A method and apparatus for radar echo identification of low-speed, small targets based on optical flow.
By using a radar echo identification method based on optical flow, multi-frame images are generated in real time and analyzed hierarchically to identify and capture low, slow, and small targets. This solves the problem of insufficient radar detection of low, slow, and small targets in ship surveillance, and improves target acquisition efficiency and management accuracy.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- ZHEJIANG SOS TECH
- Filing Date
- 2023-01-05
- Publication Date
- 2026-05-26
AI Technical Summary
Radar's ability to identify low, slow, and small targets in ship surveillance is insufficient, resulting in low target acquisition efficiency. Furthermore, ARPA functionality cannot effectively acquire small targets, leading to inaccurate management.
A radar echo identification method for low, slow, and small targets based on optical flow is adopted. By receiving radar echo data in real time to form multi-frame images, the pyramid LK optical flow method is used for hierarchical analysis to identify dynamic and static targets, and unidentified targets are judged to be low, slow, and small targets. The latitude and longitude of the targets are calculated by combining the radar position.
It improves the detection and acquisition efficiency of low, slow, and small targets, solves the problems of area and computing power limitations, and enhances the accuracy of target management.
Smart Images
Figure CN116047451B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of radar technology, and in particular to a method and apparatus for identifying radar echoes of low-speed, small targets based on optical flow. Background Technology
[0002] When radar is used for ship surveillance, the acquired echo images often contain complex moving and static targets. These echo images need to be combined with data changes from nautical charts and radar images to roughly determine the ship's position, resulting in insufficient accuracy. Furthermore, due to the limitations of radar's ability to identify low-speed, small targets, the radar's ARPA function often fails to detect some small moving targets, leading to omissions and inaccuracies in management. Therefore, it is necessary to provide a method to enhance the detection capability of low-speed, small targets and improve ship surveillance. Summary of the Invention
[0003] The technical problem to be solved by the embodiments of the present invention is to provide a method and device for radar echo identification of low, slow and small targets based on optical flow method, which solves the shortcomings of insufficient radar detection capability for low, slow and small targets and improves target acquisition efficiency.
[0004] To address the aforementioned technical problems, embodiments of the present invention provide a method for identifying radar echoes of low-speed, small targets based on optical flow, which is implemented on a device interconnected with a radar. The method includes the following steps:
[0005] The system receives radar echo data in real time and redraws the images to form continuous multi-frame radar echo images; wherein, the radar echo data includes radar reflection data of all dynamic and static targets within a predetermined range; the dynamic targets include dynamic large targets and dynamic small targets;
[0006] Using the pyramid LK optical flow method, the generated multi-frame radar echo images are sequentially analyzed in layers to obtain the optical flow vector of each radar echo image.
[0007] By comparing the optical flow vectors of each frame of radar echo images, dynamic and static targets in each frame of radar echo images can be determined.
[0008] Based on the dynamic and static target recognition results fed back by the radar, it is determined whether there is one or more dynamic targets in each frame of radar echo image that have not been recognized by the radar. If so, the dynamic targets that have not been recognized by the radar in each frame of radar echo image are identified as low, slow and small targets and are captured.
[0009] The method further includes:
[0010] Based on the optical flow vectors of each radar echo image, the movement trajectories of each low, slow, and small target are delineated, and the relative azimuth and distance between each low, slow, and small target and the radar in each radar echo image are obtained. Furthermore, combined with the known latitude and longitude of the radar's current fixed location, the latitude and longitude of each low, slow, and small target are calculated.
[0011] The radar is a pulse radar, which acquires radar reflection data of all moving and static targets within the predetermined range by adjusting the range, rain clutter, and sea clutter.
[0012] The specific steps of using the pyramid LK optical flow method to sequentially perform layered analysis on the formed multi-frame radar echo images to obtain the optical flow vector of each frame of radar echo image include:
[0013] All the generated multi-frame radar echo images were smoothed using Gaussian filtering.
[0014] The smoothed radar echo images are sequentially layered, and a predetermined pixel scaling standard is applied to the radar echo images from the second to the bottom layer to form a pyramid structure with the fewest pixels at the top layer and the most pixels at the bottom layer. The pixel scaling standard is to reduce the pixel size of the current layer image by a factor of N based on the pixel size of the previous layer image.
[0015] The optical flow vectors of all moving and static targets in the top-level radar echo image are calculated using the Taylor formula. Based on these vectors, the optical flow vectors of all moving and static targets in the second to the bottom-level radar echo images are iteratively calculated. In each iteration, the optical flow vectors of all moving and static targets in the previous layer are multiplied by N as the initial value for the current layer. The Taylor formula is then used to obtain the optical flow deviation of all moving and static targets in the current layer, so that the optical flow vector of the current layer is equal to N times the optical flow vector of the previous layer plus the calculated optical flow deviation.
[0016] After the iteration is complete, the optical flow vector of the underlying radar echo image is output.
[0017] The formula Dn-1 = N(Dn + dn) + dn-1 is used to obtain the optical flow vectors and optical flow deviations of all moving and static targets in the radar echo images from the second to the bottom layer. Here, Dn is the optical flow vector of a moving or static target T in the previous layer image; Dn-1 is the optical flow vector of the moving or static target T in the current layer image; dn is the optical flow deviation of the moving or static target T in the previous layer image; dn-1 is the optical flow deviation of the moving or static target T in the current layer image; and N is the pixel reduction factor during layering.
[0018] The dynamic and static targets in each frame of radar echo image are identified by comparing the difference between the optical flow vectors of moving points and stationary points in each frame of radar echo image.
[0019] This invention also provides a device connected to a radar, comprising:
[0020] The radar echo image rendering unit is used to receive the radar echo data in real time and redraw the image to form a continuous multi-frame radar echo image; wherein, the radar echo data includes radar reflection data of all dynamic and static targets within a predetermined range; the dynamic targets include dynamic large targets and dynamic small targets.
[0021] The optical flow vector layering calculation unit is used to perform layered analysis on the formed multi-frame radar echo images in sequence using the pyramid LK optical flow method to obtain the optical flow vector of each frame of radar echo image.
[0022] The dynamic and static target identification unit is used to compare the optical flow vectors of each frame of radar echo images to determine the dynamic and static targets in each frame of radar echo images;
[0023] The low-slow-small target identification and acquisition unit is used to determine, based on the dynamic and static target identification results fed back by the radar, whether there is one or more dynamic targets in each frame of radar echo image that have not been identified by the radar. If so, the unit determines that the dynamic targets not identified by the radar in each frame of radar echo image are all low-slow-small targets and acquires them.
[0024] This also includes:
[0025] The low-speed-small target latitude and longitude calculation unit is used to outline the movement trajectory of each low-speed-small target based on the optical flow vector of each frame of radar echo image, and to obtain the relative azimuth and distance between each low-speed-small target and the radar in each frame of radar echo image. Furthermore, it combines the known latitude and longitude of the radar's current fixed position to calculate the latitude and longitude of each low-speed-small target.
[0026] The radar is a pulse radar, which acquires radar reflection data of all moving and static targets within the predetermined range by adjusting the range, rain clutter, and sea clutter.
[0027] Implementing the embodiments of the present invention has the following beneficial effects:
[0028] This invention redraws radar echo data into a continuous multi-frame radar echo image and uses the pyramid LK optical flow method to calculate the optical flow vector layer by layer. This can more accurately detect moving targets, solve the problems of regional limitation and radar local computing power limitation, thereby overcoming the shortcomings of radar in detecting low, slow and small targets and improving target acquisition efficiency. Attached Figure Description
[0029] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, obtaining other drawings based on these drawings without creative effort still falls within the scope of the present invention.
[0030] Figure 1 A flowchart illustrating a method for identifying radar echoes of small, slow targets based on optical flow, provided in an embodiment of the present invention;
[0031] Figure 2 This is a schematic diagram of the structure of a device provided in an embodiment of the present invention. Detailed Implementation
[0032] To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings.
[0033] like Figure 1 As shown in the figure, an embodiment of the present invention provides a radar echo identification method for low-speed, small targets based on optical flow, which is implemented on a device interconnected with a radar. The method includes the following steps:
[0034] Step S1: Receive the radar echo data in real time and redraw the image to form a continuous multi-frame radar echo image; wherein, the radar echo data includes radar reflection data of all dynamic and static targets within a predetermined range; the dynamic targets include dynamic large targets and dynamic small targets;
[0035] Step S3: Using the pyramid LK optical flow method, perform layered analysis on the formed multi-frame radar echo images in sequence to obtain the optical flow vector of each frame of radar echo image.
[0036] Step S3: Compare the optical flow vectors of each frame of radar echo images to determine the dynamic and static targets in each frame of radar echo images;
[0037] Step S4: Based on the dynamic and static target recognition results fed back by the radar, determine whether there is one or more dynamic targets in each frame of radar echo image that have not been recognized by the radar. If so, determine that the dynamic targets that have not been recognized by the radar in each frame of radar echo image are all low, slow and small targets and capture them.
[0038] The specific process is as follows: Before step S1, a pulse radar with latitude and longitude (X0, Y0) is fixed, its range information is set, and parameters such as rain clutter and sea clutter are configured. Then, radar echo data is acquired within a radius of the radar's center and a predetermined range. This data includes radar reflection data of all static and dynamic targets within this range (e.g., radar reflection data of static targets such as ports and hills, and radar reflection data of dynamic targets such as large ships and slow, small ships). At this point, the radar's ARPA technology will capture static targets and large dynamic targets (i.e., static and dynamic target identification results) and send them to the device. However, slow, small targets may be missed due to the radar's inherent performance limitations.
[0039] In step S1, the device receives radar echo data in real time and redraws the image to form a continuous multi-frame radar echo image. Each frame of the radar echo image contains radar reflection data of all moving and static targets, such as ports, hills, large ships, and low-speed, small targets.
[0040] Now assume that a small, slow target A is scanned by radar and its coordinate position is A(X1,Y1). In the next frame of radar echo image, the position of A changes as (X1+Δx, Y1+Δy), where Δx and Δy are the changes of target A on the X-axis and Y-axis, respectively.
[0041] In step S2, the pyramid LK optical flow method is used to sequentially perform layered analysis on multiple frames of radar echo images to obtain the optical flow vector of each frame of radar echo image. The specific steps include:
[0042] (1) All the generated multi-frame radar echo images are smoothed by Gaussian filtering.
[0043] (2) The smoothed radar echo images are sequentially layered, and a predetermined pixel scaling standard is applied to the radar echo images from the second to the bottom layer to form a pyramid structure with the fewest pixels at the top and the most pixels at the bottom. The pixel scaling standard is to reduce the pixel size of the current layer image by a factor of N (e.g., each reduction is a factor of one time) based on the pixel size of the previous layer image. This reduces the velocity vector of each optical flow in each radar echo image proportionally, so that the movement in each radar echo image meets the "small motion" restriction requirement of the optical flow method. This is to solve the problem that the target movement range in adjacent radar echo images may be too large due to the problem of radar mechanical scanning rate, which would not meet the "small motion" calculation principle of the optical flow method. It should be noted that all moving and static targets in the radar echo image can be represented by optical flow points on the image.
[0044] (3) The optical flow vectors of all moving and static targets in the top-level radar echo image are calculated using the Taylor formula. Based on the optical flow vectors of all moving and static targets in the top-level radar echo image, the optical flow vectors and optical flow deviations of all moving and static targets in the radar echo images from the second layer to the bottom layer are iteratively calculated. In each iteration, the optical flow vectors of all moving and static targets in the previous layer image are multiplied by N as the initial value of the current layer. Then, the optical flow deviations of all moving and static targets in the current layer image are obtained using the Taylor formula. The optical flow vector of the current layer is equal to N times the optical flow vector of the previous layer plus the calculated optical flow deviation.
[0045] For example, the optical flow vectors and optical flow deviations of all moving and static targets in the radar echo images from the second to the bottom layer can be obtained using the formula Dn-1=N(Dn+dn)+dn-1; where Dn is the optical flow vector of a certain moving or static target T in the previous layer image; Dn-1 is the optical flow vector of the moving or static target T in the current layer image; dn is the optical flow deviation of the moving or static target T in the previous layer image, dn-1 is the optical flow deviation of the moving or static target T in the current layer image, and N is the pixel reduction factor when layering;
[0046] It should be noted that due to the pyramid-style image scaling, the optical flow vector of the top layer image with the smallest pixel approaches zero. By using the principle of consistent optical flow vectors in the neighborhood, N times the optical flow value of this layer is calculated as the initial value of the next layer (the previous scaling layer with the largest scaling ratio). Then, the optical flow value of this layer is calculated to obtain its residual value. Thus, the optical flow value of the moving target in this layer is the top layer optical flow value plus N times the residual value. Similarly, by analogy, it can be deduced that the optical flow value of the moving target at the bottom layer with the largest pixel is the sum of the optical flow values of each layer.
[0047] (4) After the iteration is completed, output the optical flow vector of the bottom radar echo image.
[0048] In step S3, the optical flow vectors of each radar echo image are compared to determine the dynamic and static targets in each radar echo image. That is, by using the optical flow vectors of all dynamic and static targets in the obtained bottom-layer radar echo image, the differences between the optical flow vectors of moving points and stationary points in each radar echo image are compared to identify which points are moving points. By performing optical flow calculations on more radar echo images, the changes in optical flow vectors in each image can be obtained.
[0049] In step S4, the radar's ARPA technology only captures static targets and large dynamic targets (i.e., static and dynamic target identification results) and sends them to the device. Based on the identification results fed back by the radar, the device determines whether there is one or more dynamic targets in each frame of radar echo image that have not been identified by the radar. If so, the device considers the unidentified dynamic targets in each frame of radar echo image to be low, slow and small targets and captures them, thereby improving the overall system's ability to capture small targets and greatly improving the overall system's management capabilities.
[0050] It should be noted that the optical flow method used in this invention requires the prerequisites of "invariant grayscale" and "small motion". "Invariant grayscale" means that the grayscale of the object does not change when the image moves between different frames, so the changes of the object can be identified by relying on the grayscale. The principle of radar is to reflect the scanned 3D object onto 2D data. The data of the object itself has been reduced in dimensionality, so there is no requirement for the grayscale of the object itself. "Small motion" means that the change of time will not cause a drastic change in the position of the target in the image. The detection of low, slow and small targets, coupled with the scanning cycle rate of radar, ensures that small changes of moving targets are detected. It can be seen that the combination of radar and optical flow technologies can solve their technical defects and achieve better results.
[0051] In this embodiment of the invention, the method further includes:
[0052] Based on the optical flow vectors of each radar echo image, the movement trajectories of each low, slow, and small target are delineated, and the relative azimuth and distance between each low, slow, and small target and the radar in each radar echo image are obtained. Furthermore, combined with the known latitude and longitude of the radar's current fixed location, the latitude and longitude of each low, slow, and small target are calculated.
[0053] like Figure 2 As shown in the illustration, an apparatus provided in an embodiment of the present invention is connected to a radar and includes:
[0054] The radar echo image rendering unit 110 is used to receive the radar echo data in real time and redraw the image to form a continuous multi-frame radar echo image; wherein, the radar echo data includes radar reflection data of all dynamic and static targets within a predetermined range; the dynamic targets include dynamic large targets and dynamic small targets.
[0055] The optical flow vector layering calculation unit 120 is used to perform layered analysis on the formed multi-frame radar echo images in sequence using the pyramid LK optical flow method to obtain the optical flow vector of each frame of radar echo image.
[0056] The dynamic and static target identification unit 130 is used to compare the optical flow vectors of each frame of radar echo images to determine the dynamic and static targets in each frame of radar echo images;
[0057] The low-slow-small target identification and acquisition unit 140 is used to determine, based on the dynamic and static target identification results fed back by the radar, whether there is one or more dynamic targets in each frame of radar echo image that have not been identified by the radar. If so, it is determined that the dynamic targets not identified by the radar in each frame of radar echo image are all low-slow-small targets and are acquired.
[0058] This also includes:
[0059] The low-speed-small target latitude and longitude calculation unit is used to outline the movement trajectory of each low-speed-small target based on the optical flow vector of each frame of radar echo image, and to obtain the relative azimuth and distance between each low-speed-small target and the radar in each frame of radar echo image. Furthermore, it combines the known latitude and longitude of the radar's current fixed position to calculate the latitude and longitude of each low-speed-small target.
[0060] The radar is a pulse radar, which acquires radar reflection data of all moving and static targets within the predetermined range by adjusting the range, rain clutter, and sea clutter.
[0061] Implementing the embodiments of the present invention has the following beneficial effects:
[0062] This invention redraws radar echo data into a continuous multi-frame radar echo image and uses the pyramid LK optical flow method to calculate the optical flow vector layer by layer. This can more accurately detect moving targets, solve the problems of regional limitation and radar local computing power limitation, thereby overcoming the shortcomings of radar in detecting low, slow and small targets and improving target acquisition efficiency.
[0063] It is worth noting that the various units included in the above system embodiments are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be achieved; in addition, the specific names of each functional unit are only for easy differentiation and are not used to limit the scope of protection of the present invention.
[0064] Those skilled in the art will understand that all or part of the steps in the methods of the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, such as ROM / RAM, disk, optical disk, etc.
[0065] The above description discloses only preferred embodiments of the present invention and should not be construed as limiting the scope of the present invention. Therefore, equivalent variations made in accordance with the claims of the present invention are still within the scope of the present invention.
Claims
1. A method for identifying radar echoes of low-speed, small targets based on optical flow, characterized in that, It is implemented on a device interconnected with a radar, and the method includes the following steps: The radar echo data is received in real time and the image is redrawn to form a continuous multi-frame radar echo image; wherein, the radar echo data includes radar reflection data of all dynamic and static targets within a predetermined range; dynamic targets include dynamic large targets and dynamic small targets; Using the pyramid LK optical flow method, each frame of the multi-frame radar echo image is analyzed in sequence to obtain the optical flow vector of each frame of the radar echo image. By comparing the optical flow vectors of each frame of radar echo images, dynamic and static targets in each frame of radar echo images can be determined. Based on the dynamic and static target recognition results fed back by the radar, it is determined whether there is one or more dynamic targets in each frame of radar echo image that have not been recognized by the radar. If so, the dynamic targets that have not been recognized by the radar in each frame of radar echo image are identified as low, slow and small targets and are captured.
2. The method for identifying low-speed, small-target radar echoes based on optical flow as described in claim 1, characterized in that, The method further includes: Based on the optical flow vectors of each radar echo image, the movement trajectories of each low, slow, and small target are delineated, and the relative azimuth and distance between each low, slow, and small target and the radar in each radar echo image are obtained. Furthermore, combined with the known latitude and longitude of the radar's current fixed location, the latitude and longitude of each low, slow, and small target are calculated.
3. The method for identifying low-speed, small-target radar echoes based on optical flow as described in claim 2, characterized in that, The radar is a pulse radar, and it acquires radar reflection data of all moving and static targets within the predetermined range by adjusting the range, rain clutter, and sea clutter.
4. The method for identifying low-speed, small-target radar echoes based on optical flow as described in claim 1, characterized in that, The specific steps of using the pyramid LK optical flow method to sequentially analyze each frame of the multi-frame radar echo image to obtain the optical flow vector of each frame include: All the generated multi-frame radar echo images were smoothed using Gaussian filtering. The smoothed radar echo images are sequentially layered, and a predetermined pixel scaling standard is applied to the radar echo images from the second to the bottom layer to form a pyramid structure with the fewest pixels at the top layer and the most pixels at the bottom layer. The pixel scaling standard is to reduce the pixel size of the current layer image by a factor of N based on the pixel size of the previous layer image. Using Taylor's formula, the optical flow vectors of all moving and static targets in the top-level radar echo image are calculated. Based on these optical flow vectors, the optical flow vectors and optical flow deviations of all moving and static targets in the second to the bottom-level radar echo images are iteratively calculated. In each iteration, the optical flow vectors of all moving and static targets in the previous layer are multiplied by N as the initial value for the current layer. Then, using Taylor's formula, the optical flow deviations of all moving and static targets in the current layer are obtained. Thus, the optical flow vector of the current layer is equal to N times the optical flow vector of the previous layer plus the calculated optical flow deviation. After the iteration is complete, the optical flow vector of the underlying radar echo image is output.
5. The method for identifying low-speed, small targets using radar echoes based on optical flow as described in claim 4, characterized in that, Through formula D n-1 = N(D n +d n )+d n-1 This yields the optical flow vectors of all moving and static targets in the radar echo images from the second to the bottom layer; where D n D is the optical flow vector of a dynamic or static target T in the previous image layer; n-1 d is the optical flow vector of the dynamic or static target T in this layer of the image; n d represents the optical flow deviation of a dynamic or static target T in the previous image layer. n-1 denoted as the optical flow deviation of the dynamic or static target T in this layer of the image, and N is the factor by which the pixels are reduced during layering.
6. The method for identifying low-speed, small targets using radar echoes based on optical flow as described in claim 5, characterized in that, The dynamic and static targets in each frame of radar echo image are identified by comparing the difference between the optical flow vectors of moving points and stationary points in each frame of radar echo image.
7. A radar echo identification device for low-speed, small targets based on optical flow, characterized in that, It is connected to radar, including: The radar echo image rendering unit is used to receive the radar echo data in real time and redraw the image to form a continuous multi-frame radar echo image; wherein, the radar echo data includes radar reflection data of all dynamic and static targets within a predetermined range; dynamic targets include dynamic large targets and dynamic small targets. The optical flow vector layering calculation unit is used to perform layered analysis on each frame of the multi-frame radar echo image in sequence using the pyramid LK optical flow method to obtain the optical flow vector of each frame of the radar echo image. The dynamic and static target identification unit is used to compare the optical flow vectors of each frame of radar echo images to determine the dynamic and static targets in each frame of radar echo images; The low-slow-small target identification and acquisition unit is used to determine, based on the dynamic and static target identification results fed back by the radar, whether there is one or more dynamic targets in each frame of radar echo image that have not been identified by the radar. If so, the unit determines that the dynamic targets not identified by the radar in each frame of radar echo image are all low-slow-small targets and acquires them.
8. The low-speed, small target radar echo identification device based on optical flow method as described in claim 7, characterized in that, Also includes: The low-speed-small target latitude and longitude calculation unit is used to outline the movement trajectory of each low-speed-small target based on the optical flow vector of each frame of radar echo image, and to obtain the relative azimuth and distance of each low-speed-small target to the radar in each frame of radar echo image. Furthermore, it combines the known latitude and longitude of the radar's current fixed position to calculate the latitude and longitude of each low-speed-small target.
9. The low-speed, small target radar echo identification device based on optical flow method as described in claim 8, characterized in that, The radar is a pulse radar, and it acquires radar reflection data of all moving and static targets within the predetermined range by adjusting the range, rain clutter, and sea clutter.