A method, system, electronic device and medium for collaborative interpretation of remote sensing images

Abstract

This invention discloses a method, system, electronic device, and medium for collaborative interpretation of remote sensing images, belonging to the field of remote sensing image processing technology. The method includes: acquiring remote sensing images of different reconnaissance and detection resources; dividing the remote sensing images into different regions to obtain remote sensing images of multiple regions; performing target identification based on a deep learning algorithm on the remote sensing images of each region, and fusing the identified targets in the overlapping parts of the remote sensing images of different regions; the targets include aircraft and ships; labeling the remote sensing images of different regions based on the targets, and integrating the labeled remote sensing images of different regions. This invention can ensure the timeliness, accuracy, and speed of target interpretation for large-scale remote sensing images.

Description

Technical Field

[0001] This invention relates to the field of remote sensing image processing technology, and in particular to a method, system, electronic device, and medium for collaborative interpretation of remote sensing images. Background Technology

[0002] Existing remote sensing image interpretation systems still rely primarily on visual interpretation. While local processing methods have enabled automated computer interpretation, their role is limited to reducing the workload of visual interpretation to a certain extent. They are still unable to achieve real-time, rapid, and accurate target interpretation of large-scale reconnaissance remote sensing images.

[0003] Given the extremely rapid update speed of remote sensing image data, continuing to rely on manual visual interpretation methods for target identification in remote sensing images no longer meets the practical needs of highly intelligent information processing. Researching how to achieve real-time, efficient, and automatic target identification and extraction from optical remote sensing images through collaborative interaction is the future trend. Summary of the Invention

[0004] Based on this, the present invention provides a method, system, electronic device and medium for collaborative interpretation of remote sensing images.

[0005] To achieve the above objectives, the present invention provides the following solution:

[0006] A method for collaborative interpretation of remote sensing images includes:

[0007] Acquire remote sensing images of different reconnaissance and detection resources;

[0008] The remote sensing images are divided into different regions to obtain remote sensing images of multiple regions;

[0009] For the remote sensing images of each region, target identification is performed based on deep learning algorithms, and the targets identified in the remote sensing images of overlapping areas of different regions are fused; the targets include aircraft and ships.

[0010] Based on the target, remote sensing images of different regions are labeled, and the labeled remote sensing images of different regions are integrated.

[0011] Optionally, after dividing the remote sensing image into different regions to obtain remote sensing images of multiple regions, the method further includes:

[0012] The remote sensing images of each region are subjected to correction matching and feature enhancement processing.

[0013] Optionally, targets identified in remote sensing images of overlapping areas are fused, specifically including:

[0014] Target recognition is performed on remote sensing images of overlapping areas based on deep learning algorithms to obtain the recognition probability of different targets;

[0015] The recognition probabilities of different targets are weighted and averaged to obtain the final recognition result.

[0016] The present invention also provides a remote sensing image collaborative interpretation system, comprising:

[0017] The remote sensing image acquisition module is used to acquire remote sensing images of different reconnaissance and detection resources;

[0018] The segmentation module is used to divide the remote sensing image into different regions to obtain remote sensing images of multiple regions.

[0019] The target recognition and fusion module is used to perform target recognition on the remote sensing images of each region based on a deep learning algorithm, and to perform fusion processing on the targets identified in the remote sensing images of overlapping areas of different regions; the targets include aircraft and ships.

[0020] The annotation and integration module is used to annotate remote sensing images of different regions based on the target, and to integrate the annotated remote sensing images of different regions.

[0021] Optionally, it also includes an image processing module for performing correction matching and feature enhancement processing on the remote sensing images of each of the regions.

[0022] The present invention also provides an electronic device, including a memory and a processor, wherein the memory is used to store a computer program, and the processor runs the computer program to enable the electronic device to perform the above-described remote sensing image collaborative interpretation method.

[0023] The present invention also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described remote sensing image collaborative interpretation method.

[0024] According to specific embodiments provided by the present invention, the present invention discloses the following technical effects:

[0025] Compared to traditional methods, this invention divides remote sensing images into multiple regions based on their location. It then uses deep learning algorithms to automatically interpret and identify targets in the remote sensing images of multiple regions simultaneously, and fuses targets in overlapping areas. This fully utilizes the immediacy of the collaborative interaction method to ensure the timeliness, accuracy, and speed of target interpretation for large-scale remote sensing images. Attached Figure Description

[0026] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0027] Figure 1 This is a flowchart of the remote sensing image collaborative interpretation method provided in Embodiment 1 of the present invention. Detailed Implementation

[0028] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0029] The purpose of this invention is to provide a method, system, electronic device, and medium for collaborative interpretation of remote sensing images, so as to ensure the timeliness, accuracy, and speed of target interpretation in large-scale remote sensing images.

[0030] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0031] Example 1

[0032] This embodiment provides a method for collaborative interpretation of remote sensing images, such as... Figure 1 As shown, the method includes the following steps:

[0033] Step 101: Acquire remote sensing images of different reconnaissance and detection resources.

[0034] Step 102: Divide the remote sensing images into different regions to obtain remote sensing images of multiple regions.

[0035] Following step 102, the remote sensing images of each region undergo correction matching and feature enhancement processing. Specifically, this includes:

[0036] (1) Radiometric and orthorectification processing was performed on the remote sensing images of each region using GCP sampling points and standard radiometric sources;

[0037] (2) Remote sensing image output is performed by resampling to obtain remote sensing image resources that match geographic coordinates;

[0038] (3) Enhance the features of remote sensing images by image transformation, image filtering, image edge processing, etc., and improve the image difference features around the target;

[0039] (4) Process remote sensing images by image cropping and mosaicking.

[0040] Step 103: For remote sensing images of each region, target identification is performed based on deep learning algorithms, and the identified targets in the remote sensing images of overlapping areas of different regions are fused; the targets include aircraft and ships.

[0041] Specifically, the DarkNet-53 backbone network and FPN network based on deep learning algorithms are used to extract features from remote sensing images and extract feature maps. The head network of deep learning algorithms is used to extract the center point coordinates, width, and height parameters of the predicted target boxes through regression methods. The logistic regression method of deep learning algorithms is used to extract the confidence score of each predicted target box belonging to each category. Based on the confidence score being greater than a certain threshold, the predicted target boxes are filtered. Then, the non-maximum suppression (NMS) method is used to extract the final target boxes from all predicted target boxes, and the argmax method is used to obtain the target category.

[0042] For overlapping areas in different regions, multiple target recognition results may exist, which need to be fused. Specifically, this includes:

[0043] When using deep learning algorithms for identification, the target category and confidence level are given (e.g., the probability that a target belongs to category A is a, the probability that it belongs to category B is b, the probability that it belongs to category C is c, and a+b+c=1).

[0044]

[0045] Where n is the total number of categories, p i Let be the probability of belonging to the i-th category.

[0046] The final classification of the target is obtained using a probability-weighted method, whereby the overall confidence score of a category is the average of the confidence scores given by all overlapping components, and the category with the highest confidence score is the final classification category.

[0047]

[0048] Where C is the set of categories, c is the subcategory in the set, and k is the number of overlapping regions belonging to different regions.

[0049] Step 104: Annotate the remote sensing images of different regions based on the target, and integrate the annotated remote sensing images of different regions.

[0050] Compared to traditional methods, this invention divides remote sensing images into multiple regions based on their location. It then uses deep learning algorithms to automatically interpret and identify targets in the remote sensing images of multiple regions simultaneously, and fuses targets in overlapping areas. This fully utilizes the immediacy of the collaborative interaction method to ensure the timeliness, accuracy, and speed of target interpretation for large-scale remote sensing images.

[0051] Example 2

[0052] In order to implement the method corresponding to Embodiment 1 above and achieve the corresponding functions and technical effects, a remote sensing image collaborative interpretation system is provided below.

[0053] The system includes:

[0054] The remote sensing image acquisition module is used to acquire remote sensing images of different reconnaissance and detection resources.

[0055] The segmentation module is used to divide remote sensing images into different regions, resulting in remote sensing images of multiple regions.

[0056] The target recognition and fusion module is used to identify targets in remote sensing images of various regions based on deep learning algorithms, and to fuse the identified targets in remote sensing images of overlapping areas of different regions; the targets include aircraft and ships.

[0057] The annotation and integration module is used to annotate remote sensing images of different regions based on the target, and to integrate the annotated remote sensing images of different regions.

[0058] The system also includes an image processing module for correcting, matching, and enhancing the features of remote sensing images of various regions.

[0059] Example 3

[0060] Embodiment 3 of the present invention provides an electronic device, including a memory and a processor. The memory is used to store a computer program, and the processor runs the computer program to enable the electronic device to perform the remote sensing image collaborative interpretation method of Embodiment 1.

[0061] The aforementioned electronic device may be a server.

[0062] Example 4

[0063] Embodiment 4 of the present invention provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the remote sensing image collaborative interpretation method of Embodiment 1.

[0064] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the systems disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the descriptions are relatively simple; relevant parts can be referred to the method section.

[0065] This article uses specific examples to illustrate the principles and implementation methods of the invention. The description of the above embodiments is only for the purpose of helping to understand the method and core ideas of the present invention. The described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

Claims

1. A method for collaborative interpretation of remote sensing images, characterized in that, The method comprises the following steps: acquiring remote sensing images of different reconnaissance and detection resources; dividing the remote sensing images according to different regions to obtain remote sensing images of multiple regions; performing target recognition on the remote sensing images of each region based on a deep learning algorithm, and performing fusion processing on the recognized targets in the remote sensing images of the overlapping parts of different regions; specifically, performing feature extraction on the remote sensing images based on the backbone network DarkNet-53 and the FPN network of the deep learning algorithm to extract feature maps; extracting the center point coordinates, width and height parameters of the predicted target frame by the Head network of the deep learning algorithm through a regression method; extracting the confidence of each predicted target frame belonging to each category by the logistic regression method of the deep learning algorithm; screening the predicted target frame based on the confidence being greater than a certain threshold, then extracting the final target frame by the non-maximum suppression (NMS) method, and obtaining the target category by the argmax method; the targets include aircraft and ships; annotating the remote sensing images of different regions based on the targets, and integrating the annotated remote sensing images of different regions; specifically, performing target recognition on the remote sensing images of the overlapping parts of different regions based on the deep learning algorithm to obtain the recognition probability of different targets; performing weighted average on the recognition probability of different targets to obtain the final recognition result.

2. The method of claim 1, wherein, After the remote sensing images are divided according to different regions to obtain remote sensing images of multiple regions, the method further comprises the following steps: performing correction matching and feature enhancement processing on the remote sensing images of each region.

3. A remote sensing image collaborative interpretation system, characterized in that, The system is applied to the remote sensing image collaborative interpretation method of any one of claims 1-2, and the system comprises: a remote sensing image acquisition module for acquiring remote sensing images of different reconnaissance and detection resources; a division module for dividing the remote sensing images according to different regions to obtain remote sensing images of multiple regions; a target recognition and fusion module for performing target recognition on the remote sensing images of each region based on a deep learning algorithm, and performing fusion processing on the recognized targets in the remote sensing images of the overlapping parts of different regions; the targets include aircraft and ships; a labeling and integration module for annotating the remote sensing images of different regions based on the targets, and integrating the annotated remote sensing images of different regions.

4. The remote sensing image collaborative interpretation system of claim 3, wherein, The system further comprises: an image processing module for performing correction matching and feature enhancement processing on the remote sensing images of each region.

5. An electronic device, comprising: The electronic device comprises a memory and a processor, the memory is used to store a computer program, and the processor runs the computer program to enable the electronic device to perform the remote sensing image collaborative interpretation method of any one of claims 1-2.

6. A computer readable storage medium characterized by, The electronic device has a computer program stored therein, and the computer program is executed by the processor to implement the remote sensing image collaborative interpretation method of any one of claims 1-2.

Images