Method for extracting glacier lake from remote sensing image, computer equipment and storage medium

Abstract

The invention relates to the technical field of image detection, and discloses a remote sensing image glacier lake extraction method, computer equipment and a storage medium, the remote sensing image glacier lake extraction method comprises the following steps: obtaining a trained glacier lake extraction network, the glacier lake extraction network comprising a feature extraction network and a geometric primitive extraction network; obtaining a to-be-detected remote sensing image, wherein the to-be-detected remote sensing image comprises a glacier image and / or a lake image; and inputting a to-be-detected remote sensing image into the glacier lake extraction network, extracting glacier lake features in the to-be-detected remote sensing image by using the feature extraction network, and segmenting the glacier lake features by using the geometric primitive extraction network to obtain a glacier lake region image. The glacier lake can be automatically extracted, and the extraction accuracy is improved.

Description

Technical Field

[0001] The present invention relates to the field of image detection technology, and in particular to a method for extracting glacial lakes from remote sensing images, a computer device and a storage medium. Background Art

[0002] Understanding glacier changes is of great significance for studying global climate change. Traditional glacier survey methods require field surveys, which are time-consuming and labor-intensive, and difficult to survey over a large area. Monitoring methods based on remote sensing images can quickly and accurately obtain the spatial distribution of glaciers at a low cost, so they have become a research hotspot for glacier monitoring.

[0003] At present, there are two main types of glacier monitoring methods based on remote sensing images: band combination threshold method, such as the normalized snow index (NDSI) index and the automated glacier extraction index (AGEI), which combines the characteristic bands sensitive to glaciers by mathematical operations, and automatically or semi-automatically extracts the glacier range by setting thresholds; traditional machine learning classification methods, which use machine learning classifiers such as maximum likelihood, decision tree classification, neural networks, and support vector machines, rely on manually selected sample pixels for learning, and then use them for glacier extraction, and have achieved certain experimental results. The above two methods can achieve good results in local areas or single images, but for large-scale ranges, due to complex scenes such as cloud shadows, mountain shadows, and water surface ice, it is difficult to achieve good robustness and high accuracy. Summary of the invention

[0004] In view of this, the present invention provides a method for extracting glacial lakes from remote sensing images, a computer device and a storage medium to achieve automatic extraction of glacial lakes and improve extraction accuracy.

[0005] In a first aspect, the present invention provides a method for extracting glacial lakes from remote sensing images, which comprises: obtaining a trained glacial lake extraction network, the glacial lake extraction network comprising a feature extraction network and a geometric primitive extraction network; obtaining a remote sensing image to be detected, the remote sensing image to be detected comprising a glacier image and / or a lake image; inputting the remote sensing image to be detected into the glacial lake extraction network, extracting glacial lake features in the remote sensing image to be detected using the feature extraction network, and segmenting the glacial lake area image from the glacial lake features using the geometric primitive extraction network.

[0006] In this implementation method, this application combines the feature extraction network and the geometric primitive extraction network, and can combine GIS-assisted mapping technology on the basis of algorithm design to achieve fool-proof automatic extraction of large-area remote sensing images, and can perform overall extraction of provincial-level images, realize automatic extraction of glaciers and lakes, and improve extraction accuracy.

[0007] In one possible implementation, the steps of training a glacial lake extraction network include: constructing a data set, the data set including a training set, a validation set and a test set; constructing a basic glacial lake extraction network; based on a transfer learning algorithm, using the data set to train the basic glacial lake extraction network to obtain a glacial lake extraction network.

[0008] In one possible implementation, constructing a data set includes: acquiring multiple sample images, the sample images including glacier images and lake images; extracting a glacier index and a water system index of the sample images; annotating glaciers and lakes on the sample images based on the glacier index and the water system index to obtain sample annotation data; and constructing a data set by combining multiple groups of sample images and corresponding sample annotation data.

[0009] In one possible implementation, based on a transfer learning algorithm, a basic glacial lake extraction network is trained using a data set to obtain the glacial lake extraction network, including: obtaining training parameters, the training parameters including a learning rate, a number of iterations, a block size, and a graphics processor type; selecting a corresponding graphics processor based on the graphics processor type; and calling the graphics processor to train the basic glacial lake extraction network based on the training set.

[0010] In a possible implementation, the method for extracting glacial lakes from remote sensing images also includes: verifying the glacial lake extraction network using a validation set, and testing the glacial lake extraction network using a detection set.

[0011] In this implementation, the present invention adopts a small sample and weakly supervised learning method, uses technical means to achieve sample transformation and sample expansion, and combines transfer learning technology to achieve small sample weakly supervised learning. By manually making a small number of samples, the training of glaciers and lakes can be completed.

[0012] In a possible implementation, extracting glacial lake features in a remote sensing image to be detected using a feature extraction network includes: using a backbone network to respectively extract spectral features and spatial features of the remote sensing image to be detected.

[0013] In one possible implementation, a geometric primitive extraction network includes a detector, a segmenter, and a decoder. Using the geometric primitive extraction network to segment glacial lake features to obtain a glacial lake area image includes: identifying a bounding box of a glacial lake image in a remote sensing image to be detected based on the detector; segmenting the glacial lake image according to the bounding box based on the segmenter; and determining a sequence of glacial lake images based on the decoder to obtain a glacial lake area image.

[0014] In a possible implementation, the remote sensing image glacial lake extraction method further includes: converting the glacial lake area image into a glacial lake vector image; deleting the background area in the glacial lake vector image to obtain a glacial lake target image.

[0015] In this implementation, the extraction results are directly converted into vector layers, which enhances practical performance.

[0016] In a second aspect, the present invention provides a computer device, comprising: a memory and a processor, the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the method for extracting glacial lakes in remote sensing images of the first aspect or any corresponding embodiment thereof by executing the computer instructions.

[0017] In a third aspect, the present invention provides a computer-readable storage medium having computer instructions stored thereon, the computer instructions being used to enable a computer to execute the method for extracting glacial lakes from remote sensing images of the first aspect or any corresponding embodiment thereof. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0019] Figure 1 is a flow chart of a method for extracting glacial lakes from remote sensing images according to an embodiment of the present invention;

[0020] Figure 2 is a flowchart of training a glacial lake extraction network according to an embodiment of the present invention;

[0021] Figure 3 is a flow chart of another method for extracting glacial lakes from remote sensing images according to an embodiment of the present invention;

[0022] Figure 4is a schematic structural diagram of a device for extracting glacial lakes from remote sensing images according to an embodiment of the present invention;

[0023] Figure 5 Schematic diagram of the hardware structure of a computer device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0024] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0025] According to an embodiment of the present invention, an embodiment of a method for extracting glacial lakes from remote sensing images is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in an order different from that shown here.

[0026] In this embodiment, a method for extracting glacial lakes from remote sensing images is provided. Figure 1 is a flowchart of a method for extracting glacial lakes from remote sensing images according to an embodiment of the present invention. It should be noted that if there are substantially the same results, this embodiment is not limited to Figure 1 The process sequence shown is limited. Figure 1 As shown, the process includes the following steps:

[0027] Step S101, obtaining a trained glacial lake extraction network.

[0028] Among them, the glacial lake extraction network includes a feature extraction network and a geometric primitive segmentation network.

[0029] The feature extraction network is used to extract the characteristic information of glacial lakes in the image; the geometric primitive extraction network is used to automatically identify and extract the depth information of geometric primitives (such as basic shapes such as lines, circles, rectangles, etc.) from the image, specifically, to automatically identify and extract the geometric primitive information of glacial lakes from the image.

[0030] In one implementation, the glacial lake extraction network is a P2PFormer.

[0031] Exemplarily, the glacial lake extraction network is a Mask2Former network. Mask2Former is a model for general image segmentation that combines a masked attention mechanism and a mask transformer architecture. This model can handle a variety of image segmentation tasks such as panoptic segmentation, instance segmentation, and semantic segmentation.

[0032] The basic glacial lake extraction network is trained using the model training method to obtain the trained glacial lake extraction network.

[0033] Step S102: acquiring a remote sensing image to be detected.

[0034] The remote sensing images to be detected include glacier images and / or lake images.

[0035] Among them, remote sensing images (RS) can record films or photos of electromagnetic wave sizes of various ground objects, mainly divided into aerial photos and satellite photos. Remote sensing images can be photos within the visible light range or data records of other electromagnetic spectra (such as infrared, microwave, etc.). According to different needs, remote sensing images can be divided into many types, such as optical images, radar images (SAR), thermal infrared images, etc.

[0036] Remote sensing images can be applied to environmental monitoring: monitoring changes in forest cover, land use, wetland, marine pollution, glacier changes, etc.; agriculture: assessing crop health, estimating yields, planning irrigation and fertilization, etc.; disaster management: rapid response to natural disasters (such as floods, fires, earthquakes), providing maps and analysis before, during and after disasters; urban planning: helping with urban expansion analysis, infrastructure construction planning, etc.; resource exploration: used for the detection of natural resources such as minerals, oil and natural gas; climate change research: long-term monitoring of changes in climate patterns, such as rising temperatures and sea levels.

[0037] This application mainly targets remote sensing images with glacial lakes. Specifically, images with glacial lake information on the earth's surface are acquired from a long distance through sensors carried on satellites, airplanes or other aircraft to obtain remote sensing images to be detected.

[0038] Step S103, inputting the remote sensing image to be detected into the glacial lake extraction network, using the feature extraction network to extract the glacial lake features in the remote sensing image to be detected, and using the geometric primitive extraction network to segment the glacial lake features to obtain the glacial lake area image.

[0039] The remote sensing image to be detected is input into the glacier lake extraction network, and is processed successively by the feature extraction network and the geometric primitive extraction network.

[0040] Specifically, the feature extraction network can use a convolutional neural network to automatically learn the feature representation of glacial lakes in the remote sensing images to be detected, which can capture spatial patterns of different scales and complexities and provide a basis for subsequent geometric primitive detection.

[0041] Furthermore, the geometric primitive extraction network is based on the geometric reasoning module, performs geometric reasoning according to the features learned by the feature extraction network, extracts and segments the primitive data of the glacial lake, including parameters such as specific location, size, angle or other graph structures, and outputs an image of the glacial lake area.

[0042] The method for extracting glaciers and lakes from remote sensing images provided in this embodiment, by combining a feature extraction network and a geometric primitive extraction network, can be combined with GIS-assisted mapping technology on the basis of algorithm design to achieve fool-proof automatic extraction of large-area remote sensing images. It can perform overall extraction on provincial-level images, realize automated extraction of glaciers and lakes, and improve extraction accuracy.

[0043] Before performing the above-mentioned remote sensing image glacial lake extraction method, a glacial lake extraction network is pre-trained. Therefore, in this embodiment, a glacial lake extraction network training method is provided. Figure 2 is a flowchart of training a glacial lake extraction network according to an embodiment of the present invention. It should be noted that if there are substantially the same results, this embodiment does not use Figure 2 The process sequence shown is limited. Figure 2 As shown, the process includes the following steps:

[0044] Step S201, constructing a data set.

[0045] The dataset includes training set, validation set and test set.

[0046] Specifically, a plurality of pseudo-color sample images including glacier images and lake images are obtained, and the ice and snow index of the sample images is calculated, wherein the ice and snow index includes the glacier index and the water system index of the sample images.

[0047] A water body map is generated by calculating the glacier index and water system index, and on this basis, the sample images are annotated with glaciers and lakes to obtain sample annotation data. A data set is constructed by combining multiple groups of sample images and the corresponding sample annotation data.

[0048] The dataset is divided into training set, validation set and test set.

[0049] Among them, the training set is the data set used to train the model; the validation set is used to adjust the model's hyperparameters (for example, learning rate, regularization strength, etc.) and select the best model architecture; the test set is used to evaluate the generalization ability of the final model, that is, the performance of the model on new data that has not been seen.

[0050] In one implementation, the data set is divided into a training set, a validation set, and a test set in proportion. The specific division ratio can be adjusted according to the amount of data and the test requirements.

[0051] For example, 70% of the data is used as a training set, 15% as a validation set, and 15% as a test set.

[0052] In another implementation, cross-validation is used to divide the data set into a training set, a validation set, and a test set.

[0053] For example, k-fold cross-validation is used. This method divides the data into k subsets, and uses one of the subsets as a validation set in each iteration, and the rest as a training set. This can make better use of limited data resources and obtain more reliable performance estimates.

[0054] Step S202, constructing a basic glacier lake extraction network.

[0055] The glacier lake extraction network is constructed by combining the feature extraction network and the geometric primitive extraction network.

[0056] The geometric primitive extraction network includes an input layer, a feature extraction layer, a geometric reasoning module and an output layer.

[0057] The input layer receives a color or grayscale image as input data, and the feature extraction layer uses a convolutional neural network to learn feature representations in the input image.

[0058] Among them, the geometric reasoning module adopts different strategies to predict geometric parameters, or uses classification methods to assign pixels to predefined geometric categories. In some implementations, the geometric reasoning module also includes an attention mechanism to focus more on key areas related to geometric primitives in the image. The preliminary geometric primitive prediction results are optimized using optimization algorithms, such as non-maximum suppression (NMS), geometric filtering, merging similar primitives, and other technologies.

[0059] Among them, the output layer is used to output information describing the geometric primitives in the image, such as the specific position, size, angle and other parameters of the primitives, or a more abstract graph structure representation.

[0060] In one implementation, the geometric primitive extraction network is a Mask2Former network. The Mask2Former network includes a Transformer-based encoding-decoding architecture, a multi-scale feature fusion module, a binary mask prediction module, a multi-modal cross attention module, and a dynamic convolution module.

[0061] The encoder in the Transformer-based encoding-decoding architecture is used to extract features from the input image, and the decoder is used to generate the target output, that is, the category label or mask corresponding to each pixel. The encoder usually uses a pre-trained visual transformer (ViT), such as the Swin Transformer, which can enhance the learning ability of fine-grained features. The decoder includes a binary mask prediction module for predicting binary foreground / background masks, which can simplify the model and improve the segmentation accuracy of irregularly shaped objects.

[0062] Among them, the multi-scale feature fusion module can integrate feature maps at multiple levels. It can locate small objects more accurately and maintain the overall consistency of large objects. The multimodal cross-attention module can establish connections between different modalities, and the dynamic convolution module can adaptively adjust its receptive field according to different contexts, enhancing the focus on local details and helping to improve segmentation quality.

[0063] Among them, Mask2Former uses anchor-free detection, which can reduce the workload of hyperparameter tuning and improve the flexibility of the model.

[0064] Step S203, based on the transfer learning algorithm, the basic glacial lake extraction network is trained using the data set to obtain the glacial lake extraction network.

[0065] Obtain training parameters, including learning rate, number of iterations, block size and graphics processor type, and select a corresponding graphics processor (Graphics Processing Unit, GPU) based on the graphics processor type; call the graphics processor to train the basic glacial lake extraction network based on the training set.

[0066] Among them, the use of GPU computing greatly improves the detection efficiency. The present invention uses GPU accelerated processing. Taking a single window (1000×1000 pixels) as an example, a single GPU detection time only takes 0.4 seconds. By increasing the number of GPUs and adopting a parallel processing architecture, the detection efficiency can be improved.

[0067] Among them, a loss function is constructed for training, and the loss function includes one or more combinations of a cross entropy loss function, an L2 loss function, etc. Training is performed using constraints, such as shape priors or topological consistency.

[0068] Furthermore, the validation set was used to verify the trained glacial lake extraction network and adjust the model's hyperparameters. The detection set was used to test the trained glacial lake extraction network and evaluate the model's detection accuracy.

[0069] Traditional supervised learning methods require a large number of learning samples for training. The present invention adopts small sample and weak supervised learning methods, uses technical means to achieve sample transformation and sample expansion, and combines transfer learning technology to achieve small sample weak supervised learning. By manually making a small number of samples, the training of glaciers and lakes can be completed.

[0070] On this basis, the trained glacial lake extraction network is used to extract glacial lakes. In this embodiment, another method for extracting glacial lakes from remote sensing images is provided. Figure 3 is a flowchart of another method for extracting glacial lakes from remote sensing images according to an embodiment of the present invention. It should be noted that if there are substantially the same results, this embodiment is not limited to the above. Figure 3 The process sequence shown is limited. Figure 3 As shown, the process includes the following steps:

[0071] Step S301, obtaining the trained glacial lake extraction network.

[0072] Specifically, using Figure 2 The method shown trains a glacial lake extraction network.

[0073] Step S302: Acquire a remote sensing image to be detected.

[0074] The remote sensing images to be detected include glacier images and / or lake images. Figure 1 The step S102 shown is not described in detail here.

[0075] Step S303, input the remote sensing image to be detected into the glacial lake extraction network, use the feature extraction network to extract the glacial lake features in the remote sensing image to be detected, and use the geometric primitive extraction network to segment the glacial lake features to obtain the glacial lake area image.

[0076] Specifically, in one implementation, step S303 specifically includes:

[0077] Step S3031, using the backbone network to extract the spectral features and spatial features of the remote sensing image to be detected.

[0078] Step S3032, using the detector of the geometric primitive extraction network to identify the bounding box of the glacial lake image in the remote sensing image to be detected, segmenting the glacial lake image according to the bounding box based on the segmentor, determining the sequence of the glacial lake image based on the decoder, and obtaining the glacial lake area image.

[0079] Specifically, based on the method of geometric primitive extraction network, P2PFormer is introduced, that is, primitive to polygon using Transformer. P2PFormer includes detector, segmenter and decoder.

[0080] Among them, the detector is used to identify the bounding boxes of glaciers and lakes, the segmenter is used to segment the primitives within the bounding boxes, and the decoder is used to determine the order of primitives, and the combined output is a glacier lake area image, wherein the glacier lake area image is an index map of the same size as the original image.

[0081] Furthermore, the glacier lake area image is converted into a glacier lake vector image, the background area in the glacier lake vector image is deleted, and the glacier and lake areas are retained to obtain the glacier lake target image.

[0082] In this implementation, the extraction results are directly converted into vector layers, which enhances practical performance.

[0083] In this implementation, the implicit update strategy is used to embed the query position, which can make the query focus more on the correct position, thereby improving the quality of primitive segmentation, and combining the spatial and spectral characteristics of the remote sensing image itself to successfully achieve intelligent extraction of glaciers and lakes. Taking high-resolution optical remote sensing images as the object, based on deep learning methods, using the new image segmentation model Mask2Former, and using the limited sample transfer learning method to achieve high-precision detection of typical targets in remote sensing images and high-precision classification of surface coverage in remote sensing images, we carry out large-scale and massive remote sensing image glacier and lake extraction based on deep learning.

[0084] In this embodiment, a device for extracting glacial lakes from remote sensing images is also provided, and the device is used to implement the above-mentioned embodiments and preferred implementation modes, and the details that have been described will not be repeated. As used below, the term "module" can be a combination of software and / or hardware that implements a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceivable.

[0085] This embodiment provides a remote sensing image glacial lake extraction device, such as Figure 4 As shown, Figure 4 1 is a schematic diagram of a structure of a device for extracting glacial lakes from remote sensing images according to an embodiment of the present invention. The device for extracting glacial lakes from remote sensing images comprises:

[0086] The first acquisition module 401 is used to acquire a trained glacial lake extraction network, where the glacial lake extraction network includes a feature extraction network and a geometric primitive extraction network.

[0087] The second acquisition module 402 is used to acquire the remote sensing images to be detected, where the remote sensing images to be detected include glacier images and / or lake images.

[0088] The extraction module 403 is used to input the remote sensing image to be detected into the glacial lake extraction network, use the feature extraction network to extract the glacial lake features in the remote sensing image to be detected, and use the geometric primitive extraction network to segment the glacial lake features to obtain the glacial lake area image.

[0089] The further functional description of each of the above modules is the same as that of the above corresponding embodiments and will not be repeated here.

[0090] The remote sensing image glacial lake extraction device in this embodiment is presented in the form of a functional unit, where the unit refers to an ASIC (Application Specific Integrated Circuit) circuit, a processor and memory that executes one or more software or fixed programs, and / or other devices that can provide the above functions.

[0091] The embodiment of the present invention also provides a computer device having the above Figure 4 The adaptive lane detection device shown.

[0092] See also Figure 5 , Figure 5 is a schematic diagram of the structure of a computer device provided by an optional embodiment of the present invention, such as Figure 5 As shown, the computer device includes: one or more processors 10, a memory 20, and interfaces for connecting various components, including high-speed interfaces and low-speed interfaces. Various components are connected to each other using different buses for communication, and can be installed on a common mainboard or installed in other ways as needed. The processor can process instructions executed in the computer device, including instructions stored in or on the memory to display the graphical information of the GUI on an external input / output device (such as a display device coupled to the interface). In some optional embodiments, if necessary, multiple processors and / or multiple buses can be used together with multiple memories. Similarly, multiple computer devices can be connected, and each device provides some necessary operations (for example, as a server array, a group of blade servers, or a multi-processor system). Figure 5 A processor 10 is taken as an example.

[0093] The processor 10 may be a central processing unit, a network processor or a combination thereof. The processor 10 may further include a hardware chip. The hardware chip may be a dedicated integrated circuit, a programmable logic device or a combination thereof. The programmable logic device may be a complex programmable logic device, a field programmable gate array, a general purpose array logic or any combination thereof.

[0094] The memory 20 stores instructions executable by at least one processor 10, so that the at least one processor 10 executes the method shown in the above embodiment.

[0095] The memory 20 may include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function; the data storage area may store data created according to the use of the computer device, etc. In addition, the memory 20 may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some optional embodiments, the memory 20 may optionally include a memory remotely arranged relative to the processor 10, and these remote memories may be connected to the computer device via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0096] The memory 20 may include a volatile memory, such as a random access memory; the memory may also include a non-volatile memory, such as a flash memory, a hard disk or a solid state drive; the memory 20 may also include a combination of the above types of memory.

[0097] The computer device also includes an input device 30 and an output device 40. The processor 10, the memory 20, the input device 30 and the output device 40 may be connected via a bus or other means. Figure 5 The example of connecting through bus is taken in the following.

[0098] The input device 30 can receive input digital or character information, and generate key signal input related to the user settings and function control of the computer device, such as a touch screen, a keypad, a mouse, a track pad, a touch pad, an indicator bar, one or more mouse buttons, a trackball, a joystick, etc. The output device 40 may include a display device, an auxiliary lighting device (e.g., an LED) and a tactile feedback device (e.g., a vibration motor), etc. The above-mentioned display device includes but is not limited to a liquid crystal display, a light emitting diode, a display and a plasma display. In some optional embodiments, the display device can be a touch screen.

[0099] The embodiment of the present invention also provides a computer-readable storage medium. The method according to the embodiment of the present invention can be implemented in hardware, firmware, or can be implemented as a computer code that can be recorded in a storage medium, or can be implemented as a computer code that is originally stored in a remote storage medium or a non-temporary machine-readable storage medium and will be stored in a local storage medium through a network download, so that the method described herein can be stored in such software processing on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. Among them, the storage medium can be a magnetic disk, an optical disk, a read-only storage memory, a random access memory, a flash memory, a hard disk or a solid-state hard disk, etc.; further, the storage medium can also include a combination of the above types of memories. It can be understood that a computer, a processor, a microprocessor controller, or programmable hardware includes a storage component that can store or receive software or computer code. When the software or computer code is accessed and executed by a computer, a processor, or hardware, the method shown in the above embodiment is implemented.

[0100] A part of the present invention may be applied as a computer program product, such as a computer program instruction, which, when executed by a computer, can call or provide the method and / or technical solution according to the present invention through the operation of the computer. Those skilled in the art should understand that the existence of the computer program instruction in a computer-readable medium includes, but is not limited to, a source file, an executable file, an installation package file, etc., and accordingly, the way in which the computer program instruction is executed by the computer includes, but is not limited to: the computer directly executes the instruction, or the computer compiles the instruction and then executes the corresponding compiled program, or the computer reads and executes the instruction, or the computer reads and installs the instruction and then executes the corresponding installed program. Here, the computer-readable medium may be any available computer-readable storage medium or communication medium accessible to the computer.

[0101] Although the embodiments of the present invention have been described in conjunction with the accompanying drawings, those skilled in the art may make various modifications and variations without departing from the spirit and scope of the present invention, and such modifications and variations are all within the scope defined by the appended claims.

Claims

1. A method for extracting glacial lakes from remote sensing images, characterized in that: The method comprises: Acquire a trained glacial lake extraction network, wherein the glacial lake extraction network includes a feature extraction network and a geometric primitive extraction network; Acquire a remote sensing image to be detected, wherein the remote sensing image to be detected includes a glacier image and / or a lake image; The remote sensing image to be detected is input into the glacial lake extraction network, the glacial lake features in the remote sensing image to be detected are extracted using the feature extraction network, and the geometric primitive extraction network is used to segment the glacial lake features to obtain a glacial lake area image.

2. The method for extracting glacial lakes from remote sensing images according to claim 1, characterized in that: The steps of training the glacial lake extraction network include: Constructing a data set, wherein the data set includes a training set, a validation set, and a test set; Construct a basic glacial lake extraction network; Based on the transfer learning algorithm, the basic glacial lake extraction network is trained using the data set to obtain the glacial lake extraction network.

3. The method for extracting glacial lakes from remote sensing images according to claim 2, characterized in that: The constructing of the data set comprises: Acquire a plurality of sample images, wherein the sample images include glacier images and lake images; extracting a glacier index and a water system index of the sample image; Based on the glacier index and the water system index, the sample image is annotated with glaciers and lakes to obtain sample annotated data; The data set is constructed by combining multiple groups of sample images and corresponding sample annotation data.

4. The method for extracting glacial lakes from remote sensing images according to claim 2, characterized in that: The transfer learning algorithm is based on the data set to train the basic glacial lake extraction network, and the glacial lake extraction network is obtained, which includes: Obtaining training parameters, wherein the training parameters include a learning rate, a number of iterations, a block size, and a graphics processor type; Selecting a corresponding graphics processor based on the graphics processor type; The graphics processor is called to train the basic glacial lake extraction network based on the training set.

5. The method for extracting glacial lakes from remote sensing images according to claim 2, characterized in that: The method further comprises: The glacial lake extraction network is verified using the verification set, and the glacial lake extraction network is tested using the detection set.

6. The method for extracting glacial lakes from remote sensing images according to claim 1, characterized in that: The extracting of glacial lake features in the remote sensing image to be detected by using a feature extraction network comprises: The backbone network is used to extract the spectral features and spatial features of the remote sensing image to be detected.

7. The method for extracting glacial lakes from remote sensing images according to claim 6, characterized in that: The geometric primitive extraction network includes a detector, a segmenter and a decoder, and the segmenting of the glacial lake feature by using the geometric primitive extraction network to obtain a glacial lake area image includes: Identifying a bounding box of a glacial lake image in the remote sensing image to be detected based on the detector; Segmenting the glacial lake image according to the bounding box based on the segmenter; The sequence of the glacial lake images is determined based on the decoder to obtain the glacial lake area image.

8. The method for extracting glacial lakes from remote sensing images according to claim 7, characterized in that: The method further comprises: Converting the glacial lake region image into a glacial lake vector image; The background area in the glacier lake vector image is deleted to obtain a glacier lake target image.

9. A computer device, characterized in that: include: A memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the remote sensing image glacier lake method according to any one of claims 1 to 8 by executing the computer instructions.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a computer to execute the remote sensing image glacier lake method described in any one of claims 1 to 8.

Images