technical solution of the present invention, the steps of acquiring satellite images of the application area at different accuracies, extracting the image features of the satellite images at different accuracies in the target area, and using them as target features include:

[0079] Acquire satellite images of the application area at different accuracies;

[0080] For each satellite image at a certain accuracy, read the target area calibrated by the requester, and intercept the sub-image corresponding to the target area in the satellite image;

[0081] Perform Gaussian convolution on the sub-image at different scales, and synchronously calculate the texture complexity;

[0082] Select the scale corresponding to the maximum texture complexity, and use this scale and the maximum texture complexity as the target features corresponding to the accuracy.

[0083] The above content specifically defines step S200. Acquiring satellite images of the application area at different accuracies. For each satellite image at a certain accuracy, reading the target area calibrated by the requester and intercepting the sub-image corresponding to the target area in the satellite image. This process can be understood as the selection tool in the PS tool. The target area is a selection area. After determining a selection area, the same selection area is applied to all layers. Intercepting the corresponding part of the target area in the satellite image is called a sub-image.

[0084] Perform Gaussian convolution on the sub-image at different scales and synchronously calculate the texture complexity. This process is to process the image, and then calculate the texture complexity. After performing Gaussian convolution on the image to different degrees, there are multiple processed images. Select the one with the most obvious texture complexity among these processed images as the feature of the sub-image, and statistically analyze the feature and its scale, which are collectively referred to as the target feature.

[0085] Specifically, the content of performing Gaussian convolution on the sub-image at different scales and synchronously calculating the texture complexity includes:

[0086] L(x, y, σ) = G(x, y, σ) * I(x, y);

[0087]

[0088] In the formula, L(x, y, σ) is the value of the point (x, y) in the sub-image after Gaussian convolution, I(x, y) is the value of the point (x, y) in the original sub-image; G(x, y, σ) is the Gaussian convolution kernel, σ is the preset scale, and m and n are both the sizes of the preset Gaussian convolution kernels;

[0089] In the formula, μ is the texture complexity, z i is the number of the i-th gray level in the gray histogram corresponding to the sub-image after Gaussian convolution, p(z i ) is the proportion of z i in the gray histogram, L is the total number of gray levels; m is the mean value of z i .

[0090] As a preferred embodiment of the technical solution of the present invention, in the application stage, the steps of obtaining satellite images based on a preset acquisition frequency, positioning the target in the satellite images based on the target feature table, and obtaining the target position include:

[0091] In the application stage, randomly select an accuracy from all precisions as the acquisition accuracy;

[0092] Obtain satellite images based on a preset acquisition frequency;

[0093] Traverse the satellite images in sequence based on a preset target size and read the images to be compared;

[0094] Read the target features in the target feature table based on the acquisition accuracy, perform Gaussian convolution on the images to be compared according to the scale in the target features, and synchronously calculate the texture complexity;

[0095] Compare the texture complexity of the target features with that of the images to be compared and calculate the difference rate;

[0096] When the difference rate is less than a preset difference threshold, obtain the position of the image to be compared as the target position.

[0097] The above content describes the application stage in detail, and its working principle is as follows:

[0098] First, randomly determine an accuracy as the acquisition accuracy of the satellite image; the acquisition frequency of the satellite image is preset and fixed, and a satellite image is obtained every fixed period of time. Each acquisition requires randomly determining an acquisition accuracy, and the satellite image is obtained at this acquisition accuracy. In the satellite image, parts with the same size as the target area are sequentially intercepted from left to right and from top to bottom, which are called images to be compared (essentially a part of the satellite image).

[0099] Read the target features in the target feature table according to the determined acquisition accuracy, read the scale in the target features, process the satellite image at this acquisition accuracy based on the same scale, then calculate the texture complexity of the obtained processed image, compare the texture complexity of the image to be compared and the texture complexity of the target features, calculate the difference rate. The smaller the difference rate, the higher the similarity between the two. If it is small enough, it is considered that the area corresponding to the image to be compared is the target area, that is, the target that the demander wants to locate.

[0100] Among them, the difference rate can be the absolute value of the difference between the texture complexity of the image to be compared and the texture complexity of the target features divided by the texture complexity of the target features.

[0101] It is worth mentioning that for the above process of randomly determining the acquisition accuracy, the simplest and most conventional solution is that the selection probability of each acquisition accuracy is the same; on this basis, other expansions can also be carried out. For example, starting from the current moment, a retrospective time is determined with a preset duration as the span, and the selection probability of various acquisition accuracies is determined according to the number of target areas matched within the retrospective time; for example, the more target areas are matched, the greater the selection probability of high acquisition accuracy. The higher the acquisition accuracy, the greater the magnification. This function can be achieved with the help of the exponential function or the reciprocal of the logarithmic function. It should be noted that the sum of the selection probabilities of all acquisition accuracies is one.

[0102] As a preferred embodiment of the technical solution of the present invention, the step of determining the flight path of the drone based on the target position, sending the flight path to the drone, and the drone obtaining supplementary video data of the target includes:

[0103] Obtain the acquisition height and acquisition wide angle of the drone, and calculate the acquisition radius according to the acquisition height and acquisition wide angle;

[0104] Select 90% of the acquisition radius as the equal division length, and equally divide the length of each target position in a preset direction based on the equal division length to obtain equal division line segments;

[0105] Obtain the perpendicular bisector of each equal division line segment, and use the obtained perpendicular bisector as the running path;

[0106] Send the running path to the drone, and the drone obtains supplementary video data of the target.

[0107] Obtain the acquisition height and acquisition wide angle of the drone, calculate the acquisition radius according to the acquisition height and acquisition wide angle, select 90% of the acquisition radius as the equal division length, and equally divide the length of each target position in a preset direction based on the equal division length to obtain equal division line segments; Since the equal division line segments are 90% of the acquisition radius, when the drone moves along the perpendicular bisector of the equal division line segments, the acquisition range is larger than the equal division line segments, which enables the drone to fully obtain data, and there will be some overlapping parts, and this overlapping part can also be used to verify whether there are errors in the acquisition process of the drone.

[0108] Further, the step of sending the running path to the drone and the drone obtaining supplementary video data of the target includes:

[0109] Send the running path to the drone and receive the video data real-time feedback by the drone;

[0110] Convert the video data into image frames and perform target recognition on the image frames;

[0111] When the target recognition is successful, determine the target position according to the image position of the target and the position of the drone at the corresponding moment of the image frame;

[0112] Compare the target position determined by the drone and the target position determined by the satellite image, and calculate the coverage rate.

[0113] In the technical solution of the present invention, an auxiliary solution is introduced. The purpose of the drone is to obtain the video at the target location. However, during its movement, it will inevitably collect videos at other locations. The above content is the solution for processing the videos at these other locations. Receive the video data fed back by the drone in real time, convert the video data into image frames, perform target recognition on the image frames based on a preset recognition algorithm. When the target recognition is successful, determine the target location according to the image position of the target (relative position in the image) and the drone position (GPS coordinates) at the corresponding moment of the image frame. Compare the target location determined by the drone with the target location determined by the satellite image, calculate the ratio of the number of target locations determined by the satellite image to the number of target locations determined by the drone to obtain the coverage rate. The larger the coverage rate, the more accurate the satellite recognition process is. If the coverage rate is insufficient, calibration data can be supplemented in the target feature table to optimize the recognition process based on the satellite side.

[0114] Figure 2 FIG. is a structural block diagram of a target positioning system combining vision and satellite navigation. In a preferred embodiment of the technical solution of the present invention, a target positioning system combining vision and satellite navigation is further provided. The system 10 includes:

[0115] A target calibration module 11, configured to receive the application area selected by the requester, obtain and display the satellite image with the highest accuracy of the application area, and receive the target area calibrated by the requester;

[0116] A feature extraction module 12, configured to obtain satellite images of the application area at different accuracies, and extract the image features of the satellite images at different accuracies in the target area as target features;

[0117] A feature statistics module 13, configured to count the target features and their corresponding accuracies to obtain a target feature table;

[0118] A position acquisition module 14, configured to, in the application stage, obtain satellite images based on a preset acquisition frequency, locate the target in the satellite images based on the target feature table, and obtain the target position; wherein, each time a satellite image is obtained, the accuracy is randomly determined;

[0119] A supplementary data acquisition module 15, configured to determine the operation path of the drone based on the target position, send the operation path to the drone, and the drone obtains supplementary video data of the target; wherein, during the operation of the drone, determine the coverage rate of the target feature table and feedback it to the requester.

[0120] Further, the feature extraction module 12 includes:

[0121] A satellite image acquisition unit, configured to obtain satellite images of the application area at different accuracies;

[0122] An image cropping unit, which is used to read the target area calibrated by the requester for the satellite image at each precision, and crop the sub-image corresponding to the target area in the satellite image;

[0123] An image processing unit, which is used to perform Gaussian convolution on the sub-image at different scales and synchronously calculate the texture complexity;

[0124] A data selection unit, which is used to select the scale corresponding to the maximum texture complexity, and use this scale and the maximum texture complexity as the target features at the corresponding precision.

[0125] Specifically, the position acquisition module 14 includes:

[0126] A precision determination unit, which is used to randomly select a precision from all precisions as the acquisition precision during the application stage;

[0127] An image acquisition unit, which is used to acquire satellite images based on a preset acquisition frequency;

[0128] An image traversal unit, which is used to traverse the satellite image in sequence based on a preset target size and read the image to be compared;

[0129] A transcoding comparison unit, which is used to read the target features from the target feature table based on the acquisition precision, perform Gaussian convolution on the image to be compared according to the scale in the target features, and synchronously calculate the texture complexity;

[0130] A difference calculation unit, which is used to compare the texture complexity of the target features and the texture complexity of the image to be compared and calculate the difference rate;

[0131] A position determination unit, which is used to obtain the position of the image to be compared as the target position when the difference rate is less than a preset difference threshold.

[0132] Furthermore, the supplementary data acquisition module 15 includes:

[0133] An acquisition parameter acquisition unit, which is used to acquire the acquisition height and acquisition wide angle of the unmanned aerial vehicle, and calculate the acquisition radius according to the acquisition height and acquisition wide angle;

[0134] A position equal division unit, which is used to select 90% of the acquisition radius as the equal division length, and equally divide the length of each target position in a preset direction based on the equal division length to obtain equal division line segments;

[0135] A path generation unit, which is used to obtain the perpendicular bisector of each equal division line segment and use the obtained perpendicular bisector as the operation path;

[0136] A path application unit, which is used to send the operation path to the unmanned aerial vehicle, and the unmanned aerial vehicle acquires the supplementary video data of the target.

[0137] The above are only the preferred embodiments of the present invention, and do not limit the patent scope of the present invention accordingly. Any equivalent structure or equivalent process transformation made by using the content of the specification and drawings of the present invention, or directly or indirectly applied in other related technical fields, shall similarly be included within the patent protection scope of the present invention.

Claims

1. A target positioning method combining vision and satellite navigation, characterized in that The method includes: Receiving the application area selected by the requester, obtaining and displaying the satellite image with the highest precision of the application area, and receiving the target area calibrated by the requester; Obtaining satellite images of the application area at different precisions, extracting the image features of the satellite images at different precisions in the target area as target features; Statistically analyzing the target features and their corresponding precisions to obtain a target feature table; In the application stage, obtaining satellite images based on a preset acquisition frequency, locating the target in the satellite images based on the target feature table, and obtaining the target position; wherein, each time a satellite image is obtained, the precision is randomly determined; Determining the operation path of the unmanned aerial vehicle based on the target position, sending the operation path to the unmanned aerial vehicle, and obtaining supplementary video data of the target by the unmanned aerial vehicle; wherein, during the operation of the unmanned aerial vehicle, determining the coverage rate of the target feature table and feeding it back to the requester; The steps of obtaining satellite images based on a preset acquisition frequency, locating the target in the satellite images based on the target feature table, and obtaining the target position in the application stage include: In the application stage, randomly selecting a precision from all precisions as the acquisition precision; Obtaining satellite images based on a preset acquisition frequency; Traversing the satellite images in sequence based on a preset target size and reading the images to be compared; Reading the target features in the target feature table based on the acquisition precision, performing Gaussian convolution on the images to be compared according to the scale in the target features, and synchronously calculating the texture complexity; Comparing the texture complexity of the target features with that of the images to be compared and calculating the difference rate; When the difference rate is less than a preset difference threshold, obtaining the position of the image to be compared as the target position; The determination process of the acquisition precision is: Taking the current moment as the starting point, determining a retrospective time with a preset duration as the span, and determining the selection probabilities of various acquisition precisions according to the number of target areas matched within the retrospective time; the more target areas are matched, the greater the selection probability of high acquisition precision.

2. The method for target positioning by combining vision and satellite navigation according to claim 1, wherein The steps of obtaining satellite images of the application area at different precisions, extracting the image features of the satellite images at different precisions in the target area as target features include: Obtaining satellite images of the application area at different precisions; For each satellite image at a precision, reading the target area calibrated by the requester and intercepting the sub-image corresponding to the target area in the satellite image; Performing Gaussian convolution on the sub-image at different scales and synchronously calculating the texture complexity; Selecting the scale corresponding to the maximum texture complexity and taking the scale and the maximum texture complexity as the target features at the corresponding precision.

3. The target positioning method combining vision and satellite navigation according to claim 2, wherein, The content of performing Gaussian convolution on the sub-image at different scales and synchronously calculating the texture complexity includes: ； ； In the formula, is the value of the point in the sub-image after Gaussian convolution, is the value of the point in the atomic image; is the Gaussian convolution kernel, is the preset scale, and are both the sizes of the Gaussian convolution kernel of the prelude; ; Wherein, is the texture complexity, is the number of the -th gray level in the gray histogram corresponding to the sub-image after Gaussian convolution, is 's proportion in the gray histogram, is the total number of gray levels; is 's mean value, .

4. The target positioning method combining vision and satellite navigation according to claim 1, characterized in that, The steps of determining the operation path of the unmanned aerial vehicle based on the target position, sending the operation path to the unmanned aerial vehicle, and obtaining supplementary video data of the target by the unmanned aerial vehicle include: Obtaining the acquisition height and acquisition wide angle of the unmanned aerial vehicle, and calculating the acquisition radius according to the acquisition height and acquisition wide angle; Selecting 90% of the acquisition radius as the equal division length, equally dividing the length of each target position in a preset direction based on the equal division length to obtain equal division line segments; Obtain the perpendicular bisector of each equally divided line segment and use the obtained perpendicular bisector as the running path; Send the running path to the drone, and the drone obtains supplementary video data of the target.

5. The method for target positioning by combining vision and satellite navigation according to claim 4, wherein The step of sending the running path to the drone and the drone obtaining supplementary video data of the target includes: Send the running path to the drone and receive the video data real-time feedback by the drone; Convert the video data into image frames and perform target recognition on the image frames; When the target recognition is successful, determine the target position according to the image position of the target and the position of the drone at the corresponding moment of the image frame; Compare the target position determined by the drone and the target position determined by the satellite image, and calculate the coverage rate.

6. A target positioning system combining vision and satellite navigation, characterized in that, The system includes: A target calibration module, which is used to receive the application area selected by the requester, obtain and display the satellite image with the highest precision of the application area, and receive the target area calibrated by the requester; A feature extraction module, which is used to obtain satellite images of the application area at different precisions, extract the image features of the satellite images at different precisions in the target area as target features; A feature statistics module, which is used to count the target features and their corresponding precisions to obtain a target feature table; A position acquisition module, which is used in the application stage to obtain satellite images based on a preset acquisition frequency, locate the target in the satellite image based on the target feature table, and obtain the target position; where, each time a satellite image is obtained, the precision is randomly determined; A supplementary data acquisition module, which is used to determine the running path of the drone based on the target position, send the running path to the drone, and the drone obtains supplementary video data of the target; where, during the working process of the drone, determine the coverage rate of the target feature table and feedback it to the requester; The position acquisition module includes: A precision determination unit, which is used in the application stage to randomly select a precision from all precisions as the acquisition precision; An image acquisition unit, which is used to obtain satellite images based on a preset acquisition frequency; An image traversal unit, which is used to traverse the satellite image in sequence based on a preset target size and read the image to be compared; A transcoding comparison unit, which is used to read the target features in the target feature table based on the acquisition precision, perform Gaussian convolution on the image to be compared according to the scale in the target features, and synchronously calculate the texture complexity; A difference calculation unit, which is used to compare the texture complexity of the target features and the texture complexity of the image to be compared and calculate the difference rate; A position determination unit, which is used to obtain the position of the image to be compared as the target position when the difference rate is less than a preset difference threshold; The determination process of the acquisition precision is: Taking the current moment as the starting point, determine a retrospective time with a preset duration as the span, and determine the selection probability of various acquisition precisions according to the number of target areas matched within the retrospective time; the more the number of target areas matched, the greater the selection probability of the high acquisition precision.

7. The visual and satellite navigation combined target positioning system according to claim 6, characterized in that, The feature extraction module includes: A satellite image acquisition unit, which is used to obtain satellite images of the application area at different precisions; An image interception unit, which is used to, for each satellite image at each precision, read the target area calibrated by the requester and intercept the sub-image corresponding to the target area in the satellite image; An image processing unit for performing Gaussian convolution on sub-images at different scales and synchronously calculating texture complexity; A data selection unit for selecting the scale corresponding to the maximum texture complexity and using the scale and the maximum texture complexity as target features at the corresponding accuracy.

8. The visual and satellite navigation combined target positioning system according to claim 6, characterized in that, The supplementary data acquisition module includes: An acquisition parameter acquisition unit for acquiring the acquisition height and acquisition wide angle of the drone and calculating the acquisition radius according to the acquisition height and acquisition wide angle; A position equalization unit for selecting 90% of the acquisition radius as the equalization length and equally dividing the length of each target position in a preset direction based on the equalization length to obtain equalized line segments; A path generation unit for obtaining the perpendicular bisector of each equalized line segment and using the obtained perpendicular bisector as the running path; A path application unit for sending the running path to the drone, and the drone acquires supplementary video data of the target.

Images