Radar target tracking device and radar target tracking program
The radar target tracking system improves the SC ratio and accurately tracks targets by using autocorrelation processing to generate a probability distribution of the target's predicted position, effectively addressing the challenge of high clutter interference.
Patent Information
- Application Number
- JP2023074549
- Authority / Receiving Office
- JP · JP
- Patent Type
- Applications
- Current Assignee / Owner
- Filing Date
- 2023-04-28
- Publication Date
- 2024-11-08
AI Technical Summary
Conventional radar target tracking systems fail to accurately calculate the velocity of targets when clutter intensity is high, leading to a decrease in the Signal-to-Clutter Ratio (SC ratio) and inability to track targets effectively.
The system employs autocorrelation processing to calculate an autocorrelation image for a time series of past target images, generating a probability distribution of the target's predicted position, and uses weighted addition to improve the SC ratio and track targets even in high clutter conditions.
The system enhances the SC ratio and accurately tracks targets by maintaining redundancy in the probability distribution of the target's predicted position, even in high clutter environments, preventing target loss and reducing clutter interference.
Smart Images

Figure 2024158918000001_ABST
Abstract
Description
[Technical field]
[0001] The present disclosure relates to a technique for tracking a target using a radar. [Background technology]
[0002] A technique for tracking a target using a radar is disclosed in Patent Document 1 etc. In Patent Document 1, a target is tracked after improving the SC ratio by using inter-scan correlation.
[0003] FIG. 1 shows an overview of the radar target tracking process of the prior art. In FIG. 1, it is assumed that the strength of clutter (sea clutter, etc.) is low. In the previous inter-scan correlation image CP, the speed v of the target T (ship, etc.) is calculated by a three-dimensional Fourier transform. In the latest predicted distribution image PL, a predicted distribution of the target T is generated by shifting each cell by the speed v of the target T × scan interval t with respect to the previous inter-scan correlation image CP. In the latest scanned image SL, the target T is detected at almost the same position as compared with the latest predicted distribution image PL. In the latest inter-scan correlation image CL, the latest predicted distribution image PL and the latest scanned image SL are weighted and added. Therefore, when the strength of clutter is low, the speed v of the target T can be calculated with high accuracy, and the SC ratio can be improved by the weighted addition of the latest predicted distribution image PL and the latest scanned image SL, and the target T can be tracked. [Prior art documents] [Patent documents]
[0004] [Patent Document 1] JP 2012-103197 A Summary of the Invention [Problem to be solved by the invention]
[0005] The problem of the radar target tracking process of the conventional technology is shown in FIG. 2. In FIG. 2, it is assumed that the strength of the clutter C is high. In the previous inter-scan correlation image CP, the velocity v of the target T is calculated by a three-dimensional Fourier transform. In the latest predicted distribution image PL, the predicted distribution of the target T is generated by shifting each cell by the velocity v of the target T × the scan interval t with respect to the previous inter-scan correlation image CP. In the latest scanned image SL, the target T is detected at a position slightly shifted compared to the latest predicted distribution image PL. In the latest inter-scan correlation image CL, the latest predicted distribution image PL and the latest scanned image SL are weighted and added. However, when the strength of the clutter C is high, the velocity v of the target T cannot be calculated with high accuracy, and the SC ratio cannot be improved even by weighting and adding the latest predicted distribution image PL and the latest scanned image SL, and the target T cannot be tracked.
[0006] Therefore, in order to solve the above problem, an object of the present disclosure is to improve the SC ratio by using inter-scan correlation and track a target even when the clutter intensity is high. [Means for solving the problem]
[0007] In order to solve the above problem, an autocorrelation image is calculated for a time series of past target images. Here, the autocorrelation image corresponds to a probability distribution of a predicted target position. The probability distribution of the predicted target position has high redundancy when the clutter intensity is high. Therefore, the probability distribution of the predicted target position and the target on the scan image have sufficient overlap even when the clutter intensity is high. By performing a weighted addition of the probability distribution of the predicted target position and the target on the scan image, the SC ratio can be improved and the target can be tracked.
[0008] Specifically, the present disclosure relates to a radar target tracking device that tracks a target using a radar, the radar target tracking device comprising: a target image extraction unit that extracts a target image having a predetermined size centered on the position of the target in a previous inter-scan correlation image from N past inter-scan correlation images; an autocorrelation processing unit that calculates an autocorrelation image for an inter-scan lag n for a time series of the N past target images based on the square of the amplitude spectrum in the time-spatial frequency domain or based on a convolution integral of target images in the time-spatial domain; and a predictive distribution generation unit that generates a probability distribution of a predicted position of the target n scans after the previous scan based on the autocorrelation image for the inter-scan lag n.
[0009] According to this configuration, even when the clutter intensity is high, it is possible to improve the SC ratio by using inter-scan correlation based on autocorrelation processing, and to track a target.
[0010] The present disclosure also provides a radar target tracking device, characterized in that the autocorrelation processing unit calculates an autocorrelation image with an inter-scan lag of 1, and the prediction distribution generating unit generates a probability distribution of a predicted position of the target in the latest scan that is next to the previous scan, based on the autocorrelation image with an inter-scan lag of 1.
[0011] According to this configuration, the probability distribution of the predicted position of the target in the latest scan can be generated with higher accuracy than the probability distribution of the predicted position of the target n scans (n≧2) after the previous scan. Also, for targets with acceleration and targets moving at high speed, the probability distribution of the predicted position of the target in the latest scan can be generated with higher accuracy.
[0012] The present disclosure also provides a radar target tracking device further comprising an autocorrelation detection unit that determines the presence of the target n scans after the previous scan when a maximum correlation value in an autocorrelation image of the inter-scan lag n is equal to or greater than a predetermined threshold that is set higher (lower) as the intensity of clutter other than the target is lower (higher), and inputs the autocorrelation image of the inter-scan lag n to the predictive distribution generation unit.
[0013] According to this configuration, when the intensity of the clutter is high, it is possible to prevent the target from being lost, and when the intensity of the clutter is low, it is possible to eliminate the clutter.
[0014] The present disclosure also provides a radar target tracking device, further comprising a weighting addition unit that, when calculating each cell value of an inter-scan correlation image n scans after the previous scan by performing a weighting addition on each cell value of a scan image n scans after the previous scan and each cell value of a probability distribution of a predicted position of the target n scans after the previous scan, sets a weighting coefficient of each cell value of the probability distribution to 0 or a smaller value when each cell value of the scan image is less than a predetermined threshold value, compared to when each cell value of the scan image is equal to or greater than the predetermined threshold value.
[0015] According to this configuration, the redundant components of the probability distribution of the predicted position of the target can be maintained for the overlapping portion between the probability distribution of the predicted position of the target and the target on the scan image. Then, the SC ratio of the target of interest among the multiple targets separated from each other can be improved and the target can be tracked, while the SC ratio of the other targets can be reduced.
[0016] The present disclosure also provides a radar target tracking program for causing a computer to execute each processing step executed by each processing unit included in the above-described radar target tracking device.
[0017] According to this configuration, it is possible to provide a program having the above-described effects.
[0018] The above disclosed inventions can be combined to the greatest extent possible. Effect of the Invention
[0019] In this manner, the present disclosure is capable of tracking a target by improving the SC ratio using scan-to-scan correlation, even when the clutter intensity is high. [Brief description of the drawings]
[0020] [Figure 1] FIG. 1 is a diagram showing an overview of a radar target tracking process according to the prior art; [Diagram 2] FIG. 1 is a diagram illustrating a problem with radar target tracking processing of the prior art. [Diagram 3] FIG. 1 is a diagram showing a configuration of a radar target tracking system according to the present disclosure. [Figure 4] FIG. 2 is a diagram showing a procedure of a radar target tracking process according to the present disclosure. [Diagram 5] FIG. 2 is a diagram showing a procedure of a radar target tracking process according to the present disclosure. [Figure 6] FIG. 2 is a diagram showing an overview of radar target tracking processing according to the present disclosure. [Figure 7] FIG. 2 is a diagram showing an overview of radar target tracking processing according to the present disclosure. [Figure 8] 11A to 11C are diagrams illustrating a specific example of an autocorrelation image calculation process according to the present disclosure. [Figure 9] 11A to 11C are diagrams illustrating a specific example of an autocorrelation image calculation process according to the present disclosure. [Figure 10] 11A to 11C are diagrams illustrating a specific example of an autocorrelation image calculation process according to the present disclosure. [Figure 11] 11A to 11C are diagrams illustrating a specific example of an autocorrelation image calculation process according to the present disclosure. [Figure 12] 5A to 5C are diagrams illustrating a specific example of a target presence determination process according to the present disclosure. [Figure 13] FIG. 11 is a diagram showing a specific example of weighted addition processing according to the present disclosure. [Figure 14] FIG. 11 is a diagram showing a specific example of weighted addition processing according to the present disclosure. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
[0021] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS The present disclosure will be described with reference to the accompanying drawings. The embodiments described below are examples of the present disclosure, and the present disclosure is not limited to the following embodiments.
[0022] (Configuration of the radar target tracking system of the present disclosure) The configuration of the radar target tracking system of the present disclosure is shown in Fig. 3. The radar target tracking system R includes a radar transceiver 1, a radar target tracking device 2, and a radar display device 3. The radar transceiver 1 irradiates a radar transmission signal to a target T and receives a radar reflection signal from the target T. The radar target tracking device 2 uses inter-scan correlation to improve the SC ratio and then tracks the target T. The radar display device 3 displays the tracking result of the target T.
[0023] The radar target tracking device 2 includes a scan image acquisition unit 21, an inter-scan correlation image storage unit 22, a target image extraction unit 23, an autocorrelation processing unit 24, an autocorrelation detection unit 25, a prediction distribution generation unit 26, a weighting addition unit 27, and a clustering processing unit 28, and can be realized by installing the radar target tracking programs shown in FIGS. 4 and 5 in a computer.
[0024] That is, an autocorrelation image is calculated for a time series of past target images. Here, the autocorrelation image corresponds to a probability distribution of a predicted position of a target T (such as a ship). The probability distribution of the predicted position of the target T has high redundancy when the intensity of clutter (such as sea clutter) is high. Therefore, the probability distribution of the predicted position of the target T and the target T on the scan image have sufficient overlap even when the intensity of clutter is high. The SC ratio can be improved by weighted addition of the probability distribution of the predicted position of the target T and the target T on the scan image, and the target T can be tracked. Details of the radar target tracking process of the present disclosure will be described below.
[0025] (Procedure of radar target tracking processing according to the present disclosure) The procedure of the radar target tracking process of the present disclosure is shown in Figures 4 and 5. An overview of the radar target tracking process of the present disclosure is shown in Figures 6 and 7. In Figures 4 to 7, the radar target tracking process is executed each time a scan image is acquired and each time a target T is tracked.
[0026] Every time the scan image is acquired, the scan image acquisition unit 21 performs scaling (e.g., from 0 to 1) of each cell value of each scan image, taking into consideration the probability distribution of the predicted position of the target T and the weighted addition of the target T on the scan image. The inter-scan correlation image storage unit 22 stores past inter-scan correlation images and the position of the target T (e.g., the center of gravity position).
[0027] The target image extraction unit 23 extracts a target image having a predetermined size centered on the position of the target T in the previous inter-scan correlation image CP from the past N inter-scan correlation images (step S1). In the upper left and upper right columns of FIG. 6, the target image extraction unit 23 extracts the position (x t , y t ) and extracts target images T1 to T8 having a predetermined size (taking into consideration the speed of the ship or the like).
[0028] The autocorrelation processor 24 calculates an autocorrelation image with an inter-scan lag n (-N / 2 to N / 2-1) for the time series of the past N target images (step S2). In the upper right and lower left columns of Fig. 6, the autocorrelation processor 24 calculates autocorrelation images A(-4) to A(+3) with an inter-scan lag -4 to +3 for the time series of the past eight target images T1 to T8. Specific examples of the processing by the autocorrelation processor 24 will be described later with reference to Figs. 8 to 11.
[0029] When the maximum correlation value in the autocorrelation image of the scan lag n is equal to or greater than (less than) a predetermined threshold (step S3, YES (NO)), the autocorrelation detection unit 25 determines the presence (absence) of the target T after n scans from the previous scan (step S4 (S5)). A specific example of the process of the autocorrelation detection unit 25 will be described later with reference to FIG. 12.
[0030] The predictive distribution generating unit 26 generates a probability distribution of the predicted position of the target T n scans after the previous scan based on the autocorrelation image with the inter-scan lag n (step S6). In the lower left and lower right columns of FIG. 6, the predictive distribution generating unit 26 generates a probability distribution of the predicted position of the target T in the latest scan after the previous scan (in the latest predictive distribution image PL, (x t , y t A specific example of the process of the predictive distribution generating unit 26 will be described later with reference to FIGS.
[0031] The weighting addition unit 27 performs weighted addition of each cell value of the scan image n scans after the previous scan and each cell value of the probability distribution of the predicted position of the target T n scans after the previous scan, and calculates each cell value of the inter-scan correlation image n scans after the previous scan (step S7). In Fig. 7, the weighting addition unit 27 performs weighted addition of each cell value of the latest scan image SL in the latest scan next to the previous scan, and each cell value of the probability distribution (latest predicted distribution image PL) of the predicted position of the target T in the latest scan next to the previous scan, and calculates each cell value of the latest inter-scan correlation image CL in the latest scan next to the previous scan. A specific example of the process of the weighting addition unit 27 will be described later with reference to Figs. 13 and 14.
[0032] The clustering processing unit 28 executes clustering on the inter-scan correlation image after n scans from the previous scan, and detects the position (for example, the center of gravity position) of the target T (step S11). Then, the process returns from step S11 to step S1.
[0033] In this way, even when the clutter intensity is high, the target T can be tracked after improving the SC ratio by using the inter-scan correlation based on the autocorrelation processing.
[0034] (Specific example of autocorrelation image calculation process according to the present disclosure) Specific examples of the autocorrelation image calculation process of the present disclosure are shown in Fig. 8 to Fig. 11. In Fig. 8, the autocorrelation processor 24 calculates autocorrelation images A(-4) to A(+3) with inter-scan lags of -4 to +3 based on a three-dimensional Fourier transform, the square of the amplitude spectrum in the time-frequency and spatial-frequency domains, and a three-dimensional inverse Fourier transform for the time series of eight past target images T1 to T8 (step S2). Here, the autocorrelation processor 24 scales each cell value of the autocorrelation images A(-4) to A(+3) by the peak value of the autocorrelation image A(0).
[0035] Autocorrelation images A(+1) to A(+3) with inter-scan lags of +1 to +3 correspond to the probability distribution of the predicted position of target T +1 to +3 scans after the previous scan. Autocorrelation image A(0) with inter-scan lag 0 superimposes not only the autocorrelation image of target T but also the autocorrelation image of clutter. Autocorrelation images A(+1) to A(+3) with inter-scan lags of +1 to +3 do not superimpose autocorrelation images of breaking waves, etc., nor autocorrelation images of swells, etc., because the sizes of the past eight target images T1 to T8 are small compared to the period of swells, etc.
[0036] 9, the autocorrelation processor 24 calculates an autocorrelation image A(0) with an inter-scan lag of 0 based on the convolution integrals between target images in the time and space domains and their average for a time series of eight past target images T1 to T8 (step S2). Here, the autocorrelation processor 24 executes the convolution integrals between target images Tm, Tm (m=1 to 8).
[0037] In FIG. 10, the autocorrelation processor 24 calculates an autocorrelation image A(+1) with a lag between scans of 1 based on the convolution integrals between target images in the time-space domain and their average for the time series of eight past target images T1 to T8 (step S2). Here, the autocorrelation processor 24 executes the convolution integrals between target images Tm and Tm+1 (m=1 to 8), but executes the convolution integral between target image T8 and target image T1. However, in order to calculate the autocorrelation image A(+1) with high accuracy, the autocorrelation processor 24 may not execute the convolution integral between target image T8 and target image T1, although the number of samples of the convolution integral is reduced. Then, the autocorrelation processor 24 scales each cell value of the autocorrelation image A(+1) by the peak value of the autocorrelation image A(0).
[0038] In FIG. 11, the autocorrelation processor 24 calculates an autocorrelation image A(+2) with a lag between scans of 2 based on the convolution integrals between target images in the time-space domain and their average for the time series of eight past target images T1 to T8 (step S2). Here, the autocorrelation processor 24 executes the convolution integrals between target images Tm and Tm+2 (m=1 to 8) and executes the convolution integrals between target images T1 and T2 for target images T7 and T8. However, in order to calculate the autocorrelation image A(+2) with high accuracy, the autocorrelation processor 24 may not execute the convolution integrals between target images T1 and T2 for target images T7 and T8, although the number of samples for the convolution integral is reduced. Then, the autocorrelation processor 24 scales each cell value of the autocorrelation image A(+2) with the peak value of the autocorrelation image A(0).
[0039] In this way, the probability distribution of the predicted position of the target T in the latest scan can be generated with high accuracy compared to the probability distribution of the predicted position of the target T n scans (n≧2) after the previous scan. Also, for the target T having acceleration and the target T moving at high speed, since only one scan after the previous scan is considered, the probability distribution of the predicted position of the target T in the latest scan can be generated with high accuracy.
[0040] In the conventional technology, the velocity v of the target T is calculated by a three-dimensional Fourier transform in the previous inter-scan correlation image CP. Then, for a target T having acceleration, the plane in the time frequency and spatial frequency domain varies, and the velocity v of the target T cannot be calculated with high accuracy. Furthermore, for a target T moving at high speed, aliasing occurs in the time frequency and spatial frequency domain, and the velocity v of the target T cannot be calculated with high accuracy.
[0041] (Specific example of target presence determination process according to the present disclosure) A specific example of the target presence determination process of the present disclosure is shown in Fig. 12. The autocorrelation detection unit 25 sets a predetermined threshold for the maximum correlation value in the autocorrelation images A(+1) to A(+3) of inter-scan lags +1 to +3 to a higher (lower) value as the intensity of clutter other than the target T decreases (increases). When the maximum correlation value in the autocorrelation images A(+1) to A(+3) is equal to or greater than (less than) the predetermined threshold (step S3, YES (NO)), the autocorrelation detection unit 25 determines the presence (absence) of the target T after +1 to +3 scans from the previous scan (step S4 (S5)), and inputs the autocorrelation images A(+1) to A(+3) to the predictive distribution generation unit 26.
[0042] In the upper left and upper right columns of Fig. 12, the autocorrelation detection unit 25 sets a predetermined threshold for the maximum correlation value in the autocorrelation images A(+1) to A(+3) high because the intensity of clutter other than the target T is low (step S3). In the lower left and lower right columns of Fig. 12, the autocorrelation detection unit 25 sets a predetermined threshold for the maximum correlation value in the autocorrelation images A(+1) to A(+3) low because the intensity of clutter other than the target T is high (step S3).
[0043] Then, in the upper left and lower left columns of FIG. 12, since the maximum correlation value in the autocorrelation images A(+1) to A(+3) is greater than or equal to a predetermined threshold value (step S3, YES), the autocorrelation detection unit 25 determines the presence of target T after +1 to +3 scans from the previous scan (step S4), and scales each cell value of the autocorrelation images A(+1) to A(+3) by the peak value of the previous inter-scan correlation image CP (to stabilize the intensity of target T according to changes over time).
[0044] On the other hand, in the upper right and lower right columns of FIG. 12, the maximum correlation value in the autocorrelation images A(+1) to A(+3) is less than the predetermined threshold value (step S3, NO), so the autocorrelation detection unit 25 determines that the target T is not present after +1 to +3 scans from the previous scan (step S5) and multiplies each cell value of the autocorrelation images A(+1) to A(+3) by 0.
[0045] In this way, when the intensity of the clutter is high, it is possible to prevent the target T from being lost, and when the intensity of the clutter is low, it is possible to eliminate the clutter.
[0046] (Specific example of weighted addition processing according to the present disclosure) Specific examples of the weighting addition process of the present disclosure are shown in Figs. 13 and 14. The weighting addition unit 27 calculates the cell values A xy and each cell value A of the probability distribution of the predicted position of the target T from +1 to +3 scans after the previous scan. xy ~ (tilde) and the weighted sum of the inter-scan correlation image for +1 to +3 scans after the previous scan, A xy ― (bar) is calculated (step S7).
[0047] Here, the weighting addition unit 27 calculates each cell value A xy is a given threshold A th When it is less than (step S8, YES), each cell value A xy is a given threshold A thCompared with the case where the probability distribution is 0 or more (step S8, NO), xy ~ The weighting coefficients of the two are set to 0 or smaller values (steps S9 vs. S10).
[0048] That is, the weighting addition unit 27 weights each cell value A xy is a given threshold A th When it is less than (step S8, YES), each cell value A of the inter-scan correlation image xy ― =α A xy +(1-α)·β·A xy ~ (where β=γ (0≦γ<1)) (step S9). On the other hand, the weighting addition unit 27 calculates each cell value A xy is a given threshold A th If it is equal to or greater than this (step S8, NO), the cell value A xy ― =α A xy +(1-α)·β·A xy ~ (where β=1) is calculated (step S10).
[0049] Here, the closer α is to 0, the smaller the cell value A of the probability distribution. xy ~ The weighting coefficient of α becomes larger, and clutter is suppressed. On the other hand, the closer α is to 1, the smaller the weighting coefficient of each cell value A xy The weighting coefficient of becomes larger, and the predictive distribution is suppressed. And the closer β is to 0, the more the cell value A of the probability distribution becomes smaller. xy ~ The weighting coefficient of each cell value A in the probability distribution becomes smaller. xy ~ On the other hand, the closer β is to 1, the more the redundancy of each cell value A in the probability distribution is suppressed. xy ~ The weighting coefficient of the probability distribution A becomes larger. xy ~ Basically, γ is set close to 0, but γ may be set close to 1 to prevent loss of the target T.
[0050] In the upper left column of Figure 13, the cell values A xy and each cell value A of the probability distribution xy ~ The weighting and adding unit 27 calculates the two-dimensional distribution of each cell value A of the scanned image. xy is a given threshold A th In the two-dimensional range above (step S8, NO), each cell value A of the probability distribution xy ~ On the other hand, the weighted addition unit 27 maintains the redundancy of each cell value A of the scanned image (step S10). xy is a given threshold A th In the two-dimensional range where the probability distribution is less than (step S8, YES), xy ~ Therefore, in the lower left column of FIG. 13, the redundancy of each cell value A of the inter-scan correlation image is suppressed (step S9). xy ― The cell value A of the scanned image in the two-dimensional distribution of xy and each cell value A of the probability distribution xy ~ In most of the areas where the two-dimensional distribution of xy ~ redundancy can be maintained.
[0051] In the upper right column of Fig. 13, the cell values A xy and each cell value A of the probability distribution xy ~ The two-dimensional distribution of the scanned image corresponds to only a part of the image. xy is a given threshold A th In the two-dimensional range above (step S8, NO), each cell value A of the probability distribution xy ~ On the other hand, the weighted addition unit 27 maintains the redundancy of each cell value A of the scanned image (step S10). xy is a given threshold A thIn the two-dimensional range where the probability distribution is less than (step S8, YES), xy ~ Therefore, in the lower right column of FIG. 13, the redundancy of each cell value A of the inter-scan correlation image is suppressed (step S9). xy ― The cell value A of the scanned image in the two-dimensional distribution of xy and each cell value A of the probability distribution xy ~ The two-dimensional distribution of A is only partially consistent with the cell value A of the probability distribution. xy ~ redundancy can be maintained.
[0052] In the upper left column of Fig. 14, among multiple targets T and T' that are separated from each other in eight past target images T1 to T8, attention is focused on target T, but not on target T'. In the upper right column of Fig. 14, in an autocorrelation image A(+1) with an inter-scan lag of 1, a probability distribution of the predicted position of target T in the latest scan after the previous scan is generated, but it cannot be separated from the probability distribution of the predicted position of target T' in the latest scan after the previous scan.
[0053] In the lower left column of Fig. 14, in the latest scan image SL, target T is detected at almost the same position as in the autocorrelation image A(+1) with inter-scan lag 1, but target T' is also detected at a slightly shifted position. In the lower right column of Fig. 14, in the latest inter-scan correlation image CL, the autocorrelation image A(+1) with inter-scan lag 1 and the latest scan image SL are weighted and added. For the target T of interest, the SC ratio can be improved and the target T can be tracked, but for the target T' that is not of interest, the SC ratio can be reduced.
[0054] In this way, for the overlapping portion between the probability distribution of the predicted position of the target T and the target T on the scan image, it is possible to maintain the redundant components of the probability distribution of the predicted position of the target T. Then, for the target T (which may be T') of interest among a plurality of targets T and T' that are separated from each other, it is possible to improve the SC ratio and track the target T (which may be T'), but it is possible to reduce the SC ratio for the other targets T' (which may be T).
[0055] In the conventional technology, the velocities v and v' of multiple targets T and T' that are separated from each other are calculated by a three-dimensional Fourier transform in the previous inter-scan correlation image CP. However, for the multiple targets T and T', the planes in the time frequency and spatial frequency domains are in a twisted relationship, and it is not possible to calculate the velocities v and v' of the multiple targets T and T' with high accuracy. [Industrial Applicability]
[0056] The radar target tracking device and radar target tracking program disclosed herein can track targets (ships, etc.) by improving the SC ratio using inter-scan correlation based on autocorrelation processing, even when the intensity of clutter (sea clutter, etc.) is high. [Explanation of symbols]
[0057] R: Radar target tracking system T, T': Target 1: Radar transmitter and receiver 2: Radar target tracking device 3: Radar display device 21: Scan image acquisition unit 22: Inter-scan correlation image storage unit 23: Target image extraction unit 24: Autocorrelation processing section 25: Autocorrelation detector 26: Predictive distribution generation unit 27: Weighting addition unit 28: Clustering processing unit CP: Previous scan correlation image PL: Latest predictive distribution image SL: Latest scanned image CL: Latest scan correlation images C: Clutter T1~T8: Past target images A(-4)~A(+3): Autocorrelation image
Claims
1. A radar target tracking device that tracks a target using a radar, a target image extracting unit that extracts a target image having a predetermined size centered on the position of the target in the previous inter-scan correlation image from the past N inter-scan correlation images; an autocorrelation processing unit that calculates an autocorrelation image with an inter-scan lag of n based on the square of the amplitude spectrum in the time-frequency and spatial-frequency domains for a time series of past N target images, or based on the convolution integral of the target images in the time-spatial domain; a predictive distribution generating unit that generates a probability distribution of a predicted position of the target n scans after a previous scan based on the autocorrelation image of the nth scan between scans; A radar target tracking device comprising:
2. The autocorrelation processing unit calculates an autocorrelation image with an inter-scan lag of 1, The prediction distribution generating unit generates a probability distribution of a predicted position of the target in the latest scan next to the previous scan based on the autocorrelation image of the inter-scan lag 1.
2. The radar target tracking device according to claim 1 .
3. an autocorrelation detection unit which, when a maximum correlation value in an autocorrelation image of the inter-scan lag n is equal to or greater than a predetermined threshold which is set higher (lower) as the intensity of clutter other than the target is lower (higher), determines the presence of the target n scans after the previous scan, and inputs the autocorrelation image of the inter-scan lag n to the prediction distribution generation unit; 3. The radar target tracking device according to claim 1, further comprising:
4. a weighting addition unit which performs a weighted addition of each cell value of the scan image n scans after the previous scan and each cell value of a probability distribution of the predicted position of the target n scans after the previous scan, and then calculates each cell value of the inter-scan correlation image n scans after the previous scan, and sets a weighting coefficient of each cell value of the probability distribution to 0 or a smaller value when each cell value of the scan image is less than a predetermined threshold value, compared to when each cell value of the scan image is equal to or greater than the predetermined threshold value; 3. The radar target tracking device according to claim 1, further comprising:
5. A radar target tracking program for causing a computer to execute each processing step executed by each processing unit included in the radar target tracking device according to claim 1 or 2.
Citation Information
Patent Citations
Moving target adaptive type scan correlation method
JP2012103197A
