A multi-source remote sensing image fusion method and system based on deep learning

Abstract

The present invention is applicable to the field of image fusion technology, and in particular relates to a multi-source remote sensing image fusion method and system based on deep learning, the method comprising: acquiring all remote sensing images from different sources; preprocessing all remote sensing images from different sources, and statistically analyzing the position distribution of each pixel; dividing feature regions according to the position distribution of the pixels, and identifying regional features; matching remote sensing images from different sources according to the regional features to complete image fusion. An embodiment of the present invention provides a multi-source remote sensing image fusion method based on deep learning, which identifies multiple groups of remote sensing images from different sources, determines the features contained therein, and then determines the relative position relationship between the features and the shape features of the features themselves, thereby quickly completing the matching between features in remote sensing images from different sources, ensuring the accuracy of feature recognition and improving the efficiency of feature matching.

Description

Technical Field

[0001] The present invention belongs to the technical field of image fusion, and in particular relates to a multi-source remote sensing image fusion method and system based on deep learning. Background Art

[0002] Deep learning is a type of machine learning, which is the essential path to achieving artificial intelligence. The concept of deep learning originates from the study of artificial neural networks. Multilayer perceptrons with multiple hidden layers are an example of a deep learning architecture. Deep learning discovers distributed feature representations of data by combining lower-level features to form more abstract higher-level representations of attribute categories or features.

[0003] In remote sensing, data fusion is a type of attribute fusion. It intelligently synthesizes multi-source remote sensing image data from the same area to produce more accurate, complete, and reliable estimates and judgments than a single source. Its advantages include robustness, improved spatial resolution and clarity of images, enhanced planar mapping accuracy, improved classification precision and reliability, enhanced interpretation and dynamic monitoring capabilities, reduced ambiguity, and effectively increased the utilization of remote sensing image data.

[0004] In the current multi-source remote sensing image fusion process, determining image features is a prerequisite for ensuring the accuracy of the fused image. When identifying features in existing technologies, the recognition speed is slow, resulting in low fusion efficiency. Summary of the Invention

[0005] The purpose of the embodiments of the present invention is to provide a multi-source remote sensing image fusion method based on deep learning, aiming to solve the problem in the prior art of slow recognition speed and low fusion efficiency when identifying features.

[0006] The embodiment of the present invention is implemented as follows: a multi-source remote sensing image fusion method based on deep learning, the method comprising:

[0007] Acquire remote sensing images from all different sources;

[0008] Preprocess all remote sensing images from different sources and count the position distribution of each pixel;

[0009] Divide the feature area according to the position distribution of pixel points and identify the regional features;

[0010] Remote sensing images from different sources are matched according to regional features to complete image fusion.

[0011] Preferably, the step of pre-processing all remote sensing images from different sources and calculating the position distribution of each pixel point specifically includes:

[0012] Perform grayscale processing on remote sensing images from different sources to obtain grayscale images;

[0013] Establish a coordinate system and determine the coordinates of each pixel based on the pixel position;

[0014] Determine the grayscale value corresponding to each pixel.

[0015] Preferably, the step of dividing the feature regions according to the position distribution of the pixels and identifying the regional features specifically includes:

[0016] Select pixels one by one and use them as reference pixels, and calculate the difference between the grayscale value of each other pixel and the grayscale value of the reference pixel;

[0017] Divide a connected region composed of pixel points whose difference is less than a first preset value into a feature region, wherein the discreteness of the pixel points in the feature region is lower than a second preset value;

[0018] Determine the relative positions between each feature area, determine the outline of the feature area, and obtain the regional features.

[0019] Preferably, the step of matching remote sensing images from different sources according to regional features to complete image fusion specifically includes:

[0020] Classify regional features according to their locations in corresponding remote sensing images from different sources;

[0021] According to the correspondence between regional features of the same type, the positions of remote sensing images from different sources are adjusted to complete image fusion.

[0022] Preferably, before image fusion, remote sensing images from different sources are cropped to the same size.

[0023] Preferably, there are at least two overlapping regional features between remote sensing images from different sources.

[0024] Another object of an embodiment of the present invention is to provide a multi-source remote sensing image fusion system based on deep learning, the system comprising:

[0025] Image acquisition module, used to acquire remote sensing images from all different sources;

[0026] The pixel statistics module is used to pre-process all remote sensing images from different sources and count the position distribution of each pixel;

[0027] The feature recognition module is used to divide the feature area according to the position distribution of the pixel points and identify the regional features;

[0028] The image fusion module is used to match remote sensing images from different sources according to regional features and complete image fusion.

[0029] Preferably, the pixel statistics module includes:

[0030] An image preprocessing unit is used to perform grayscale processing on remote sensing images from different sources to obtain grayscale images;

[0031] A coordinate recognition unit is used to establish a coordinate system and determine the coordinates of each pixel based on the pixel position;

[0032] The grayscale value determining unit is used to determine the grayscale value corresponding to each pixel.

[0033] Preferably, the feature recognition module includes:

[0034] A grayscale value calculation unit is used to select pixels one by one and use them as reference pixels to calculate the difference between the grayscale value of each other pixel and the grayscale value of the reference pixel;

[0035] A region feature division unit, configured to divide a connected region consisting of pixel points whose difference value is less than a first preset value into a feature region, wherein the discreteness of the pixel points in the feature region is less than a second preset value;

[0036] The feature generation unit is used to determine the relative positions between the feature regions, determine the contours of the feature regions, and obtain the region features.

[0037] Preferably, the image fusion module includes:

[0038] A feature classification unit is used to classify regional features according to their positions in corresponding remote sensing images from different sources;

[0039] The positioning fusion unit is used to adjust the positions of remote sensing images from different sources according to the corresponding relationship between regional features of the same type to complete image fusion.

[0040] An embodiment of the present invention provides a multi-source remote sensing image fusion method based on deep learning. By identifying multiple groups of remote sensing images from different sources, the features contained in each of them are determined respectively, and then the relative position relationship between each feature and the shape characteristics of the features themselves are determined, thereby quickly completing the matching between features in remote sensing images from different sources, which not only ensures the accuracy of feature recognition but also improves the efficiency of feature matching. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 A flowchart of a multi-source remote sensing image fusion method based on deep learning provided by an embodiment of the present invention;

[0042] Figure 2 A flowchart of the steps of preprocessing all remote sensing images from different sources and counting the position distribution of each pixel provided by an embodiment of the present invention;

[0043] Figure 3A flowchart of the steps of dividing feature regions according to the position distribution of pixel points and identifying regional features provided in an embodiment of the present invention;

[0044] Figure 4 A flowchart of the steps of matching remote sensing images from different sources based on regional features and completing image fusion provided in an embodiment of the present invention;

[0045] Figure 5 An architecture diagram of a multi-source remote sensing image fusion system based on deep learning provided by an embodiment of the present invention;

[0046] Figure 6 An architectural diagram of a pixel statistics module provided in an embodiment of the present invention;

[0047] Figure 7 An architectural diagram of a feature recognition module provided by an embodiment of the present invention;

[0048] Figure 8 This is an architectural diagram of an image fusion module provided in an embodiment of the present invention. DETAILED DESCRIPTION

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

[0050] It is understood that the terms "first," "second," etc., used herein may be used to describe various elements, but unless otherwise specified, these elements are not limited by these terms. These terms are only used to distinguish a first element from another element. For example, a first xx script may be referred to as a second xx script, and similarly, a second xx script may be referred to as a first xx script without departing from the scope of this application.

[0051] In remote sensing, data fusion is a type of attribute fusion. It intelligently synthesizes multi-source remote sensing image data from the same region to produce estimates and judgments that are more accurate, complete, and reliable than those from a single source. Its advantages include robustness, improved spatial resolution and clarity of images, enhanced planar mapping accuracy, improved classification accuracy and reliability, enhanced interpretation and dynamic monitoring capabilities, reduced ambiguity, and effectively increased the utilization of remote sensing image data. In current multi-source remote sensing image fusion processes, identifying image features is a prerequisite for ensuring the accuracy of the fused image. Existing technologies for feature recognition are slow, resulting in low fusion efficiency.

[0052] In the present invention, by identifying multiple groups of remote sensing images from different sources, the features contained in each of them are determined respectively, and then the relative position relationship between each feature and the shape characteristics of the features themselves are determined, so as to quickly complete the matching between the features in remote sensing images from different sources, which not only ensures the accuracy of feature recognition but also improves the efficiency of feature matching.

[0053] like Figure 1 FIG. 1 is a flowchart of a multi-source remote sensing image fusion method based on deep learning provided by an embodiment of the present invention, wherein the method includes:

[0054] S100, acquiring all remote sensing images from different sources.

[0055] In this step, remote sensing images from all different sources are acquired. The remote sensing images can be multi-temporal images, multi-spectral images, multi-sensor images, and multi-platform images of the same area. After the above remote sensing images are fused, more reliable remote sensing images will be obtained.

[0056] S200, pre-processing all remote sensing images from different sources, and counting the position distribution of each pixel point.

[0057] In this step, all remote sensing images from different sources are preprocessed. The preprocessing includes at least grayscale processing, which converts them into grayscale images, thereby reducing the amount of data processing. Each pixel is then analyzed to determine the grayscale value of each pixel. By constructing a coordinate system, the coordinates of each pixel are determined, and the position distribution of each pixel can be obtained.

[0058] S300, dividing feature regions according to the position distribution of pixel points, and identifying and obtaining regional features.

[0059] In this step, the feature area is divided according to the position distribution of the pixel points, and the pixels are divided according to the grayscale value of each pixel point. The grayscale value difference is less than the preset value and the adjacent pixels together form a connected feature area. For example, if there is a series of buildings in the remote sensing image, then the remote sensing image has a positive area with the same pixels. By analyzing the pixel points, the corresponding feature area can be determined, and the image corresponding to the building can be determined based on the feature area, which is the regional feature.

[0060] S400: Match remote sensing images from different sources according to regional features to complete image fusion.

[0061] In this step, remote sensing images from different sources are matched according to regional features. Since there are multiple remote sensing images from different sources, each remote sensing image has corresponding regional features. For example, for the same building, there will be corresponding regional features in different remote sensing images. Then, the positional relationship between the remote sensing images is adjusted according to the positional relationship between the regional features in different remote sensing images, and then fusion processing is performed to obtain a fused image.

[0062] like Figure 2 As shown, as a preferred embodiment of the present invention, the step of preprocessing all remote sensing images from different sources and counting the position distribution of each pixel point specifically includes:

[0063] S201, performing grayscale processing on remote sensing images from different sources to obtain grayscale images.

[0064] In this step, remote sensing images from different sources are gray-scale processed, and the color of each pixel is reflected by the gray-scale value. The gray-scale processed images greatly reduce the amount of data to be processed and improve the data processing speed.

[0065] S202: Establish a coordinate system and determine the coordinates of each pixel based on the pixel position.

[0066] S203: Determine the grayscale value corresponding to each pixel.

[0067] In this step, a coordinate system is established, and a coordinate system is constructed with any pixel as the origin to determine the coordinates of each pixel. The horizontal and vertical axes of the coordinates are both integers. For a remote sensing image of 1000 pixels * 1000 pixels, its width is 1000 pixels and its height is also 1000 pixels. The coordinates of the pixels in the first row and first column are (0,0), and the coordinates of the pixels in the last row and last column are (1000,1000). At this time, each pixel point has corresponding coordinates, and the grayscale value of each pixel is determined.

[0068] like Figure 3 As shown, as a preferred embodiment of the present invention, the step of dividing the feature area according to the position distribution of the pixel points and identifying the regional features specifically includes:

[0069] S301 , selecting pixels one by one and using them as reference pixels, and calculating the difference between the grayscale value of each other pixel and the grayscale value of the reference pixel.

[0070] In this step, pixels are selected one by one, specifically according to pixel coordinates, as reference pixels. The reference pixels traverse all pixels in a set of remote sensing images. After selecting any pixel, the difference in grayscale value between the pixel and all other pixels is calculated.

[0071] S302: Divide a connected region composed of pixel points whose difference value is less than a first preset value into a feature region, wherein the discreteness of the pixel points in the feature region is lower than a second preset value.

[0072] In this step, the connected area composed of pixel points whose difference is less than the first preset value is divided into a feature area, and all pixel points whose difference is less than the first preset value are counted. The area is divided according to the connectivity relationship between the pixel points to obtain multiple feature areas, and the discreteness of the pixel points in the feature area is calculated and characterized by the standard deviation, that is, the standard deviation of the pixel points in the feature area is calculated. If the standard deviation is lower than the second preset value, it can be used as a feature area. If the number of pixels in the feature area is lower than the preset threshold, it will be discarded.

[0073] S303: Determine the relative positions between the characteristic regions, determine the contours of the characteristic regions, and obtain regional features.

[0074] In this step, the relative position between each feature area is determined. Specifically, each feature area is labeled. For example, there are three groups of remote sensing images, namely A, B and C. The three groups of remote sensing images contain three groups of feature areas, which are labeled A1, A2 and A3, B1, B2 and B3, C1, C2 and C3 respectively. The center point of each feature area is calculated, and the center points corresponding to all feature areas in the same remote sensing image are connected with line segments to obtain a feature area connection diagram. The feature area connection diagrams corresponding to different remote sensing images are made to overlap by rotating and / or scaling the feature area connection diagram. At this time, different feature areas overlap with each other, and the overlapping feature areas correspond to each other. The outline of the feature area is determined to obtain the regional features.

[0075] like Figure 4 As shown, as a preferred embodiment of the present invention, the step of matching remote sensing images from different sources according to regional features to complete image fusion specifically includes:

[0076] S401 , classifying regional features according to their positions in corresponding remote sensing images from different sources.

[0077] S402, adjusting the positions of remote sensing images from different sources based on the corresponding relationship between regional features of the same type to complete image fusion.

[0078] In this step, the regional features are classified. Specifically, the overlapping regional features on different remote sensing images are classified as the same category. By scaling, the regional features of the same category on different remote sensing images are made to overlap with each other, so that multiple remote sensing images completely overlap, and the remote sensing images from different sources are cropped to the same size. At this time, the fusion processing can be directly performed to obtain a fused image; there are at least two overlapping regional features between the remote sensing images from different sources.

[0079] like Figure 5 As shown in FIG, a multi-source remote sensing image fusion system based on deep learning is provided in an embodiment of the present invention, and the system includes:

[0080] The image acquisition module 100 is used to acquire remote sensing images from all different sources.

[0081] In this system, the image acquisition module 100 acquires remote sensing images from all different sources. The remote sensing images can be multi-temporal images, multi-spectral images, multi-sensor images, and multi-platform images of the same area. After the above remote sensing images are fused, more reliable remote sensing images will be obtained.

[0082] The pixel statistics module 200 is used to pre-process all remote sensing images from different sources and calculate the position distribution of each pixel point.

[0083] In this system, the pixel statistics module 200 preprocesses all remote sensing images from different sources. The preprocessing includes at least grayscale processing, which converts them into grayscale images, thereby reducing the amount of data processing. Each pixel is then analyzed to determine the grayscale value of each pixel point. By constructing a coordinate system, the coordinates of each pixel are determined, and the position distribution of each pixel can be obtained.

[0084] The feature recognition module 300 is used to divide feature regions according to the position distribution of pixel points and identify regional features.

[0085] In this system, the feature recognition module 300 divides the feature area according to the position distribution of the pixel points, and divides the pixels according to the grayscale value of each pixel point. The grayscale value difference is less than the preset value and the adjacent pixels together form a connected feature area. For example, if there is a series of buildings in the remote sensing image, then the remote sensing image has a positive area with the same pixels. By analyzing the pixel points, the corresponding feature area can be determined, and the image corresponding to the building can be determined based on the feature area, which is the regional feature.

[0086] The image fusion module 400 is used to match remote sensing images from different sources according to regional features to complete image fusion.

[0087] In this system, the image fusion module 400 matches remote sensing images from different sources based on regional features. Since there are multiple remote sensing images from different sources, each remote sensing image has corresponding regional features. For example, for the same building, there will be corresponding regional features in different remote sensing images. Then, the positional relationship between the remote sensing images is adjusted according to the positional relationship between the regional features in different remote sensing images, and then fusion processing is performed to obtain a fused image.

[0088] like Figure 6As shown, as a preferred embodiment of the present invention, the pixel statistics module 200 includes:

[0089] The image pre-processing unit 201 is used to perform grayscale processing on remote sensing images from different sources to obtain grayscale images.

[0090] In this module, the image preprocessing unit 201 performs grayscale processing on remote sensing images from different sources, and then reflects the color of each pixel through the grayscale value. The grayscale processed image greatly reduces the data processing amount and improves the data processing speed.

[0091] The coordinate identification unit 202 is used to establish a coordinate system and determine the coordinates of each pixel according to the pixel position.

[0092] The grayscale value determining unit 203 is configured to determine the grayscale value corresponding to each pixel.

[0093] In this module, a coordinate system is established, and a coordinate system is constructed with any pixel as the origin to determine the coordinates of each pixel. The horizontal and vertical axes of the coordinates are both integers. For a remote sensing image of 1000 pixels * 1000 pixels, its width is 1000 pixels and its height is also 1000 pixels. The coordinates of the pixels in the first row and first column are (0,0), and the coordinates of the pixels in the last row and last column are (1000,1000). At this time, each pixel point has corresponding coordinates, and the grayscale value of each pixel is determined.

[0094] like Figure 7 As shown, as a preferred embodiment of the present invention, the feature recognition module 300 includes:

[0095] The grayscale value calculation unit 301 is used to select pixels one by one and use them as reference pixels to calculate the difference between the grayscale values ​​of other pixels and the grayscale value of the reference pixel.

[0096] In this module, the grayscale value calculation unit 301 selects pixel points one by one, specifically according to the pixel coordinates, as reference pixel points. The reference pixel points traverse all pixels in a set of remote sensing images. After selecting any pixel point, the difference in grayscale value between the pixel point and all other pixel points is calculated.

[0097] The region feature division unit 302 is configured to divide a connected region consisting of pixel points whose difference value is less than a first preset value into a feature region, wherein the discreteness of the pixel points in the feature region is less than a second preset value.

[0098] In this module, the regional feature division unit 302 divides the connected area composed of pixel points whose difference is less than the first preset value into a feature area, counts all pixel points whose difference is less than the first preset value, and divides the area according to the connectivity relationship between the pixel points to obtain multiple feature areas. The discreteness of the pixel points in the feature area is calculated and characterized by the standard deviation, that is, the standard deviation of the pixel points in the feature area is calculated. If the standard deviation is lower than the second preset value, it can be used as a feature area. If the number of pixels in the feature area is lower than the preset threshold, it will be discarded.

[0099] The feature generation unit 303 is used to determine the relative positions between the feature regions, determine the contours of the feature regions, and obtain region features.

[0100] In this module, the feature generation unit 303 determines the relative position between each feature area. Specifically, each feature area is labeled. For example, there are three groups of remote sensing images, namely A, B and C. The three groups of remote sensing images respectively contain three groups of feature areas, which are labeled A1, A2 and A3, B1, B2 and B3, C1, C2 and C3 respectively. The center point of each feature area is calculated, and the center points corresponding to all feature areas in the same remote sensing image are connected with line segments to obtain a feature area connection diagram. The feature area connection diagrams corresponding to different remote sensing images are made to overlap by rotating and / or scaling the feature area connection diagram. At this time, different feature areas overlap with each other, and the overlapping feature areas correspond to each other. The outline of the feature area is determined to obtain the regional features.

[0101] like Figure 8 As shown, as a preferred embodiment of the present invention, the image fusion module 400 includes:

[0102] The feature classification unit 401 is used to classify regional features according to their positions in the corresponding remote sensing images from different sources.

[0103] The positioning fusion unit 402 is used to adjust the positions of remote sensing images from different sources based on the corresponding relationship between regional features of the same type to complete image fusion.

[0104] In this module, regional features are classified. Specifically, overlapping regional features on different remote sensing images are classified as the same category. By scaling, regional features of the same category on different remote sensing images are made to overlap with each other, so that multiple remote sensing images completely overlap, and remote sensing images from different sources are cropped to the same size. At this time, fusion processing can be performed directly to obtain a fused image; there are at least two overlapping regional features between remote sensing images from different sources.

[0105] In one embodiment, a computer device is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the following steps are performed:

[0106] Acquire remote sensing images from all different sources;

[0107] Preprocess all remote sensing images from different sources and count the position distribution of each pixel;

[0108] Divide the feature area according to the position distribution of pixel points and identify the regional features;

[0109] Remote sensing images from different sources are matched according to regional features to complete image fusion.

[0110] In one embodiment, a computer-readable storage medium is provided. The computer-readable storage medium stores a computer program. When the computer program is executed by a processor, the processor performs the following steps:

[0111] Acquire remote sensing images from all different sources;

[0112] Preprocess all remote sensing images from different sources and count the position distribution of each pixel;

[0113] Divide the feature area according to the position distribution of pixel points and identify the regional features;

[0114] Remote sensing images from different sources are matched according to regional features to complete image fusion.

[0115] It should be understood that, although the various steps in the flow chart of each embodiment of the present invention are shown in sequence according to the indication of the arrows, these steps are not necessarily performed in sequence according to the order indicated by the arrows. Unless otherwise specified herein, the execution of these steps is not strictly limited in order, and these steps can be performed in other orders. Moreover, at least a portion of the steps in each embodiment may include a plurality of sub-steps or a plurality of stages, and these sub-steps or stages are not necessarily performed at the same time, but can be performed at different times, and the execution order of these sub-steps or stages is not necessarily performed in sequence, but can be performed in turn or alternately with at least a portion of other steps or sub-steps or stages of other steps.

[0116] Those skilled in the art will appreciate that all or part of the processes in the above-described method embodiments can be implemented by instructing the relevant hardware through a computer program. The program can be stored in a non-volatile computer-readable storage medium. When executed, the program can include the processes of the above-described method embodiments. Any reference to memory, storage, database, or other media used in the various embodiments provided herein may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct RAMbus dynamic RAM (DRDRAM), and RAMbus dynamic RAM (RDRAM).

[0117] The technical features of the above-mentioned embodiments can be combined arbitrarily. In order to make the description concise, not all possible combinations of the technical features in the above-mentioned embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0118] The above-described embodiments merely illustrate several implementations of the present invention, and while their descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that a person skilled in the art would be able to make numerous variations and improvements without departing from the spirit of the present invention, all of which fall within the scope of protection of the present invention. Therefore, the scope of protection of the present invention shall be determined by the appended claims.

[0119] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A multi-source remote sensing image fusion method based on deep learning, characterized in that: The method comprises: Acquire remote sensing images from all different sources; Preprocess all remote sensing images from different sources and count the position distribution of each pixel; Divide the feature area according to the position distribution of pixel points and identify the regional features; Match remote sensing images from different sources according to regional characteristics to complete image fusion; The step of dividing the feature area according to the position distribution of the pixel points and identifying the regional features specifically includes: Select pixels one by one and use them as reference pixels, and calculate the difference between the grayscale value of each other pixel and the grayscale value of the reference pixel; Divide a connected region composed of pixel points whose difference is less than a first preset value into a feature region, wherein the discreteness of the pixel points in the feature region is lower than a second preset value; Determine the relative positions between each feature area, determine the outline of the feature area, and obtain the regional features; The step of matching remote sensing images from different sources according to regional features to complete image fusion specifically includes: Classify regional features according to their locations in corresponding remote sensing images from different sources; According to the correspondence between regional features of the same type, the positions of remote sensing images from different sources are adjusted to complete image fusion.

2. The multi-source remote sensing image fusion method based on deep learning according to claim 1, characterized in that: The step of preprocessing all remote sensing images from different sources and counting the position distribution of each pixel point specifically includes: Perform grayscale processing on remote sensing images from different sources to obtain grayscale images; Establish a coordinate system and determine the coordinates of each pixel based on the pixel position; Determine the grayscale value corresponding to each pixel.

3. The multi-source remote sensing image fusion method based on deep learning according to claim 1, characterized in that: Before image fusion, remote sensing images from different sources are cropped to the same size.

4. The multi-source remote sensing image fusion method based on deep learning according to claim 1, characterized in that: There are at least two regional features that overlap in position between remote sensing images from different sources.

5. A multi-source remote sensing image fusion system based on deep learning, characterized in that: The system comprises: Image acquisition module, used to acquire remote sensing images from all different sources; The pixel statistics module is used to pre-process all remote sensing images from different sources and count the position distribution of each pixel; The feature recognition module is used to divide the feature area according to the position distribution of the pixel points and identify the regional features; Image fusion module, used to match remote sensing images from different sources according to regional features and complete image fusion; The feature recognition module includes: A grayscale value calculation unit is used to select pixels one by one and use them as reference pixels to calculate the difference between the grayscale value of each other pixel and the grayscale value of the reference pixel; A region feature division unit, configured to divide a connected region consisting of pixel points whose difference value is less than a first preset value into a feature region, wherein the discreteness of the pixel points in the feature region is less than a second preset value; A feature generation unit is used to determine the relative positions between the feature regions, determine the contours of the feature regions, and obtain the regional features; The image fusion module includes: A feature classification unit is used to classify regional features according to their positions in corresponding remote sensing images from different sources; The positioning fusion unit is used to adjust the positions of remote sensing images from different sources according to the corresponding relationship between regional features of the same type to complete image fusion.

6. The multi-source remote sensing image fusion system based on deep learning according to claim 5, characterized in that: The pixel statistics module includes: An image preprocessing unit is used to perform grayscale processing on remote sensing images from different sources to obtain grayscale images; A coordinate recognition unit is used to establish a coordinate system and determine the coordinates of each pixel based on the pixel position; The grayscale value determining unit is used to determine the grayscale value corresponding to each pixel.

Images