Remote sensing image ship type identification method

Abstract

The invention provides a remote sensing image ship type identification method, which comprises the following steps of S1, performing preliminary pre-labeling on an acquired SAR remote sensing image, and marking a ship target position; s2, preprocessing the AIS data; s3, matching the ship target position in the SAR remote sensing image with the ship information in the AIS data; s4, after matching succeeds, detailed label information is given to the ship target in the SAR remote sensing image through AIS data, and pre-labeled raster data is formed; s5, segmenting the pre-labeled raster data into a plurality of small blocks to form labeled information; s6, enabling the training set to train a ship category recognition model, and verifying and evaluating the trained model by using the verification set; and S7, performing ship type identification on the remote sensing image by using the verified ship type identification model to obtain a detection result. The invention provides an efficient ship detection method integrating large graph labeling and AIS matching. On the premise of ensuring the detection precision, the efficiency is greatly improved.

Description

Technical Field

[0001] The present invention relates to the field of smart maritime technology, and in particular to a method for identifying ship types using remote sensing images. Background Art

[0002] In modern marine environmental monitoring and management, accurate detection and identification of ships is a crucial technical link. With the continuous increase in global maritime trade, fishing activities and military defense, how to efficiently and accurately detect the location and type of ships has become one of the key factors in improving marine monitoring capabilities. Traditionally, ship monitoring mainly relies on the Automatic Identification System (AIS), which can track ships in real time and provide detailed location information. However, AIS signals rely on ships to actively send information, so there is a problem that "dark ships" (i.e. ships that do not turn on AIS) cannot be detected. In addition, the accuracy of AIS data is limited by the working status of ship equipment and the interaction between satellites and receiving equipment. There are often errors in geographical location, and in complex marine environments, the coverage and quality of signals are also unstable.

[0003] To overcome the limitations of AIS, remote sensing technology, especially the application of multi-source data such as synthetic aperture radar (SAR), optical imaging, and infrared imaging, has gradually become an effective means of ship detection in recent years. These technologies do not rely on the signal transmission equipment carried by the ship itself, and can monitor marine activities over a wide area and at a high frequency. Among them, SAR is particularly suitable for marine monitoring under complex weather conditions due to its all-weather and all-day imaging capabilities; optical and infrared imaging can provide clearer detailed information, especially in the refined identification and classification of ships.

[0004] However, traditional remote sensing ship detection methods usually require image preprocessing (such as segmenting into small blocks) and then labeling. This process is not only time-consuming and labor-intensive, but also faces many challenges in practical applications due to the differences in the characteristics of different modal images. Summary of the invention

[0005] In view of this, the purpose of the present invention is to propose a remote sensing image ship type recognition method to solve the technical problem of low automation level of existing remote sensing ship detection methods.

[0006] The technical means adopted by the present invention are as follows:

[0007] A method for identifying ship types using remote sensing images comprises the following steps:

[0008] S1. Preliminary pre-labeling of the acquired SAR remote sensing images to mark the target location of the ship;

[0009] S2, preprocessing the AIS data so that the AIS data and the SAR remote sensing images can be compared within the same time period;

[0010] S3, by matching the ship target position in the SAR remote sensing image with the ship information in the AIS data through a shape-topology matching method, and constructing polygons for matching according to the relative positions and shapes between the ships, so that the ships in the image correspond one to one with the ships in the AIS;

[0011] S4, after successful matching, detailed label information is assigned to the ship target in the SAR remote sensing image through the AIS data to form pre-labeled raster data;

[0012] S5, dividing the pre-labeled raster data into several small blocks, each image block contains the ship target and its category label, and extracting the rectangular vertices and category of the ship from each block to form labeling information; at the same time, the SAR remote sensing image blocks and the labeling information are used as a sample set, and are divided into a training set and a validation set according to a preset ratio;

[0013] S6, using the training set to train the ship category recognition model, and using the validation set to validate and evaluate the trained model;

[0014] S7. Use the verified ship category recognition model to identify the ship type in the remote sensing image to obtain the detection result.

[0015] Further, in S2, S2 specifically includes the following steps:

[0016] Preprocess AIS data: noise removal, time alignment, and spatial interpolation;

[0017] According to the changes in the speed and position of the ship, abnormal points are detected and eliminated. The speed v of the ship between two consecutive moments is used to determine whether there are points that obviously exceed the reasonable range. A speed threshold is set. If the speed of an AIS point exceeds the threshold, it is judged as a noise point.

[0018] Obtain the specific imaging time T of SAR remote sensing images SAR , record its timestamp; filter the AIS data, extract the ship position data close to the imaging time, and select the ship position data close to the imaging time. SAR The nearest AIS point is used as the initial alignment point T AIS_initial ;

[0019] If T AIS_initial and T SAR If there is a time difference of more than 5 minutes, linear interpolation is used to estimate the ship's time at T SAR The ship’s precise position at the time of T1 and T2 is calculated by the following formula:SAR Interpolated position at time:

[0020]

[0021] Among them, P1 and P2 are the positions of the ship at T1 and T2 respectively, P SAR After time alignment, SAR The ship's position at that time.

[0022] Furthermore, in S4, the detailed tag information includes the ship type.

[0023] Furthermore, in S6, the ship category recognition model is Oriented R-CNN.

[0024] Furthermore, S3 specifically includes the following steps:

[0025] S31, shape feature calculation, including side length and internal angle;

[0026] The side length is defined as the distance between each pair of adjacent ships in the polygon and is calculated as follows:

[0027]

[0028] Among them, (x i ,y i ) and (x j ,y j ) are the center coordinates of ships i and j;

[0029] The interior angle is defined as the interior angle of each vertex in the polygon, and is calculated as follows:

[0030]

[0031] Among them, l ij and l ik are the two edges adjacent to vertex i, l jk is the length of the opposite side enclosed by these sides;

[0032] S32, calculating side length comparison and interior angle comparison;

[0033] The side length comparison is defined as comparing the length of each corresponding side of the polygon in the image and the polygon in the AIS data. The calculation formula is as follows:

[0034]

[0035] The internal angle comparison is defined as comparing the similarity of the internal angles of polygons in the image and AIS data. The calculation formula is as follows:

[0036]

[0037] Where m is the number of sides of the polygon, and are the side lengths and internal angles in the image, respectively. and are the side lengths and internal angles in the AIS data, respectively;

[0038] S33, topological feature calculation, including relative direction and relative azimuth;

[0039] The relative direction is defined as follows: for each ship pair (i, j), its direction vector relative to the image coordinate system can be defined:

[0040]

[0041] The relative azimuth is defined as the azimuth θ of ship i relative to ship j ij , the calculation formula is:

[0042]

[0043] S34, similarity calculation, calculate the shape similarity S respectively shape and topological similarity S topo , forming a comprehensive similarity S by weighting combined , the calculation formula is:

[0044] S combined =α·S shape +β·S topo

[0045] Among them, α and β are weight coefficients, S shape and S topo The calculation formula is as follows:

[0046]

[0047] Among them, w1 and w2 are weight coefficients, and the set P represents the ship pairs used for topological structure similarity calculation;

[0048] S35. Setting threshold S thresh With comprehensive similarity S combined For comparison, if S combined ≤S thresh , then the ship in the image is successfully matched with the ship in the AIS data, S combined >S thresh , the matching fails and the SAR remote sensing image of this scene is discarded.

[0049] The present invention also provides a storage medium, which includes a stored program, wherein when the program is run, any of the above remote sensing image ship type identification methods is executed.

[0050] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes any of the above remote sensing image ship type identification methods through the computer program.

[0051] Compared with the prior art, the present invention has the following advantages:

[0052] In order to improve the efficiency of ship detection and the accuracy of sample annotation, the present invention proposes a new method based on "large map annotation", that is, directly annotating the entire image without segmenting the image. Combined with AIS data, the present invention can assign more fine-grained labels to each ship, including information such as ship type, size, and purpose, thereby effectively improving the automation of sample production.

[0053] The present invention greatly improves the efficiency of sample preparation, has a high degree of automation, and the sample preparation method can be applied to a variety of remote sensing image modalities, such as SAR, optical and infrared images. With the increase in the number of remote sensing satellites, the ship type recognition method of the present invention is applicable to a variety of remote sensing image modalities, and has wide applicability and general application prospects.

[0054] Compared with the traditional method, the present invention can not only realize the recognition of ships, but also accurately identify the type of ships. For the recognition of ship types, the method has good scalability and can flexibly expand different ship type libraries to meet the needs in practical applications. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0056] Figure 1 It is a schematic diagram of the process of the present invention.

[0057] Figure 2 A flow chart is produced for the ship type identification sample of the present invention.

[0058] Figure 3 This is the ship type identification result of the present invention. DETAILED DESCRIPTION

[0059] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme 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 only 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 ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0060] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0061] like Figure 1 As shown, the present invention provides a remote sensing image ship type recognition method, comprising the following steps:

[0062] S1. Preliminary pre-labeling of the acquired SAR remote sensing images to mark the target location of the ship;

[0063] S2. Preprocessing the AIS data, including noise removal, time alignment, spatial interpolation and other steps, to ensure that the AIS data and SAR remote sensing images can be compared within the same time period;

[0064] According to the speed and position changes of the ship, abnormal points are detected and removed. The speed v of the ship between two consecutive moments is used to determine whether there are points that obviously exceed the reasonable range. A speed threshold is set. If the speed of an AIS point exceeds the threshold, it is judged as a noise point.

[0065] The acquisition time of AIS data and SAR remote sensing images is usually not completely consistent, especially in dynamic scenes, such as the movement of target ships at sea. First, the specific imaging time T of the SAR remote sensing image is obtained. SAR , record its timestamp. Filter the AIS data, extract the ship position data close to the imaging time, and select the ship position data close to the imaging time. SAR The nearest AIS point is used as the initial alignment point T AIS_initial ;

[0066] If T AIS_initial and TSAR If there is a time difference of more than 5 minutes, linear interpolation is used to estimate the ship's time at T SAR The ship’s speed between the two AIS times T1 and T2 remains stable, and the ship’s speed at T2 is calculated by the following formula SAR Interpolated position at time:

[0067]

[0068] Among them, P1 and P2 are the positions of the ship at T1 and T2 respectively, P SAR After time alignment, SAR The ship's position at that time.

[0069] S3. Match the ship targets in the SAR remote sensing image with the ship information in the AIS data through a combination of shape and topology matching. According to the relative positions and shapes of the ships, polygons are constructed for matching to ensure that the ships in the image correspond to the ships in the AIS data one by one.

[0070] S31, shape feature calculation, including side length and internal angle;

[0071] The side length is defined as the distance between each pair of adjacent ships in the polygon and is calculated as follows:

[0072]

[0073] Among them, (x i ,y i ) and (x j ,y j ) are the center coordinates of ships i and j;

[0074] The interior angle is defined as the interior angle of each vertex in the polygon, and is calculated as follows:

[0075]

[0076] Among them, l ij and l ik are the two edges adjacent to vertex i, l jk is the length of the opposite side enclosed by these sides;

[0077] S32, calculating side length comparison and interior angle comparison;

[0078] The side length comparison is defined as comparing the length of each corresponding side of the polygon in the image and the polygon in the AIS data. The calculation formula is as follows:

[0079]

[0080] The internal angle comparison is defined as comparing the similarity of the internal angles of polygons in the image and AIS data. The calculation formula is as follows:

[0081]

[0082] Where m is the number of sides of the polygon, and are the side lengths and internal angles in the image, respectively. and are the side lengths and internal angles in the AIS data, respectively;

[0083] S33, topological feature calculation, including relative direction and relative azimuth;

[0084] The relative direction is defined as follows: for each ship pair (i, j), its direction vector relative to the image coordinate system can be defined:

[0085]

[0086] The relative azimuth is defined as the azimuth θ of ship i relative to ship j ij , the calculation formula is:

[0087]

[0088] S34, similarity calculation, calculate the shape similarity S respectively shape and topological similarity S topo , forming a comprehensive similarity S by weighting combined , the calculation formula is:

[0089] S combined =α·S shape +β·S topo

[0090] Among them, α and β are weight coefficients, S shape and S topo The calculation formula is as follows:

[0091]

[0092] Among them, w1 and w2 are weight coefficients, and the set P represents the ship pairs used for topological structure similarity calculation.

[0093] S35. Setting threshold S thresh With comprehensive similarity S combined For comparison, if S combined ≤S thresh , then the ship in the image is successfully matched with the ship in the AIS data, S combined >S thresh , the matching fails and the SAR remote sensing image of this scene is discarded.

[0094] S4. After the matching is successful, the ship target in the SAR remote sensing image is given detailed label information, including the ship type, through the AIS data to form pre-labeled raster data;

[0095] S5, the pre-labeled raster data is divided into several small blocks, each of which contains the ship target and its category label, and the rectangular vertices and categories of the ship are extracted from each block to form the labeling information. At the same time, the SAR remote sensing image blocks and the labeling information are used as a sample set, and are divided into a training set and a validation set according to a preset ratio;

[0096] Figure 2 It is a flow chart of making a sample for ship type identification, and steps S1 to S5 are the sample making process;

[0097] S6, using the training set to train the ship category recognition model, and using the validation set to validate and evaluate the trained model;

[0098] The model is Oriented R-CNN;

[0099] S7. Use the verified model to identify the ship type in the remote sensing image to obtain the detection result.

[0100] Example

[0101] A method for identifying ship types using remote sensing images comprises the following steps:

[0102] S1. Preliminary pre-labeling of the acquired SAR remote sensing images to mark the target location of the ship;

[0103] Specifically, 1,000 SAR remote sensing images of the waters around Singapore were selected from the remote sensing image library, and the ships in the images were outlined using the rotating box annotation method using QGIS software and saved as vector files.

[0104] S2. Preprocessing the AIS data, including noise removal, time alignment, spatial interpolation and other steps, to ensure that the AIS data and SAR remote sensing images can be compared within the same time period;

[0105] Specifically, the AIS data corresponding to each SAR remote sensing image is obtained based on the metadata of the SAR remote sensing image. First, the abnormal points are detected and removed according to the speed and position changes of the ship. The speed v of the ship between two consecutive moments is used to determine whether there are points that obviously exceed the reasonable range. A speed threshold is set. If the speed of an AIS point exceeds the threshold, it is judged as a noise point.

[0106] The acquisition time of AIS data and SAR remote sensing images is usually not completely consistent, especially in dynamic scenes, such as the movement of target ships at sea. First, the specific imaging time T of the SAR remote sensing image is obtained. SAR , record its timestamp. Filter the AIS data, extract the ship position data close to the imaging time, and select the ship position data close to the imaging time. SAR The nearest AIS point is used as the initial alignment point T AIS_initial ;

[0107] If T AIS_initial and T SAR If there is a time difference of more than 5 minutes, linear interpolation is used to estimate the ship's time at T SAR The ship’s speed between the two AIS times T1 and T2 remains stable, and the ship’s speed at T2 is calculated by the following formula SAR Interpolated position at time:

[0108]

[0109] Among them, P1 and P2 are the positions of the ship at T1 and T2 respectively, P SAR After time alignment, SAR The ship's position at that time.

[0110] S3. Match the ship targets in the SAR remote sensing image with the ship information in the AIS data through a combination of shape and topology matching. According to the relative positions and shapes of the ships, polygons are constructed for matching to ensure that the ships in the image correspond to the ships in the AIS data one by one.

[0111] Specifically, the ship target in the SAR remote sensing image is matched with the ship information in the AIS data through a matching method combining shape and topology. The process is as follows:

[0112] S31, shape feature calculation, including side length and internal angle;

[0113] The side length is defined as the distance between each pair of adjacent ships in the polygon and is calculated as follows:

[0114]

[0115] Among them, (x i ,y i) and (x j ,y j ) are the center coordinates of ships i and j;

[0116] The interior angle is defined as the interior angle of each vertex in the polygon, and is calculated as follows:

[0117]

[0118] Among them, l ij and l ik are the two edges adjacent to vertex i, l jk is the length of the opposite side enclosed by these sides;

[0119] S32, calculating side length comparison and interior angle comparison;

[0120] The side length comparison is defined as comparing the length of each corresponding side of the polygon in the image and the polygon in the AIS data. The calculation formula is as follows:

[0121]

[0122] The internal angle comparison is defined as comparing the similarity of the internal angles of polygons in the image and AIS data. The calculation formula is as follows:

[0123]

[0124] Where m is the number of sides of the polygon, and are the side lengths and internal angles in the image, respectively. and are the side lengths and internal angles in the AIS data, respectively;

[0125] S33, topological feature calculation, including relative direction and relative azimuth;

[0126] The relative direction is defined as follows: for each ship pair (i, j), its direction vector relative to the image coordinate system can be defined:

[0127]

[0128] The relative azimuth is defined as the azimuth θ of ship i relative to ship j ij , the calculation formula is:

[0129]

[0130] S34, similarity calculation, calculate the shape similarity S respectively shape and topological similarity S topo , forming a comprehensive similarity S by weighting combined , the calculation formula is:

[0131] S combined =α·S shape +β·S topo

[0132] Among them, α and β are weight coefficients, S shape and S topo The calculation formula is as follows:

[0133]

[0134] Among them, w1 and w2 are weight coefficients, and the set P represents the ship pairs used for topological structure similarity calculation.

[0135] S35. Setting threshold S thresh With comprehensive similarity S combined For comparison, if S combined ≤S thresh , then the ship in the image is successfully matched with the ship in the AIS data, S combined >S thresh , the matching fails and the SAR remote sensing image of this scene is discarded.

[0136] S4. After the matching is successful, the ship target in the SAR remote sensing image is given detailed label information, including the ship type, through the AIS data to form pre-labeled raster data;

[0137] S5, the pre-labeled raster data is divided into several small blocks, each of which contains the ship target and its category label, and the rectangular vertices and categories of the ship are extracted from each block to form the labeling information. At the same time, the SAR remote sensing image blocks and the labeling information are used as a sample set, and are divided into a training set and a validation set according to a preset ratio;

[0138] S6, using the training set to train the ship category recognition model, and using the validation set to validate and evaluate the trained model;

[0139] Specifically, the model is Oriented R-CNN;

[0140] S7. Use the verified model to identify the ship type in the remote sensing image to obtain the detection result.

[0141] See also Figure 3 , Figure 3 This is the result of ship type recognition of the present invention. It can be seen that through the remote sensing image ship type recognition method provided by the present invention, ship type recognition on SAR images can achieve higher recognition accuracy and efficiency.

[0142] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for identifying ship types using remote sensing images, characterized in that: The steps include: S1. Preliminary pre-labeling of the acquired SAR remote sensing images to mark the target location of the ship; S2, preprocessing the AIS data so that the AIS data and the SAR remote sensing images can be compared within the same time period; S3, by matching the ship target position in the SAR remote sensing image with the ship information in the AIS data through a shape-topology matching method, and constructing polygons for matching according to the relative positions and shapes between the ships, so that the ships in the image correspond one to one with the ships in the AIS; S4, after successful matching, detailed label information is assigned to the ship target in the SAR remote sensing image through the AIS data to form pre-labeled raster data; S5, dividing the pre-labeled raster data into several small blocks, each image block contains the ship target and its category label, and extracting the rectangular vertices and category of the ship from each block to form labeling information; at the same time, the SAR remote sensing image blocks and the labeling information are used as a sample set, and are divided into a training set and a validation set according to a preset ratio; S6, using the training set to train the ship category recognition model, and using the validation set to validate and evaluate the trained model; S7. Use the verified ship category recognition model to identify the ship type in the remote sensing image to obtain the detection result.

2. The method for identifying ship types using remote sensing images according to claim 1, characterized in that: In S2, S2 specifically includes the following steps: Preprocess AIS data: noise removal, time alignment, and spatial interpolation; According to the changes in the speed and position of the ship, abnormal points are detected and eliminated. The speed v of the ship between two consecutive moments is used to determine whether there are points that obviously exceed the reasonable range. A speed threshold is set. If the speed of an AIS point exceeds the threshold, it is judged as a noise point. Obtain the specific imaging time T of SAR remote sensing images SAR , record its timestamp; The AIS data is screened to extract the ship position data close to the imaging time, and the ship position data close to the imaging time is selected. SAR The nearest AIS point is used as the initial alignment point T AIS_initial ; If T AIS_initial and T SAR If there is a time difference of more than 5 minutes, linear interpolation is used to estimate the ship's time at T SAR The ship’s precise position at the time of T1 and T2 is calculated by the following formula: SAR Interpolated position at time: Among them, P1 and P2 are the positions of the ship at T1 and T2 respectively, P SAR After time alignment, SAR The ship's position at that time.

3. The method for identifying ship types using remote sensing images according to claim 1, characterized in that: In S4, detailed label information includes the vessel type.

4. The method for identifying ship types using remote sensing images according to claim 1, characterized in that: In S6, the ship category recognition model is Oriented R-CNN.

5. The method for identifying ship types using remote sensing images according to claim 1, characterized in that: S3 specifically includes the following steps: S31, shape feature calculation, including side length and internal angle; The side length is defined as the distance between each pair of adjacent ships in the polygon and is calculated as follows: Among them, (x i ,y i ) and (x j ,y j ) are the center coordinates of ships i and j; The interior angle is defined as the interior angle of each vertex in the polygon, and is calculated as follows: Among them, l ij and l ik are the two edges adjacent to vertex i, l jk is the length of the opposite side enclosed by these sides; Among them, l ij and l ik are the two edges adjacent to vertex i, l jk is the length of the opposite side enclosed by these sides; S32, calculating side length comparison and interior angle comparison; The side length comparison is defined as comparing the length of each corresponding side of the polygon in the image and the polygon in the AIS data. The calculation formula is as follows: The internal angle comparison is defined as comparing the similarity of the internal angles of polygons in the image and AIS data. The calculation formula is as follows: Where m is the number of sides of the polygon, and are the side lengths and internal angles in the image, respectively. and are the side lengths and internal angles in the AIS data, respectively; S33, topological feature calculation, including relative direction and relative azimuth; The relative direction is defined as follows: for each ship pair (i, j), its direction vector relative to the image coordinate system can be defined: The relative azimuth is defined as the azimuth θ of ship i relative to ship j ij , the calculation formula is: S34, similarity calculation, calculate the shape similarity S respectively shape and topological similarity S topo , forming a comprehensive similarity S by weighting combined , the calculation formula is: S combined =α·S shape +β·S topo Among them, α and β are weight coefficients, S shape and S topo The calculation formula is as follows: Among them, w1 and w2 are weight coefficients, and the set P represents the ship pairs used for topological structure similarity calculation; S35. Setting threshold S thresh With comprehensive similarity S combined For comparison, if S combined ≤S thresh , then the ship in the image is successfully matched with the ship in the AIS data, S combined >S thresh , the matching fails and the SAR remote sensing image of this scene is discarded.

6. A storage medium, characterized in that: The storage medium includes a stored program, wherein when the program is run, the remote sensing image ship type recognition method described in any one of claims 1 to 5 is executed.

7. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: The processor executes the remote sensing image ship type recognition method according to any one of claims 1 to 5 by running the computer program.

Images