A lightweight full information fusion and dual-guide remote sensing image change detection method and device

Abstract

A lightweight full information fusion and dual-guide remote sensing image change detection method and device, the method comprising: inputting the obtained remote sensing image data set into the constructed lightweight full information fusion and dual-guide network Lighter for training, obtaining the trained Lighter and saving it; inputting the remote sensing image to be processed, predicting it through the trained Lighter, and obtaining a dual-phase change prediction map. The present invention significantly improves the accuracy and performance of remote sensing image change detection through an extremely lightweight design.

Description

Technical Field

[0001] The present invention belongs to the technical field of remote sensing image processing, and more specifically, relates to a lightweight full information fusion and dual-guide remote sensing image change detection method and device. Background Art

[0002] Change detection of remote sensing images is of great significance for urban construction, disaster assessment, and environmental protection. Current change detection methods for remote sensing images are often designed to be overly complex, resulting in high model parameters (Params.) and computational complexity (FLOPs), which hinders the practical application of change detection. Summary of the invention

[0003] In view of the above defects or improvement needs of the prior art, the present invention proposes a lightweight full information fusion and dual-guide remote sensing image change detection method and device. By embedding the innovatively designed full information fusion module and dual-guide difference capture module into the lightweight Lighter, pseudo-change interference is eliminated, and multi-scale and multi-shape real change areas are identified, and Lighter is successfully applied to the field of remote sensing image analysis.

[0004] To achieve the above object, according to one aspect of the present invention, a lightweight full information fusion and dual-guide remote sensing image change detection method is provided, comprising:

[0005] The obtained remote sensing image data set is input into the constructed lightweight full information fusion and dual-guide network (Lighter) for training, and the trained Lighter is obtained and saved;

[0006] Input the remote sensing image to be processed, perform prediction through the trained Lighter, and obtain the dual-phase change prediction map.

[0007] In some optional embodiments, the remote sensing image dataset obtained includes:

[0008] Pairs of dual-temporal images and their corresponding change masks are used to construct the remote sensing image dataset. To ensure the consistency and comparability of the dataset, each pair of dual-temporal images and their corresponding change masks must correspond exactly and follow a unified file naming convention. This correspondence ensures that in each pair of images, whether it is a changing area or a stable area, they show consistent geometric features in the corresponding change mask, including size, shape, and spatial position. In addition, the change mask uses a binary representation method, in which the pixel values ​​of the changing area are encoded as 1 or 255, while the pixel values ​​of the unchanged area are set to 0.

[0009] In some optional implementations, training is performed in a lightweight full information fusion and dual-guide network (Lighter), including:

[0010] Build Lighter, input the obtained dataset into Lighter, set the batch size and training rounds for each iteration to train Lighter.

[0011] In some optional implementation schemes, the construction of Lighter includes: a weight-sharing Siamese backbone, a full information fusion module, a dual-guide difference capture module, a decoder and a prediction head.

[0012] In some optional embodiments, the weight-sharing Siamese backbone includes: a parameter-sharing pre-trained MobileNetv2 constructed feature extractor, a pair of dual-phase remote sensing images is input into the structure to generate dual-time features , , and then , Enter the full information fusion module.

[0013] In some optional implementation schemes, the full information fusion module includes: dual-time information fusion, intra-layer spatial aggregation and inter-layer spatial interaction.

[0014] In some optional implementations, the dual-time information fusion includes: splicing and After convolution, we get For advanced dual time features , Time injection, firstly use the absolute value difference to extract the rich semantic difference information . Then The input is then used as the coordinate attention (CA) [Spatiotemporal Enhancement and Interlevel Fusion Network for Remote Sensing Images Change Detection, IEEE Transactions on Geoscience and Remote Sensing (Volume: 62)] and then , Multiply to get the full information feature of the first scale , Finally, , Similarly, splicing and After convolution, we get advanced It is obtained from the following formulas (1)-(3):

[0015] (1)

[0016] (2)

[0017] (3)

[0018] In some optional embodiments, the intra-layer spatial aggregation includes: setting different hierarchical spatial aggregation rates in different attention heads This method only works on small-scale features in the spatial interaction between layers. (Right now and ) to generate K and V of different particle sizes. It is obtained by the following formula (4):

[0019] (4)

[0020] In formula (4) and Represents keys and values ​​of different granularities, that is, the keys and values ​​in the jth attention head. Indicates that the aggregation rate is set to The intra-layer spatial aggregation, Indicates the features that LSA needs to process. and represents the linear projection parameters of the j-th attention head.

[0021] In some optional implementation schemes, the inter-layer spatial interaction includes: the full information features of the other two scales are obtained by the step-by-step inter-layer spatial interaction according to the following formulas (5)-(7):

[0022] (5)

[0023] (6)

[0024] (7)

[0025] In the formula Represented by The large-scale features of the layer are used as the dual-time query Q and the The multi-granularity key obtained by the small-scale features (S) of the layer, and Represents the splicing feature, value , The inter-layer interaction is obtained by the following formula (8):

[0026] (8)

[0027] In formula (8) represents the features that have been interacted with, represents the query in the jth attention head, represents the transpose of the key in the jth attention head, and d represents the dimension. After concatenation and linear projection, the full information feature is obtained. The full information feature is adjusted by learnable parameters and added to the large-scale feature , Generated , .

[0028] In some optional embodiments, the dual-guide difference capture module includes: taking two different inputs as guides: the first guide is a fusion guide obtained by concatenating full information features of the same scale and integrating rich information of different time and space positions through convolution operation, batch normalization and activation function; The other is to calculate the absolute difference of full information features of the same scale to obtain the difference guide , the difference calculation can provide a direct and contrasting feature representation, making the model more sensitive and accurate in capturing changes. After that, lightweight key information extraction is performed. Finally, after the dual-guide difference capture module, the accurate differences of three scales are captured. It is obtained from the following formulas (9)-(11):

[0029] (9)

[0030] (10)

[0031] (11)

[0032] In the above formula (9)-formula (11), , , Difference Wizard The generated query, transpose of keys, values, , , Respectively, the fusion wizard Generated query, transpose of keys, values. represents the Softmax function, represents the learnable parameters, represents the concatenation of channel dimensions, express The convolutional layer.

[0033] In some optional implementations, the decoder and prediction head include: the decoder uses the decoder in [Change Detection on Remote Sensing Images Using Dual-Branch Multilevel Intertemporal Network], and the prediction head uses the prediction head in [Remote Sensing Image Change Detection with Transformers], and there is no change, so the specific structures of the two are not repeated. Processing results in two predictions , It should be noted that Exists only during training as deep supervision, helping to generate more accurate change maps.

[0034] In some optional implementation schemes, the step of obtaining and saving a trained Lighter includes: calculating losses and updating network parameters through step-by-step iterations, completing network training, and saving a network parameter file.

[0035] In some optional implementation schemes, the prediction is performed through the trained Lighter to obtain a dual-phase change prediction map, and the dual-time remote sensing image pair that needs to be detected for change is processed to a resolution size that the model can accept, and then input into the trained Lighter for prediction to obtain a dual-phase change prediction map, thereby completing the method.

[0036] The second aspect of the present invention relates to a lightweight full information fusion and dual-guide remote sensing image change detection device, including a memory and one or more processors, wherein the memory stores executable code, and when the one or more processors execute the executable code, they are used to implement a lightweight full information fusion and dual-guide remote sensing image change detection method of the present invention.

[0037] A third aspect of the present invention relates to a computer-readable storage medium having a program stored thereon, which, when executed by a processor, implements a lightweight full information fusion and dual-guide remote sensing image change detection method of the present invention.

[0038] The working principle of the present invention is:

[0039] The core working principle of the present invention is to realize remote sensing image change detection by constructing a lightweight full information fusion and dual-guide network (Lighter). The Lighter network consists of a weight-sharing Siamese backbone, a full information fusion module, a dual-guide difference capture module, a decoder and a prediction head. First, the dual-phase remote sensing image is input into the Siamese backbone, the features are extracted and input into the full information fusion module, which injects the dynamic temporal properties, spatial relationships, details and semantic information of the scene, and enhances the multi-dimensional information of the features. Then, the dual-guide difference capture module uses two guides to reduce the interference of pseudo-changes from different perspectives to identify the actual change area. Finally, the decoder and the prediction head decode and predict the extracted features, generate a dual-phase change prediction map, and complete the change detection of the remote sensing image.

[0040] The advantages of the present invention are:

[0041] This paper proposes a lightweight full information fusion and dual-guide network (Lighter), which significantly improves the accuracy and performance of remote sensing image change detection through an extremely lightweight design. By designing a full information fusion module, we focus on injecting multi-view information in the feature interaction stage, so as to use rich semantics as clues to infer changes of various shapes. In addition, a dual-guide difference capture module is proposed in the differential acquisition stage, which uses the unique information of each guide to guide each other to reduce the interference of pseudo-changes. Finally, our lightweight architecture only requires 1.10M parameters and 2.01G FLOPs to achieve state-of-the-art performance. In summary, this paper not only theoretically confirms the significant advantages of deep learning technology in the field of remote sensing image analysis, but also demonstrates its excellent performance and wide applicability in practical applications. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 Flow chart of the method of the present invention.

[0043] Figure 2 This is a diagram of the lightweight full information fusion and dual-guide network model of the present invention. Figure 2 Part (a) is a schematic diagram of the overall architecture of the lightweight full information fusion and dual-guide network of the present invention. Figure 2 Part (b) of FIG. 2 is a schematic diagram of the full information fusion module of the present invention; Part (c) of FIG. 2 is a schematic diagram of the two-way guided search and capture module of the present invention. Figure 2 In part (a) of , represent the real dual-time remote sensing images input into the lightweight full information fusion and dual-guide network, respectively. Figure 2 In part (a) of , Respectively represent the dual-time features generated by the encoder, Figure 2In part (a) of , They represent the dual-time full-information features generated by the full-information module, Figure 2 In part (a) of Indicates the precise difference produced by the two-way search and capture module. Figure 2 In part (a) of , Represents the two predictions generated by the decoder and the prediction head. Figure 2 In part (b) of , , , Respectively , , , exist Features, Figure 2 Part (b) represents the features after concatenation and convolution. Figure 2 Part (b) represents the full information feature, Figure 2 Part (b) , represents the learnable parameters. , They represent the difference wizard and fusion wizard respectively. represents learnable parameters.

[0044] Figure 3 This is the effect diagram of the change detection of the lightweight full information fusion and dual-guide network of the present invention. The Image1 and Image2 columns are real dual-time remote sensing images, the ICIFNet, BIT, DMINet, SEIFNet, and Lighter columns are the prediction results of different networks, GT is the real change mask, and the last two columns are heat maps.

[0045] Figure 4 The figure is a graph of the generalization performance of the lightweight full information fusion and dual-guide network for change detection on the LEVIR dataset. Different colors represent different percentages of test data, and BIT, ICIFNet, DMINet, SEIFNet, and Lighter are listed as the generalization performance of change detection of different networks. The F1-Score (%) on the ordinate in Figure 4 represents the F1 score, which is a key indicator in statistics for evaluating the performance of binary classification models.

[0046] Figure 5This is a graph of the generalization performance of the lightweight full information fusion and dual-guide network for change detection on the GZ dataset. Different colors represent different percentages of test data, and BIT, ICIFNet, DMINet, SEIFNet, and Lighter are listed as the generalization performance of change detection of different networks. Figure 5 The F1-Score (%) on the vertical axis represents the F1 score, which is a key indicator in statistics for evaluating the performance of binary classification models.

[0047] Figure 6 The figure is a graph of the generalization performance of the lightweight full information fusion and dual-guide network for change detection on the WHU dataset. Different colors represent different percentages of test data, and BIT, ICIFNet, DMINet, SEIFNet, and Lighter are listed as the generalization performance of change detection of different networks. The F1-Score (%) on the ordinate in Figure 6 represents the F1 score, which is a key indicator in statistics for evaluating the performance of binary classification models.

[0048] Figure 7 It is a schematic diagram of the structure of the device of the present invention. DETAILED DESCRIPTION

[0049] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not intended to limit the scope of protection of the present application.

[0050] like Figure 1 This embodiment provides a lightweight full information fusion and dual-guide remote sensing image change detection method, including:

[0051] Step S1, obtain remote sensing image dataset; adjust the dataset to the input image size .

[0052] Step S2: construct a lightweight full information fusion and dual-guide network (Lighter), such as Figure 2 Lighter consists of a weight-shared Siamese backbone, a full information fusion module, a dual-guided difference capture module, a decoder, and a prediction head.

[0053] The weight-sharing Siamese backbone is a parameter-sharing pre-trained MobileNetv2 feature extractor, into which a pair of dual-temporal remote sensing images is input to generate dual-temporal features. , , and then , Enter the full information fusion module.

[0054] The full information fusion module includes: dual-time information fusion, intra-layer spatial aggregation and inter-layer spatial interaction. Dual-time information fusion combines and After convolution, we get For advanced dual time features , Time injection, firstly use the absolute value difference to extract the rich semantic difference information . Then The input is then used as the coordinate attention (CA) [Spatiotemporal Enhancement and Interlevel Fusion Network for Remote Sensing Images Change Detection, IEEE Transactions on Geoscience and Remote Sensing (Volume: 62)] and then combined with , Multiply to get the full information feature of the first scale , Finally, , Similarly, splicing and After convolution, the advanced :

[0055] (1)

[0056] (2)

[0057] (3)

[0058] Intra-layer spatial aggregation sets different hierarchical spatial aggregation rates in different attention heads This method only works on small-scale features in the spatial interaction between layers. (Right now and ) to generate K and V of different granularities by the following formula (4):

[0059] (4)

[0060] Inter-layer spatial interaction The full information features of the other two scales are obtained through the inter-layer spatial interaction step by step, which are given by the following formulas (5)-(7):

[0061] (5)

[0062] (6)

[0063] (7)

[0064] In formula (7) Represented by The large-scale features (L) of the layer are used as the dual-time query Q and the The multi-granularity key and value obtained by the small-scale features (S) of the layer , The inter-layer interaction is obtained by the following formula (8):

[0065] (8)

[0066] The dual-guide difference capture module uses two different inputs as guides: the first guide is a fusion guide obtained by concatenating full-information features of the same scale and integrating rich information at different time and spatial locations through convolution operations, batch normalization and activation functions. The other is to calculate the absolute difference of full information features of the same scale to obtain the difference guide , the difference calculation can provide a direct and contrasting feature representation, making the model more sensitive and accurate in capturing changes. After that, lightweight key information extraction is performed. Finally, after the dual-guide difference capture module, the accurate differences of three scales are captured. It is obtained from the following formulas (9)-(11):

[0067] (9)

[0068] (10)

[0069] (11)

[0070] The decoder uses the decoder in [Change Detection on Remote Sensing Images Using Dual-Branch Multilevel Intertemporal Network], and the prediction head uses the prediction head in [RemoteSensing Image Change Detection with Transformers], and there is no change, so the specific structures of the two are not repeated. Processing results in two predictions , It should be noted that Exists only during training as deep supervision, helping to generate more accurate change maps.

[0071] Step S3: Input the obtained remote sensing image data set into the constructed Lighter for training, calculate the loss for each batch and update the network parameters until all batches of training data are trained, and finally obtain the trained weights, and all updated parameters will be saved in the weight file. Obtain the trained Lighter and save it.

[0072] Step S4: Input the remote sensing image to be processed, and predict it through the trained Lighter to obtain the dual-phase change prediction map. The comparison of Lighter with other current optimal models in terms of parameters (Params.), computational complexity (FLOPs), and performance indicators (F1 / IOU / OA) is shown in Tables 1 to 3 below. The indicators in bold represent the optimal model:

[0073] Table 1 Quantitative comparison on the GZ dataset

[0074]

[0075] Table 2 Quantitative comparison on the LEVIR dataset

[0076]

[0077] Table 3 Quantitative comparison on the SYSU dataset

[0078]

[0079] Example 2

[0080] The present embodiment relates to a lightweight full information fusion and dual-guide remote sensing image change detection device, including a memory and one or more processors, wherein the memory stores executable code, and when the one or more processors execute the executable code, they are used to implement a lightweight full information fusion and dual-guide remote sensing image change detection method of embodiment 1.

[0081] like Figure 7 At the hardware level, the device includes a processor, an internal bus, a network interface, a memory, and a non-volatile memory, and may also include other hardware required for the business. The processor reads the corresponding computer program from the non-volatile memory into the memory and then runs it to achieve the above Figure 1 Of course, in addition to the software implementation, the present invention does not exclude other implementations, such as logic devices or a combination of software and hardware, etc., that is, the execution subject of the following processing flow is not limited to each logic unit, but can also be hardware or logic devices.

[0082] For the improvement of a technology, it can be clearly distinguished whether it is a hardware improvement (for example, improvement of the circuit structure of diodes, transistors, switches, etc.) or a software improvement (improvement of the method flow). However, with the development of technology, many improvements of the method flow today can be regarded as direct improvements of the hardware circuit structure. Designers almost always obtain the corresponding hardware circuit structure by programming the improved method flow into the hardware circuit. Therefore, it cannot be said that the improvement of a method flow cannot be implemented with a hardware entity module. For example, a programmable logic device (PLD) (such as a field programmable gate array (FPGA)) is such an integrated circuit whose logical function is determined by the user's programming of the device. Designers can "integrate" a digital system on a PLD by programming themselves, without having to ask chip manufacturers to design and make dedicated integrated circuit chips. Moreover, nowadays, instead of manually making integrated circuit chips, this kind of programming is mostly implemented by "logic compiler" software, which is similar to the software compiler used when developing and writing programs, and the original code before compilation must also be written in a specific programming language, which is called hardware description language (HDL). There is not only one kind of HDL, but many kinds, such as ABEL (Advanced Boolean Expression Language), AHDL (Altera Hardware Description Language), Confluence, CUPL (Cornell University Programming Language), HDCal, JHDL (Java Hardware Description Language), Lava, Lola, MyHDL, PALASM, RHDL (Ruby Hardware Description Language), etc. The most commonly used ones are VHDL (Very-High-Speed ​​Integrated Circuit Hardware Description Language) and Verilog. Those skilled in the art should also know that it is only necessary to program the method flow slightly in the above-mentioned hardware description languages ​​and program it into the integrated circuit, and then it is easy to obtain the hardware circuit that implements the logic method flow.

[0083] The controller may be implemented in any suitable manner, for example, the controller may take the form of a microprocessor or processor and a computer-readable medium storing a computer-readable program code (such as software or firmware) executable by the (micro)processor, a logic gate, a switch, an application-specific integrated circuit (ASIC), a programmable logic controller, and an embedded microcontroller. Examples of controllers include, but are not limited to, the following microcontrollers: ARC 625D, Atmel AT91SAM, Microchip PIC18F26K20, and Silicone Labs C8051F320. The memory controller may also be implemented as part of the control logic of the memory. It is also known to those skilled in the art that, in addition to implementing the controller in a purely computer-readable program code manner, the controller may be implemented in the form of a logic gate, a switch, an application-specific integrated circuit, a programmable logic controller, and an embedded microcontroller by logically programming the method steps. Therefore, such a controller may be considered as a hardware component, and the devices for implementing various functions included therein may also be considered as structures within the hardware component. Or even, the devices for implementing various functions may be considered as both software modules for implementing the method and structures within the hardware component.

[0084] The systems, devices, modules or units described in the above embodiments may be implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer. Specifically, the computer may be, for example, a personal computer, a laptop computer, a cellular phone, a camera phone, a smart phone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or a combination of any of these devices.

[0085] For the convenience of description, the above device is described as being divided into various units according to their functions. Of course, when implementing the present invention, the functions of each unit can be implemented in the same or multiple software and / or hardware.

[0086] Example 3

[0087] This embodiment relates to a computer-readable storage medium having a program stored thereon. When the program is executed by a processor, a lightweight full information fusion and dual-guide remote sensing image change detection method of embodiment 1 is implemented.

[0088] It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0089] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0090] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0091] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0092] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0093] The memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0094] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0095] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0096] It should be understood by those skilled in the art that the embodiments of the present invention may be provided as methods, systems or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment or an embodiment combining software and hardware aspects. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0097] The present invention may be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. The present invention may also be practiced in distributed computing environments where tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules may be located in local and remote computer storage media, including storage devices.

[0098] Each embodiment of the present invention is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0099] The above description is only an embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent substitution, improvement, etc. made within the spirit and principle of the present invention should be included in the scope of the claims of the present invention.

Claims

1. A lightweight full information fusion and dual-guide remote sensing image change detection method, characterized in that: include: The obtained remote sensing image data set is input into the constructed lightweight full information fusion and dual-guide network Lighter for training, and the trained Lighter is obtained and saved; The constructed Lighter includes a weight-sharing Siamese backbone, a full information fusion module, a dual-guide difference capture module, a decoder, and a prediction head; The weight-sharing Siamese backbone is a parameter-sharing pre-trained MobileNetv2 feature extractor, into which a pair of dual-temporal remote sensing images are input to generate dual-temporal features. , , and then , Enter the full information fusion module; The full information fusion module includes dual-time information fusion, intra-layer spatial aggregation and inter-layer spatial interaction; The dual-time information fusion includes: splicing and combining the dual-time features of the first two scales After convolution, we get ; For advanced dual time features , Time injection, firstly use the absolute value difference to extract the rich semantic difference information ; Afterwards Enter the coordinates and note that after CA , Multiply to get the full information feature of the first scale , ; Finally, , Also splice and After convolution, we get advanced ; The intra-layer spatial aggregation includes: setting different hierarchical spatial aggregation rates in different attention heads ; only acts on small-scale features in the spatial interaction between layers Thus, K and V of different granularity are generated; The inter-layer spatial interaction includes: obtaining full information features of the other two scales through inter-layer spatial interaction step by step; The dual-guide difference capture module includes: taking two different inputs as guides: the first guide is a fusion guide obtained by splicing full information features of the same scale and integrating rich information of different time and space positions through convolution operation, batch normalization and activation function. ; The other is to calculate the absolute difference of the full information features of the same scale to obtain the difference guide , the difference calculation can provide a direct and contrasting feature representation, making the model more sensitive and accurate in capturing changes; then, lightweight key information extraction is performed; finally, the accurate differences of three scales are captured after the dual-guide difference capture module ; The decoder and prediction header Processing results in two predictions , ; It only exists as deep supervision during training, which helps to generate more accurate change maps; Input the remote sensing image to be processed, perform prediction through the trained Lighter, and obtain the dual-phase change prediction map.

2. The method according to claim 1, characterized in that The remote sensing image data set obtained includes: The remote sensing image dataset includes dual-temporal remote sensing images and corresponding change masks. It is ensured that each pair of dual-temporal remote sensing images and corresponding change masks in the remote sensing image dataset correspond one-to-one and have the same file name. The one-to-one correspondence means that the changed area or unchanged area in each pair of dual-temporal remote sensing images needs to have the same size, shape, and position in the corresponding change mask. In addition, the pixel value of the changed area in each change mask in the remote sensing image dataset is 1 or 255, and the pixel value of the unchanged area is 0. The input is trained into the constructed lightweight full information fusion and dual-guide network Lighter, including: Build Lighter, input the obtained dataset into Lighter, set the batch size and training rounds for each iteration to train Lighter.

3. The method according to claim 1, characterized in that The obtaining and saving of the trained Lighter includes: calculating the loss and updating the network parameters through step-by-step iteration, completing the network training and saving the network parameter file; The trained Lighter is used to make predictions to obtain a dual-phase change prediction map, including: processing the dual-time remote sensing image pairs that need to be detected for changes into a resolution size that can be accepted by the model, and then inputting them into the trained Lighter for prediction to obtain a dual-phase change prediction map, thereby completing the method.

4. A lightweight full information fusion and dual-guide remote sensing image change detection device, characterized in that: It comprises a memory and one or more processors, wherein the memory stores executable codes, and when the one or more processors execute the executable codes, they are used to implement a lightweight full information fusion and dual-guide remote sensing image change detection method as described in any one of claims 1 to 3.

5. A computer-readable storage medium, characterized in that: A program is stored thereon, and when the program is executed by a processor, a lightweight full information fusion and dual-guide remote sensing image change detection method as described in any one of claims 1 to 3 is implemented.

Images