JP6076113B2 - Wake correlation device - Google Patents

Description

  The present invention relates to a track correlator that performs track allocation based on a track group detected by a plurality of sensors such as a radar and an optical sensor.

  In target tracking in a multi-target environment using multiple sensors, track assignments are made so that there is a one-to-one correspondence between the sensors for the track group (position, speed, etc.) detected by performing target tracking with each sensor. Wake correlation is necessary. However, in a sensor network system composed of a plurality of sensors, the main problem is that the target appearance differs among the sensors due to a bias error. In the case of different types of sensors, the range in which the target can be seen (hereinafter referred to as the covered area) is different, so the number of wakes differs between sensors, or the wake correlation is wrong due to erroneous wakes due to clutter, etc. There are challenges. For such a problem, there is an apparatus described in Non-Patent Document 1.

  In the wake correlator described in Non-Patent Document 1, a likelihood function considering the prior probability of a bias error is used as the likelihood function of wake assignment. A bias error is estimated for each hypothesis of wake assignment, and an optimum wake correlation is made possible by finding an assignment that maximizes the likelihood function of the wake assignment.

M.M. Levedahl, “An exclusive pattern matching assignment algorithm,” in Proceedings of SPIE Conference on Data Processing of Small Targets. 4728, pp. 461-469, 2002.

  However, the track correlation based on the likelihood maximization considering the bias error as in Non-Patent Document 1 has a problem that the calculation load increases exponentially as the target number increases. Although the calculation load can be reduced by using the prior distribution of the bias error, there is a problem that it is necessary to accurately estimate the prior distribution.

  The present invention has been made to solve the above-described problems, and there is a bias error or a detection target number between sensors without performing wake correlation by maximizing likelihood in consideration of the bias error. An object of the present invention is to provide a wake correlation device that enables wake correlation with a small calculation load even in different cases.

  The track correlation device according to the present invention includes a region selection unit that extracts a track group of a designated region from position information of each track group detected by a plurality of sensors, and a sensor that is extracted by the region selection unit. The contour extraction unit that extracts the contour of the wake group and the contour of the wake group of each sensor extracted by the contour extraction unit are used for collation. By matching the contour of the track group of the sensor with few sensors, the contour matching unit that selects a track pair in a predetermined combination between the track groups as a matching pair, and the track correlation based on the matching pair selected by the contour matching unit And a nearest neighbor matching unit.

Further, the wake correlator according to the present invention includes a region selection unit that extracts a wake group of a designated region from position information of each wake group detected by a plurality of sensors, and each of the regions selected by the region selection unit. The contour extraction unit that extracts the contour of the track group of the sensor and the contour of the track group of each sensor extracted by the contour extraction unit are collated, and the contour of the track group of the sensor having a large number of tracks in the region is obtained. By matching the contour of the track group of the sensor with a small number of tracks, the contour matching unit that extracts the combination of the track pairs between the track groups as the track pair hypothesis, and the position information of each track group detected by each sensor When the comparison using the observation information can be used as the track information, the observation information collating unit for extracting a combination of track pairs between track groups as track pair hypothesis, contour matching unit and the observation Based on the wake pair hypothesis extracted by the information collation unit, an integrated collation unit that selects a track pair in a predetermined combination as a collation pair, and the nearest neighbor that performs wake correlation based on the collation pair selected by the integrated collation unit And a collation unit.

  According to the present invention, since it is configured as described above, a small calculation can be performed even when a bias error exists or the number of detection targets differs between sensors without performing wake correlation by maximizing likelihood in consideration of the bias error. Track correlation is possible by load.

1 is a block diagram showing a configuration of a wake correlator according to Embodiment 1. FIG. 4 is an explanatory diagram showing a difference in coverage between sensor A and sensor B in Embodiment 1. FIG. FIG. 6 is an explanatory diagram showing a difference in the number of detected wakes due to correlation integration between sensor A and sensor C in the first embodiment. It is explanatory drawing which shows the contour extraction in the case where there are many detected track numbers in Embodiment 1. FIG. It is explanatory drawing which shows the contour extraction in case the number of detected wakes in Embodiment 1 is small. 3 is a block diagram illustrating a configuration of a contour matching unit in Embodiment 1. FIG. FIG. 6 is an explanatory diagram illustrating an example after contour extraction of the sensor A and the sensor B in the first embodiment. 6 is an explanatory diagram illustrating offset processing according to Embodiment 1. FIG. FIG. 10 is an explanatory diagram illustrating gate inside / outside determination in the first embodiment. It is explanatory drawing which shows the gate inside / outside determination result in the reference | standard track pair (2, b) in Embodiment 1. FIG. FIG. 3 is an explanatory diagram showing track pairs for each reference track in the first embodiment. 10 is a block diagram illustrating a configuration of a contour matching unit according to Embodiment 2. FIG. FIG. 10 is an explanatory diagram illustrating candidate pair extraction processing in the second embodiment. FIG. 10 is an explanatory diagram showing candidate pairs for each reference track in the second embodiment. FIG. 10 is an explanatory diagram illustrating candidate pair extraction processing in the second embodiment. FIG. 10 is a block diagram illustrating a configuration of a wake correlator according to a third embodiment. FIG. 10 is a block diagram illustrating a configuration of a contour matching unit according to Embodiment 3. 10 is a list of wake pair hypotheses output from a contour matching unit according to Embodiment 3. FIG. 10 is a block diagram illustrating a configuration of an altitude collation unit in a third embodiment. 10 is a list of wake pair hypotheses output from an altitude collating unit in the third embodiment. FIG. 10 is a block diagram showing a configuration of a speed verification unit in the third embodiment. FIG. 10 is a block diagram showing a configuration of a signal power verification unit in the third embodiment. FIG. 10 is a block diagram showing a configuration of an integrated collation unit in a third embodiment. 10 is a list of wake pair hypotheses output from an integrated wake pair extraction unit in the third embodiment. 10 is a list of wake pair hypotheses output from an integrated verification pair selection unit in the third embodiment. FIG. 10 is a block diagram showing a configuration of an altitude collation unit in a fourth embodiment. FIG. 10 is a block diagram illustrating a configuration of a wake correlator according to a fifth embodiment. FIG. 10 is a block diagram showing a configuration of a reliability determination unit in the fifth embodiment. 16 is a list showing contour formation determination results for all combinations of statistical distances between wakes by the contour formation determination unit in the fifth embodiment. FIG. 10 is a block diagram illustrating a configuration of an integrated collation unit in a fifth embodiment. 10 is a list of wake pair hypotheses output from an integrated wake pair extraction unit in the fifth embodiment.

Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. In the following description, a description will be given assuming a wake correlation between two sensors (sensor A and sensor B), but the same processing can be applied between two or more sensors. In addition, it is assumed that a relative bias error of the sensor A with respect to the sensor B has occurred.
Embodiment 1 FIG.
FIG. 1 is a block diagram showing a configuration of a wake correlation apparatus 1 according to the first embodiment.
As shown in FIG. 1, the wake correlation apparatus 1 includes a region selection unit 101 (101 ₁ , 101 ₂ ), a contour extraction unit 102 (102 ₁ , 102 ₂ ), a contour matching unit 103, and a nearest neighbor matching unit 104. It is configured.

The area selection unit 101 extracts a track group of a designated area from position information of each track group detected by a plurality of sensors. In the example of FIG. 1, the area selecting section 101 ₁ corresponding to the sensor A, it shows the case of providing the a region selection unit 101 ₂ corresponding to the sensor B.

For example, as shown in FIG. 2, the coverage is different between the sensor A and the sensor B. However, the area selection unit 101 extracts wake groups in areas designated as being common to areas near the target to be collated. Reference numeral 2 in FIG. 2 is a network. Note with respect to the sensor A in the following, the detected number of track in the sensor B in the area (number of detected track) N _B is the sensor A is detected in the number of track (detection number track) and is N _A less ( N _B <= N _A ). Further, the wake handled by the region selection unit 101 is not limited to the wake detected by the sensor alone, but may be an integrated wake obtained by correlating and integrating the wakes detected by a plurality of sensors, for example, as shown in FIG.

The contour extraction unit 102 extracts the contour of the track group of each sensor extracted by the region selection unit 101. In the example of FIG. 1, the contour extracting unit 102 ₁ corresponding to the sensor A, it shows the case of providing a contour extraction unit 102 ₂ corresponding to the sensor B.

  The contour matching unit 103 performs matching using the contours of the track groups of each sensor extracted by the contour extraction unit 102, and the contours of the track groups of the sensors having a large number of detected tracks in the region are detected with a small number of detected tracks. By matching with the contour of the track group of the sensor, a track pair in a predetermined combination between the track groups is selected as a verification pair. The configuration of the contour matching unit 103 will be described later.

For example, as shown in FIG. 4, each wake group of sensor A and sensor B has a singular point region that cannot be collated because the shape of the wake group is different. Therefore, the contour extraction unit 102 extracts the contour of each wake group, and the contour matching unit 103 performs contour matching, thereby deleting the track of the singular point region that cannot be verified.
For example, as shown in FIG. 5, when the track groups of the sensor A and the sensor B are arranged side by side, the contour may not be created. However, even in such a case, contour matching can be performed by extracting line segments as contours and performing matching between the line segments.

  The nearest neighbor matching unit 104 performs wake correlation by the NN (Nearest Neighbor) method or the GNN (Global Nearest Neighbor) method based on the matching pair selected by the contour matching unit 103.

Next, the configuration of the contour matching unit 103 will be described. FIG. 6 is a block diagram illustrating a configuration of the contour matching unit 103 according to the first embodiment.
As shown in FIG. 6, the contour matching unit 103 includes an offset processing unit 1031, a gate inside / outside determination unit 1032, and a matching pair selection unit 1033.

  The offset processing unit 1031 assembles a pair of reference wake pairs among the wake groups of each sensor for all combinations, and sets the relative positions of the wake groups of the sensors with a small number of detected wakes in order so that the positions of the wake pairs match. It is to be offset.

  The gate inside / outside determination unit 1032 assembles track pairs other than the reference track pair among the track groups of each sensor for all combinations in each positional relationship offset by the offset processing unit 1031 and obtains position information of the track pair. Based on this, it is determined whether the wake pair is within the gate by determining whether the Mahalanobis distance (statistical distance) is equal to or less than a threshold value. Then, the gate inside / outside determination unit 1032 stores the gate inside / outside determination result and the statistical distance.

  The matching pair selection unit 1033 is a combination that minimizes the sum of the statistical distances among the combinations of track pairs that are determined to be in the gate for all track groups of the sensor with a small number of detected tracks by the gate inside / outside determination unit 1032. The track pair at is selected as a verification pair.

  Next, the operation of the wake correlation device 1 according to Embodiment 1 will be described. In the following, the process performed by the contour matching unit 103 after contour extraction will be described in detail with respect to the case where the wake groups of the sensors A and B are obtained with two-dimensional position coordinates (XY position information) as shown in FIG. explain. In FIG. 7, track numbers 1 to 6 are assigned to the tracks of the track group of the sensor A, and track numbers a to c are assigned to the tracks of the track group of the sensor B. Since there is a bias error between sensor A and sensor B, it is assumed that the track group of sensor A and the track group of sensor B are at different positions when displayed in absolute coordinates. Further, since the NN method and the GNN method used in the nearest neighbor matching unit 104 after the contour matching are known techniques, description thereof is omitted.

In the processing by the contour matching unit 103, first, the offset processing unit 1031 offsets the relative position of the track group of the sensor B for each track based on the position of the track group of the sensor A. Here, FIG. 8 shows a case where the wake group of sensor B is offset based on wake 2 of sensor A. In the example of FIG. 7, the number of wakes of sensor B is three, so three offsets are performed for each wake of sensor A.
As shown in FIG. 8A, for example, when offset is performed using the track 2 of the sensor A and the track a of the sensor B as a reference track pair, the relative position of the track a with respect to the position of the track 2 is obtained and calculated. The offset processing is performed by subtracting the relative position thus obtained from the position of each track of the sensor B. Further, as shown in FIGS. 8B and 8C, the same processing is performed when the wake 2 and the wakes b and c are set as reference wake pairs, respectively. Since FIG. 8 is based on the track 2 of the sensor A, the offset process is performed in the same manner when other tracks (track numbers 1, 3 to 6) of the sensor A are referenced. In the example of FIG. 7, since the number of detected tracks of sensor A is six and the number of detected tracks of sensor B is three, a total of 18 offset processes are performed.

Next, the gate inside / outside determination unit 1032 first calculates a statistical distance for each track pair other than the reference track pair in the positional relationship offset by the offset processing unit 1031. The statistical distance is expressed by equation (1). Here, x _A is the two-dimensional position coordinate of the track of sensor A, x _B is the two-dimensional position coordinate of the track of sensor B, and S is the observation noise covariance matrix of the track of sensor A and sensor B. Calculation is performed using the observation accuracy of sensor A and sensor B. When the tracking process is obtained, a smooth value and a smooth error covariance matrix may be used.

Next, the gate inside / outside determination unit 1032 determines whether the calculated wake distance pair pair is within the gate by determining whether the calculated statistical distance of the wake pair is equal to or less than a threshold value. Equation (2) is used as an index for determining that the gate is inside. γ is a gate threshold and is set in advance with a value determined from the chi-square distribution.

FIG. 9 is a diagram illustrating a case where the gate inside / outside determination process is performed on the positional relationship in which the track group of the sensor B is offset with reference to the track 2 of the sensor A illustrated in FIG.
Here, in the positional relationship shown in FIG. 9A (when track 2 of sensor A and track a of sensor B are used as a reference track pair), a track pair of track 3 of sensor A and track b of sensor B, And the track pair of the track 4 of the sensor A and the track c of the sensor B is determined to be in the gate. Similarly, in the positional relationship shown in FIG. 9B (when the track 2 of the sensor A and the track b of the sensor B are used as a reference track pair), the track pair of the track 1 of the sensor A and the track a of the sensor B, In addition, it is determined that the wake pair of the wake 5 of the sensor A and the wake c of the sensor B is in the gate. On the other hand, in the positional relationship shown in FIG. 9C (when the track 2 of the sensor A and the track c of the sensor B are used as a reference track pair), the tracks a and b of the sensor B are the tracks 1 to 6 of the sensor A. It is determined that it is away from both and outside the gate.

FIG. 10 is a diagram illustrating an example of a list of gate inside / outside determination results and statistical distances when the track 2 of the sensor A and the track a of the sensor B illustrated in FIG. In this list, the result of the inside / outside determination of the gate and the statistical distance of the track pair other than the reference track pair are shown.
In FIG. 10, “○” (1 in the case of a numerical value) is assigned to a wake pair determined to be inside the gate, and “×” (indicated to a numerical value) for a wake pair determined to be out of the gate. In this case, 0) is added. In this case, (2, a), (3, b), (4, c) are track pairs in the gate (the above (,) indicates (track number of sensor A, track number of sensor B). ing).

When the gate inside / outside determination unit 1032 stores the gate inside / outside determination result and the statistical distance in the memory, the wake of the wake pair in the combination determined to be inside the gate for each positional relationship subjected to the offset process. Save the sum of the number and the statistical distance. For example, a list of wake pairs determined to be in the gate with reference to each wake of the sensor A is obtained by the gate inside / outside determination unit 1032 as shown in FIG. In addition, the sum of the track number and the statistical distance in the combination of all track pairs may be stored regardless of the gate inside / outside determination result.
In FIG. 11, the “reference” column indicates the track number of the sensor A that is a reference when the inside / outside determination of the gate is performed. In addition, the columns of “wake a”, “wake b”, and “wake c” indicate track numbers of the sensor A that form a track pair with the tracks a, b, and c, respectively. Further, the column “sum of statistical distances” indicates the sum of the statistical distances of the wake pairs in combinations determined to be in the gate for each wake of the sensor A serving as a reference.

  Next, the verification pair selection unit 1033 selects a track pair in the combination shown in FIG. 11 obtained by the gate inside / outside determination unit 1032 as a verification pair in the combination having the smallest statistical distance. In the example of FIG. 11, the wake pairs (2, a), (3, b), (4, c) in the combination having the smallest statistical distance sum (9.7) are selected as the matching pairs.

  As described above, according to the first embodiment, the NN method or the GNN method is performed after extracting the contour of the wake group of each sensor and performing contour matching by contour matching. The wake correlation can be performed with a small calculation load even when there is a bias error or the number of detection targets is different among sensors without performing the wake correlation by maximizing the likelihood in consideration of the error.

Embodiment 2. FIG.
Below, the outline collating part 103b in Embodiment 2 is demonstrated. FIG. 12 is a block diagram showing a configuration of the contour matching unit 103b in the second embodiment. The contour matching unit 103b in the second embodiment shown in FIG. 12 is the gate inside / outside judging unit 1032b and the matching pair selecting unit 1033 of the contour matching unit 103 in the first embodiment shown in FIG. It changes to the part 1033b, and the candidate pair extraction part 1034 is added. Note that, in the configuration of the contour matching unit 103b in the second embodiment, the same components as those in the configuration of the contour matching unit 103 in the first embodiment are denoted by the same reference numerals, and description thereof is omitted.

  The gate inside / outside determination unit 1032b is obtained by deleting the gate inside / outside determination result and the statistical distance storage function from the gate inside / outside determination unit 1032 in the first embodiment. Then, the gate inside / outside determination unit 1032b outputs the gate inside / outside determination result and the statistical distance to the candidate pair extraction unit 1034. The function stored in the memory is performed by the candidate pair extraction unit 1034.

  The candidate pair extraction unit 1034 includes, among the combinations of wake pairs determined by the gate inside / outside determination unit 1032b as being within the gate for all wake groups of the sensors having a small number of detected wakes, for each positional relationship subjected to the offset process. A track pair with a combination that minimizes the sum of statistical distances is extracted as a candidate pair. In addition, for each offset-processed positional relationship, when it is determined that the wake group is out of the gate with respect to all wake groups of the sensor having a small number of detected wakes, the combination of wake pairs determined to be out of the gate The wake pair in the combination that minimizes the sum of the statistical distances is extracted as a candidate pair. Then, the candidate pair extraction unit 34b stores the extraction result and the statistical distance in the memory.

  FIG. 13 shows an example of candidate pair extraction when the track 2 of the sensor A is used as a reference. In FIG. 13, the candidate pair extraction unit 1034 extracts a wake pair with a combination that minimizes the sum of the statistical distances as a candidate pair from among the combinations determined to be within the gate. Here, wake pairs (1, a), (2, b), (5, c) are extracted as candidate pairs. This candidate pair extraction is performed in the same manner even when the other track of the sensor A (track numbers 1, 3 to 6) is used as a reference. Thereby, a candidate pair list as shown in FIG. 14 is obtained.

  The verification pair selection unit 1033b selects a candidate pair having the smallest sum of statistical distances as a verification pair from among the candidate pairs for each positional relationship extracted by the candidate pair extraction unit 1034. In the example of FIG. 14, a candidate pair (2, a), (3, b), (4, c) having the smallest statistical distance sum (9.7) is selected as a matching pair.

  In the candidate pair extraction, for example, as shown in FIG. 15, the observation noise is large, and the track of sensor B and the track of sensor A are all in the reference track pair (1, a), (1, b), (1, c). When the wake is out of the gate, a wake pair with a combination that minimizes the sum of the statistical distances is extracted as a candidate pair from among the combinations determined that the wake pair is out of the gate.

  As described above, according to the second embodiment, the combination of wake pairs determined by the gate inside / outside determination unit 1032b determined to be within the gate for all wake groups of the sensors having a small number of detected wakes, or detected. Of the track pair combinations determined to be out of the gate for all track groups of sensors with a small number of track numbers, track pairs with combinations that minimize the sum of statistical distances for each positional relationship are extracted as candidate pairs. Since the extraction result is stored, the memory can be reduced in addition to the effects of the first embodiment.

Embodiment 3 FIG.
Below, the wake correlation apparatus 1c which concerns on Embodiment 3 is demonstrated. FIG. 16 is a block diagram showing a configuration of a wake correlation apparatus 1c according to the third embodiment. The track correlation apparatus 1c according to the third embodiment shown in FIG. 16 changes the contour matching unit 103 of the track correlation apparatus 1 according to the first embodiment shown in FIG. A verification unit 106, a signal power verification unit 107, and an integrated verification unit 108 are added. In the configuration of the wake correlation device 1c according to the third embodiment, the same components as those of the wake correlation device 1 according to the first embodiment are denoted by the same reference numerals, and the description thereof is omitted.

  The contour matching unit 103c performs matching using the contours (position information) of the track groups of each sensor extracted by the contour extracting unit 102, and detects the contours of the track groups of the sensors having a large number of detected tracks in the region. A combination of wake pairs between wake groups is extracted as a wake pair hypothesis by matching the contours of wake groups of a small number of sensors.

  The altitude collating unit 105 performs collation using the altitude information of each wake group detected by each sensor, and extracts a combination of wake pairs between the wake groups as a wake pair hypothesis.

  The speed matching unit 106 performs matching using speed information (not speed vector but speed magnitude) of each wake group detected by each sensor, and extracts a wake pair combination between wake groups as a wake pair hypothesis. Is.

The signal power verification unit 107 performs verification using the signal power information of each wake group detected by each sensor, and extracts a combination of wake pairs between the wake groups as a wake pair hypothesis.
The altitude collation unit 105, the speed collation unit 106, and the signal power collation unit 107 correspond to the observation information collation unit of the present invention.

Based on the wake pair hypothesis extracted by the contour matching unit 103c, the altitude matching unit 105, the speed matching unit 106, and the signal power matching unit 107, the integrated matching unit 108 selects a track pair in a predetermined combination as a matching pair. Is.
The nearest neighbor verification unit 104 performs wake correlation by the NN method or the GNN method based on the verification pair selected by the integrated verification unit 108.

Next, the configuration of the contour matching unit 103c will be described. FIG. 17 is a block diagram showing a configuration of the contour matching unit 103c in the third embodiment.
As shown in FIG. 17, the contour matching unit 103c includes an offset processing unit 1031, a gate inside / outside determining unit 1032 and a wake pair hypothesis extracting unit 1035. Compared with FIG. 6, since the offset processing unit 1031 and the gate inside / outside determination unit 1032 are the same, only the wake pair hypothesis extraction unit 1035 will be described.

The wake pair hypothesis extraction unit 1035 rearranges the combinations of wake pairs determined by the gate inside / outside determination unit 1032 as being in the gate with respect to all the tracks of the sensor having a small number of detected wakes, in ascending order of the sum of the statistical distances. A wake pair hypothesis is generated. At this time, for the combinations with overlapping wake pairs, only the wake pair combinations having the smallest statistical distance sum are left, and the other wake pair combinations are deleted. For example, in the list of wake pairs shown in FIG. 11, wake pairs (1, a), (2, b), (5, c) on the reference wakes 1, 2, 5 and the reference wakes 2, 3, 4 Since there is an overlap in the wake pair (2, a), (3, b), (4, c), the wake pair (1, a) in the reference wake 2 with the smallest sum of statistical distances. , (2, b), (5, c) and the track pair (2, a), (3, b), (4, c) in the reference track 4 are left, and the others are deleted. Alternatively, the wake pair hypotheses may be generated by rearranging the statistic distances in ascending order of the combination of all wake pairs regardless of the gate inside / outside determination result.
FIG. 18 shows a list of track pair hypotheses generated by the track pair hypothesis extraction unit 1035 with respect to the list of track pairs shown in FIG. As shown in FIG. 18, numbers “hypothesis 1” and “hypothesis 2” are assigned to the list in ascending order of the sum of statistical distances.

Next, the configuration of the altitude collation unit 105 will be described. FIG. 19 is a block diagram showing the configuration of the altitude collating unit 105 in the third embodiment.
As shown in FIG. 19, the altitude collation unit 105 includes a gate inside / outside determination unit 1051 and a wake pair hypothesis extraction unit 1052. The gate inside / outside determination unit 1051 and the wake pair hypothesis extraction unit 1052 are functionally equivalent to the gate inside / outside determination unit 1032 and the wake pair hypothesis extraction unit 1035 of the contour matching unit 103c, but the altitude matching unit 105 has a contour matching unit 103c. There is no offset processing unit 1031 shown in FIG. 5, and gate inside / outside determination processing is performed using altitude information of each track for all combinations of track pairs.

  The gate inside / outside determination unit 1051 assembles wake pairs for all combinations between wake groups, and determines whether the Mahalanobis distance is equal to or less than a threshold based on altitude information of the wake pairs, so that the wake pair is inside the gate. This is a judgment.

  The wake pair hypothesis extraction unit 1052 generates a wake pair hypothesis by rearranging combinations of wake pairs determined to be in the gate by the gate inside / outside determination unit 1051 in ascending order of the statistical distance. Alternatively, the wake pair hypotheses may be generated by rearranging the statistic distances in ascending order of the combination of all wake pairs regardless of the gate inside / outside determination result.

In contour matching, it is assumed that the relative position between wakes is preserved, so there is no need to consider replacement between wakes, but in altitude matching, it is necessary to consider replacement because only the altitude is checked. FIG. 20 is a list of wake pair hypotheses output from the wake pair hypothesis extraction unit 1052.
Compared with FIG. 18, since the altitude collation considers the replacement of the track of sensor A corresponding to sensor B, for example, the track of sensor A (2, 3, 3) with respect to the track of sensor B (a, b, c). Statistical distance is calculated for all combinations of 4). In the example of FIG. 20, since the wakes (2, 3, 4) have the same altitude, the sum of the statistical distances in hypothesis 1 to hypothesis 6 is the same.

Next, the configuration of the speed verification unit 106 will be described. FIG. 21 is a block diagram showing the configuration of the speed verification unit 106 in the third embodiment.
As shown in FIG. 21, the speed verification unit 106 includes a gate inside / outside determination unit 1061 and a wake pair hypothesis extraction unit 1062. The gate inside / outside determination unit 1061 and the track pair hypothesis extraction unit 1062 are the same as the gate inside / outside determination unit 1051 and the track pair hypothesis extraction unit 1052 of the altitude collation unit 105 except that the altitude information of each track is replaced with speed information. Therefore, the description thereof is omitted.

Next, the configuration of the signal power verification unit 107 will be described. FIG. 22 is a block diagram showing the configuration of the signal power verification unit 107 in the third embodiment.
As shown in FIG. 22, the signal power verification unit 107 includes a gate inside / outside determination unit 1071 and a wake pair hypothesis extraction unit 1072. The gate inside / outside determination unit 1071 and the wake pair hypothesis extraction unit 1072 are the same as the gate inside / outside determination unit 1051 and the wake pair hypothesis extraction unit 1052 of the altitude collation unit 105 except that the altitude information of each track is replaced with signal power information. Therefore, the description thereof is omitted.

Next, the configuration of the integrated verification unit 108 will be described. FIG. 23 is a block diagram showing the configuration of the integrated collation unit 108 in the third embodiment.
As shown in FIG. 23, the integrated collation unit 108 includes an integrated track pair extraction unit 1081 and an integrated collation pair selection unit 1082.

The integrated wake pair extraction unit 1081 extracts a combination of wake pairs from the wake pair hypotheses extracted by the contour matching unit 103c, the altitude matching unit 105, the speed matching unit 106, and the signal power matching unit 107.
FIG. 24 is a list of track pair combinations extracted by the integrated track pair extraction unit 1081. As shown in FIG. 24, in the list, in addition to the combination of wake pairs output by each matching unit 103c, 105-107, the sum of the statistical distances calculated by each matching unit 103c, 105-107 is stored. The

The integrated collation pair selection unit 1082 includes, for each statistical distance in the contour collation unit 103c, the altitude collation unit 105, the speed collation unit 106, and the signal power collation unit 107, among the wake pair combinations extracted by the integrated pair extraction unit 81. A track pair with a combination that minimizes the sum of the sums is selected as a matching pair.
FIG. 25 is a list showing added values of the sums of statistical distances calculated by the matching units 103c and 105 to 107 with respect to the list shown in FIG. Then, in the example shown in FIG. 25, the integrated verification pair selection unit 1082 has a wake pair (2, a), (3,3) in the combination of hypothesis 1 in which the addition value of the sum of statistical distances is the minimum (27.2). b), (4, c) are selected as collation pairs.

  As described above, according to the third embodiment, in addition to the contour matching, it is configured to perform the matching using the altitude information, the speed information, and the signal power information. In addition to the effects in the first embodiment, Even when collation cannot be performed at the relative position of the contour, collation is possible if there is an altitude difference, speed difference, or signal power difference between wake groups.

  In Embodiment 3 described above, altitude information, speed information, and signal power information are used as observation information other than position information. However, the present invention is not limited to this, and any observation information that can be used as wake information. It is possible to add to this embodiment.

Embodiment 4 FIG.
Below, the altitude collation part 105d in Embodiment 4 is demonstrated. FIG. 26 is a diagram showing a configuration of the altitude collating unit 105d in the fourth embodiment. The altitude collating unit 105d in the fourth embodiment shown in FIG. 26 is obtained by adding a wake candidate setting unit 1053 and a wake candidate extracting unit 1054 to the altitude collating unit 105 in the third embodiment shown in FIG. Note that in the configuration of the altitude collation unit 105d in the fourth embodiment, the same components as those in the altitude collation unit 105 in the third embodiment are denoted by the same reference numerals, and description thereof is omitted.

  The wake candidate setting unit 1053 sets arbitrary values used in the wake candidate extraction unit 1054 in advance.

The wake candidate extraction unit 1054 has a value larger than a value obtained by adding a predetermined value to the maximum value of the altitude information of the wake group of the sensor having a small number of detected wakes and the minimum of the altitude information for the wake group of the sensor having a large number of detected wakes A wake group in which wakes having altitude information that is smaller than a value obtained by subtracting a predetermined value from the value is deleted is extracted.
In other words, the wake candidate extraction unit 1054 uses the value set by the wake candidate setting unit 1053 and the altitude information of the wake group of the sensor B (the sensor having a small number of detected wakes), to obtain the following equations (3) and (4): The track of sensor A (sensor with a large number of detected tracks) having altitude information that satisfies is deleted. Here, μ is a value set in advance by the wake candidate setting unit 1053, h _A indicates altitude information of all wakes of sensor A, and h _B indicates altitude information of all wakes of sensor B. max [h _B ] indicates that the maximum altitude is extracted from the altitude information of all tracks of sensor B, and min [h _B ] is the minimum altitude of altitude information of all tracks of sensor B. It shows that.

  The gate inside / outside determination unit 1051 performs gate inside / outside determination processing using the wake group extracted by the wake candidate extraction unit 1054 for the wake group of sensors having a large number of detected wakes.

  As described above, according to the fourth embodiment, the altitude separated from the maximum value and the minimum value of the altitude information by the sensor having a small number of detected tracks with respect to the track group of the sensors having a small number of detected tracks. Since the configuration is such that the track having information is deleted, the amount of calculation can be reduced.

  In the fourth embodiment, the altitude collating unit 105d has been described. However, similar processing can be applied to other observation collating units (such as the speed collating unit 106 and the signal power collating unit 107).

Embodiment 5. FIG.
Below, the wake correlation apparatus 1e which concerns on Embodiment 5 is demonstrated. FIG. 27 is a diagram showing a configuration of a wake correlator 1e according to the fifth embodiment. The track correlation apparatus 1e according to the fifth embodiment shown in FIG. 27 changes the integrated verification unit 108 of the track correlation apparatus 1c according to the third embodiment shown in FIG. (Sensor A observation data storage unit 109 ₁ , sensor B observation data storage unit 109 ₂ ) and reliability determination unit 110 are added. In the configuration of the wake correlator 1e according to the fifth embodiment, the same components as those of the wake correlator 1c according to the third embodiment are denoted by the same reference numerals, and the description thereof is omitted.

The observation item storage unit 109 stores observation items such as the observation accuracy and resolution of each known sensor in advance. In the example of FIG. 27, a sensor A observed specification storage unit 109 ₁ corresponding to the sensor A, shows the case of providing the a sensor B observations specifications storage unit 109 ₂ corresponding to the sensor B.

  The reliability determination unit 110 is based on the observation information (XY position information, altitude information, speed information, and signal power information) detected by each sensor and the observation data of each sensor stored in the observation data storage unit 109. The reliability of each observation information is determined. Details of the reliability determination unit 110 will be described later.

  Based on the reliability determination result by the reliability determination unit 110, the integrated verification unit 108e selects a track pair in a predetermined combination from the track pair hypotheses extracted by the verification units 103c and 105 to 107 as a verification pair. Is. Details of the integrated collating unit 108e will be described later.

Next, the configuration of the reliability determination unit 110 will be described. FIG. 28 is a block diagram showing a configuration of reliability determination section 110 in the fifth embodiment.
As shown in FIG. 28, the reliability determination unit 110 includes a contour formation determination unit 1101, an altitude difference determination unit 1102, a speed difference determination unit 1103, a signal power difference determination unit 1104, and a reliability determination threshold storage unit 1105. Has been.

  The contour formation determination unit 1101 is based on the XY position information of the wake group detected by each sensor and output via the contour extraction unit 102, and the observation specifications of each sensor stored in the observation specification storage unit 109. The contour formation determination rate of the wake group obtained by the sensor is calculated, and by determining whether the determination rate exceeds the reliability determination threshold stored in the reliability determination threshold storage unit 1105, the contour of the wake group is formed. It is determined whether it is possible.

  The altitude difference determination unit 1102 detects each sensor based on the altitude information of the wake group detected by each sensor and output via the contour extraction unit 102 and the observation specifications of each sensor stored in the observation specification storage unit 109. By calculating the altitude difference determination rate of the wake groups obtained by the above, and determining whether the determination rate exceeds the reliability determination threshold stored in the reliability threshold storage unit 135, there is an altitude difference between the wake groups This is a judgment.

  The speed difference determination unit 1103 detects each sensor based on the speed information of the wake group detected by each sensor and output via the contour extraction unit 102 and the observation data of each sensor stored in the observation data storage unit 109. By calculating the speed difference determination rate of the wake group obtained by the above, and determining whether the determination rate exceeds the reliability determination threshold stored in the reliability threshold storage unit 135, there is a speed difference between the wake groups This is a judgment.

The signal power difference determination unit 1104 is based on the signal power information of the track group detected by each sensor and output via the contour extraction unit 102 and the observation specifications of each sensor stored in the observation specification storage unit 109. By calculating the signal power difference determination rate of the wake groups obtained by each sensor and determining whether the determination rate exceeds the reliability determination threshold stored in the reliability threshold storage unit 135, signals between the wake groups are calculated. This is to determine whether there is a power difference.
The altitude difference determination unit 1102, the speed difference determination unit 1103, and the signal power difference determination unit 1104 correspond to the observation information difference determination unit of the present invention.

  The reliability determination threshold storage unit 1105 stores a threshold used when the determination units 1101 to 1104 in the reliability determination unit 110 perform reliability determination.

Next, details of processing by the contour formation determination unit 1101 will be described.
The contour formation determination unit 1101 first uses the XY position information from the contour extraction unit 102 of the sensor A, the XY position information from the contour extraction unit 102 of the sensor B, and the observation accuracy obtained from the observation specification storage unit 109. Then, it is determined whether or not there is a possibility that the outline of each wake group obtained for each sensor is destroyed by observation noise. The contour formation determination formulas for sensor A are shown in (5) and (6), and the contour formation determination formulas for sensor B are shown in (7) and (8). Here, x ^p _A is the XY position vector of track number p of sensors A, x ^q _A is the XY position vector of the track number q of the sensor A, S _A is the track of the sensor A measurement noise covariance matrix It is calculated using the observation accuracy of sensor A. Further, x ^p _B is the XY position vector of track number p of sensors B, x ^q _B is XY position vector of the track number q of the sensor B, S _B in wake of observation noise covariance matrix of the sensor B Calculation is performed using the observation accuracy of sensor B.

Equations (5) and (7) indicate the statistical distance between wakes, and equations (6) and (8) indicate determination equations. When the statistical distance between wakes satisfies the formulas (6) and (8), the distance between wakes is sufficiently far away, and the contours will not be greatly collapsed due to the change in the arrangement relationship between wakes due to observation noise. It shows that.
When determination is made for all combinations of wakes, a determination table as shown in FIG. 29 is obtained. FIG. 29 shows an example of a determination table of sensor A. If it is determined that there is no possibility of switching between tracks, it will be '○' (1 if indicated by a numerical value), and if it is determined that there is a possibility of switching between tracks, it will be '×' (0 if indicated by a numerical value). It becomes.

  Then, the contour formation determination unit 1101 performs contour determination using the determination table obtained for each sensor, and calculates the contour formation determination rate from the number of times it is determined that there is no possibility of switching between wakes. In the example of the contour formation determination table of sensor A in FIG. 29, the number of times that it is determined that there is no possibility of switching between wakes for all 15 combinations is 12, so that 12 ÷ 15 = 0.8 is that of sensor A. This is the contour formation determination rate. This process is similarly performed for the sensor B. Eventually, when the contour formation determination rate of each sensor exceeds the reliability determination threshold obtained from the reliability determination threshold storage unit 1105, it is determined that the contour can be formed, and the integrated collation unit 108e receives a determination value “◯” ( When the numerical value is indicated, 1) is output.

Next, details of processing by the altitude difference determination unit 1102 will be described.
The altitude difference determination unit 1102 first uses the altitude information obtained via the contour extraction unit 102 and the observation accuracy obtained from the observation specification storage unit 109 to determine whether there is an altitude difference for each track obtained for each sensor. Determine. The content of the process is the same as that of the contour formation determination unit 1101 except that altitude information is used as the observation information, and is therefore omitted.
Finally, if the altitude difference determination rate of each sensor exceeds the reliability determination threshold obtained from the reliability determination threshold storage unit 1105, it is considered that there is an altitude difference and is sent to the integrated verification unit 108e as a determination value “◯” ( When the numerical value is indicated, 1) is output.

Next, details of processing by the speed difference determination unit 1103 will be described.
First, the speed difference determination unit 1103 uses the speed information obtained via the contour extraction unit 102 and the observation accuracy obtained from the observation specification storage unit 109 to determine whether there is a speed difference for each track obtained for each sensor. Determine. The content of the processing is the same as that of the contour formation determination unit 1101 except that velocity information is used as observation information, and thus the description thereof is omitted.
Finally, when the speed difference determination rate of each sensor exceeds the reliability determination threshold obtained from the reliability determination threshold storage unit 1105, it is considered that there is a speed difference, and the determination value to the integrated verification unit 108e is “O” ( When the numerical value is indicated, 1) is output.

Next, details of the processing by the signal power difference determination unit 1104 will be described.
First, the signal power difference determination unit 1104 uses the signal power information obtained via the contour extraction unit 102 and the observation accuracy obtained from the observation specification storage unit 109 to determine the signal power difference for each track obtained for each sensor. Determine if there is any. The details of the process are the same as those of the contour formation determination unit 1101 except that the signal power information is used as the observation information, and the description thereof will be omitted.
Finally, when the signal power difference determination rate of each sensor exceeds the reliability determination threshold obtained from the reliability determination threshold storage unit 1105, it is considered that there is a signal power difference, and the determination value to the integrated collation unit 108e is “O”. '(1 if numerical value is indicated) is output.

  Next, details of the integrated verification unit 108e will be described. FIG. 30 is a block diagram showing a configuration of the integrated collating unit 108e in the fifth embodiment. The integrated collating unit 108e in the fifth embodiment shown in FIG. 30 is obtained by changing the integrated wake pair extracting unit 1081 of the integrated collating unit 108c in the third embodiment shown in FIG. 23 into an integrated wake pair extracting unit 1081e. Note that in the configuration of the integrated collation unit 108e according to Embodiment 5, the same components as those of the integrated collation unit 108c according to Embodiment 3 are denoted by the same reference numerals, and description thereof is omitted.

The integrated track pair extraction unit 1081e is based on the determination results by the reliability determination unit 110 (contour formation determination unit 1101, altitude difference determination unit 1102, speed difference determination unit 1103, and signal power difference determination unit 1104). The wake pair in a predetermined combination is extracted from the wake pair hypotheses extracted by the altitude collation unit 105, the speed collation unit 106, and the signal power collation unit 107.
That is, the integrated wake pair extraction unit 1081e is the contour formation determination unit of the reliability determination unit 110 among the wake pair hypotheses output from the contour verification unit 103c, the altitude verification unit 105, the speed verification unit 106, and the signal power verification unit 107. 1101, a wake pair with a hypothetical combination in which the determination values by the altitude difference determination unit 1102, the speed difference determination unit 1103, and the signal power difference determination unit 1104 are “◯” (1 in the case of numerical values) is extracted.

FIG. 31 shows a list of integrated wake pairs output from the integrated wake pair extracting unit 1081e when the determination values by the contour formation determining unit 1101 and the speed difference determining unit 1103 are “◯” (1 in the case of numerical values). It is.
In the example shown in FIG. 31, in addition to the combination of wake pairs whose determination value is “◯” (1 in the case of numerical values), the sum of the statistical distances is stored.

  As described above, according to the fifth embodiment, since it is configured to perform the reliability determination based on the observation information and the observation data of each sensor, it is possible to extract the integrated track pair using the information with high reliability. Thus, highly accurate track correlation is possible.

  In the fifth embodiment, the case where the reliability determination is performed using the altitude information, the speed information, and the signal power information as the observation information has been described. However, the present invention is not limited to this, and the reliability is determined using other observation information. The degree determination may be performed.

  Further, within the scope of the present invention, the invention of the present application can be freely combined with each embodiment, modified with any component in each embodiment, or omitted with any component in each embodiment. .

1, 1c, 1e track correlator, 2 network, 101, 101 _1, 101 ₂ region selection unit, 102, 102 _1, 102 ₂ contour extraction unit, 103 and 103 b, 103c contour matching unit, 104 closest matching unit, 105 , 105d Altitude verification unit, 106 Speed verification unit, 107 Signal power verification unit, 108, 108e Integrated verification unit, 109, 109 ₁ , 109 ₂ Observation specification storage unit, 110 Reliability determination unit, 1031 Offset processing unit, 1032 1032b Gate inside / outside determination unit, 1033, 1033b Collation pair selection unit, 1034 Candidate pair extraction unit, 1035 Track pair hypothesis extraction unit, 1051 Gate inside / outside determination unit, 1052 Track pair hypothesis extraction unit, 1053 Track candidate setting unit, 1054 Track candidate extraction 1061 Gate inside / outside determination unit 1062 Track pair hypothesis extraction unit 1 71 Gate inside / outside determination unit, 1072 Track pair hypothesis extraction unit, 1081, 1081e Integrated track pair extraction unit, 1082 Integrated collation pair selection unit, 1101 Contour formation determination unit, 1102 Altitude difference determination unit, 1103 Speed difference determination unit, 1104 Signal power Difference determination unit, 1105 Reliability determination threshold storage unit.

Claims (8)

An area selection unit that extracts a wake group of a specified area from position information of each wake group detected by a plurality of sensors;
A contour extracting unit that extracts the contour of the wake group of each sensor extracted by the region selecting unit;
Collation is performed using the contours of the wake groups of each sensor extracted by the contour extraction unit, and the contours of the wake groups of the sensors having a large number of wakes in the region are changed to the contours of the wake groups of the sensors having a small number of wakes. By combining, a contour matching unit that selects a track pair in a predetermined combination between track groups as a matching pair;
A wake correlator comprising: a nearest neighbor matching unit that performs wake correlation based on a matching pair selected by the contour matching unit. The contour matching unit
An offset processing unit that assembles a pair of wake pairs as a reference between the wake groups for all combinations, and offsets the relative positions of the wake groups of the sensors having a small number of wakes in order so that the positions of the wake pairs match;
In each positional relationship offset by the offset processing unit, wake pairs other than the reference wake pair are assembled for all combinations between the wake groups, and the Mahalanobis distance is equal to or less than a threshold based on the position information of the wake pair. By determining whether the wake pair is in the gate,
Of the combination of wake pairs determined to be in the gate for all wakes of the sensor with a small number of wakes by the inside / outside determination unit of the gate, or among all the combinations of wake pairs regardless of the inside / outside determination result of the gate, The track correlation apparatus according to claim 1, further comprising: a matching pair selection unit that selects a track pair with a combination that minimizes the sum of the Mahalanobis distances as the matching pair. The contour matching unit
An offset processing unit that assembles a pair of wake pairs as a reference between the wake groups for all combinations, and offsets the relative positions of the wake groups of the sensors with a small number of wakes in order so that the positions of the wake pairs match;
In each positional relationship offset by the offset processing unit, wake pairs other than the reference wake pair are assembled for all combinations between the wake groups, and the Mahalanobis distance is equal to or less than a threshold based on the position information of the wake pair. By determining whether the wake pair is in the gate,
Of the combination of wake pairs determined to be in the gate for all wakes of the sensor with a small number of wakes by the inside / outside determination unit, or out of the gate for all wakes of the sensor with a small number of wakes A candidate pair extraction unit that extracts a wake pair in a combination that minimizes the sum of the Mahalanobis distances for each positional relationship among the determined combinations of wake pairs;
A verification pair selection unit that selects a candidate pair having a minimum sum of the Mahalanobis distances among the candidate pairs for each positional relationship extracted by the candidate pair extraction unit, as the verification pair. Item 4. The wake correlator according to item 1. An area selection unit that extracts a wake group of a specified area from position information of each wake group detected by a plurality of sensors;
A contour extracting unit that extracts the contour of the wake group of each sensor extracted by the region selecting unit;
Collation is performed using the contours of the wake groups of each sensor extracted by the contour extraction unit, and the contours of the wake groups of the sensors having a large number of wakes in the region are changed to the contours of the wake groups of the sensors having a small number of wakes. By combining, a contour matching unit that extracts a combination of wake pairs between wake groups as a wake pair hypothesis,
Observation information collation that uses observation information that can be used as wake information other than the position information of each wake group detected by each sensor, and extracts a combination of wake pairs between wake groups as a wake pair hypothesis And
Based on the track pair hypothesis extracted by the contour matching unit and the observation information matching unit, an integrated matching unit that selects a track pair in a predetermined combination as a matching pair;
A wake correlator comprising: a nearest neighbor collation unit that performs wake correlation based on a collation pair selected by the integrated collation unit. The contour matching unit
An offset processing unit that assembles a pair of wake pairs as a reference between the wake groups for all combinations, and offsets the relative positions of the wake groups of the sensors with a small number of wakes in order so that the positions of the wake pairs match;
In each positional relationship offset by the offset processing unit, wake pairs other than the reference wake pair are assembled for all combinations between the wake groups, and the Mahalanobis distance is equal to or less than a threshold based on the position information of the wake pair. A first gate inside / outside determination unit that determines whether the wake pair is in the gate by determining
The first gate inside / outside determination unit determines a combination of wake pairs determined to be in the gate for all wakes of the sensor with a small number of wakes, or a combination of all wake pairs regardless of the gate inside / outside determination result. A wake pair hypothesis extraction unit that generates the wake pair hypotheses sorted in ascending order of the sum of the Mahalanobis distances,
The observation information matching unit
By combining all the wake pairs between the wake groups, and determining whether the Mahalanobis distance is less than the threshold based on observation information that can be used as wake information other than the position information of the wake pairs, the wake pairs a second gate in the outdoor decision unit but determines that it is within the gate,
The combination of wake pairs determined to be inside the gate by the second gate inside / outside determination unit, or all the combinations of wake pairs regardless of the gate inside / outside determination result, were rearranged in ascending order of the sum of the Mahalanobis distances. A wake pair hypothesis extraction unit that generates the wake pair hypothesis,
The integrated verification unit
From the wake pair hypothesis generated by the contour matching unit and the observation information matching unit, an integrated wake pair extracting unit that extracts a combination of wake pairs;
Of the wake pair combinations extracted by the integrated wake pair extracting unit, the wake pair in the combination that minimizes the sum of the sums of Mahalanobis distances calculated by the contour matching unit and the observation information matching unit, respectively. The track correlation apparatus according to claim 4, further comprising an integrated collation pair selection unit that selects as a collation pair. The observation information collating unit has a value larger than a value obtained by adding a predetermined value to the maximum value of the observation information of the track group of the sensor having a small number of tracks for the track group of the sensor having a large number of tracks, and the observation information A wake candidate extraction unit for extracting a wake group in which a wake having the observation information that is smaller than a value obtained by subtracting a predetermined value from the minimum value of
The second gate inside / outside determination unit performs gate inside / outside determination processing using a wake group extracted by the wake candidate extraction unit for a wake group of sensors having a large number of wakes. Wake correlator. A reliability determination unit that determines the reliability of the observation information detected by each of the sensors;
The wake correlation device according to any one of claims 4 to 6, wherein the integrated collation unit selects the collation pair based on a determination result by the reliability determination unit. The reliability determination unit
A contour formation determination unit that determines whether the contour of the wake group can be formed based on the position information of the wake group detected by each sensor and the observation data of each sensor;
An observation information difference determination unit that determines whether there is a difference in observation information between the wake groups based on observation information other than the position information of the wake groups detected by the sensors and the observation specifications of the sensors. Prepared,
The wake correlation apparatus according to claim 7, wherein the integrated collation unit selects the collation pair based on determination results by the contour formation determination unit and the observation information difference determination unit.

Images