Remote sensing image moving ship target tracking method, system, equipment and storage medium

Abstract

This invention relates to a method for tracking moving ship targets in remote sensing images, applied to remote sensing scenarios with a large field of view and low frame rate. The method compensates for the limitations of single-frame detection features by combining spatial information of the marine ship target with local details of its motion and multi-frame correlation. The process includes continuously acquiring multiple frames of remote sensing images and selecting ship targets to be tracked; determining the identity of the selected ship targets in two consecutive frames; and generating the trajectory of the moving ship target based on its position in the multi-frame images. This achieves high detection performance with relatively low complexity.

Description

Technical Field

[0001] This invention relates to the field of satellite optical remote sensing image ship target tracking technology, specifically to a method, system, device, and storage medium for tracking moving ship targets in remote sensing images. Background Technology

[0002] Satellite optical remote sensing images (hereinafter referred to as remote sensing images) have the characteristics of large field of view and wide coverage, which have significant advantages in the application of monitoring and tracking moving ships. However, due to the large size of a single video image and the relatively small size of the target, as well as the interference from clouds, waves, reefs, etc., how to accurately and quickly locate the target position and perform accurate target comparison to complete the tracking is a difficult problem.

[0003] For wide-swath imaging by ocean surveillance satellites, remote sensing images typically have low resolution, resulting in unclear texture details of ship targets. This can easily lead to unstable target features and insufficient feature quantity. Furthermore, the low frame rate of remote sensing images can cause changes in target features to lack continuity, thus affecting detection and tracking performance. Traditional tracking algorithms are usually designed for natural images such as pedestrian and vehicle tracking, and are suitable for high-frame-rate, high-resolution, and detailed scenarios. They are not suitable for direct application in wide-field-of-view, low-frame-rate remote sensing scenarios.

[0004] For the application of ship target tracking in large field-of-view remote sensing images from satellite optical remote sensing, this paper proposes a tracking method that combines local detail information with multi-frame correlation. By combining spatial information and motion information of marine ship targets, this method makes up for the deficiencies of single-frame detection features. This method achieves high detection performance with low complexity and is suitable for on-orbit applications of optical satellites. Summary of the Invention

[0005] This invention provides a method, system, device, and storage medium for tracking moving ship targets in remote sensing images.

[0006] To solve the above problems, the present invention adopts the following technical solution.

[0007] One aspect of the present invention provides a method for tracking moving ship targets in remote sensing images, applicable to remote sensing scenarios with a large field of view and low frame rate. The tracking method is achieved through local detail information and multi-frame correlation. This method for tracking moving ship targets in remote sensing images includes the following steps:

[0008] Multiple frames of remote sensing images were continuously acquired and the tracked ship targets were filtered out one by one;

[0009] Determine the identity of the tracked ship targets selected in two consecutive frames of remote sensing images;

[0010] Based on the target position in the multi-frame images of the tracked ship target, the navigation trajectory of the moving ship target is generated.

[0011] According to an embodiment of the present invention, the step of continuously acquiring multiple frames of remote sensing images and respectively filtering the tracked ship targets includes the following steps:

[0012] An anomaly detection model was established based on the distribution characteristics of ship samples in marine images;

[0013] Acquire a frame of remote sensing image, and detect abnormal targets in the frame of remote sensing image using an anomaly detection model;

[0014] By combining multiple features to filter out abnormal targets in the frame of remote sensing image, the ship target is determined;

[0015] According to the predetermined task, the ship targets tracked in the frame of remote sensing image are determined based on the size and speed priority of the ship targets.

[0016] According to an embodiment of the present invention, determining the identity of the tracked ship target selected in two consecutive frames of remote sensing images includes the following steps:

[0017] Obtain the first adaptive directional gradient histogram features of the ship target tracked in the frame;

[0018] Predict the range of motion of the tracked ship target in the next frame of remote sensing image;

[0019] All targets to be matched within the motion range in the next frame of the remote sensing image are obtained, and the adaptive directional gradient histogram features of all targets to be matched are obtained and compared with the first adaptive directional gradient histogram features respectively to determine the tracked ship target in the next frame of the remote sensing image.

[0020] According to an embodiment of the present invention, the anomaly detection model is as shown in formula (1):

[0021]

[0022] Among them, f s The grayscale value of pixel 's' represents the frequency of occurrence in a frame of remote sensing image; 'r' represents the neighboring pixel region centered on 's'; S r The variance of region r is represented by m. r This represents the average frequency of each gray value within region r; w1 and w2 are weight values ​​preset based on the sea surface background analysis structure.

[0023] The establishment of the anomaly detection model for detecting abnormal targets in a frame of remote sensing image includes determining whether the output result s′ of the anomaly detection model is greater than a first threshold. If s′ is greater than the first threshold, it is determined that there is an abnormal target in the remote sensing image.

[0024] The distribution characteristics include: the statistical features of the image's grayscale and texture;

[0025] The anomalous target refers to a non-sea surface object.

[0026] According to one embodiment of the present invention, the multiple features include the area, aspect ratio, compactness, and regularity index of the anomalous target;

[0027] The process of filtering abnormal targets in the frame of remote sensing image by combining multiple features to determine ship targets specifically includes using area criterion P. s Aspect Ratio Criterion P R Compactness Criterion P Cmp Regular shape index criterion P RsSI Perform the weighted summation P as shown in formula (2) CON Determine the P CON Whether it is greater than the second threshold, take P. CON When the value exceeds the second threshold, the ship target is identified.

[0028] P CON =l1P S +l2P R +l3P Cmp +l4P RSSI (2)

[0029] Wherein, l1, l2, l3, and l4 are respectively area criteria P S Aspect Ratio Criterion P R Compactness Criterion P Cmp Regular shape index criterion P RSSI The weight.

[0030] According to one embodiment of the present invention, the area represents the area of ​​the abnormal target, which is obtained by multiplying the total number of pixels by the size unit represented by a single pixel;

[0031] The area is shown in formula (3):

[0032] S=(∑I i )×P r 2 (3)

[0033] Where S represents the area, I i Let P represent the i-th pixel. r For pixel resolution; when the area S is within a preset first value range, the area criterion P... S Record it as 1, otherwise the area criterion P S Recorded as 0;

[0034] The aspect ratio refers to the aspect ratio of the minimum bounding rectangle of the abnormal target, which is the ratio of the width of the minimum bounding rectangle to the length of the minimum bounding rectangle, as shown in formula (4):

[0035]

[0036] The minimum circumscribed rectangle is used to mark the outline of the abnormal target; W represents the width of the minimum circumscribed rectangle of the abnormal target, L represents the length of the minimum circumscribed rectangle of the abnormal target, and R represents the aspect ratio of the minimum circumscribed rectangle of the abnormal target; when the aspect ratio R is within a preset second value range, the aspect ratio criterion P... R Record it as 1, otherwise the aspect ratio criterion P R Recorded as 0;

[0037] The compactness is used to describe the circularity of the target, that is, to judge the degree to which the target approximates a circle. The compactness Cmp is obtained by formula (5):

[0038]

[0039] Where, N A N represents the number of pixels contained in the potential region of an abnormal target. P The number of edge pixels in the potential region of the abnormal target; when the compactness Cmp is within the preset third value range, the compactness criterion P... Cmp Record it as 1, otherwise the compactness criterion P Cmp Recorded as 0;

[0040] The regular shape index (RSSI) is used to describe the similarity between a target and a regular shape. The RSSI is obtained by formula (6):

[0041]

[0042] Where, N A Area(MABS(F)) represents the number of pixels contained in the potential region of an abnormal target, and Area(MABS(F)) represents the number of pixels in the minimum bounding regular shape. When the regularity index RSSI is within the preset fourth value range, the regularity index criterion P... RSSI It is denoted as 1, otherwise the regular shape index criterion P RSSI Record it as 0.

[0043] According to an embodiment of the present invention, determining the tracked ship target in the frame of remote sensing image according to the size and speed priority of the ship target, according to a predetermined task, specifically includes:

[0044] Select all ship targets to be tracked according to the predetermined mission;

[0045] The tracking priority of all ship targets to be tracked is determined based on the size and speed parameters of the ship targets.

[0046] The ship targets to be tracked are determined based on the tracking priority.

[0047] According to an embodiment of the present invention, obtaining the first adaptive directional gradient histogram feature of the ship target tracked in the frame specifically includes the following steps:

[0048] Based on the minimum circumscribed rectangle of the tracked ship target, the midpoints of the two short sides of the minimum circumscribed rectangle of the tracked ship target are obtained as the two endpoints. The two endpoints are connected to obtain the line segment L', where the length of the line segment L' is the same as the length L of the minimum circumscribed rectangle.

[0049] A rectangle is obtained using line segment L' as the diagonal, the tracked ship target is marked, and the height Hs and width Ws of the rectangle are determined.

[0050] The size of the adaptive sliding window block is determined based on the height Hs and width Ws of the rectangle.

[0051] The cell is divided according to the determined size of the sliding window block;

[0052] Calculate I(x,y) for each pixel in each unit, and determine the horizontal gradient Ix and the vertical gradient Iy;

[0053] The gradient magnitude m(x,y) and direction a(x,y) are determined based on the horizontal gradient Ix and the vertical gradient Iy.

[0054] The number of features is counted according to the gradient direction of the pixels in each unit, and the four units in the block image are cascaded to obtain the feature histogram of the sliding window block image.

[0055] The feature histograms extracted from the sliding window block image are sequentially concatenated to form the A-HOG histogram feature of the ship target.

[0056] According to an embodiment of the present invention, the height Hs and width Ws of the rectangular frame are as shown in formulas (7) and (8):

[0057] Hs=L'cos(α)(7)

[0058] Ws=L'sin(α) (8)

[0059] Where α represents the angle between line segment L' and the short side of the rectangle.

[0060] According to an embodiment of the present invention, predicting the motion range of the tracked ship target in the next frame of remote sensing image includes the following steps:

[0061] The rotation angle range of the tracked ship target in the next frame is determined based on the maximum yaw angular velocity of the tracked ship target, the satellite imaging time interval, and the current position of the tracked ship target.

[0062] The movement distance range of the tracked ship target in the next frame is determined based on the speed range of the tracked ship target, the current position of the tracked ship target, the satellite imaging time interval, and the image resolution.

[0063] The motion range of the tracked ship target in the next frame is determined based on the rotation angle range and the motion distance range.

[0064] According to an embodiment of the present invention, the range of rotation angles is as shown in formula (9):

[0065] α=ω max t (9)

[0066] The range of movement distance is shown in formula (10):

[0067]

[0068] Where, ω max The maximum yaw angular velocity of the tracked ship target is represented by t, the satellite imaging time interval is represented by t, and [v1,v2] represents the range of the tracked ship target's speed.

[0069] According to an embodiment of the present invention, the steps of obtaining all targets to be matched within the motion range in the next frame of remote sensing image, obtaining the adaptive directional gradient histogram features of all targets to be matched, and comparing them with the first adaptive directional gradient histogram features respectively to determine the tracked ship target in the next frame of remote sensing image include the following steps:

[0070] Determine whether any ship target with the same adaptive directional gradient histogram features as the first adaptive directional gradient histogram features exists among all the targets to be matched within the motion range in the next frame of the remote sensing image.

[0071] If they exist, all targets to be matched that have the same features as the first adaptive directional gradient histogram are identified as the same ship target and stored. Then, for all the same ship targets within the motion range, the motion range of the next frame is predicted and the same ship target in the next frame is determined.

[0072] If it does not exist, it means that no ship target was found. A new frame of remote sensing image is acquired to filter the ship targets to be tracked, and the identity of the ship targets to be tracked in the two consecutive frames of remote sensing images is determined.

[0073] According to an embodiment of the present invention, generating the navigation trajectory of the moving ship target based on the target position in the multi-frame image of the tracked ship target includes: generating the motion trajectory or deleting the trajectory of the ship target to be tracked based on the target position in the multi-frame image of the tracked ship target.

[0074] Another aspect of the present invention provides a remote sensing image moving ship target tracking system, applied to remote sensing scenes with a large field of view and low frame rate. The tracking method is implemented through local detail information and multi-frame correlation. This remote sensing image moving ship target tracking system is used to implement the remote sensing image moving ship target tracking method described above. The target tracking system includes:

[0075] The target selection module is used to continuously acquire multiple frames of remote sensing images and select the ship targets to be tracked.

[0076] The identity determination module is used to determine the identity of the tracked ship targets selected in two consecutive frames of remote sensing images;

[0077] The navigation trajectory generation module is used to generate the navigation trajectory of the moving ship target based on the target position in the multi-frame images of the tracked ship target.

[0078] Another aspect of the present invention provides a computer device. The computer device includes a memory and a processor, the memory storing a processing program; the processor, when executing the processing program, implements the remote sensing image moving ship target tracking method according to an embodiment of the present invention.

[0079] A final aspect of the invention is to provide a readable storage medium. The readable storage medium stores a processing program, which, when executed by a processor, implements the remote sensing image moving ship target tracking method according to an embodiment of the invention.

[0080] The advantages of this invention are:

[0081] Compared with existing technologies, the advantages of the remote sensing image moving ship target tracking method, system, device, and storage medium provided by this invention are as follows: This invention achieves tracking and determines the ship target's trajectory by combining spatial information of the marine ship target with local detail information of motion information and multi-frame correlation. This enables application in remote sensing scenarios with large field of view and low frame rate, achieving high detection performance with relatively low complexity. An anomaly detection model is established based on the distribution characteristics of ship samples in marine images, enabling preliminary detection of non-sea surface objects such as broken clouds, islands, reefs, ships, and marine life in the current frame of the remote sensing image. By combining multiple features to filter abnormal targets in the frame of the remote sensing image, all ship targets in that frame are identified. According to a predetermined task, the ship targets to be tracked are determined based on their size and speed priority. The first adaptive directional gradient histogram feature of the ship targets tracked in the frame is obtained as a reference for ship targets in the next frame. The system predicts the range of motion of the tracked ship target within the next frame of remote sensing image; it acquires all targets to be matched within the range of motion in the next frame, obtains the adaptive directional gradient histogram features of all targets to be matched, and compares them with the first adaptive directional gradient histogram features to determine the tracked ship target in the next frame. This process identifies which ships in the next frame might be the same target as the tracked ship target in the current frame. Based on the target position of the tracked ship target in multiple frames, the system generates the navigation trajectory of the moving ship target, and deletes navigation trajectories where the target position disappears in multiple frames, thus determining the unique and correct navigation trajectory of the ship. Attached Figure Description

[0082] Figure 1 This is a flowchart of a remote sensing image moving ship target tracking method according to an embodiment of the present invention;

[0083] Figure 2 This is a flowchart of a method for screening and tracking ship targets according to an embodiment of the present invention;

[0084] Figure 3 This is a flowchart of a method for determining a tracked ship target in a frame of remote sensing image according to an embodiment of the present invention;

[0085] Figure 4 This is a flowchart of a method for determining the identity of tracked ship targets selected in two consecutive frames of remote sensing images according to an embodiment of the present invention.

[0086] Figure 5 This is an embodiment of the present invention for obtaining the first adaptive directional gradient histogram feature of the ship target being tracked in the current frame;

[0087] Figure 6This is a flowchart of a method for predicting the range of motion of a tracked ship target in the next frame of a remote sensing image, according to an embodiment of the present invention.

[0088] Figure 7 This is a flowchart of a method for comparing whether a target to be matched is a ship target that needs to be tracked, according to an embodiment of the present invention.

[0089] Figure 8 This is a schematic diagram of drawing the minimum circumscribed rectangle for an abnormal target in one embodiment of the present invention;

[0090] Figure 9 This is a schematic diagram of drawing a rectangular frame around a tracked ship target in one embodiment of the present invention;

[0091] Figure 10 This is a schematic diagram illustrating the determination of the position of the tracked ship target in the next frame according to an embodiment of the present invention;

[0092] Figure 11 This is a schematic diagram illustrating the generation or deletion of the motion trajectory of a ship target to be tracked in one embodiment of the present invention;

[0093] Figure 12 This is a schematic diagram of the structure of a remote sensing image moving ship target tracking system according to an embodiment of the present invention;

[0094] Figure 13 This is a schematic diagram of the structure of a remote sensing image moving ship target tracking device according to an embodiment of the present invention;

[0095] Figure 14 This is a schematic diagram of the structure of a computer-readable storage medium according to an embodiment of the present invention. Detailed Implementation

[0096] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of the embodiments of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0097] For wide-swath imaging by ocean surveillance satellites, remote sensing images typically have low resolution, resulting in unclear texture details of ship targets and easily leading to unstable target features and insufficient feature quantity. Low frame rates further exacerbate the problem. Low frame rates can cause changes in target features to lack continuity, thus affecting detection and tracking performance. Existing tracking algorithms are generally designed for natural images such as pedestrian and vehicle tracking, suitable for high-frame-rate, high-resolution, and detailed scenarios, but not suitable for direct application in wide-field-of-view, low-frame-rate remote sensing scenarios.

[0098] Compared with existing technologies, the advantages of the remote sensing image moving ship target tracking method, system, device and storage medium provided by the present invention are as follows: The present invention realizes the tracking method and determines the ship target's navigation trajectory by combining the spatial information of the ocean ship target with the local detail information of the motion information and multi-frame correlation, and realizes the application in remote sensing scenes with large field of view and low frame rate, achieving high detection performance with low complexity.

[0099] To address the technical problems in existing technologies, Embodiment 1 of the present invention provides a method for tracking moving ship targets in remote sensing images. This method is applied to remote sensing scenarios with a large field of view and low frame rate, and achieves tracking through local detail information and multi-frame correlation. (See also...) Figures 1-11 The method for tracking moving ship targets in remote sensing images includes the following steps:

[0100] S100: Continuously acquire multiple frames of remote sensing images and filter the tracked ship targets respectively;

[0101] Specifically, S100 includes the following steps:

[0102] S110: Establish an anomaly detection model based on the distribution characteristics of ship samples in marine images.

[0103] For example, the anomaly detection model is as shown in formula (1):

[0104]

[0105] Among them, f s The grayscale value of pixel 's' represents the frequency of occurrence in a frame of remote sensing image; 'r' represents the neighboring pixel region centered on 's'; S r The variance of region r is represented by m. r This represents the average frequency of each gray value within region r; w1 and w2 are preset weight values ​​based on the sea surface background analysis structure, for example, w1 is 0.3 and w2 is 0.7.

[0106] An anomaly detection model is established to detect anomalous targets in a frame of remote sensing image. This includes determining whether the output result S′ of the anomaly detection model is greater than a first threshold, for example, 0.65. If s′ is greater than the first threshold, it is determined that an anomalous target exists in the remote sensing image. The larger s′ is, the greater the probability that it is an anomaly.

[0107] This distribution characteristic includes the statistical features of the image's grayscale and texture. The anomalous target represents non-sea-surface objects such as islands, reefs, broken clouds, ships, animals, etc.

[0108] S120: Acquire a frame of remote sensing image and detect abnormal targets in the frame using an anomaly detection model. For example, the anomaly detection model initially filters out abnormal targets such as non-sea surface objects, islands, reefs, broken clouds, ships, and animals in the frame of remote sensing image.

[0109] S130: By combining multiple features to filter anomalous targets in this frame of remote sensing image, the ship target is determined. These features include the area, aspect ratio, compactness, and regularity index of the anomalous target. Specifically, determining the ship target by combining multiple features to filter anomalous targets in the frame of remote sensing image includes using area criterion P... S Aspect Ratio Criterion P R Compactness Criterion P Cmp Regular shape index criterion P RSSI Perform the weighted summation P as shown in formula (2) CON Determine the P CON Whether it is greater than the second threshold, for example, 0.5, take P. CON When the value exceeds the second threshold, the ship target is identified.

[0110] P CON =l1P S +l2P R +l3P Cmp +l4P RSSI (2)

[0111] Where l1, l2, l3, and l4 are the area criteria P S Aspect Ratio Criterion P R Compactness Criterion P Cmp Regular shape index criterion P RSSI The weights, for example, l1, l2, l3, and l4 are set to 0.1, 0.4, 0.2, and 0.3 respectively.

[0112] For example,

[0113] The area represents the area of ​​the abnormal target, which is obtained by multiplying the total number of pixels by the size unit represented by a single pixel. The area is shown in formula (3):

[0114] S=(∑I i )×P r 2 (3)

[0115] Where S represents the area, I i Let P represent the i-th pixel. r For pixel resolution; when the area S is within a preset first value range, such as [1000, 40000], then the area criterion P... S Record it as 1, otherwise the area criterion P SRecord it as 0.

[0116] The aspect ratio represents the aspect ratio of the minimum bounding rectangle of the abnormal target. It is the ratio of the width of the minimum bounding rectangle to the length of the minimum bounding rectangle, as shown in formula (4):

[0117]

[0118] The minimum circumscribed rectangle is used to mark the outline of the abnormal target; W represents the width of the minimum circumscribed rectangle of the abnormal target, L represents the length of the minimum circumscribed rectangle of the abnormal target; R represents the aspect ratio of the minimum circumscribed rectangle of the abnormal target; when the aspect ratio R is within a preset second value range, such as [5,7], then the aspect ratio criterion P... R Record it as 1, otherwise the aspect ratio criterion P R Recorded as 0;

[0119] The compactness is used to describe the circularity of the target, that is, to judge the degree to which the target approximates a circle. The compactness Cmp is obtained by formula (5):

[0120]

[0121] Where, N A N represents the number of pixels contained in the potential region of an abnormal target. P The number of edge pixels in the potential region of the abnormal target; when the compactness Cmp is within a preset third value range, such as [30, 36], then the compactness criterion P Cmp Record it as 1, otherwise the compactness criterion P Cmp Recorded as 0;

[0122] The regular shape index (RSSI) is used to describe the similarity between a target and a regular shape. The RSSI is obtained by formula (6):

[0123]

[0124] Where, N A This indicates the number of pixels contained in the potential region of an abnormal target.

[0125] Area(MABS(F)) represents the number of pixels in the smallest bounding regular shape; when the regular shape index RSSI is within the preset fourth value range, such as [11, 15], then the regular shape index criterion P... RSSI It is denoted as 1, otherwise the regular shape index criterion P RSSI Record it as 0.

[0126] S140: According to the predetermined task, determine the ship target being tracked in the frame of remote sensing image based on the size and speed priority of the ship target.

[0127] Specifically, S140 includes:

[0128] S141: Select all ship targets to be tracked according to the predetermined mission.

[0129] S142: Determine the tracking priority of all ship targets to be tracked based on their size and speed parameters. For example, among the ship targets to be tracked, the ship with the longest length and fastest speed has the highest priority; the ship with the fastest speed has the next highest priority; and the ship with the longest length has the lowest priority.

[0130] S143: Determine the ship target to be tracked based on the tracking priority.

[0131] S200: Determine the identity of the tracked ship targets selected in two consecutive frames of remote sensing images;

[0132] Specifically, S200 includes the following steps:

[0133] S210: Obtain the first adaptive directional gradient histogram feature of the ship target tracked in the frame.

[0134] S210 includes the following steps:

[0135] S211: Based on the minimum circumscribed rectangle of the tracked ship target, obtain the midpoints of the two short sides of the minimum circumscribed rectangle of the tracked ship target as two endpoints, and connect the two endpoints to obtain line segment L', wherein the length of line segment L' is the same as the length L of the minimum circumscribed rectangle.

[0136] The shorter side represents the ship's width, and the longer side represents the ship's length.

[0137] S212: Obtaining a rectangle using line segment L' as the diagonal may include marking the tracked ship target and determining the height Hs and width Ws of the rectangle.

[0138] For example, the height Hs and width Ws of the rectangle are shown in formulas (7) and (8):

[0139] Hs=L'cos(α)(7)

[0140] Ws=L'sin(α) (8)

[0141] Where α represents the angle between line segment L' and the short side of the rectangle.

[0142] S213: Determine the size of the adaptive sliding window block based on the height Hs and width Ws of the rectangle.

[0143] S214: Divide the unit into cells according to the determined size of the sliding window block.

[0144] S215: Calculate I(x,y) for each pixel in each unit and determine the horizontal gradient Ix and the vertical gradient Iy.

[0145] S216: Determine the gradient magnitude m(x,y) and direction a(x,y) based on the horizontal gradient Ix and the vertical gradient Iy.

[0146] S217: Count the number of features according to the gradient direction of each pixel in each unit, and cascade the four units in the block image to obtain the feature histogram of the sliding window block image.

[0147] S218: Sequentially connect the feature histograms extracted from the sliding window block image to form the A-HOG histogram feature of the ship target.

[0148] For example,

[0149] The size of the adaptive sliding window block specifically includes:

[0150] 2 i ≤Hs<2 i+1 ,2 j ≤Ws<2 j+1 , where i and j are integers greater than or equal to 0.

[0151] If (i≤3, j≤3), then according to the size of 2 i ×2 j The sliding window is a block image, and each block can also be called a cell;

[0152] If (i≤3, j>3), then according to the size of 2 i A ×8 sliding window is a block image, and each block can also be called a cell.

[0153] If (3 < i, j ≤ 3), then the size is 8 × 2 j The sliding window is a block image, and each block can also be called a cell;

[0154] If (3 < i ≤ 6, 3 < j ≤ 6), then the rectangle to be detected is divided into multiple block images according to a sliding window of size 16×16, and each block image is divided into 8×8 cells.

[0155] If (6 < i, 6 < j), then the rectangle to be detected is divided into multiple block images according to a sliding window of size 32×32, and each block image is divided into 16×16 cells.

[0156] Calculate I(x,y) for each pixel in each unit, and determine the horizontal gradient Ix and the vertical gradient Iy, as shown in formulas (11) and (12):

[0157] I x (x,y)=I(x+1,y)-I(x-1,y) (11)

[0158] I y (x,y)=I(x,y+1)-I(x,y-1) (12)

[0159] The gradient magnitude m(x,y) and direction a(x,y) are shown in equations (11) and (12):

[0160]

[0161]

[0162] The gradient direction is adaptively determined based on the ship target's navigation direction. Based on the angle α between the length L of the minimum circumscribed rectangle of the abnormal target in step S3 and the horizontal direction of the image, α and α+90 are selected as gradient directions, generating two gradient direction intervals [α-15, α+15] and [α+75, α+105]. Then, the number of features is counted according to the gradient direction interval of each pixel within each cell. Finally, the four cells within the block image are concatenated to obtain the feature histogram of the block image. Following the above steps, the feature histograms extracted from all block images are sequentially connected to form the A-HOG histogram feature of the ship target.

[0163] S220: Predict the range of motion of the tracked ship target in the next frame of remote sensing image.

[0164] S220 includes the following steps:

[0165] S221: Determine the rotation angle range of the tracked ship target in the next frame based on the maximum yaw angular velocity of the tracked ship target, the satellite imaging time interval, and the current position of the tracked ship target.

[0166] The range of rotation angles is shown in formula (9):

[0167] α=ωm ax t (9)

[0168] Where, ω max The maximum yaw angular velocity of the tracked ship target is represented by t; the satellite imaging time interval is represented by t; the rotation angle range α = m1 + m2 represents the sum of the maximum rotation angles of the tracked ship target in the left and right directions. In this case, the tracked ship target may have already turned to one side, and m1 and m2 are different; when the tracked ship target has not turned to one side, m1 and m2 are the same.

[0169] S222: Determine the range of motion distance of the tracked ship target in the next frame based on the range of motion speed of the tracked ship target, the current position of the tracked ship target, the satellite imaging time interval and the image resolution.

[0170] The range of the movement distance is shown in formula (10):

[0171]

[0172] Where, ω max The maximum yaw angular velocity of the tracked ship target is represented by t, the satellite imaging time interval is represented by t, and [v1,v2] represents the range of the tracked ship target's speed.

[0173] S223: Determine the motion range H of the tracked ship target in the next frame based on the rotation angle range and the motion distance range.

[0174] S230: Obtain all targets to be matched within the motion range in the next frame of remote sensing image, obtain the adaptive directional gradient histogram features of all targets to be matched, and compare them with the first adaptive directional gradient histogram features respectively to determine the tracked ship target in the next frame of remote sensing image.

[0175] Specifically, it includes:

[0176] Determine whether any ship target with the same adaptive directional gradient histogram features as the first adaptive directional gradient histogram features exists among all the targets to be matched within the motion range in the next frame of the remote sensing image.

[0177] If they exist, all targets to be matched that have the same features as the first adaptive directional gradient histogram are identified as the same ship target and stored. Then, for all the same ship targets within the motion range, the process returns to step S220 to predict the motion range of the tracked ship target in the next frame of remote sensing image, that is, to continue to predict the motion range of the next frame and determine the same ship target in the next frame.

[0178] If it does not exist, it means that no ship target was found. Then return to step S100, reacquire multiple consecutive frames of remote sensing images and filter the ship targets to be tracked respectively. Acquire one frame of remote sensing image to filter the ship targets to be tracked, and determine the identity of the ship targets to be tracked in two consecutive frames of remote sensing images.

[0179] For example, such as Figure 10 As shown, the motion range of the ship target to be tracked in the next frame N+1 is determined in the current frame N remote sensing image, that is, the possible location of the ship target to be tracked in the next frame is predicted. In the N+1 frame remote sensing image, it is determined that there are 3 targets to be matched within the predicted motion range (i.e., the fan-shaped window area). Among them, objects outside the predicted motion range are ignored. The second, third, and fourth adaptive directional gradient (ADR) histogram features of three targets A11, A12, and A13 within the predicted motion range, acquired in the (N+1)th frame of the remote sensing image, are obtained respectively. These features are then compared with the first ADR histogram feature. If the second and third ADR features of targets A11 and A12 are identical to the first ADR histogram feature, then A11 and A12 are considered to be the same ship target to be tracked, thus confirming the ship target identity. The possible trajectory curves of A11 and A12 as the ship target to be tracked are stored, for example, trajectory A-A11 and trajectory A-A12. Then, using A11 and A12 as the ship targets to be tracked, return to step S220 to continue predicting the motion range of N+2 frames and determining the same ship target in N+2 frames.

[0180] If it is determined that there is no target to be matched within the predicted motion range (i.e., the fan-shaped window area), then return to S120 to reacquire a frame of remote sensing image to filter the ship targets to be tracked, and determine the identity of the ship targets to be tracked in the two consecutive frames of remote sensing images.

[0181] S300: Generates the trajectory of the moving ship target based on the target position in the multi-frame images of the tracked ship target. Specifically, it generates or deletes the trajectory of the ship target to be tracked based on the target position in the multi-frame images of the tracked ship target.

[0182] For example, such as Figure 11As shown, when target A11 in frame N+1 determines the motion range of target A11 in the next frame N+2; in the remote sensing image of frame N+2, if it is determined that there is no target to be matched in the predicted fan-shaped window area, then trajectory A-A11 is deleted. When target A12 in frame N+1 determines the motion range of target A12 in the next frame N+2; in the remote sensing image of frame N+2, it is determined that there are two targets A121 and A122 to be matched within the predicted fan-shaped window area. The fifth adaptive directional gradient histogram (ADR) features and the sixth adaptive directional gradient histogram (ADR) features of the targets to be matched are obtained. The fifth and sixth adaptive directional gradient histogram features are compared with the first adaptive directional gradient histogram features. If the fifth adaptive directional gradient histogram feature of A121 is the same as the first adaptive directional gradient histogram feature, then it is considered that A121 may be the same ship target as the ship target to be tracked. That is, the identity of the ship target is determined, and A121 is stored as a possible trajectory curve of the ship target to be tracked, that is, the trajectory A-A12-A121 is generated. Then, using A121 as the target ship to be tracked, return to step S220 to continue predicting the motion range of N+3 frames and determining the same target ship in N+3 frames.

[0183] Embodiment 2 of the present invention provides a specific implementation of a remote sensing image moving ship target tracking system 500, applied to remote sensing scenes with a large field of view and low frame rate. The tracking method is achieved through local detail information and multi-frame correlation. See [link to documentation]. Figure 12 The system includes:

[0184] Tracking target filtering module 510: used to continuously acquire multiple frames of remote sensing images and filter the tracked ship targets respectively;

[0185] Identity determination module 520: used to determine the identity of the tracked ship targets selected in two consecutive frames of remote sensing images;

[0186] The navigation trajectory generation module 530 is used to generate the navigation trajectory of the moving ship target based on the target position in the multi-frame images of the tracked ship target.

[0187] It is understood that various aspects of the technical solution of the present invention can be implemented as a system, method, or program product. Therefore, various aspects of the technical solution of the present invention can be specifically implemented in the following forms: a completely hardware implementation, a completely software implementation (including firmware, microcode, etc.), or an implementation combining hardware and software aspects, which can be collectively referred to here as a "circuit", "unit", or "platform".

[0188] Those skilled in the art should understand that the various units or steps of the present invention described above can be implemented using general-purpose computing devices. They can be centralized on a single computing device or distributed across a network of multiple computing devices. Optionally, they can be implemented using computing device-executable program code, thereby storing them in a storage medium for execution by the computing device. In some cases, the steps shown or described can be performed in a different order than those described herein, or they can be fabricated as separate integrated circuit units, or multiple units or steps can be fabricated as a single integrated circuit unit.

[0189] Embodiment 3 of the present invention provides a schematic diagram of the structure of a remote sensing image moving ship target tracking device 600. Refer below for... Figure 13 This section describes in detail the remote sensing image moving ship target tracking device 600 implemented according to the embodiments described herein. It is understood that... Figure 13 The remote sensing image of the moving ship target tracking device 600 shown is merely an example and should not impose any limitation on the function and scope of use of any embodiment of the technical solution of the present invention.

[0190] like Figure 13 As shown, the remote sensing image moving ship target tracking device 600 is presented in the form of a general-purpose computing device. The components of the remote sensing image moving ship target tracking device 600 may include, but are not limited to: at least one processing unit 610, at least one storage unit 620, a bus 630 connecting different platform components (including storage unit 620 and processing unit 610), a display unit 640, etc.

[0191] The storage unit stores program code, which can be executed by the processing unit 610, enabling the processing unit 610 to implement the various functional modules in the remote sensing image moving ship target tracking system described above in this embodiment.

[0192] Storage unit 620 may include a readable medium in the form of a volatile storage unit, such as random access unit (RAM) 6201 and / or cache storage unit 6202, and may further include read-only storage unit (ROM) 6203.

[0193] Storage unit 620 may also include a program / utility 6204 having a set (at least one) program module 6205, such program module 6205 including but not limited to: operating system, one or more application programs, other program modules and program data, each of these examples or some combination thereof may include an implementation of a network environment.

[0194] Bus 630 can represent one or more of several types of bus structures, including a memory cell bus or memory cell controller, a peripheral bus, an image acceleration port, a processing unit, or a local bus using any of the various bus structures.

[0195] The remote sensing image moving ship target tracking device 600 can also communicate with one or more external devices 700 (e.g., keyboard, pointing device, Bluetooth device, etc.), and with one or more devices that enable a user to interact with the remote sensing image moving ship target tracking device 600, and / or with any device that enables the electronic device to communicate with one or more other computing devices (e.g., router, modem, etc.). This communication can be performed via input / output (I / O) interface 650. Furthermore, the remote sensing image moving ship target tracking device 600 can also communicate with one or more networks (e.g., local area network (LAN), wide area network (WAN), and / or public networks, such as the Internet) via network adapter 660. Network adapter 660 can communicate with other modules of the electronic device 600 via bus 630. It should be understood that, although... Figure 13 As not shown in the diagram, other hardware and / or software modules may be used in conjunction with the electronic device 600, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage platforms.

[0196] Embodiment 4 of the present invention also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, can realize the various functional units in the remote sensing image moving ship target tracking system disclosed above.

[0197] Although other specific implementation methods are not listed exhaustively in this embodiment, in some possible implementation methods, various aspects of the technical solution described in this invention can also be implemented in the form of a program product, which includes program code. When the program product is run on a terminal device, the program code is used to cause the terminal device to execute the steps of the implementation methods in various embodiments of the technical solution of this invention, as described in the remote sensing image moving ship target tracking method of this invention.

[0198] Figure 14 A schematic diagram of the structure of a computer-readable storage medium is shown according to some embodiments of the present invention. For example... Figure 14As shown, a program product 800 for implementing the above-described method according to an embodiment of the technical solution of the present invention is described. This product may employ a portable compact disc read-only memory (CD-ROM) and include program code, and can run on a terminal device, such as a personal computer or a satellite computer. Of course, the program product generated according to this embodiment is not limited to this. In the technical solution of the present invention, the readable storage medium can be any tangible medium containing or storing a program, which can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0199] The program product may employ any combination of one or more readable media. A readable medium may be a readable signal medium or a readable storage medium. A readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples (a non-exhaustive list) of readable storage media include: electrical connections having one or more wires, portable disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0200] Computer-readable storage media may include data signals propagated in baseband or as a carrier wave region, carrying readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A readable storage medium may also be any readable medium other than a readable storage medium that can transmit, propagate, or transfer a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the readable storage medium may be transmitted using any suitable medium, including but not limited to wireless, wired, optical fiber, RF, etc., or any suitable combination thereof.

[0201] Program code for executing the technical solutions of this invention can be written in any combination of one or more programming languages. These programming languages ​​include object-oriented programming languages—such as Java and C++—as well as conventional procedural programming languages—such as C or similar languages. The program code can execute entirely on the user's computing device, locally on the user's device, as a standalone software package, locally on the user's computing device, locally on a remote computing device, or entirely on a remote computing device or server. In cases involving remote computing devices, the remote computing device can be connected to the user's computing device via any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computing device (e.g., via the Internet using an Internet service provider).

[0202] In summary, the technical solution proposed in this invention enables a tracking method and determination of the ship's trajectory by combining spatial information of marine ship targets with local details of motion information and multi-frame correlation. This method is applicable to remote sensing scenarios with large field of view and low frame rate, achieving high detection performance with relatively low complexity. An anomaly detection model is established based on the distribution characteristics of ship samples in marine images, enabling preliminary detection of non-sea surface objects such as broken clouds, islands, ships, and marine life in the current frame of remote sensing image. By combining multiple features to filter abnormal targets in the frame of remote sensing image, all ship targets in that frame are identified. According to a predetermined task, the ship targets to be tracked are determined based on their size and speed priority. The first adaptive directional gradient histogram feature of the ship targets tracked in the frame is obtained as a reference for ship targets in the next frame. The system predicts the motion range of the tracked ship target within the next frame of remote sensing image. It then acquires all targets within the motion range in the next frame, obtains the adaptive directional gradient histogram (AGP) features of each target, and compares them with the first AGP feature to determine the tracked ship target in the next frame. This process identifies which ships in the next frame might be the same target as the tracked ship target in the current frame. Based on the target position in multiple frames of images, the system generates the ship target's trajectory and deletes trajectories where the target position disappears from the multi-frame images, thus determining the unique and correct trajectory of the ship.

[0203] The above description is merely a description of a preferred embodiment of the technical solution of the present invention and is not intended to limit the scope of the technical solution of the present invention in any way. Any changes or modifications made by those skilled in the art based on the above disclosure shall fall within the protection scope of the claims.

Claims

1. A method for tracking moving ship targets in remote sensing images, applied to remote sensing scenes with a large field of view and low frame rate, which achieves tracking through local detail information and multi-frame correlation, characterized in that... Includes the following steps: The method of continuously acquiring multiple frames of remote sensing images and filtering the tracked ship targets includes the following steps: establishing an anomaly detection model based on the distribution characteristics of ship samples in ocean images; acquiring a frame of remote sensing image and detecting abnormal targets in the frame of remote sensing image using the anomaly detection model; filtering abnormal targets in the frame of remote sensing image by combining multiple features to determine ship targets; and determining the tracked ship targets in the frame of remote sensing image according to a predetermined task based on the size and speed priority of the ship targets. Determining the identity of the tracked ship targets selected in two consecutive frames of remote sensing images includes the following steps: obtaining the first adaptive directional gradient histogram features of the tracked ship targets in the frame of remote sensing images, including: obtaining the midpoints of the two short sides of the minimum bounding rectangle of the tracked ship targets as two endpoints, connecting the two endpoints to obtain a line segment L', where the length of line segment L' is the same as the length L of the minimum bounding rectangle; dividing the area into cells according to the determined size of the sliding window block; calculating each... For pixel I(x,y), determine the horizontal gradient Ix and vertical gradient Iy; sequentially connect the feature histograms extracted from the sliding window block image to form the A-HOG histogram feature of the ship target; predict the motion range of the tracked ship target in the next frame of remote sensing image; obtain all targets to be matched within the motion range in the next frame of remote sensing image, obtain the adaptive directional gradient histogram features of all targets to be matched, and compare them with the first adaptive directional gradient histogram feature to determine the tracked ship target in the next frame of remote sensing image; Based on the target position in the multi-frame images of the tracked ship target, the navigation trajectory of the moving ship target is generated; Predicting the range of motion of the tracked ship target in the next frame of remote sensing image includes the following steps: The rotation angle range of the tracked ship target in the next frame is determined based on the maximum yaw angular velocity of the tracked ship target, the satellite imaging time interval, and the current position of the tracked ship target. The movement distance range of the tracked ship target in the next frame is determined based on the speed range of the tracked ship target, the current position of the tracked ship target, the satellite imaging time interval, and the image resolution. The motion range of the tracked ship target in the next frame is determined based on the rotation angle range and the motion distance range.

2. The method for tracking moving ship targets in remote sensing images according to claim 1, characterized in that, The anomaly detection model is as shown in formula (1): Among them, f s The grayscale value of pixel 's' represents the frequency of occurrence in a frame of the remote sensing image; 'r' represents the adjacent pixel region centered on 's'; S r The variance of region r is represented by m. r This represents the average frequency of each gray value within region r; w1 and w2 are weight values ​​preset based on the sea surface background analysis structure. The establishment of the anomaly detection model for detecting abnormal targets in a frame of remote sensing image includes determining whether the output result s' of the anomaly detection model is greater than a first threshold. If s' is greater than the first threshold, it is determined that there is an abnormal target in the remote sensing image. The distribution characteristics include: the statistical features of the image's grayscale and texture; The anomalous target refers to a non-sea surface object.

3. The method for tracking moving ship targets in remote sensing images according to claim 2, characterized in that, The various features include the area, aspect ratio, compactness, and regularity index of the anomalous target; The process of filtering abnormal targets in the frame of remote sensing image by combining multiple features to determine ship targets specifically includes using area criterion P. S Aspect Ratio Criterion P R Compactness Criterion P Cmp Regular shape index criterion P RSSI Perform the weighted summation P as shown in formula (2) CON Determine the P CON Whether it is greater than the second threshold, take P. CON When the value exceeds the second threshold, the ship target is identified. P CON =l1P S +l2P R +l3P Cmp +l4P RSSI (2) Wherein, l1, l2, l3, and l4 are respectively area criteria P S Aspect Ratio Criterion P R Compactness Criterion P Cmp Regular shape index criterion P RSSI The weight.

4. The method for tracking moving ship targets in remote sensing images according to claim 3, characterized in that, The area represents the area of ​​the abnormal target, which is obtained by multiplying the total number of pixels by the size unit represented by a single pixel; The area is shown in formula (3): S=(∑I i )×P r 2 (3) Where S represents the area, I i Let P represent the i-th pixel. r For pixel resolution; when the area S is within a preset first value range, the area criterion P... S Record it as 1, otherwise the area criterion P S Recorded as 0; The aspect ratio refers to the aspect ratio of the minimum bounding rectangle of the abnormal target, which is the ratio of the width of the minimum bounding rectangle to the length of the minimum bounding rectangle, as shown in formula (4): The minimum circumscribed rectangle is used to mark the outline of the abnormal target; W represents the width of the minimum circumscribed rectangle of the abnormal target, L represents the length of the minimum circumscribed rectangle of the abnormal target, and R represents the aspect ratio of the minimum circumscribed rectangle of the abnormal target; when the aspect ratio R is within a preset second value range, the aspect ratio criterion P... R Record it as 1, otherwise the aspect ratio criterion P R Recorded as 0; The compactness is used to describe the circularity of the target, that is, to judge the degree to which the target approximates a circle. The compactness Cmp is obtained by formula (5): Where, N A N represents the number of pixels contained in the potential region of an abnormal target. P The number of edge pixels in the potential region of the abnormal target; when the compactness Cmp is within the preset third value range, the compactness criterion P... Cmp Record it as 1, otherwise the compactness criterion P Cmp Recorded as 0; The regular shape index (RSSI) is used to describe the similarity between a target and a regular shape. The RSSI is obtained by formula (6): Where, N A Area(MABS(F)) represents the number of pixels contained in the potential region of an abnormal target, and Area(MABS(F)) represents the number of pixels in the minimum bounding regular shape. When the regularity index RSSI is within the preset fourth value range, the regularity index criterion P... RSSI It is denoted as 1, otherwise the regular shape index criterion P RSSI Record it as 0.

5. The method for tracking moving ship targets in remote sensing images according to claim 4, characterized in that, The step of determining the tracked ship target in the frame of remote sensing image according to the predetermined task and based on the size and speed priority of the ship target specifically includes: Select all ship targets to be tracked according to the predetermined mission; The tracking priority of all ship targets to be tracked is determined based on the size and speed parameters of the ship targets. The ship targets to be tracked are determined based on the tracking priority.

6. The method for tracking moving ship targets in remote sensing images according to claim 5, characterized in that, The specific steps for obtaining the first adaptive directional gradient histogram feature of the tracked ship target in the frame remote sensing image include the following: Based on the minimum circumscribed rectangle of the tracked ship target, the midpoints of the two short sides of the minimum circumscribed rectangle of the tracked ship target are obtained as the two endpoints. The two endpoints are connected to obtain the line segment L', where the length of the line segment L' is the same as the length L of the minimum circumscribed rectangle. Obtain a rectangular frame using line segment L' as the diagonal, and determine the height Hs and width Ws of the rectangular frame; The size of the adaptive sliding window block is determined based on the height Hs and width Ws of the rectangle. The cell is divided according to the determined size of the sliding window block; Calculate I(x,y) for each pixel in each unit, and determine the horizontal gradient Ix and the vertical gradient Iy; The gradient magnitude m(x,y) and direction a(x,y) are determined based on the horizontal gradient Ix and the vertical gradient Iy. The number of features is counted according to the gradient direction of the pixels in each unit, and the four units in the block image are cascaded to obtain the feature histogram of the sliding window block image. The feature histograms extracted from the sliding window block image are sequentially concatenated to form the A-HOG histogram feature of the ship target.

7. The method for tracking moving ship targets in remote sensing images according to claim 6, characterized in that, The height Hs and width Ws of the rectangle are shown in formulas (7) and (8): Hs=L'cos(α) (7) Ws=L'sin(α) (8) Where α represents the angle between line segment L' and the short side of the rectangle.

8. The method for tracking moving ship targets in remote sensing images according to claim 7, characterized in that, The range of rotation angles is shown in formula (9): a = oh max t (9) The range of movement distance is shown in formula (10): Where, ω max The maximum yaw angular velocity of the tracked ship target is represented by t, the satellite imaging time interval is represented by t, and [v1,v2] represents the range of the tracked ship target's speed.

9. A method for tracking moving ship targets in remote sensing images according to claim 8, characterized in that, The steps of obtaining all targets to be matched within the motion range in the next frame of the remote sensing image, obtaining the adaptive directional gradient histogram features of all targets to be matched, and comparing them with the first adaptive directional gradient histogram features to determine the tracked ship target in the next frame of the remote sensing image include the following steps: Determine whether any ship target with the same adaptive directional gradient histogram features as the first adaptive directional gradient histogram features exists among all the targets to be matched within the motion range in the next frame of the remote sensing image. If they exist, all targets to be matched that have the same features as the first adaptive directional gradient histogram are identified as the same ship target and stored. Then, for all the same ship targets within the motion range, the motion range of the next frame is predicted and the same ship target in the next frame is determined. If it does not exist, it means that no ship target has been found. A new frame of remote sensing image is acquired to filter the ship targets to be tracked, and the identity of the ship targets to be tracked in the two consecutive frames of remote sensing images is determined.

10. A method for tracking moving ship targets in remote sensing images according to claim 9, characterized in that, The step of generating the navigation trajectory of the moving ship target based on the target position in the multi-frame images of the tracked ship target includes: generating or deleting the trajectory of the ship target to be tracked based on the target position in the multi-frame images of the tracked ship target.

11. A remote sensing image moving ship target tracking system, wherein a tracking method is applied to remote sensing scenes with a large field of view and low frame rate, and the remote sensing image moving ship target tracking method as described in any one of claims 1 to 10 is realized through local detail information and multi-frame correlation, characterized in that: include: The target selection module is used to continuously acquire multiple frames of remote sensing images and select the ship targets to be tracked. The identity determination module is used to determine the identity of the tracked ship targets selected in two consecutive frames of remote sensing images; The navigation trajectory generation module is used to generate the navigation trajectory of the moving ship target based on the target position in the multi-frame images of the tracked ship target.

12. A computer device, characterized in that, include: The memory is used to store the processing program; A processor, which, when executing the processing program, implements the remote sensing image moving ship target tracking method as described in any one of claims 1 to 10.

13. A readable storage medium, characterized in that, The readable storage medium stores a processing program, which, when executed by a processor, implements the remote sensing image moving ship target tracking method as described in any one of claims 1 to 10.

Images