Target image processing device and target image processing program

Abstract

This disclosure aims to detect small targets separately from clutter, even when the small targets are in close proximity to clutter, when clustering cells in which the radar received power exceeds a "small received power threshold" in order to detect not only large targets but also small targets using a radar device. [Solution] In this disclosure, a Doppler velocity distribution is created for all cells in each cluster, and when the Doppler velocity distribution has multiple peaks in each cluster, the cluster is separated into multiple clusters corresponding to the multiple peaks. Specifically, in each cluster, when the Doppler velocity distribution changes velocity, passing through a first threshold, then a second threshold smaller than the first threshold, and then returning to the first threshold, the cluster is separated into multiple clusters corresponding to the multiple peaks of the Doppler velocity distribution.

Description

[Technical Field]

[0001] This disclosure relates to a technique for detecting targets separately from clutter using a radar device. [Background technology]

[0002] A technology for detecting targets such as ships by separating them from clutter such as waves using a radar system is disclosed in Patent Document 1, etc. In Patent Document 1, in the target image acquired from the radar system, cells in which the radar received power exceeds the received power threshold are clustered using DBSCAN (Density-Based Spatial Clustering of Applications with Noise), and clusters are extracted. [Prior art documents] [Patent Documents]

[0003] [Patent Document 1] Japanese Patent Publication No. 2024-086562 [Overview of the Initiative] [Problems that the invention aims to solve]

[0004] Figure 1 shows the problems that conventional cluster extraction processing methods have to solve. Patent Document 1 requires clustering cells where the radar received power exceeds a "small received power threshold" in order to detect not only large targets such as ships, but also small targets such as ships V or buoys. However, when small targets such as ships V or buoys are in close proximity to clutter such as waves W, it is not possible to detect the small targets such as ships V or buoys separately from the clutter such as waves W.

[0005] Consequently, when tracking a small vessel V or buoy or other target, the vessel V or buoy or other target becomes integrated with clutter such as waves W, causing the characteristic quantities of the tracked target (center of gravity, illumination area, maximum received power, or average Doppler velocity, etc.) to shift, resulting in lower accuracy of target tracking.

[0006] Therefore, in order to solve the aforementioned problems, this disclosure aims to detect small targets separately from clutter when clustering cells in which the radar received power exceeds a "small received power threshold" in order to detect not only large targets but also small targets using a radar device, even when the small targets are in close proximity to the clutter. [Means for solving the problem]

[0007] To solve the aforementioned problem, a Doppler velocity distribution was created for all cells in each cluster, and when the Doppler velocity distribution in each cluster had multiple peaks, the cluster was separated into multiple clusters corresponding to the multiple peaks.

[0008] Specifically, the present disclosure is a target image processing device characterized by comprising: a cluster extraction unit that clusters cells in a target image acquired from a radar device in which the radar received power exceeds a received power threshold and extracts the clusters; a velocity distribution creation unit that creates a Doppler velocity distribution for all cells in each cluster; and a cluster separation unit that, when the Doppler velocity distribution in each cluster has multiple peaks, separates each cluster into multiple clusters corresponding to the multiple peaks, and maintains each cluster without separation when the Doppler velocity distribution has a single peak.

[0009] With this configuration, even when a small target is in close proximity to the clutter, it can be detected separately from the clutter. On the other hand, even when a small target is not in close proximity to the clutter and is moving in a straight line or turning, it can be detected without excessive separation.

[0010] Further, in the present disclosure, the cluster separation unit separates each cluster into a plurality of clusters corresponding to the adjacent peaks when the Doppler velocity distribution has a deep bottom between the adjacent peaks in each cluster, and maintains each cluster without separating it into a plurality of clusters corresponding to the adjacent peaks when the Doppler velocity distribution has a shallow bottom between the adjacent peaks. The target image processing apparatus is characterized in that.

[0011] According to this configuration, even when a small target is close to clutter, the small target can be separated from the clutter and detected. On the other hand, even when a small target has a smaller irradiation area near the center than near both ends, is not close to clutter, and turns around the center, the small target can be detected without being excessively separated.

[0012] Further, in the present disclosure, the cluster separation unit separates each cluster into a plurality of clusters corresponding to the adjacent peaks when the Doppler velocity distribution has a small frequency difference between the adjacent peaks in each cluster, and maintains each cluster without separating it into a plurality of clusters corresponding to the adjacent peaks when the Doppler velocity distribution has a large frequency difference between the adjacent peaks. The target image processing apparatus is characterized in that.

[0013] According to this configuration, even when a small target is close to clutter, the small target can be separated from the clutter and detected. On the other hand, even when a small target has a smaller irradiation area near one end than near the other end, is not close to clutter, and turns around the center, the small target can be detected without being excessively separated.

[0014] In addition, in the present disclosure, the cluster separation unit separates each cluster into a plurality of clusters corresponding to the adjacent peaks when the Doppler velocity distribution has a large velocity difference between the adjacent peaks in each cluster, and maintains each cluster without separating it into a plurality of clusters corresponding to the adjacent peaks when the Doppler velocity distribution has a small velocity difference between the adjacent peaks. The present disclosure relates to a target image processing apparatus characterized by this.

[0015] According to this configuration, even when a small target is close to a large target and clutter (for example, clutter such as wake waves left by a large ship), the small target can be separated from the large target and clutter and detected. On the other hand, even when a small target has a smaller irradiation area near one end than near the other end, is not close to clutter, and rotates around the center near the center, the small target can be detected without being overly separated.

[0016] To solve the above problems, after creating a Doppler velocity distribution in all cells of each cluster, in each cluster, when the Doppler velocity distribution passes through a second threshold smaller than the first threshold after passing through the first threshold as the velocity changes and then returns to the first threshold, each cluster is separated into a plurality of clusters corresponding to the plurality of peaks of the Doppler velocity distribution.

[0017] Specifically, the present disclosure provides a target image processing apparatus including: a cluster extraction unit that clusters cells in a target image acquired from a radar device and extracts clusters when the radar reception power exceeds a reception power threshold; a velocity distribution creation unit that creates a Doppler velocity distribution in all cells of each cluster; and in each cluster, (1) when the Doppler velocity distribution passes through a second threshold smaller than the first threshold after passing through the first threshold as the velocity changes and then returns to the first threshold, each cluster is separated into a plurality of clusters corresponding to the plurality of peaks of the Doppler velocity distribution, and (2) when the Doppler velocity distribution does not show the frequency change described in (1) as the velocity changes, a cluster separation unit that maintains each cluster without separation.

[0018] With this configuration, even when a small target is in close proximity to the clutter, it can be detected separately from the clutter. On the other hand, even when a small target is not in close proximity to the clutter and is moving in a straight line or turning, it can be detected without excessive separation.

[0019] Furthermore, even when a small target has a smaller illumination area near the center compared to the areas near both ends, is not in close proximity to clutter, and rotates around the center, the small target can be detected without excessive separation.

[0020] Furthermore, the present disclosure is a target image processing apparatus characterized in that, in each cluster, (1) when the Doppler velocity distribution exceeds a predetermined multiple of the length of the tail of the strong peak, which is determined based on at least one of the first threshold and the second threshold as the velocity difference between a strong peak that exceeds the first threshold and a weak peak that does not exceed the first threshold, the cluster is separated into a plurality of clusters corresponding to the strong peak and the weak peak, and (2) when the Doppler velocity distribution does not show the velocity difference described in (1), the cluster is not separated and is maintained.

[0021] With this configuration, even when a small target is in close proximity to a large target and clutter (for example, clutter such as the wake left by a large target such as a ship), the small target can be detected separately from the large target and clutter. On the other hand, even when a small target has a smaller illumination area near one end compared to the other end, is not in close proximity to clutter, and rotates around the center, the small target can be detected without excessive separation.

[0022] Furthermore, this disclosure is a target image processing program that causes a computer to sequentially execute each processing step performed by each processing unit of the target image processing apparatus described above.

[0023] This configuration makes it possible to provide a program that has the effects described above.

[0024] Furthermore, the inventions disclosed above can be combined as much as possible. [Effects of the Invention]

[0025] Thus, this disclosure enables the detection of small targets not only by using a radar device, but also by clustering cells where the radar received power exceeds a "small received power threshold" in order to detect small targets. This allows for the detection of small targets separately from clutter, even when the small targets are in close proximity to clutter. [Brief explanation of the drawing]

[0026] [Figure 1] This figure shows the problems that conventional cluster extraction methods have failed to solve. [Figure 2] This figure shows the configuration of the target image processing system disclosed herein. [Figure 3] This figure shows the solution for the cluster extraction process described herein. [Figure 4] This figure shows the first step of the cluster isolation process of this disclosure. [Figure 5] This figure shows the second step of the cluster isolation process of this disclosure. [Figure 6] This figure shows a first specific example of the cluster isolation process of this disclosure. [Figure 7] This figure shows a second specific example of the cluster isolation process of this disclosure. [Figure 8] This figure shows a third specific example of the cluster isolation process of this disclosure. [Figure 9] This figure shows a fourth specific example of the cluster isolation process of this disclosure. [Figure 10]This figure shows a fifth specific example of the cluster isolation process of this disclosure. [Figure 11] This figure shows a sixth specific example of the cluster isolation process of this disclosure. [Figure 12] This figure shows the results of the cluster extraction process using the prior art and the present disclosure. [Modes for carrying out the invention]

[0027] Embodiments of the present disclosure will be described with reference to the attached drawings. The embodiments described below are examples of the implementation of the present disclosure, and the present disclosure is not limited to these embodiments.

[0028] (Configuration of the target image processing system in this disclosure) Figure 2 shows the configuration of the target image processing system of this disclosure. The target image processing system S comprises a radar device 1, a target image processing device 2, and a radar display device 3. The radar device 1 transmits a radar irradiation signal and receives a radar reflection signal. The target image processing device 2 comprises a target image acquisition unit 21, a cluster extraction unit 22, a velocity distribution creation unit 23, and a cluster separation unit 24, and can be implemented by installing the target image processing program shown in Figures 4 and 5 on a computer. The radar display device 3 displays the target's position, velocity, and tracking results.

[0029] Figure 3 shows the solution for the cluster extraction process in this disclosure. In this disclosure as well, it is necessary to cluster cells in which the radar received power exceeds a "small received power threshold" in order to detect not only large targets such as ships, but also small targets such as ships V or buoys. However, when small targets such as ships V or buoys are in close proximity to clutter such as waves W, it is not possible to detect the small targets such as ships V or buoys separately from the clutter such as waves W.

[0030] In this disclosure, a Doppler velocity distribution is created for all cells in each cluster, and when the Doppler velocity distribution in each cluster has multiple peaks, the cluster is separated into multiple clusters corresponding to the multiple peaks (targets such as small vessels V or buoys and clutter such as waves W). Specifically, in each cluster, when the Doppler velocity distribution changes in velocity, passing through a first threshold, then a second threshold smaller than the first threshold, and then returning to the first threshold, the cluster is separated into multiple clusters corresponding to the multiple peaks of the Doppler velocity distribution (targets such as small vessels V or buoys and clutter such as waves W).

[0031] As a result, when tracking a small vessel V or buoy or other target, the target is not integrated with clutter such as waves W, and the characteristic quantities of the tracked target (center of gravity position, illumination area, maximum received power, or average Doppler velocity, etc.) are not distorted, thus improving the accuracy of target tracking.

[0032] In Figure 3, the two clutters, such as a small vessel V or a target like a buoy, and the wave W in close proximity, are caused by spatially uniform sea surface wind and surface currents. Therefore, they have a spatially uniform Doppler velocity distribution, and although the generation process is random, the velocity distribution is stable.

[0033] Furthermore, while Figure 3 shows that small vessels V or buoys are detected separately from clutter such as waves W, as a modified example, small helicopters (whose rotor rotation can be detected) or aircraft may be detected separately from clutter such as rain clouds.

[0034] The first step of the cluster isolation process in this disclosure is shown in Figure 4. The second step of the cluster isolation process in this disclosure is shown in Figure 5. Figure 4 shows the "conceptual procedure" of the cluster isolation process in this disclosure. Figure 5 shows the "specific procedure" of the cluster isolation process in this disclosure.

[0035] The target image acquisition unit 21 acquires a target image from the radar device 1 for each scan. The cluster extraction unit 22 uses DBSCAN (Density-Based Spatial Clustering of Applications with Noise) or the like to cluster cells in the target image acquired from the radar device 1 where the radar received power exceeds the received power threshold, and extracts the clusters (steps S1, S11).

[0036] The velocity distribution creation unit 23 creates a Doppler velocity distribution for all cells in each cluster (steps S2, S12). Here, it is desirable for the velocity distribution creation unit 23 to set the velocity width of each class so that the frequency of each class is sufficiently large according to the number of cells in each cluster, and it is desirable to set the velocity width of each class according to the resolution of the Doppler velocity.

[0037] The cluster separation unit 24 determines, based on the Doppler velocity distribution (steps S3-S6, S13-S16), whether to separate each cluster into multiple clusters corresponding to multiple peaks (steps S7, S18) or to maintain each cluster without separation (steps S8, S17). A specific example of the processing of the cluster separation unit 24 is described below.

[0038] (Specific examples of the cluster isolation process described herein) A first specific example of the cluster isolation process of this disclosure is shown in Figure 6. In Figure 6, the target of a small vessel V is in close proximity to only the clutter of low waves W and is moving in a straight line.

[0039] In the case of Figure 6, the Doppler velocity distribution in each cluster is: Doppler velocity V d Around it, a narrow Doppler velocity width V w It has a single peak forming (step S3, NO). Here, the Doppler velocity distribution has a narrow Doppler velocity width V. w This occurs because the target of the small vessel V appears to the radar device 1 as a single, unified target moving together.

[0040] Alternatively, the Doppler velocity distribution does not show a frequency change in which, as the velocity changes, it passes through a first threshold, then a second threshold smaller than the first threshold, and then returns to the first threshold (NO in steps S13 and S14). Furthermore, the Doppler velocity distribution does not show a velocity difference between strong peaks exceeding the first threshold and weak peaks not exceeding the first threshold that exceeds a predetermined multiple of the length of the tail of the strong peak, which is determined based on at least one of the first and second thresholds (NO in steps S15 and S16).

[0041] Therefore, the cluster separation unit 24 maintains each cluster without separating them in the case of Figure 6 (steps S8, S17). Thus, even when a small vessel V or a target such as a buoy is traveling in a straight line without coming into close proximity to clutter such as waves W, the small vessel V or the target such as a buoy can be detected without excessive separation. The first and second thresholds are set to a predetermined multiple (less than 1) of the maximum frequency of the Doppler velocity distribution so that the case of Figure 6 can be distinguished from the cases of Figures 7 to 11.

[0042] A second specific example of the cluster separation process of this disclosure is shown in Figure 7. In Figure 7, the target of the small vessel V has a larger illumination area near the center (wheelhouse) compared to the areas near both ends (bow and stern), is in close proximity to only low wave clutter W, and is circling around the center.

[0043] In the case of Figure 7, the Doppler velocity distribution in each cluster is: Doppler velocity V d Around it, a wide Doppler velocity width V w It has a single peak forming (step S3, NO). Here, the Doppler velocity distribution has a wide Doppler velocity width V. w This occurs because the target of the small vessel V is moving so that its bow / stern is moving away from / towards the radar device 1.

[0044] Alternatively, the Doppler velocity distribution does not show a frequency change such that, as the velocity changes, after passing through the first threshold, it passes through a second threshold smaller than the first threshold and then returns to the first threshold (NO in steps S13 and S14). And the Doppler velocity distribution does not show a velocity difference such that, as the velocity difference between a strong peak exceeding the first threshold and a weak peak not exceeding the first threshold, it exceeds a predetermined multiple with respect to the length of the base of the strong peak determined based on at least one of the first threshold and the second threshold (NO in steps S15 and S16).

[0045] Therefore, the cluster separation unit 24 maintains each cluster in the case of FIG. 7 without separating it (steps S8 and S17). Thus, even when a target such as a small ship V or a buoy is not close to clutter such as waves W and is turning, the target such as a small ship V or a buoy can be detected without being overly separated. Note that the first threshold and the second threshold are set to a predetermined multiple (less than 1 times) of the maximum frequency of the Doppler velocity distribution so that the case of FIG. 7 is distinguished from the cases of FIGS. 6, 8 to 11. Also, threshold processing for overly separating each cluster is not performed separately for values that are more than a certain distance away from the average value or the median value of the Doppler velocity distribution and values that fall within a certain range.

[0046] (Third and fourth specific examples of the cluster separation process of the present disclosure) A third specific example of the cluster separation process of the present disclosure is shown in FIG. 8. In FIG. 8, the target of the small ship V is close to the clutter of the high waves W and is moving straight ahead.

[0047] In each cluster in the case of FIG. 8, the Doppler velocity distribution has a plurality of peaks around the Doppler velocities V d1 、V d2 (step S3, YES), has a deep bottom between adjacent peaks (step S4, YES), and has a small frequency difference between adjacent peaks (step S5, YES). Here, the Doppler velocity distribution is the Doppler velocities V d1 、V d2The presence of multiple peaks around is because the small vessel V target is moving as a single target from the perspective of the radar device 1, and clutter such as high waves W is moving at a spatially uniform velocity from the perspective of the radar device 1. Furthermore, the deep bottom between adjacent peaks in the Doppler velocity distribution is due to the Doppler velocity V d1 , V d2 This is because the multiple peaks around it are spaced far apart.

[0048] Alternatively, the Doppler velocity distribution, as the velocity changes, passes through a first threshold (detecting the first peak), then a second threshold smaller than the first threshold (detecting the first bottom), and then returns to the first threshold (detecting the second peak) (YES in steps S13 and S14).

[0049] Therefore, the cluster separation unit 24 separates each cluster in the case of Figure 8 into multiple clusters (here, two clusters) corresponding to multiple peaks (here, two peaks) (steps S7, S18). For example, the cluster separation unit 24 sets the intermediate velocity between the two peaks or the lowest frequency velocity of the Doppler velocity distribution as the velocity threshold V th Threshold processing can then be performed. Therefore, even when a small vessel V or a target such as a buoy is in close proximity to clutter such as waves W, the small vessel V or the target such as a buoy can be detected separately from the clutter such as waves W. The first and second thresholds are set to a predetermined multiple (less than 1) of the maximum frequency of the Doppler velocity distribution so that the case in Figure 8 is distinguished from the cases in Figures 6, 7, 9 to 11.

[0050] A fourth specific example of the cluster isolation process of this disclosure is shown in Figure 9. In Figure 9, the target of the small vessel V has a smaller illumination area near the center (cargo hold) compared to the areas near both ends (bow and stern), is in close proximity to only low wave clutter W, and is circling around the center.

[0051] In the case of Figure 9, the Doppler velocity distribution in each cluster is: Doppler velocity V d1 , V d2There are multiple peaks around (Step S3, YES) and shallow bottoms between adjacent peaks (Step S4, NO). Here, the Doppler velocity distribution is such that the Doppler velocity V d1 , V d2 The presence of multiple peaks around the point indicates that the target, a small vessel V, is moving away from / towards the bow / stern from the radar device 1. Furthermore, the shallow bottom between adjacent peaks in the Doppler velocity distribution indicates that, among the targets of the small vessel V, the cargo hold is moving at an intermediate speed (Doppler velocity V) compared to the bow and stern. d3 This is because they are traveling by )

[0052] Alternatively, the Doppler velocity distribution does not show a frequency change in which, as the velocity changes, it passes through a first threshold, then a second threshold smaller than the first threshold, and then returns to the first threshold (NO in steps S13 and S14). Furthermore, the Doppler velocity distribution does not show a velocity difference between strong peaks exceeding the first threshold and weak peaks not exceeding the first threshold that exceeds a predetermined multiple of the length of the tail of the strong peak, which is determined based on at least one of the first and second thresholds (NO in steps S15 and S16).

[0053] Therefore, the cluster separation unit 24 maintains each cluster without separating them in the case of Figure 9 (steps S8, S17). Thus, even when a small vessel V or target such as a buoy has a smaller illumination area near the center compared to the areas near both ends, does not come into close proximity with clutter such as waves W, and rotates around the center, the small vessel V or target such as a buoy can be detected without excessive separation. The first and second thresholds are set to a predetermined multiple (less than 1) of the maximum frequency of the Doppler velocity distribution so that the case of Figure 9 can be distinguished from the cases of Figures 6-8, 10, and 11.

[0054] (Specific examples of the fifth and sixth cluster isolation processes of this disclosure) A fifth specific example of the cluster separation process of this disclosure is shown in Figure 10. In Figure 10, the target of the small vessel V has a smaller illumination area near the center (cargo hold) compared to the areas near both ends (bow and stern), and a smaller illumination area near one end (pole) compared to the area near the other end (wheelhouse), and is in close proximity to only the clutter of low waves W, and is rotating around the center.

[0055] In the case of Figure 10, the Doppler velocity distribution in each cluster is: Doppler velocity V d1 , V d2 The peak has multiple peaks around it (Step S3, YES), a deep bottom between adjacent peaks (Step S4, YES), a large frequency difference between adjacent peaks (Step S5, NO), and a small velocity difference between adjacent peaks (Step S6, NO).

[0056] Here, the Doppler velocity distribution is the Doppler velocity V d1 , V d2 The presence of multiple peaks around the point indicates that the target, a small vessel V, is moving away from / towards the bow / stern from the radar device 1. Furthermore, the deep bottom between adjacent peaks in the Doppler velocity distribution is due to the cargo hold of the small vessel V having an intermediate velocity (Doppler velocity V) compared to the bow and stern. d3 This is because it is moving at a slow speed (Doppler velocity V). Furthermore, the reason why the Doppler velocity distribution has small velocity differences between adjacent peaks is that, among the targets of the small vessel V, the bow and stern (small and large illumination areas) are moving at slower speeds (Doppler velocity V). d2 , V d1 This is because it is turning in that direction.

[0057] Alternatively, the Doppler velocity distribution does not show a frequency change in which, as the velocity changes, it passes through a first threshold, then a second threshold smaller than the first threshold, and then returns to the first threshold (NO in steps S13 and S14). Furthermore, the Doppler velocity distribution does not show a velocity difference between strong peaks exceeding the first threshold and weak peaks not exceeding the first threshold that exceeds a predetermined multiple of the length of the tail of the strong peak, which is determined based on at least one of the first and second thresholds (NO in steps S15 and S16).

[0058] Here, the length of the tail of the strong peak may be (1) the velocity difference between the velocity at which the curve of the strong peak intersects the first threshold and the velocity at which the curve of the strong peak intersects the second threshold (see Figure 10), (2) the velocity difference between the velocity that gives the maximum of the strong peak and the velocity at which the curve of the strong peak intersects the second threshold, or (3) the velocity difference between the velocity that gives the maximum of the strong peak and the velocity at which the curve of the strong peak intersects the first threshold.

[0059] Therefore, the cluster separation unit 24 maintains each cluster in the case of Figure 10 without separating them (steps S8, S17). Thus, even when a small vessel V or buoy or other target has a smaller illumination area near one end compared to the other end, does not come into close proximity with clutter such as waves W, and rotates around the center, the small vessel V or buoy or other target can be detected without excessive separation. The first and second thresholds are set to a predetermined multiple (less than 1) of the maximum frequency of the Doppler velocity distribution so that the case of Figure 10 can be distinguished from the cases of Figures 6-9 and 11. Also, the predetermined multiple of the tail length is set so that the case of Figure 10 can be distinguished from the case of Figure 11.

[0060] A sixth specific example of the cluster isolation process of this disclosure is shown in Figure 11. In Figure 11, the target of the small vessel V2 is located behind the target of the large vessel V1, which is traveling in a straight line, and is in close proximity to the clutter of the wave W (wake wave) left behind by the target of the large vessel V1, and is traveling in a straight line.

[0061] In the case of Figure 11, the Doppler velocity distribution in each cluster is: Doppler velocity V d1 , V d2 The peak has multiple peaks around it (Step S3, YES), a deep bottom between adjacent peaks (Step S4, YES), a large frequency difference between adjacent peaks (Step S5, NO), and a large velocity difference between adjacent peaks (Step S6, YES).

[0062] Here, the Doppler velocity distribution is the Doppler velocity V d1 , V d2 The presence of multiple peaks around the point is because the target of the small vessel V2 and the target of the large vessel V1 are moving at different speeds as seen from the radar device 1. Furthermore, the deep bottom between adjacent peaks in the Doppler velocity distribution is due to the clutter of the wave W (wake wave) left behind by the target of the large vessel V1, which, as seen from the radar device 1, is moving at an intermediate speed (Doppler velocity V) compared to the targets of the small vessel V2 and the large vessel V1. d3 This is because they are moving at different speeds. Furthermore, the reason the Doppler velocity distribution has a large velocity difference between adjacent peaks is that the target of the small vessel V2 and the target of the large vessel V1 are moving at different speeds as seen from the radar device 1.

[0063] Alternatively, the Doppler velocity distribution does not show a frequency change in which, as the velocity changes, it passes through a first threshold, then a second threshold smaller than the first threshold, and then returns to the first threshold (NO in steps S13 and S14). Furthermore, the Doppler velocity distribution shows a strong peak above the first threshold (Doppler velocity V). d1 ) and a weak peak that does not exceed the first threshold (Doppler velocity V) d2 ) and the velocity difference between them (Doppler velocity difference V) d2 -V d1 ) is defined as exceeding a predetermined multiple of the length of the tail of a strong peak, based on at least one of the first threshold and the second threshold (YES in steps S15 and S16).

[0064] Here, the length of the tail of the strong peak may be (1) the velocity difference between the velocity at which the curve of the strong peak intersects the first threshold and the velocity at which the curve of the strong peak intersects the second threshold (see Figure 11), (2) the velocity difference between the velocity that gives the maximum of the strong peak and the velocity at which the curve of the strong peak intersects the second threshold, or (3) the velocity difference between the velocity that gives the maximum of the strong peak and the velocity at which the curve of the strong peak intersects the first threshold.

[0065] Therefore, the cluster separation unit 24 separates each cluster in the case of Figure 11 into multiple clusters (here, two clusters) corresponding to multiple peaks (here, two peaks) (steps S7, S18). For example, the cluster separation unit 24 sets the intermediate velocity between the two peaks or the lowest frequency velocity of the Doppler velocity distribution to a velocity threshold V th Threshold processing can be performed accordingly. Therefore, even when a small vessel V2 or a target such as a buoy is in close proximity to a large vessel V1 or other target and clutter such as waves W (wake waves) left behind by the large vessel V1 or other target, the small vessel V2 or the buoy can be detected separately from the large vessel V1 or other target and the clutter such as waves W. The first and second thresholds are set to a predetermined multiple (less than 1) of the maximum frequency of the Doppler velocity distribution so that the case in Figure 11 can be distinguished from the cases in Figures 6 to 10. In addition, the predetermined multiple of the tail length is set so that the case in Figure 11 can be distinguished from the case in Figure 10.

[0066] (Results of the prior art and the cluster extraction process of this disclosure) Figure 12 shows the results of the prior art and the cluster extraction process of this disclosure. In Figure 12, a simulation is realized of a situation in which a target such as a small vessel is in close proximity to clutter such as waves.

[0067] In the left column of Figure 12, as with conventional technology, cells where the radar received power exceeds the received power threshold are clustered and the clusters are extracted. As a result, clusters containing targets such as small vessels and clutter such as waves are treated as a single cluster.

[0068] In the right column of Figure 12, as shown in this disclosure, in each cluster, it is determined whether to separate each cluster into multiple clusters based on the Doppler velocity distribution. As a result, clusters containing targets such as small vessels and clutter such as waves are separated into two clusters.

[0069] In the right-hand column of Figure 12, the Doppler velocity distribution is created for "each cluster only," not for the "entire target image." Therefore, the computational cost of the Doppler velocity distribution can be reduced to a minimum.

[0070] In Figures 6-11, when determining whether to separate each cluster into "two clusters," the determinations in steps S4-S6 and S13-S16 are made for "one pair of adjacent peaks" in the Doppler velocity distribution. As a variation, when determining whether to separate each cluster into "three or more clusters," the determinations in steps S4-S6 and S13-S16 may be made for "each of two or more pairs of adjacent peaks" in the Doppler velocity distribution.

[0071] Examples of situations in which we might determine whether to separate each cluster into "three or more clusters" include when ships are linked together in waves, when a ship pulls up a buoy in waves, or when birds are distinguished from each other in a rain cloud.

[0072] In Figures 6-11, when the number of cells in each cluster is "sufficiently large," the decision of whether to separate each cluster into multiple clusters is made based on the "Doppler velocity distribution." As a variation, when the number of cells in each cluster is "not sufficiently large," it is difficult to create a "Doppler velocity distribution," so the decision of whether to separate each cluster into multiple clusters may be made based on "thresholding that considers the mean or median of the Doppler velocity." [Industrial applicability]

[0073] The target image processing apparatus and target image processing program of this disclosure, when clustering cells in which the radar received power exceeds a "small received power threshold" in order to detect not only large targets but also small targets using a radar apparatus, can detect small targets separately from clutter even when the small targets are in close proximity to clutter. [Explanation of Symbols]

[0074] S: Target image processing system 1: Radar equipment 2: Target Image Processing Device 3: Radar display device 21: Target image acquisition unit 22: Cluster extraction unit 23: Velocity distribution creation section 24: Cluster Isolation Unit V, V1, V2: Vessel W:Wave

Claims

1. A cluster extraction unit that clusters cells in a target image acquired from a radar device where the radar received power exceeds a received power threshold, and extracts the clusters, A velocity distribution generation unit that generates a Doppler velocity distribution for all cells in each cluster, A cluster separation unit that, when the Doppler velocity distribution has multiple peaks in each of the aforementioned clusters, separates each cluster into multiple clusters corresponding to the multiple peaks, and maintains each cluster without separation when the Doppler velocity distribution has a single peak, A target image processing apparatus characterized by comprising:

2. The cluster separation unit separates each cluster into multiple clusters corresponding to adjacent peaks when the Doppler velocity distribution has a deep bottom between adjacent peaks, and maintains each cluster without separating it into multiple clusters corresponding to adjacent peaks when the Doppler velocity distribution has a shallow bottom between adjacent peaks. The target image processing apparatus according to claim 1, characterized in that

3. The cluster separation unit separates each cluster into multiple clusters corresponding to adjacent peaks when the Doppler velocity distribution has a small frequency difference between adjacent peaks, and maintains each cluster without separating it into multiple clusters corresponding to adjacent peaks when the Doppler velocity distribution has a large frequency difference between adjacent peaks. The target image processing apparatus according to claim 2, characterized in that

4. The cluster separation unit separates each cluster into multiple clusters corresponding to adjacent peaks when the Doppler velocity distribution has a large velocity difference between adjacent peaks, and maintains each cluster without separating it into multiple clusters corresponding to adjacent peaks when the Doppler velocity distribution has a small velocity difference between adjacent peaks. The target image processing apparatus according to claim 3, characterized in that

5. A cluster extraction unit that clusters cells in a target image acquired from a radar device where the radar received power exceeds a received power threshold, and extracts the clusters, A velocity distribution generation unit that generates a Doppler velocity distribution for all cells in each cluster, In each of the aforementioned clusters, (1) when the Doppler velocity distribution, as the velocity changes, passes through a first threshold, then a second threshold smaller than the first threshold, and then returns to the first threshold, the cluster is separated into multiple clusters corresponding to multiple peaks in the Doppler velocity distribution, and (2) when the Doppler velocity distribution, as the velocity changes, does not show the frequency change described in case (1), the cluster is maintained without separation. A target image processing apparatus characterized by comprising:

6. The cluster separation unit, in each cluster, (1) when the Doppler velocity distribution exceeds a predetermined multiple of the length of the tail of the strong peak, which is determined based on at least one of the first threshold and the second threshold as the velocity difference between a strong peak that exceeds the first threshold and a weak peak that does not exceed the first threshold, the unit separates each cluster into multiple clusters corresponding to the strong peak and the weak peak, and (2) when the Doppler velocity distribution does not show the velocity difference described in case (1), the unit maintains each cluster without separating it. The target image processing apparatus according to claim 5, characterized in that

7. A target image processing program for causing a computer to sequentially execute each processing step performed by each processing unit of the target image processing apparatus according to any one of claims 1 to 6.

Images