Weak and small target association method, equipment and system for scanning imaging

By converting the target position to the earth coordinate system and adopting a clustering method based on density and direction, the problem of weak target association in scanning scenarios is solved, and the accuracy and real-time improvement of target association in scanning scenarios is achieved.

CN119091172AActive Publication Date: 2024-12-06HUAZHONG UNIV OF SCI & TECH
View PDF 4 Cites 0 Cited by

Patent Information

Application Number
CN202411197928.3
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-29
Publication Date
2024-12-06
Estimated Expiration
2044-08-29

AI Technical Summary

Technical Problem

The prior art is difficult to achieve the association of weak targets in scanning scenarios, mainly because the target motion characteristics are different from those in gaze scenarios, resulting in the lack of continuous trajectory and timing information in scanning scenarios. The existing methods are insufficient in real-time and accuracy.

Method used

By converting the target position detected by the detector to the earth coordinate system or the earth polar coordinate system, the absolute position is obtained, and the density and direction-based clustering method is used to cluster the target to ensure that the angle of the target in the same category does not exceed the angle of the target in the same category. Preset the angle threshold to achieve weak target association in the scanning scenario.

Benefits of technology

The accurate correlation of weak targets in scanning scenarios is achieved, real-time and accuracy are improved, the problems of unstable target motion trajectory and noise interference are overcome, and the effectiveness of target association is enhanced.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119091172A_ABST
    Figure CN119091172A_ABST
Patent Text Reader

Abstract

The invention discloses a weak and small target association method, device and system for scanning imaging, and belongs to the technical field of target association, and the method comprises the steps: converting the position of each target in a current frame image into an earth coordinate system or an earth polar coordinate system as an absolute position, forming a data set D with the absolute positions of the previously detected targets, executing clustering based on density and direction, and determining the targets of the same category as associated targets; the clustering comprises the following steps: for the current category, sequentially accessing unaccessed objects in the D until no unaccessed object exists or the currently accessed object d0 is a core object; if the object which is not accessed does not exist in the D, ending clustering; if not, determining that the E neighborhood of the d0 and the E neighborhood of each core object di in the E neighborhood of the d0 meet the included angle theta ilt between the # imgabs0 # and the # imgabs1 #; and the object dj of the theta th is added into the current category, and then clustering of the next category is carried out. According to the invention, weak and small target association can be effectively realized in a scanning scene.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the technical field of target association, and more specifically, to a method, device and system for associating small targets in scanning imaging. Background Art

[0002] After using the detector to detect the target, quickly associating the target to know the number and location of the target is crucial to the ability to perceive the environment. When the detector detects the target, due to the long distance, the detected target is often very small, so it is generally called a weak target. Unlike large targets, which can directly extract features and complete target detection and target association based on the extracted features, weak targets are small in size and lack sufficient texture information, so feature extraction is impossible, which makes it difficult to associate weak targets.

[0003] The association of targets is a spatial-temporal combination, which integrates the timing information contained in the image sequence, which is also the motion information of the target. This method is called TBD (Track Before Detect), which is mainly used in scenes with low signal noise. It uses the continuity of target motion to screen trajectories and match targets belonging to the same trajectory at different times. It can assign numbers to them according to actual conditions.

[0004] In the actual detection process, the detector is divided into different scenarios, namely, staring scenario and scanning scenario. In the staring scenario, Figure 1As shown in the figure, the center of the detector's field of view is relatively fixed, the target's motion trajectory is continuous, and an obvious and continuous trajectory can be accumulated from the image sequence, making the association relatively simple. For staring scenes, representative traditional TBD algorithms include dynamic programming, three-dimensional matched filter method, inter-frame correlation method, maximum likelihood estimation, particle filtering, and multi-level hypothesis testing method, which can achieve good association performance in scenes with low signal-to-noise ratio. This is mainly due to the fact that they not only rely on the spatial information of the current frame, but also integrate the time series information of the previous N frames for energy accumulation, thereby effectively suppressing interference such as noise. However, due to the large amount of data to be processed and the relatively complex algorithm, such methods have poor real-time performance, which is not conducive to practical application. On the other hand, the effectiveness of such methods is premised on accurately accumulating the energy of moving targets in multiple consecutive frames, which requires overcoming the problem of unstable target motion trajectory, including but not limited to target occlusion, trajectory breakage caused by search platform jitter, etc., especially on airborne platforms, the detector often needs to continuously scan the target area in search mode, and the available information of the target between adjacent frames is minimal, especially when the detector is scanning quickly, adjacent frames may not overlap or only have very small parts overlap. At this time, the purpose of associating the target is to "recognize" the target and associate it with the previously detected target when the detector scans the target next time. This requires decoupling the motion of the entire platform, which requires the fusion of airborne inertial navigation information and detector hardware information, such as azimuth, pitch angle, etc., thereby introducing errors in the entire system and increasing the difficulty of target association.

[0005] Different from the staring scene, in the detector scanning scene, the center of the field of view will change periodically, so the target may appear briefly at the edge of the field of view and disappear, waiting for the detector to rotate to the angle corresponding to the target next time. This makes the detection of the target exist in a "window period". The image sequence contains multiple frames of images without target information, the position of the target in the image is very different, and there is a lack of timing information. Figure 2 The figure shows a scanning scene. In each scanning cycle, the position of the target in the image is different, the target has no continuous trajectory, and there is an N-frame interval between two consecutive detections of the target (depending on the angular velocity of the scan). When the detector detects the current target, it is difficult to associate it with the previously detected target. At present, there is no effective means of associating weak targets in scanning scenes. Due to the different target motion characteristics in different scenes, the weak target association method proposed for the staring scene cannot effectively achieve weak target association in the scanning scene. Summary of the invention

[0006] In view of the defects of the prior art and the need for improvement, the present invention provides a method, device and system for associating small targets in scanning imaging. The purpose is to propose a method that can effectively achieve small target association in a scanning scene based on the motion characteristics of the target in the scanning scene.

[0007] To achieve the above object, according to one aspect of the present invention, a method for associating small targets in scanning imaging is provided, which is used to associate small targets in a frame image sequence detected by a detector in a scanning scene; the method comprises:

[0008] Perform target detection on the current frame image, obtain the position of each target in the image, and then convert it to the earth coordinate system or the earth polar coordinate system to obtain the absolute position of each target;

[0009] Perform density and direction based clustering on the data set D consisting of the absolute positions of each target in the current frame image and the absolute positions of all previously detected targets, and determine the targets belonging to the same category in the clustering results as related targets; clustering based on density and direction includes:

[0010] Step S1: Initialize category label k=1;

[0011] Step S2: sequentially access the unaccessed objects in the data set D until there are no unaccessed objects in the data set D, or the currently accessed object d0 is a core object;

[0012] Step S3: If there is no unvisited object in the data set D, go to step S6; if the currently visited object d0 is a core object, mark the category labels of the objects in the E neighborhood of object d0 as k, and go to step S4;

[0013] Step S4: For each core object d in the E neighborhood of object d0 i , and its E neighborhood satisfies and Angle θ i <θ th Object d j The category label is marked as k; θ th is the preset angle threshold;

[0014] Step S5: After updating the category number k according to k=k+1, go to step S2;

[0015] Step S6: All categories together form a clustering result, and clustering is completed.

[0016] In some optional embodiments, the absolute position is the position of the target in the earth coordinate system; and, for any target, converting its position information in the image to the earth coordinate system includes:

[0017] according to Calculate the target's azimuth angle h in the carrier's polar coordinate system b , pitch angle v b and distance ρ b, get the target carrier polar coordinates (h b ,v b ,ρ b ); h0, v0 and ρ0 represent the azimuth, pitch and distance of the detector field of view center, α and β represent the field of view angles of image height and width, M and N represent the number of pixels of image height and width, Δr and Δc represent the displacement of the target position in the image relative to the center of the image in the height and width directions, respectively;

[0018] according to Convert the target from the carrier polar coordinate system to the carrier rectangular coordinate system to obtain the target coordinate x in the carrier rectangular coordinate system b =(x b ,y b ,z b );

[0019] according to Convert the target position from the carrier rectangular coordinate system to the geographic coordinate system to obtain the target coordinate x in the geographic coordinate system g =(x g ,y g ,z g ); is the transformation matrix from the carrier coordinate system to the geographic coordinate system;

[0020] according to Convert the target position from the geographic coordinate system to the earth coordinate system to obtain the coordinate x of the target in the earth coordinate system e =(x e ,y e ,z e ) and use it as the absolute position of the target; It is the transformation matrix from geographic coordinate system to earth coordinate system.

[0021] In some optional embodiments, the absolute position is the position of the target in the Earth's polar coordinate system; and, for any target, converting its position information in the image into the Earth's polar coordinate system includes:

[0022] according to Calculate the target's azimuth angle h in the carrier's polar coordinate system b , pitch angle v b and distance ρ b , get the target carrier polar coordinates (h b ,v b ,ρ b); h0, v0 and ρ0 represent the azimuth, pitch and distance of the detector field of view center, α and β represent the field of view angles of image height and width, M and N represent the number of pixels of image height and width, Δr and Δc represent the displacement of the target position in the image relative to the center of the image in the height and width directions, respectively;

[0023] according to Convert the target from the carrier polar coordinate system to the carrier rectangular coordinate system to obtain the target coordinate x in the carrier rectangular coordinate system b =(x b ,y b ,z b );

[0024] according to Convert the target position from the carrier rectangular coordinate system to the geographic coordinate system to obtain the target coordinate x in the geographic coordinate system g =(x g ,y g ,z g ); is the transformation matrix from the carrier coordinate system to the geographic coordinate system;

[0025] according to Convert the target position from the geographic coordinate system to the earth coordinate system to obtain the coordinate x of the target in the earth coordinate system e =(x e ,y e ,z e ); is the transformation matrix from geographic coordinate system to earth coordinate system;

[0026] The coordinates of the target in the earth coordinate system are converted to the earth polar coordinate system as the absolute position of the target.

[0027] Furthermore, the scanned imaging small target association method provided by the present invention further includes: after obtaining the absolute position of each target in the current frame image, caching these absolute positions in a preset memory.

[0028] According to another aspect of the present invention, a computer program product is provided, comprising a computer program; when the computer program is executed by a processor, the above-mentioned scanning imaging small target association method provided by the present invention is implemented.

[0029] According to another aspect of the present invention, a computer-readable storage medium is provided, comprising a stored computer program; when the computer program is executed by a processor, the above-mentioned scanning imaging small target association method provided by the present invention is implemented.

[0030] According to another aspect of the present invention, a scanning imaging small target association device is provided, comprising a computer-readable storage medium and a processor; the computer-readable storage medium is used to store a computer program; the processor is used to read the computer program stored in the computer-readable storage medium to implement the above-mentioned computer-readable storage medium provided by the present invention.

[0031] According to another aspect of the present invention, a scanning imaging detection system is provided, comprising: a detector, and the above-mentioned scanning imaging small target association device provided by the present invention.

[0032] In general, the above technical solutions conceived by the present invention can achieve the following beneficial effects:

[0033] After acquiring the position of each target in the current frame image through target detection, the present invention converts it into the earth coordinate system or the earth polar coordinate system as the absolute position of the target. The absolute position thus obtained is unique and is not affected by the posture of the detector. Further, according to the motion characteristics of the target and the carrier in the process of movement, the motions made are almost linear motions. A clustering method based on density and direction is proposed, and the absolute positions of the targets in the current frame image and the absolute positions of the targets detected before are clustered based on the clustering method. While paying attention to the density accessibility between the data, the directional accessibility between the data is also paid attention to, ensuring that the angle between the core objects in the same category does not exceed the preset angle threshold. Finally, based on the motion characteristics of the target, the association of weak targets in the scanning scene can be accurately and effectively achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 It is a schematic diagram of the frame image sequence detected by the detector in the staring scene;

[0035] Figure 2 It is a schematic diagram of a frame image sequence detected by the detector in a scanning scene;

[0036] Figure 3 A flow chart of a method for associating small targets in scanning imaging provided by an embodiment of the present invention;

[0037] Figure 4 This is a schematic diagram of the earth coordinate system;

[0038] Figure 5 This is a schematic diagram of the geographic coordinate system;

[0039] Figure 6 Schematic diagram of the carrier coordinate system;

[0040] Figure 7 A schematic diagram of the position of a target in a frame image provided by an embodiment of the present invention;

[0041] Figure 8Schematic diagram of a density- and direction-based clustering method provided in an embodiment of the present invention; wherein (a) is a noise recognition schematic diagram, and (b) is a directional accessibility schematic diagram;

[0042] Fig. 9 Schematic diagrams of clustering results of different clustering methods provided in embodiments of the present invention; wherein (a) is a schematic diagram of clustering results of the DBSCAN clustering method, and (b) is a schematic diagram of clustering results of the density- and direction-based clustering method;

[0043] Fig.10 Frame images detected by the detector in a scanning scene in an embodiment of the present invention; wherein (a) to (f) are 6 frames of images detected successively;

[0044] Fig.11 This is a schematic diagram of clustering results when no noise is added according to an embodiment of the present invention;

[0045] Fig.12 This is a schematic diagram of clustering results after adding noise 1 in an embodiment of the present invention;

[0046] Fig.13 This is a schematic diagram of clustering results after adding noise 2 according to an embodiment of the present invention. DETAILED DESCRIPTION

[0047] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0048] In the present invention, the terms "first", "second", etc. (if any) in the present invention and the accompanying drawings are used to distinguish similar objects but not necessarily to describe a specific order or sequence.

[0049] In order to solve the technical problem that the existing small target association method cannot effectively realize small target association in a scanning scene, the present invention provides a small target association method, device and system for scanning imaging, and the overall idea is to convert the position information of the detected target in the image into the earth coordinate system or the earth polar coordinate system, so that the target has unique absolute position information, thereby eliminating the influence of the scanning scene on the target motion trajectory, and further combining the motion characteristics of the target, a density and direction based clustering method is proposed to cluster the absolute position of the target, so as to realize accurate and effective small target association.

[0050] The density and direction based clustering method proposed in the present invention is a new clustering method obtained by adding a directional index based on the motion characteristics of the target on the basis of the traditional density based clustering method DBSCAN. Before explaining the technical solution of the present invention in detail, the DBSCAN clustering method and related concepts are briefly described as follows:

[0051] The DBSCAN algorithm is based on density clustering. Its core idea is to continuously search for data points that meet the requirements in a search area with a fixed radius. When there are no points that meet the requirements, a new class is generated. The search continues until no new classes are generated. There are two hyperparameters involved: Eps and MinPts. Ep is the radius of the search area, and MinPts represents the minimum number of data in a neighborhood that is judged to be of the same class. There are some key concepts in DBSCAN:

[0052] E neighborhood: the area within a given Eps radius;

[0053] Core object: data whose number in the neighborhood of E is greater than MinPts;

[0054] Direct density reachability: The core object can directly reach all data in its E neighborhood through density; direct density reachability is not symmetric because the data in the E neighborhood of the core object is not necessarily the core object;

[0055] Density reachability: Direct density reachability is transitive. If two data can be connected through several core objects, they are density reachable. Density reachability is not symmetrical because direct density reachability is not symmetrical.

[0056] Density connection: Two data in the E neighborhood of the same core object but not directly density-reachable are called density connected, and density connection has symmetry.

[0057] The process of the DBSCAN algorithm can be described by the pseudo code shown in Table 1 and Table 2. Specifically, for the data set D to be clustered, the objects in it have not been visited at the initial time. Traverse the objects in the data set D. If the object d i If the object d has not been visited, the number of data in its neighborhood E is determined according to the area radius Eps. If it is greater than or equal to the number threshold MinPts, the object d is i Mark as a core object and save the object d i The data in the E neighborhood of the cluster are divided into the current k-th category and are expanded by ExpandCluster (D i ,Eps,MinPts,k) function completes the expansion of the current k-th category. ExpandCluster(D i,Eps,MinPts,k) function further traverses the objects in the current k-th category. If it has not been visited and is a core object, the data in the E neighborhood of the object is further added to the current k-th category, and the objects in the current k-th category are traversed again until no new objects can be added to the current k-th category. After that, the category number is increased by 1, and the next category is determined in the same way.

[0058] Table 1

[0059]

[0060] Table 2

[0061]

[0062] When clustering data in the present invention, the terms used have the same meanings as those mentioned above, and in the final clustering results, those belonging to the same category are related targets.

[0063] In view of the problem that the positions of the targets in the images are not the same and there is no continuous trajectory in the scanning scene, the present invention will first convert the position information of the target in the image to the earth coordinate system or the earth polar coordinate system before clustering, so that the position information of the target is unique. It is easy to understand that the coordinate conversion can be completed based on the servo information provided by the servo system of the detector and the inertial navigation information provided by the inertial navigation system. When the position of the target is converted to the earth coordinate system, the target will have unique three-dimensional coordinate information. When the position of the target is converted to the earth polar coordinate system, the target will have unique azimuth and pitch angle information. Considering that it is relatively simple to calculate the azimuth and pitch angle when clustering, therefore, without loss of generality, in the following embodiments, the earth polar coordinate system is used as an example for explanation.

[0064] The following are examples.

[0065] Embodiment 1:

[0066] A scanning imaging weak and small target association method is used to associate weak and small targets in a frame image sequence detected by a detector in a scanning scene.

[0067] like Figure 3 As shown, the scanned small target association method provided in this embodiment includes:

[0068] Perform target detection on the current frame image, obtain the position of each target in the image, and then convert it to the earth coordinate system or the earth polar coordinate system to obtain the absolute position of each target;

[0069] Density- and direction-based clustering is performed on a data set D consisting of the absolute positions of each target in the current frame image and the absolute positions of all previously detected targets, and targets belonging to the same category in the clustering results are determined as related targets.

[0070] In this embodiment, any target detection method based on local contrast, background estimation, principal component analysis, etc. can be used to complete target detection. After obtaining the position of each target in the image, the position can be converted to the earth's polar coordinate system according to the servo information provided by the servo system of the detector, which involves multiple navigation coordinate systems and the conversion between them. Figure 4 is the Earth coordinate system, referred to as the e system. Figure 5 It is the geographic coordinate system, referred to as the G system. Figure 6 is the carrier coordinate system, referred to as b system. The specific process of coordinate transformation includes:

[0071] First, the coordinates of the target in the image are converted to the carrier polar coordinate system to obtain the carrier polar coordinates of the target (h b ,v b ,ρ b ), the relevant calculation expressions are as follows:

[0072]

[0073] ρ b =ρ0 (3)

[0074] like Figure 7 As shown, h b 、v b and ρ b Respectively represent the azimuth, pitch angle and distance of the target in the carrier polar coordinate system, h0, v0 and ρ0 represent the azimuth, pitch angle and distance of the detector field of view center, α and β represent the field of view of the image height and width, M and N represent the number of pixels of the image height and width, (r0, c0) represent the image center coordinates, (r, c) represent the detected target position, Δr and Δc represent the displacement of the target position in the image relative to the image center in the height and width directions, respectively;

[0075] After that, the target is converted from the carrier polar coordinate system to the carrier rectangular coordinate system to obtain the target coordinate x in the carrier rectangular coordinate system. b =(x b ,y b ,z b ), the relevant calculation expressions are as follows:

[0076] x b =ρ b ×cosv b ×sinh b (4)

[0077] y b =ρ b ×cosv b ×cosh b (5)

[0078] z b =ρ b ×sinv b (6)

[0079] Next, the target position is converted from the carrier rectangular coordinate system to the geographic coordinate system to obtain the coordinate x of the target in the geographic coordinate system. g =(x g ,y g ,z g ), the relevant calculation expressions are as follows:

[0080]

[0081] in, is the transformation matrix from the carrier coordinate system to the geographic coordinate system, and its expression is:

[0082]

[0083] The angle variable in the formula is Figure 6 As shown, ψ represents the pitch angle, θ represents the azimuth angle, and γ represents the roll angle.

[0084] Furthermore, the target position is converted from the geographic coordinate system to the earth coordinate system to obtain the coordinate x of the target in the earth coordinate system. e =(x e ,y e ,z e ), the relevant calculation expressions are as follows:

[0085]

[0086] in, is the transformation matrix from geographic coordinate system to earth coordinate system, and its expression is:

[0087]

[0088] The angle variable in the formula is Figure 5 As shown, where λ represents the angle between the longitude line where the carrier is located and the prime meridian, It indicates the angle between the latitude line of the carrier, the line connecting the center of the sphere and the equator.

[0089] Finally, the position information of the target in the earth coordinate system is converted to the earth polar coordinate system to obtain the direction angle and pitch angle of the target in the earth polar coordinate system. These information together constitute the absolute position of the target.

[0090] In the actual detection process, the movement of the target and the carrier is similar to linear movement, so the angle between the adjacent positions is not too large. Therefore, when realizing the association of weak targets through clustering, it is necessary to consider both density accessibility and direction accessibility. In this embodiment, the clustering based on density and direction adds a direction indicator on the basis of the traditional DBSCAN clustering method, which specifically includes:

[0091] Step S1: Initialize category label k=1;

[0092] Step S2: sequentially access the unaccessed objects in the data set D until there are no unaccessed objects in the data set D, or the currently accessed object d0 is a core object;

[0093] Step S3: If there is no unvisited object in the data set D, go to step S6; if the currently visited object d0 is a core object, mark the category labels of the objects in the E neighborhood of object d0 as k, and go to step S4;

[0094] Step S4: For each core object d in the E neighborhood of object d0 i , and its E neighborhood satisfies and Angle θ i <θ th Object d j The category label is marked as k; θ th is a preset angle threshold. In practical applications, it can be set according to specific motion parameters of the carrier and the target. Optionally, in this embodiment, θ th is set to 40°;

[0095] Step S5: After updating the category number k according to k=k+1, go to step S2;

[0096] Step S6: All categories together form a clustering result, and clustering is completed.

[0097] Table 3

[0098]

[0099] The overall process of the density and direction-based clustering method provided in this embodiment is similar to the process shown in Table 1, except that this embodiment improves the ExpandCluster() function so that it can fully consider directional accessibility when expanding data in a category. The process of the improved ExpandCluster() function in this embodiment can be described by the pseudo code shown in Table 3.

[0100] Since this embodiment adds the consideration of directional reachability on the basis of the DBSCAN algorithm, noise points can be effectively identified during the clustering process, such as Figure 8 Specifically, when considering the data in the density accessibility and direction accessibility expansion categories at the same time, this embodiment will not perform clustering iterations for data points that do not have obvious directionality, such as Figure 8 As shown in (b) in .

[0101] Further Fig. 9 As shown in Figure 1, the dataset originally containing three targets is divided into one category after using density-based reachable clustering. There are obvious errors, such as Fig. 9 As shown in (a) in , after adding the direction reachability indicator, it is divided into three categories, such as Fig. 9 As shown in (b), this shows that after adding the directional reachability indicator, this embodiment can mine the directionality in the data that meets the density requirement, which significantly increases the accuracy of target association.

[0102] Fig.10 As shown in the figure, when the detector detects two targets in the scanning scene, the detector detects 6 frames of images successively, as shown in the figure. Fig.10 As shown in (a) to (f) in FIG. 1 , in each frame image, the carrier azimuth and elevation angle of the center of the field of view, the target offset relative to the center of the field of view, and the converted azimuth and elevation angle of the target in the Earth's polar coordinate system are shown as follows: Figure 4 shown.

[0103] Table 4

[0104]

[0105] Based on the density and direction based clustering method provided in this embodiment, clustering is performed, and finally the two targets can be accurately detected and the association between the targets can be achieved, such as Fig.11 To further verify its effectiveness, this embodiment Fig.10 After adding different degrees of Gaussian noise to the frame images shown in the figure, target association is performed and the association results are shown in Figure 2. Fig.12 and Fig.13 shown. Fig.12 and Fig.13 The shown association results all indicate that this embodiment can effectively identify noise, and can also effectively identify noise that is close to the target, and ultimately accurately cluster different targets.

[0106] After the current frame image is processed, the next frame image can be processed based on the same process until the frame image sequence detected by the detector is processed. Considering that when clustering the current frame, it is necessary to obtain the target position information detected previously, in order to facilitate the acquisition of this information, as a preferred implementation, this embodiment also includes: after obtaining the absolute position of each target in the current frame image, these absolute positions are cached in a preset memory for subsequent access.

[0107] The present invention decouples the offset of the target between consecutive frames caused by the scanning motion of the detector by combining the attitude information of the inertial navigation system, projects the position of the target into the earth coordinate system, and clusters the projected coordinate points using the DBSCAN clustering method with the added directional reachability property, which can accurately and effectively associate the targets detected in different scanning cycles.

[0108] Embodiment 2:

[0109] A computer program product includes a computer program. When the computer program is executed by a processor, the method for associating small targets in scanning imaging provided in the above-mentioned embodiment 1 is implemented.

[0110] Embodiment 3:

[0111] A computer-readable storage medium includes a stored computer program; when the computer program is executed by a processor, the method for associating small targets in scanning imaging provided in the above-mentioned embodiment 1 is implemented.

[0112] Embodiment 4:

[0113] A scanning imaging small target association device includes a computer-readable storage medium and a processor; the computer-readable storage medium is used to store a computer program; the processor is used to read the computer program stored in the computer-readable storage medium to implement the computer-readable storage medium provided in the above-mentioned embodiment 1.

[0114] Embodiment 5:

[0115] A scanning imaging detection system includes: a detector, and the scanning imaging small target association device provided in the above-mentioned embodiment 4.

[0116] It will be easily understood by those skilled in the art that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for associating small targets in scanning imaging, used to associate small targets in a frame image sequence detected by a detector in a scanning scene; characterized in that: The method comprises: Perform target detection on the current frame image, obtain the position of each target in the image, and then convert it to the earth coordinate system or the earth polar coordinate system to obtain the absolute position of each target; Perform density and direction based clustering on the data set D consisting of the absolute positions of each target in the current frame image and the absolute positions of all previously detected targets, and determine the targets belonging to the same category in the clustering results as related targets; the density and direction based clustering includes: Step S1: Initialize category label k=1; Step S2: sequentially accessing the unaccessed objects in the data set D until there are no unaccessed objects in the data set D, or the currently accessed object d0 is a core object; Step S3: If there is no unvisited object in the data set D, proceed to step S6; if the currently visited object d0 is a core object, the category labels of the objects in the E neighborhood of the object d0 are all marked as k, and proceed to step S4; Step S4: For each core object d in the E neighborhood of object d0 i , and its E neighborhood satisfies and Angle θ i <θ th Object d j The category label is marked as k; θ th is the preset angle threshold; Step S5: After updating the category number k according to k=k+1, go to step S2; Step S6: All categories together form a clustering result, and clustering is completed.

2. The scanned small target association method according to claim 1, characterized in that: The absolute position is the position of the target in the earth coordinate system; and, for any target, the position information of the target in the image is converted to the earth coordinate system, including: according to Calculate the target's azimuth angle h in the carrier's polar coordinate system b , pitch angle v b and distance ρ b , get the target carrier polar coordinates (h b ,v b ,ρ b ); h0, v0 and ρ0 represent the azimuth, pitch and distance of the detector field of view center, α and β represent the field of view angles of image height and width, M and N represent the number of pixels of image height and width, Δr and Δc represent the displacement of the target position in the image relative to the center of the image in the height and width directions, respectively; according to Convert the target from the carrier polar coordinate system to the carrier rectangular coordinate system to obtain the target coordinate x in the carrier rectangular coordinate system b =(x b ,y b ,z b ); according to Convert the target position from the carrier rectangular coordinate system to the geographic coordinate system to obtain the target coordinate x in the geographic coordinate system g =(x g ,y g ,z g ); is the transformation matrix from the carrier coordinate system to the geographic coordinate system; according to Convert the target position from the geographic coordinate system to the earth coordinate system to obtain the coordinate x of the target in the earth coordinate system e =(x e ,y e ,z e ) and use it as the absolute position of the target; It is the transformation matrix from geographic coordinate system to earth coordinate system.

3. The scanned small target association method according to claim 1, characterized in that: The absolute position is the position of the target in the Earth's polar coordinate system; and, for any target, converting its position information in the image to the Earth's polar coordinate system includes: according to Calculate the target's azimuth angle h in the carrier's polar coordinate system b , pitch angle v b and distance ρ b , get the target carrier polar coordinates (h b ,v b ,ρ b ); h0, v0 and ρ0 represent the azimuth, pitch and distance of the detector field of view center, α and β represent the field of view angles of image height and width, M and N represent the number of pixels of image height and width, Δr and Δc represent the displacement of the target position in the image relative to the center of the image in the height and width directions, respectively; according to Convert the target from the carrier polar coordinate system to the carrier rectangular coordinate system to obtain the target coordinate x in the carrier rectangular coordinate system b =(x b ,y b ,z b ); according to Convert the target position from the carrier rectangular coordinate system to the geographic coordinate system to obtain the target coordinate x in the geographic coordinate system g =(x g ,y g ,z g ); is the transformation matrix from the carrier coordinate system to the geographic coordinate system; according to Convert the target position from the geographic coordinate system to the earth coordinate system to obtain the coordinate x of the target in the earth coordinate system e =(x e ,y e ,z e ); is the transformation matrix from geographic coordinate system to earth coordinate system; The coordinates of the target in the earth coordinate system are converted to the earth polar coordinate system as the absolute position of the target.

4. The method for associating small targets in scanning imaging according to any one of claims 1 to 3, characterized in that: Also includes: After obtaining the absolute positions of each target in the current frame image, these absolute positions are cached in a preset memory.

5. A computer program product, characterized in that It comprises a computer program; when the computer program is executed by a processor, it implements the scan imaging small target association method as described in any one of claims 1 to 4.

6. A computer-readable storage medium, characterized in that: The invention comprises a stored computer program; when the computer program is executed by a processor, the method for associating small targets in scanning imaging as described in any one of claims 1 to 4 is implemented.

7. A scanning imaging small target association device, characterized in that: The invention comprises a computer-readable storage medium and a processor; the computer-readable storage medium is used to store a computer program; the processor is used to read the computer program stored in the computer-readable storage medium, and implement the computer-readable storage medium according to any one of claims 1 to 4.

8. A scanning imaging detection system, characterized in that: include: A detector and a small target association device for scanning imaging as described in claim 7.

Citation Information

Patent Citations

  • Flight path association method and electronic device

    CN107063259A

  • Underwater small target dynamic fine detection three-dimensional imaging device and method

    CN111090103A

  • Density-based fuzzy clustering method and device

    CN115423019A

  • Method and system for improved unresolved target detection using multiple frame association

    US20050185824A1