Remote Sensing Multi-Target Detection Method, Device, Electronic Device and Storage Medium

Abstract

The present invention provides a remote sensing multi-target detection method, device, electronic device and storage medium, belonging to the technical field of remote sensing monitoring. The method includes: detecting each detection target in the remote sensing image to be detected one by one, and generating a target detection result description file for each of the detection targets; the target detection result description file is used to describe the image coordinate information of the detection target in the remote sensing image to be detected; based on each of the target detection result description files, generating a target vector data file for each of the detection targets; and superimposing each of the target vector data files on the remote sensing image to be detected to obtain the multi-target detection result of the remote sensing image to be detected. The present invention can greatly reduce the storage space occupied by the target detection result, and at the same time can realize superimposing multiple target detection results on a single remote sensing image, achieving the multi-target detection effect of satellite remote sensing images.

Description

Technical Field

[0001] The present invention relates to the field of remote sensing monitoring technology, and in particular to a remote sensing multi-target detection method, device, electronic equipment and storage medium. Background Art

[0002] A satellite remote sensing image contains a lot of information. Different targets need to be extracted according to different user needs. In some scenarios, users need to extract multiple targets on a remote sensing image. However, the current traditional target detection method uses machine learning methods, which can only frame a single detection target on the remote sensing image. Multiple target detection results cannot be superimposed on a remote sensing image, thus failing to achieve the multi-target detection effect of satellite remote sensing images.

[0003] Therefore, how to better realize remote sensing multi-target detection has become a technical problem that needs to be solved urgently in the industry. Summary of the invention

[0004] The present invention provides a remote sensing multi-target detection method, device, electronic equipment and storage medium, which are used to better realize remote sensing multi-target detection.

[0005] The present invention provides a remote sensing multi-target detection method, comprising:

[0006] Detect each detection target in the remote sensing image to be detected one by one, and generate a target detection result description file for each detection target; the target detection result description file is used to describe the image coordinate information of the detection target in the remote sensing image to be detected;

[0007] Based on each of the target detection result description files, generating a target vector data file for each of the detected targets;

[0008] Each target vector data file is superimposed on the remote sensing image to be detected to obtain a multi-target detection result of the remote sensing image to be detected.

[0009] According to a remote sensing multi-target detection method provided by the present invention, generating a target vector data file for each detected target based on each target detection result description file includes:

[0010] Extracting the image coordinate information covered by each of the detected targets in the remote sensing image to be detected from each of the target detection result description files, and acquiring the affine transformation parameters of the remote sensing image to be detected;

[0011] Based on the affine transformation parameters and the image coordinate information covered by each of the detection targets in the remote sensing image to be detected, generating the longitude and latitude information corresponding to each of the detection targets;

[0012] Based on the longitude and latitude information corresponding to each of the detected targets, determine the polygon information corresponding to each of the detected targets; the polygon information is used to describe the geographical location information of each of the detected targets;

[0013] Using the polygon information corresponding to each of the detected targets as elements, generate the target vector data file for each of the detected targets.

[0014] According to a remote sensing multi-target detection method provided by the present invention, before detecting each detected target in the to-be-detected remote sensing image one by one and generating the target detection result description file for each of the detected targets, the method further includes:

[0015] Obtain the target detection requirement information for the to-be-detected remote sensing image;

[0016] Based on the target detection requirement information, respectively determine the target remote sensing data and the target detection algorithm for detecting each of the detected targets from a preset remote sensing database and a preset algorithm library, so as to use the target remote sensing data and the target detection algorithm to detect each of the detected targets in the to-be-detected remote sensing image one by one; the preset algorithm library includes a variety of detection algorithms, and the preset remote sensing database includes multiple types of remote sensing data corresponding to the to-be-detected remote sensing image.

[0017] According to a remote sensing multi-target detection method provided by the present invention, the target detection requirement information includes multiple detected targets, the applicability data information of each detection algorithm with respect to each detected target, and the timeliness data information of each type of remote sensing data with respect to each detected target; the determining the target remote sensing data and the target detection algorithm for detecting each of the detected targets from a preset remote sensing database and a preset algorithm library based on the target detection requirement information includes: based on a variety of the detection algorithms and multiple types of the remote sensing data, determine multiple data groups, and each data group includes one of the detection algorithms and one of the types of remote sensing data;

[0018] For any one of the multiple detected targets, based on the applicability data information of each detection algorithm with respect to the any one of the detected targets and the timeliness data information of each type of remote sensing data with respect to the any one of the detected targets, determine the comprehensive adaptation score of each data group with respect to the any one of the detected targets, and obtain the comprehensive adaptation scores of each data group with respect to each of the detected targets;

[0019] With the goal of maximizing the total score of the comprehensive adaptation scores of all the detected targets, based on the comprehensive adaptation scores of each data group with respect to each of the detected targets, determine the target data group from each of the data groups;

[0020] Based on the target data group, the target remote sensing data and the target detection algorithm are obtained.

[0021] According to a remote sensing multi-target detection method provided by the present invention, after superimposing each of the target vector data files on the remote sensing image to be detected to obtain the multi-target detection result of the remote sensing image to be detected, the method further includes:

[0022] At a preset slicing level, the original image corresponding to the multi-target detection result of the remote sensing image to be detected is sliced into a full-map sliced image, and the full-map sliced image is published as a web map tile service;

[0023] When a web map tile service request is received, the full-map sliced image is sent to the image display interface at the front end for display.

[0024] According to a remote sensing multi-target detection method provided by the present invention, after sending the full-map sliced image to the image display interface at the front end for display when a web map tile service request is received, the method further includes: receiving a first input from a user, where the first input is used to zoom in on the full-map sliced image;

[0025] In response to the first input, the target longitude and latitude information corresponding to the four top corner points of the image display interface after zooming in on the full-map sliced image in the original image is obtained, and based on the target geographical area covered by the target longitude and latitude information corresponding to the four top corner points, a target image corresponding to the target geographical area is cut out from the original image;

[0026] Through the web map tile service, the target image is sent to the image display interface at the front end for display.

[0027] The present invention also provides a remote sensing multi-target detection device, including:

[0028] A first generation module, configured to detect each detection target in the remote sensing image to be detected one by one, and generate a target detection result description file for each of the detection targets; the target detection result description file is used to describe the image coordinate information of the detection target in the remote sensing image to be detected;

[0029] A second generation module, configured to generate a target vector data file for each of the detection targets based on each of the target detection result description files;

[0030] A first detection module, configured to superimpose each of the target vector data files on the remote sensing image to be detected to obtain the multi-target detection result of the remote sensing image to be detected.

[0031] A remote sensing multi-target detection device provided by the present invention, the second generation module is specifically further configured to:

[0032] Extract the image coordinate information covered by each detection target in the to-be-detected remote sensing image from each of the target detection result description files, and obtain the affine transformation parameters of the to-be-detected remote sensing image;

[0033] Generate the longitude and latitude information corresponding to each detection target based on the affine transformation parameters and the image coordinate information covered by each detection target in the to-be-detected remote sensing image;

[0034] Determine the polygon information corresponding to each detection target based on the longitude and latitude information corresponding to each detection target; the polygon information is used to describe the geographical location information of each detection target;

[0035] Generate the target vector data file of each detection target with the polygon information corresponding to each detection target as the element.

[0036] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the program, it implements the remote sensing multi-target detection method as described in any one of the above.

[0037] The present invention also provides a non-transitory computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it implements the remote sensing multi-target detection method as described in any one of the above.

[0038] The present invention also provides a computer program product, including a computer program. When the computer program is executed by a processor, it implements the remote sensing multi-target detection method as described in any one of the above.

[0039] The remote sensing multi-target detection method, device, electronic device, and storage medium provided by the present invention simplify the multi-target recognition problem into multiple single-target detection problems by detecting each detection target in the to-be-detected remote sensing image one by one. The training is simple and can be flexibly combined. By extracting the image coordinate information of each detection target in the remote sensing image, a target detection result description file of all detection target position information is generated. Furthermore, the target detection result description file can be converted into a corresponding target vector data file to describe the actual geographical locations of each detection target. Since the vector data file occupies less memory, each target vector data file can be well superimposed on the to-be-detected remote sensing image to obtain the multi-target detection result of the to-be-detected remote sensing image, greatly reducing the storage space occupied by the target detection result. At the same time, it can realize the superposition of multiple target detection results on a remote sensing image, achieving the multi-target detection effect of satellite remote sensing images. Description of the Drawings

[0040] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0041] Figure 1 It is one of the schematic flowcharts of the remote sensing multi-target detection method provided by the present invention.

[0042] Figure 2 It is another schematic flowchart of the remote sensing multi-target detection method provided by the present invention.

[0043] Figure 3 It is the schematic structural diagram of the remote sensing multi-target detection device provided by the present invention.

[0044] Figure 4 It is the schematic physical structure diagram of the electronic device provided by the present invention. Detailed implementation manners

[0045] To make the objectives, technical solutions and advantages of the present invention clearer, the following will clearly and completely describe the technical solutions in the present invention in conjunction with the drawings in the present invention. Obviously, the described embodiments are some but not all of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art without creative efforts based on the embodiments in the present invention belong to the scope of protection of the present invention.

[0046] In the description of the present invention, it should be noted that unless otherwise clearly defined and limited, the terms "installation", "connection", and "connection" should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be directly connected or indirectly connected through an intermediate medium, and it can be the communication inside two components. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific situations.

[0047] The following will describe Figures 1 - 4 the remote sensing multi-target detection method, device, electronic device and storage medium of the present invention.

[0048] Figure 1 It is the schematic flowchart of the remote sensing multi-target detection method provided by the present invention. As Figure 1 shown, it includes:

[0049] Step 110: Detect each detection target in the remote sensing image to be detected one by one, and generate a target detection result description file for each detection target; the target detection result description file is used to describe the image coordinate information of the detection target in the remote sensing image to be detected.

[0050] Step 120: Generate a target vector data file for each detection target based on each target detection result description file.

[0051] Step 130: Superimpose each target vector data file on the remote sensing image to be detected to obtain the multi-target detection result of the remote sensing image to be detected.

[0052] Specifically, the remote sensing image to be detected described in the embodiments of the present invention refers to a remote sensing image for multi-target detection, which contains multiple detection targets. For example, the remote sensing image to be detected is a remote sensing image of a ship port area to be detected, which may include detection targets such as ships, airplanes, and oil tanks.

[0053] The target detection result description file described in the embodiments of the present invention is used to describe the image coordinate information of each detection target in the remote sensing image to be detected. Specifically, it can be a description file formed by using a target detection algorithm to detect the remote sensing image and then extracting the image coordinate information of each detection target in the remote sensing image.

[0054] In the embodiments of the present invention, the target detection result description file can specifically use the TXT form to record information, which may contain multiple lines of information. Each line of recorded information corresponds to a detection target, and each line of record corresponds to the position of the detection target relative to the upper left corner of the original image. That is, the pixel point in the upper left corner of the original image is the origin, and the coordinates are represented as (0,0). (x, y) represents the coordinates of the pixel points included in a certain detection target in the original image, where x represents the distance from the coordinate horizontal direction to the origin, with the right direction as the positive direction, and y represents the distance from the coordinate vertical direction to the origin, with the downward direction as the positive direction.

[0055] The target vector data file described in the embodiments of the present invention refers to the boundary file (Shapefile) of the vector data of a single detection target, which is the most common vector data format and is supported by all commercial and open-source GIS software. The Shapefile consists of three necessary files (.shp,.shx,.dbf) and several other optional files.

[0056] Among them, the.shp file stores vector map data and records the spatial position information of each feature; the.shx file is an index file used to store the positions of the features in the.shp file to speed up data access; the.dbf file stores the attribute information of the vector data, such as the name, type, etc. of each point on the map.

[0057] In an embodiment of the present invention, in step 110, each detection target in the remote sensing image to be detected is detected one by one through one or more pre-determined target detection algorithms, the position information of each detection target required by the user in the remote sensing image is determined, and then the image coordinate information of each detection target in the remote sensing image is extracted to generate a corresponding target detection result description file.

[0058] Further, in an embodiment of the present invention, in step 120, after obtaining the target detection result description file of each detection target, the method of converting the pixel coordinates of the remote sensing image to spatial geographic coordinates can be used to convert the image coordinate information of the detection target in each target detection result description file into the corresponding spatial geographic coordinate information, so as to generate the target vector data file of each detection target, that is, to obtain the Shape file of each target detection result.

[0059] In an embodiment of the present invention, in step 130, the Shape files of each target detection result generated in step 120 are superimposed on the remote sensing image to be detected in the spatial dimension. Since remote sensing satellite images above level 1 usually contain geographic coordinate information, if the target detection result is in the form of a Shape file, it can be superimposed on the remote sensing image. In this way, the Shape files of multiple target detection results can be superimposed on the original remote sensing image to be detected, so as to obtain the multi-target detection result of the remote sensing image to be detected for the user to browse and view.

[0060] The remote sensing multi-target detection method of the embodiment of the present invention simplifies the multi-target recognition problem into multiple single-target detection problems by detecting each detection target in the remote sensing image to be detected one by one. The training is simple and can be flexibly combined. By extracting the image coordinate information of each detection target in the remote sensing image, a target detection result description file containing the position information of all detection targets is generated. Furthermore, the target detection result description file can be converted into the corresponding target vector data file to describe the actual geographical location of each detection target. Since the vector data file occupies less memory, each target vector data file can be well superimposed on the remote sensing image to be detected to obtain the multi-target detection result of the remote sensing image to be detected, greatly reducing the storage space occupied by the target detection result. At the same time, multiple target detection results can be superimposed on a single remote sensing image, realizing the multi-target detection effect of satellite remote sensing images.

[0061] Based on the content of the above embodiment, as an alternative embodiment, generating the target vector data file of each detection target based on each target detection result description file includes:

[0062] Extract the image coordinate information covered by each detected object in the remote sensing image to be detected from each target detection result description file, and obtain the affine transformation parameters of the remote sensing image to be detected;

[0063] Generate the longitude and latitude information corresponding to each detected object based on the affine transformation parameters and the image coordinate information covered by each detected object in the remote sensing image to be detected;

[0064] Determine the polygon information corresponding to each detected object based on the longitude and latitude information corresponding to each detected object; the polygon information is used to describe the geographical location information of each detected object;

[0065] Generate the target vector data file of each detected object with the polygon information corresponding to each detected object as the feature.

[0066] Specifically, the affine transformation parameters described in the embodiments of the present invention refer to the affine transformation parameters used for converting the pixel coordinates of the remote sensing image to be detected into geographical coordinates. The affine transformation parameters are stored in the affine matrix and can be denoted as the gt array. Specifically, it can include 6 parameters, which are used to describe the relationship between the raster row and column numbers of the image and the geographical coordinates.

[0067] Among them, the 6 affine transformation parameters can be expressed as gt[0], gt[1], gt[2], gt[3], gt[4], and gt[5]. Here, gt[0] represents the x coordinate of the upper left corner of the image; gt[1] represents the resolution in the east-west direction of the image; gt[2] represents the rotation angle, and if the north of the image is upward, this value is 0; gt[3] represents the y coordinate of the upper left corner of the image; gt[4] represents the rotation angle, and if the north of the image is upward, this value is 0; gt[5] represents the resolution in the north-south direction of the image.

[0068] In the embodiments of the present invention, after generating the target detection result description files of each detected object, the image coordinate information of each pixel point in the area covered by each detected object in the remote sensing image to be detected can be extracted from each target detection result description file, including the coordinate information of the contour pixel points corresponding to each detected object. At the same time, based on the format parameters of the remote sensing image to be detected, the affine transformation parameters of the remote sensing image to be detected can be obtained.

[0069] Further, in the embodiments of the present invention, the longitude and latitude information corresponding to each detected object is generated based on the affine transformation parameters and the image coordinate information covered by each detected object in the remote sensing image to be detected. For any pixel coordinate point (x 1 , y 1 ) of the detected object in the remote sensing image to be detected, calculate the longitude and latitude of this pixel coordinate point according to the affine transformation parameters. Assume that geox 1 represents the longitude and latitude of this pixel coordinate point (x 1, y 1 ) longitude, geoy 1 Indicates the longitude of the pixel coordinate point (x 1 , y 1 ), and the latitude can be specifically calculated by the following formula:

[0070] double geox 1 = gt[0] + gt[1] * x 1 + gt[2] * y 1 ;

[0071] double geoy 1 = gt[3] + gt[4] * x 1 + gt[5] * y 1 ;

[0072] Thus, the longitude and latitude information corresponding to the pixel points included in each detection target can be generated.

[0073] In the embodiments of the present invention, the longitude and latitude information corresponding to each detection target can be further used to determine the polygon information corresponding to each detection target. That is, one line in the target detection result description file corresponds to generating one polygon information, and the Nth line in the target detection result description file is represented as polygon N . polygon N The generation method is as follows:

[0074] polygon N = POLYGON((geox 1 geoy 1 , geox 2 geoy 2 ...... geox N geoy N , geox 1 ; geoy 1 ))

[0075] Furthermore, in the embodiments of the present invention, taking the polygon information corresponding to each detection target as an element, the geographical space position information corresponding to each detection target is recorded and stored as vector map data, thereby generating the target vector data file of each detection target.

[0076] The method according to the embodiment of the present invention obtains the target detection result description files of each detection target, and uses affine transformation to generate the longitude and latitude information corresponding to the pixels included in each detection target, so as to obtain the polygon information corresponding to each detection target. Using the polygon information as elements, a target vector data file with a smaller memory space is generated, which is convenient for better superimposing multiple target detection results on the same remote sensing image, and is beneficial to realizing more efficient multi-target detection.

[0077] It should be noted that it is difficult for the target detection algorithm to adapt to all satellite images, and it is usually only applicable to the images of certain sensors of certain satellites. Therefore, in the embodiment of the present invention, tags can be set for each specific target detection algorithm. The tags mainly include the target detection type and the list of satellite sensors that can be adapted, and describe the detection accuracy rate of each satellite sensor. A value can be initially set for the detection accuracy rate of each satellite sensor, and the detection accuracy rate can be updated according to the results of manual verification detection, so as to determine the detection accuracy rate of the target detection algorithm.

[0078] Based on the content of the above embodiment, as an optional embodiment, before detecting each detection target in the remote sensing image to be detected one by one and generating the target detection result description file of each detection target, the method further includes:

[0079] Obtain the target detection requirement information for the remote sensing image to be detected;

[0080] Based on the target detection requirement information, respectively determine the target remote sensing data and the target detection algorithm for detecting each detection target from the preset remote sensing database and the preset algorithm library, so as to use the target remote sensing data and the target detection algorithm to detect each detection target in the remote sensing image to be detected one by one; the preset algorithm library includes a variety of detection algorithms, and the preset remote sensing database includes multiple types of remote sensing data corresponding to the remote sensing image to be detected.

[0081] Specifically, the target detection requirement information described in the embodiment of the present invention refers to the requirement information submitted by an external user terminal or the system internal for multi-target detection of the remote sensing image to be detected, and specifically may include requirements information such as multiple detection targets to be detected, data timeliness, and detection accuracy rate.

[0082] In the embodiment of the present invention, according to the timeliness requirements of each type of detection target in the requirements, retrieve the remote sensing data that meets the timeliness in the preset remote sensing database, and according to the retrieved remote sensing data, retrieve the detection algorithm with the accuracy rate meeting the requirements in the preset algorithm library. Retrieve the remote sensing data and detection algorithms that meet the requirements for each type of detection target in turn.

[0083] Among them, the preset algorithm library includes a variety of detection algorithms, such as detection algorithms for ship targets, detection algorithms for aircraft targets, detection algorithms for oil tank targets, etc. The preset remote sensing database includes various types of remote sensing data corresponding to the remote sensing images to be detected, such as multispectral remote sensing data, radar remote sensing data, etc.

[0084] In an embodiment of the present invention, according to the target detection requirement information for the remote sensing image to be detected, the weights of the detection algorithms and remote sensing data for the detection accuracy and data timeliness of each type of detection target can be increased, indicating the importance of the accuracy and timeliness corresponding to this type of target. Then, a weight is set for each type of detection target, and finally, according to the total score, the optimal solution with the highest score is selected. Thus, the target remote sensing data and target detection algorithm corresponding to each detection target are respectively determined from the preset remote sensing database and the preset algorithm library, so as to use the target remote sensing data and target detection algorithm of each detection target to perform multi-target detection on the remote sensing image to be detected.

[0085] The method of the embodiment of the present invention realizes dynamic selection of appropriate satellite remote sensing data and target detection algorithms for multi-target detection according to user requirements by designing the detection requirements of the front-end input user and scheduling the use of target detection algorithms and remote sensing data as needed, meeting the needs of different users for different target detection tasks.

[0086] Based on the content of the above embodiment, as an optional embodiment, the target detection requirement information includes multiple detection targets, applicability data information of each detection algorithm relative to each detection target, and timeliness data information of each type of remote sensing data relative to each detection target; based on the target detection requirement information, determining the target remote sensing data and target detection algorithm for detecting each of the detection targets from the preset remote sensing database and the preset algorithm library respectively includes:

[0087] Based on a variety of detection algorithms and multiple types of remote sensing data, multiple data groups are determined, and each data group includes a detection algorithm and a type of remote sensing data;

[0088] For any one of the multiple detection targets, based on the applicability data information of each detection algorithm relative to any one of the detection targets and the timeliness data information of each type of remote sensing data relative to any one of the detection targets, the comprehensive adaptation score of each data group relative to any one of the detection targets is determined, and the comprehensive adaptation scores of each data group relative to each detection target are obtained;

[0089] With the goal of maximizing the total score of the comprehensive adaptation scores of all detection targets, based on the comprehensive adaptation scores of each data group relative to each detection target, the target data group is determined from each data group;

[0090] Based on the target data group, the target remote sensing data and the target detection algorithm are obtained.

[0091] Specifically, the applicability data information described in the embodiments of the present invention is used to describe the accuracy of the detection algorithm for target detection of the detection target, and can be represented by the accuracy rate.

[0092] The timeliness data information described in the embodiments of the present invention is used to describe the timeliness required by the user for remote sensing data. The calculation method for the timeliness of a certain type of remote sensing data can be as follows: the time required by the user minus the current time (in hours) minus the time of the data from the current time, and then divided by the timeliness required by the user. For example, if the timeliness required by the user is data for the next 12 hours, then the timeliness of the data for the next 5 hours is (12 - 5) / 12.

[0093] In the embodiments of the present invention, according to a variety of detection algorithms in the preset algorithm library and multiple types of remote sensing data in the preset remote sensing database, a detection algorithm and a type of remote sensing data can be determined as a data group, and multiple data groups can be obtained. For example, multiple detection algorithms are represented as {A1, A2,..., Ap}, and multiple types of remote sensing data are represented as {B1, B2,..., Bm}, then multiple data groups can be determined and can be represented as {A1, B1}, {A1, B2},..., {A2, B1}, {A2, B2},..., {Ap, Bm}.

[0094] Furthermore, in the embodiments of the present invention, for any one of the multiple detection targets, according to the preset weight information, based on the applicability data information of each detection algorithm relative to any one of the detection targets and the timeliness data information of each type of remote sensing data relative to any one of the detection targets, the comprehensive adaptation score of each data group relative to any one of the detection targets is determined.

[0095] More specifically, for the data group {A1, B1}, its applicability data information for detection target 1 is the accuracy rate accuracy t1 , and the timeliness data information is time t1 . The weight information corresponding to detection target 1 preset by the combined algorithm is Weight t1 , the corresponding accuracy rate weight information is Weight act1 , and the timeliness weight is Weight tt1 . In this way, the comprehensive adaptation score of the data group {A1, B1} relative to detection target 1 can be calculated through the following formula:

[0096] Weight t1 ×[accuracy t1 ×Weight at1 +time t1 ×Weight tt1 ;

[0097] According to the above principle, the comprehensive adaptation score of the data group {A1, B1} relative to the detection target n can be calculated by the following formula:

[0098]

[0099] Therefore, the total score of the comprehensive adaptation scores of the data group {A1, B1} relative to n detection targets can be calculated by the following formula:

[0100]

[0101] Similarly, the comprehensive adaptation scores of other data groups {A1, B2}, …, {A2, B1}, {A2, B2}, …, {Ap, Bm} relative to each detection target can be calculated in turn.

[0102] Furthermore, in the embodiments of the present invention, according to the comprehensive adaptation scores of each data group relative to each detection target, the total scores of the comprehensive adaptation scores of each data group {A1, B1}, {A1, B2}, …, {A2, B1}, {A2, B2}, …, {Ap, Bm} relative to n detection targets are calculated. Taking the maximum total score of the comprehensive adaptation scores of all detection targets as the goal, the data group with the highest total score is selected. That is, the target data group with the highest total score can be determined from each data group. Assuming the target data group is {A2, B1}, the target remote sensing data B1 and the target detection algorithm A2 for detecting each detection target can be obtained according to the target data group {A2, B1}.

[0103] The method of the embodiments of the present invention can efficiently select the target remote sensing data and the target detection algorithm for multi-target remote sensing detection by using the applicability data information of each detection algorithm relative to any detection target and the timeliness data information of each type of remote sensing data relative to any detection target, and comprehensively analyzing and selecting. It can conveniently enable the user to flexibly and specifically perform multi-target remote sensing recognition and detection.

[0104] In the prior art, the results of traditional target detection are represented in the form of a single picture, and the detection results are obvious at a glance, but it is not convenient for the superposition display of multiple target detection results.

[0105] Based on the content of the above embodiments, as an optional embodiment, after superimposing each target vector data file on the remote sensing image to be detected to obtain the multi-target detection results of the remote sensing image to be detected, the method further includes:

[0106] Slice the original image corresponding to the multi-object detection result of the remote sensing image to be detected according to the preset slicing level to generate a full-image sliced image, and publish the full-image sliced image as a web map tile service;

[0107] When a web map tile service request is received, send the full-image sliced image to the image display interface at the front end for display.

[0108] Specifically, the preset slicing level described in the embodiments of the present invention can be set to the default slicing level, such as level 10. The higher the level, the higher the resolution of the image.

[0109] In the embodiments of the present invention, by superimposing the target vector data file of each detection target on the remote sensing image to be detected, the Shape files of multiple detection targets can be superimposed on the remote sensing image to be detected to obtain the multi-object detection result of the remote sensing image to be detected. Then, the Shape files of each target detection result can be superimposed with the Web Map Tile Service (WMTS), and by publishing the WMTS service, the multi-object detection result can be displayed in the form of an image and presented to the user client.

[0110] Specifically, according to the preset slicing level, such as level 10, slice the original image corresponding to the multi-object detection result of the remote sensing image to be detected, so as to cut the original image into multiple small tiles, and each tile represents a specific area on the map, thereby generating a full-image sliced image. Further, by configuring the parameters of the WMTS service, such as the service URL, layer name, coordinate reference system, etc., the full-image sliced image can be published as a WMTS service for efficient loading and display on the client. Among them, the original image corresponding to the multi-object detection result can be represented as a remote sensing image marked with multiple polygon frames on the remote sensing image to be detected, and each polygon frame corresponds to the Shape file of a detection target.

[0111] Further, in the embodiments of the present invention, when a WMTS service request input by the user at the front end is received, in response to the user request, the full-image sliced image can be sent to the image display interface at the front end for display for the user to browse the multi-object detection result.

[0112] The method of the embodiments of the present invention can achieve the superimposed display of the multi-object detection result by superimposing the Shape file of the multi-object detection result with the WMTS service and publishing the WMTS service to present the multi-object detection result to the user client in the form of a sliced image.

[0113] In the prior art, the traditional method of displaying the target detection results in the browser / server (B / S) mode is to slice the remote sensing image block by block for the whole image and then publish it. The slicing of the whole image takes a long time, has a slow response, and provides a poor user experience.

[0114] Based on the content of the above embodiments, as an alternative embodiment, after receiving a web map tile service request input by the user at the front end and sending the sliced full-image to the image display interface at the front end for display, the method further includes:

[0115] Receiving a first input from the user, where the first input is used to zoom in on the sliced full-image;

[0116] In response to the first input, obtaining the target longitude and latitude information corresponding to the four vertex points of the image display interface after zooming in on the sliced full-image in the original image, and cutting out a target image corresponding to the target geographical area from the original image based on the target geographical area covered by the target longitude and latitude information corresponding to the four vertex points;

[0117] Sending the target image to the image display interface at the front end for display through the WMTS service.

[0118] Specifically, the first input described in the embodiments of the present invention refers to an operation for zooming in on the sliced full-image.

[0119] It should be noted that the first input can be presented in at least one of the following ways:

[0120] Firstly, the first input can be presented as a touch input, including but not limited to click input, swipe input, press input, etc.

[0121] In this embodiment, receiving the first input from the user can be presented as receiving the first input from the user in the image display interface on the front-end display screen.

[0122] Secondly, the first input can be presented as a physical button input.

[0123] In this embodiment, the terminal device is connected to an external device equipped with physical buttons, such as a mouse or a keyboard. Receiving the first input from the user can be presented as receiving the first input from the user pressing the physical buttons of the external device for zooming in on the sliced full-image.

[0124] Thirdly, the first input can be presented as a voice input.

[0125] In this embodiment, the terminal can trigger the operation of zooming in on the sliced full-image through voice recognition.

[0126] Of course, in other embodiments, the first input may also be presented in other forms, including but not limited to character input, etc., which can be determined according to actual needs, and the embodiments of the present invention do not limit this.

[0127] Further, in the embodiments of the present invention, after receiving the first input of the user for the full-map sliced image, in response to the first input, the pixel coordinate information corresponding to the four vertex points of the image display interface after magnifying the full-map sliced image is obtained. Then, through the conversion calculation between the geographic coordinates and the pixel coordinates, the target longitude and latitude information corresponding to the four vertex points of the image display interface in the original image is determined. Thus, the geographical location corresponding to the quadrilateral area formed by the four vertex points of the image display interface can be determined, and the target geographical area covered by it can be obtained.

[0128] Further, in the embodiments of the present invention, the target image corresponding to the target geographical area can be cut out from the original image to ensure that the cut image block matches the current magnification level and the size of the display interface. Further, through the aforementioned WMTS service method, according to the current magnification level and the size of the image display interface, the cut target image can be further cut into multiple tiles, and the size and format of each tile should meet the requirements of the WMTS service. Finally, the sliced tiles are published as a WMTS service. After ensuring the successful publication of the WMTS service, the target image can be sent to the front end in response to the WMTS service request input by the front-end user, and the cut target image is displayed in the front-end image display interface.

[0129] The method of the embodiments of the present invention quickly determines the image area that the user wants to view by locking the position information of the four vertex points of the image display interface after the user magnifies the remote sensing image, realizes the on-demand slicing method, and thus can quickly respond to the WMTS service request of the user, and quickly display the multi-target detection area image that the user wants to view on the front end. Compared with the traditional method of slicing the full map block by block, it can greatly save the slicing time and improve the user browsing experience.

[0130] Figure 2 is the second flow diagram of the remote sensing multi-target detection method provided by the present invention, as Figure 2As shown, in an embodiment of the present invention, according to the user requirements input at the front end, the target detection requirement information of the user for the input remote sensing image is obtained, and multiple detection targets to be detected are determined, including ships, airplanes, oil tanks, etc. At the same time, according to the data timeliness requirement and detection accuracy requirement determined by the target detection requirement information, remote sensing data and target detection algorithms applicable to each detection target are selected to perform multi-target detection, including ship detection, airplane detection, oil tank detection, etc. By detecting each detection target in the input remote sensing image one by one, a target detection result description file for each detection target is extracted, and a result vector file corresponding to each detection target is generated, that is, a target vector data file is obtained. Finally, each target vector data file is superimposed on the original remote sensing image, so that the Shape files of multiple target detection results are superimposed on the original remote sensing image to be detected together, thereby obtaining the multi-target detection result of the remote sensing image to be detected for the user to browse and view.

[0131] The method of the embodiment of the present invention simplifies the multi-target detection problem into multiple single-target detection problems, determines the Shape files of each detection target, is simple to train, and can be flexibly combined to realize the combination of any detection target types.

[0132] The remote sensing multi-target detection device provided by the present invention will be described below. The remote sensing multi-target detection device described below can be correspondingly referred to the remote sensing multi-target detection method described above.

[0133] Figure 3 is a schematic structural diagram of the remote sensing multi-target detection device provided by the present invention. As shown in the figure, it includes: a first generation module 310, a second generation module 320, and a first detection module 330 that are sequentially linked.

[0134] Among them, the first generation module 310 is used to detect each detection target in the remote sensing image to be detected one by one, and generate a target detection result description file for each detection target; the target detection result description file is used to describe the image coordinate information of the detection target in the remote sensing image to be detected;

[0135] The second generation module 320 is used to generate a target vector data file for each detection target based on each target detection result description file;

[0136] The first detection module 330 is used to superimpose each target vector data file on the remote sensing image to be detected to obtain the multi-target detection result of the remote sensing image to be detected.

[0137] The remote sensing multi-object detection device according to the embodiment of the present invention simplifies the multi-object recognition problem into multiple single-object detection problems by detecting each detection target in the remote sensing image to be detected one by one. The training is simple and can be flexibly combined. By extracting the image coordinate information of each detection target in the remote sensing image, a target detection result description file of all detection target position information is generated. Furthermore, the target detection result description file can be converted into a corresponding target vector data file to describe the actual geographical locations of each detection target. Since the vector data file occupies less memory, each target vector data file can be well superimposed on the remote sensing image to be detected to obtain the multi-object detection result of the remote sensing image to be detected, greatly reducing the storage space occupied by the target detection result. At the same time, it can realize superimposing multiple target detection results on one remote sensing image, achieving the multi-object detection effect of satellite remote sensing images.

[0138] Based on the content of the above embodiment, as an optional embodiment, the second generation module 320 is specifically further configured to: extract the image coordinate information covered by each detection target in the remote sensing image to be detected from each target detection result description file, and obtain the affine transformation parameters of the remote sensing image to be detected;

[0139] Generate the longitude and latitude information corresponding to each detection target based on the affine transformation parameters and the image coordinate information covered by each detection target in the remote sensing image to be detected;

[0140] Determine the polygon information corresponding to each detection target based on the longitude and latitude information corresponding to each detection target; the polygon information is used to describe the geographical location information of each detection target;

[0141] Generate a target vector data file for each detection target with the polygon information corresponding to each detection target as an element.

[0142] Based on the content of the above embodiment, as an optional embodiment, the device further includes:

[0143] A first acquisition module, configured to acquire target detection requirement information for the remote sensing image to be detected;

[0144] A first processing module, configured to determine the target remote sensing data and target detection algorithm for detecting each detection target from a preset remote sensing database and a preset algorithm library respectively based on the target detection requirement information, so as to use the target remote sensing data and the target detection algorithm to detect each detection target in the remote sensing image to be detected one by one; the preset algorithm library includes multiple detection algorithms, and the preset remote sensing database includes multiple types of remote sensing data corresponding to the remote sensing image to be detected.

[0145] Based on the content of the above embodiments, as an alternative embodiment, the target detection requirement information includes multiple detection targets, the applicability data information of each detection algorithm with respect to each detection target, and the timeliness data information of each type of remote sensing data with respect to each detection target; the first processing module is further specifically configured to:

[0146] Based on multiple detection algorithms and multiple types of remote sensing data, determine multiple data groups, where a data group includes one detection algorithm and one type of remote sensing data;

[0147] For any one of the multiple detection targets, based on the applicability data information of each detection algorithm with respect to any one detection target and the timeliness data information of each type of remote sensing data with respect to any one detection target, determine the comprehensive adaptation score of each data group with respect to any one detection target, and obtain the comprehensive adaptation scores of each data group with respect to each detection target;

[0148] With the goal of maximizing the total score of the comprehensive adaptation scores of all detection targets, based on the comprehensive adaptation scores of each data group with respect to each detection target, determine the target data group from each data group;

[0149] Based on the target data group, obtain the target remote sensing data and the target detection algorithm.

[0150] Based on the content of the above embodiments, as an alternative embodiment, the device is further specifically configured to:

[0151] Slice the original image corresponding to the multi-target detection result of the remote sensing image to be detected in full according to a preset slicing level to generate a full-image sliced image, and publish the full-image sliced image as a WMTS service;

[0152] When receiving a WMTS service request, send the full-image sliced image to the image display interface at the front end for display.

[0153] Based on the content of the above embodiments, as an alternative embodiment, the device is further specifically configured to:

[0154] Receive a first input from the user, where the first input is used to magnify the full-image sliced image;

[0155] In response to the first input, obtain the target longitude and latitude information in the original image corresponding to the four top corner points of the image display interface after magnifying the full-image sliced image, and based on the target geographical area covered by the target longitude and latitude information corresponding to the four top corner points, cut out the target image corresponding to the target geographical area from the original image;

[0156] Send the target image to the image display interface at the front end for display through the WMTS service.

[0157] The remote sensing multi-target detection device described in this embodiment can be used to execute the remote sensing multi-target detection method embodiment described above. Their principles and technical effects are similar and will not be elaborated here.

[0158] Figure 4 is a schematic physical structure diagram of the electronic device provided by the present invention. As Figure 4 shown, the electronic device may include: a processor 410, a communication interface 420, a memory 430, and a communication bus 440. Among them, the processor 410, the communication interface 420, and the memory 430 complete mutual communication through the communication bus 440. The processor 410 can call the logical instructions in the memory 430 to execute the remote sensing multi-target detection method provided by each of the above methods. The method includes: detecting each detection target in the remote sensing image to be detected one by one, and generating a target detection result description file for each said detection target; the target detection result description file is used to describe the image coordinate information of the detection target in the remote sensing image to be detected; based on each said target detection result description file, generating a target vector data file for each said detection target; superimposing each said target vector data file on the remote sensing image to be detected to obtain the multi-target detection result of the remote sensing image to be detected.

[0159] In addition, when the logical instructions in the above-mentioned memory 430 are implemented in the form of software functional units and sold or used as an independent product, they can be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The foregoing storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), magnetic disks, or optical discs that can store program codes.

[0160] On the other hand, the present invention also provides a computer program product, which includes a computer program. The computer program can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the remote sensing multi-target detection method provided by each of the above methods. The method includes: detecting each detection target in the remote sensing image to be detected one by one, and generating a target detection result description file for each detection target; the target detection result description file is used to describe the image coordinate information of the detection target in the remote sensing image to be detected; based on each target detection result description file, generating a target vector data file for each detection target; and superimposing each target vector data file on the remote sensing image to be detected to obtain the multi-target detection result of the remote sensing image to be detected.

[0161] In another aspect, the present invention also provides a non-transitory computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it is implemented to execute the remote sensing multi-target detection method provided by each of the above methods. The method includes: detecting each detection target in the remote sensing image to be detected one by one, and generating a target detection result description file for each detection target; the target detection result description file is used to describe the image coordinate information of the detection target in the remote sensing image to be detected; based on each target detection result description file, generating a target vector data file for each detection target; and superimposing each target vector data file on the remote sensing image to be detected to obtain the multi-target detection result of the remote sensing image to be detected.

[0162] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place, or may be distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art can understand and implement it without creative efforts.

[0163] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a necessary general hardware platform, and of course, also by hardware. Based on such an understanding, the above technical solutions, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. The computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disc, etc., and includes several instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0164] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A remote sensing multi-target detection method, characterized in that: include: Detect each detection target in the remote sensing image to be detected one by one, and generate a target detection result description file for each detection target; The target detection result description file is used to describe the image coordinate information of the detected target in the remote sensing image to be detected; Based on each of the target detection result description files, generating a target vector data file for each of the detected targets; Superimposing each of the target vector data files with the remote sensing image to be detected to obtain a multi-target detection result of the remote sensing image to be detected; The step of generating a target vector data file for each detected target based on each target detection result description file comprises: Extracting the image coordinate information covered by each of the detected targets in the remote sensing image to be detected from each of the target detection result description files, and acquiring the affine transformation parameters of the remote sensing image to be detected; Based on the affine transformation parameters and the image coordinate information covered by each of the detection targets in the remote sensing image to be detected, generating the longitude and latitude information corresponding to each of the detection targets; Based on the latitude and longitude information corresponding to each of the detection targets, determine the polygon information corresponding to each of the detection targets; the polygon information is used to describe the geographical location information of each of the detection targets; Using polygon information corresponding to each of the detected targets as elements, a target vector data file of each of the detected targets is generated.

2. The remote sensing multi-target detection method according to claim 1, characterized in that: Before detecting each detection target in the remote sensing image to be detected one by one and generating a target detection result description file for each detection target, the method further includes: Acquiring target detection requirement information for the remote sensing image to be detected; Based on the target detection requirement information, the target remote sensing data and the target detection algorithm for detecting each of the detection targets are respectively determined from a preset remote sensing database and a preset algorithm library, so that each detection target in the remote sensing image to be detected is detected one by one by using the target remote sensing data and the target detection algorithm; the preset algorithm library includes multiple detection algorithms, and the preset remote sensing database includes multiple categories of remote sensing data corresponding to the remote sensing image to be detected.

3. The remote sensing multi-target detection method according to claim 2, characterized in that: The target detection requirement information includes multiple detection targets, applicability data information of each detection algorithm relative to each detection target, and timeliness data information of each type of remote sensing data relative to each detection target; and determining the target remote sensing data and target detection algorithm for detecting each detection target from a preset remote sensing database and a preset algorithm library based on the target detection requirement information, respectively, includes: Based on the plurality of detection algorithms and the plurality of types of remote sensing data, determining a plurality of data groups, wherein the data groups include one detection algorithm and one type of remote sensing data; For any detection target among the multiple detection targets, based on the applicability data information of each detection algorithm relative to the any detection target and the timeliness data information of each type of remote sensing data relative to the any detection target, determine the comprehensive adaptation score of each data group relative to the any detection target, and obtain the comprehensive adaptation score of each data group relative to each detection target; Taking the maximum total score of the comprehensive adaptation scores of all the detection targets as the goal, determining a target data group from each of the data groups based on the comprehensive adaptation score of each of the data groups relative to each of the detection targets; Based on the target data group, the target remote sensing data and the target detection algorithm are obtained.

4. The remote sensing multi-target detection method according to any one of claims 1 to 3, characterized in that: After superimposing each of the target vector data files with the remote sensing image to be detected to obtain a multi-target detection result of the remote sensing image to be detected, the method further includes: According to the preset slicing level, the original image corresponding to the multi-target detection result of the remote sensing image to be detected is fully sliced ​​to generate a full-image slice image, and the full-image slice image is published as a web map tile service; When a web map tile service request is received, the full map tile image is sent to the front-end image display interface for display.

5. The remote sensing multi-target detection method according to claim 4, characterized in that: In the case of receiving a web map tile service request, after sending the full map tile image to the image display interface of the front end for display, the method further includes: receiving a first input from a user, wherein the first input is used to enlarge the full-image slice image; In response to the first input, obtaining target longitude and latitude information in the original image corresponding to four vertex points of the image display interface after the full-image slice image is enlarged, and based on the target geographical area covered by the target longitude and latitude information corresponding to the four vertex points, cutting out a target image corresponding to the target geographical area from the original image; The target image is sent to the image display interface of the front end for display through the web map tile service.

6. A remote sensing multi-target detection device, characterized in that: include: The first generating module is used to detect each detection target in the remote sensing image to be detected one by one, and generate a target detection result description file for each detection target; The target detection result description file is used to describe the image coordinate information of the detected target in the remote sensing image to be detected; A second generating module, configured to generate a target vector data file for each detected target based on each target detection result description file; A first detection module, used for superimposing each of the target vector data files with the remote sensing image to be detected to obtain a multi-target detection result of the remote sensing image to be detected; The second generating module is further specifically used for: Extracting the image coordinate information covered by each of the detected targets in the remote sensing image to be detected from each of the target detection result description files, and acquiring the affine transformation parameters of the remote sensing image to be detected; Based on the affine transformation parameters and the image coordinate information covered by each of the detection targets in the remote sensing image to be detected, generating the longitude and latitude information corresponding to each of the detection targets; Determine polygon information corresponding to each of the detection targets based on the latitude and longitude information corresponding to each of the detection targets; The polygon information is used to describe the geographical location information of each of the detection targets; Using polygon information corresponding to each of the detected targets as elements, a target vector data file of each of the detected targets is generated.

7. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the remote sensing multi-target detection method as described in any one of claims 1 to 5 is implemented.

8. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the remote sensing multi-target detection method according to any one of claims 1 to 5 is implemented.

Images