JP5694002B2 - Target detection device, guidance device, and target detection method - Google Patents

Description

  Embodiments described herein relate generally to a target detection device, a guidance device, and a target detection method that accompany radar detection.

  In target identification using radar, not only the shape in the target range direction, that is, the shape in the distance direction, but also the cross range, that is, the shape in the angular direction, is detected, and what kind of object it is from the detected vertical and horizontal shapes. Is identified.

  The shape in the range direction is identified using a high resolution radar. For the shape in the cross-range direction, a method of performing high-resolution angle detection using an array antenna, a synthetic aperture using radar motion, and a reverse synthetic aperture using target motion are used to image a target. It is often identified by a method or the like.

  If the number of antennas per direction that the radar can use is limited and a sufficient number of antennas cannot be secured, an array antenna such as a method of performing imaging of a target by applying a synthetic aperture or a reverse synthetic aperture is required. It is necessary to use a method that does not. However, in these methods, it becomes difficult to detect if there is an unknown fluctuation in the target with respect to the detection direction.

  By the way, in monopulse angle measurement with two antennas, an angle measurement error called glint noise occurs due to interference of a plurality of reflection points included in the same range. In Non-Patent Documents 1 and 2, assuming that the number of reflection points included in the same range is sufficiently large, the distribution of angle measurement errors is applied to the predicted probability density distribution to estimate the target spread.

  When a high-resolution radar that can sufficiently identify the target shape is used, the number of reflection points included in the same range is about several. When the number of reflection points is about several, the distribution shape of the probability density of the angle measurement error in each range depends on the configuration of the reflection coefficient of the multiple reflection points included, and the probability density distribution uniquely determined by the cross-range spread Not. As a result, when estimation is performed by a method such as Non-Patent Documents 1 and 2, assuming that the number of reflection points included is very large, the error is large and correct estimation cannot be performed.

  Further, since the SNR (Signal to Noise Ratio) at the time of angle measurement is lower than the SNR at the time of range detection, in the method using the glint noise as in Non-Patent Documents 1 and 2, even if an attempt is made to detect the glint noise, It is indistinguishable from angle measurement error due to thermal noise, and correct estimation may not be performed.

R. H. DeLano, Proceedings of the IRE, Vol.41, No.12, 1953, pp.1778-1784 B. H. Borden, Transactions on Antennas and Propagation, Vol.43, No.8, 1995, pp.759-765

  As described above, in the method of performing imaging of a target by applying a synthetic aperture or a reverse synthetic aperture, the shape in the cross-range direction cannot be estimated when the target has an unknown fluctuation. In the method using the distribution of angle measurement errors, when the SNR is low, it cannot be distinguished from the angle measurement error due to thermal noise, and the shape in the cross range direction cannot be accurately estimated.

  Therefore, the object is to provide a target detection device, a guidance device, and a target detection method capable of estimating a shape including a target cross-range direction even when there is target fluctuation with a small number of antennas and a low SNR. It is in.

  According to the embodiment, the target detection apparatus includes a radar unit, a main reflection position extraction unit, a power recording unit, a fluctuation range determination unit, a frequency count unit, and a relative spread calculation unit. The radar unit acquires a plurality of ranges showing significant amplitude within the range measurement range, and amplitude values of the significant amplitude. The main reflection position extraction unit selects a main reflection position in the range direction within the target based on the reception status of the range and the amplitude value. The power recording unit records the amplitude value for each main reflection position during a preset processing period. The fluctuation range determination unit determines, for each main reflection position, whether or not the fluctuation range of the amplitude data recorded in the processing period exceeds a preset threshold value, and the main reflection position where the fluctuation range exceeds the threshold value Output amplitude data. The frequency counting unit counts the variation frequency of the amplitude data of the main reflection position from the variation range determination unit. The relative spread calculation unit associates the main reflection position and the count result with the range position within the target and the relative spread in the angular direction.

It is a block diagram which shows the function structure of the target detection apparatus which concerns on 1st Embodiment. It is a figure which shows an example of the target detected by the target detection apparatus of FIG. It is a block diagram which shows an example of a function structure of the radar process part of FIG. It is a figure which shows an example of the pulse structure of the transmission pulse transmitted from the radar process part 112 of FIG. It is a figure which shows an example of the table produced when the main reflection position extraction part of FIG. 1 extracts a main reflection position. It is a block diagram which shows the function structure of the frequency calculation part of FIG. It is a figure which shows the correction | amendment of the power by the power correction part of FIG. It is a figure which shows the low-pass filtering by the low-pass filter part of FIG. It is a figure explaining the principle of the other count algorithm by the frequency count part of FIG. It is a figure which shows the calculation result at the time of calculating a cost function based on FIG. It is a figure which shows the flowchart about the processing operation of the target detection apparatus of FIG. It is a figure which shows the target used by the simulation which concerns on 1st Embodiment. It is a figure which shows the range correction power output from the power correction part of FIG. It is a figure which shows the relative breadth of the target output from the relative breadth calculation part of FIG. It is a block diagram which shows the function structure of the target detection apparatus which concerns on 2nd Embodiment. It is a block diagram which shows an example of a function structure of the radar process part of FIG. It is a block diagram which shows the function structure of the frequency calculation part of FIG. It is a figure which shows the range correction power output from the power correction part of FIG. It is a figure which shows the power fluctuation waveform output from the low-pass filter part of FIG. It is a figure which shows the flowchart about the processing operation of the target detection apparatus of FIG. It is a figure which shows the other example of the target detection apparatus of FIG. It is a figure which shows the other example of the target detection apparatus of FIG. It is a block diagram which shows the function structure of the guidance device which concerns on 3rd Embodiment.

  Hereinafter, embodiments will be described with reference to the drawings.

(First embodiment)
FIG. 1 is a block diagram illustrating a functional configuration of a target detection apparatus 10 according to the first embodiment. A target detection apparatus 10 shown in FIG. 1 includes a high-resolution radar unit 11 and a relative spread detection unit 12.

  The high resolution radar unit 11 includes an antenna 111 and a radar processing unit 112.

  When a target having a complicated shape is detected by the radar, the radar wave is reflected at several main points having a large reflection cross section for the radar wave in the target. The antenna 111 receives the reflected radar wave.

  The radar processing unit 112 detects a range having significant power based on the received radar wave by an appropriate method such as peak detection. At this time, an amplitude waveform in which the amplitude of the range corresponding to the main reflection point has a peak is generated in the radar processing unit 112. As a result, the high resolution radar unit 11 outputs a range having significant power extracted by appropriate processing and at least the amplitude of the range to the relative spread detection unit 12.

  The relative spread detection unit 12 includes a main reflection position extraction unit 121, a power recording unit 122, a frequency calculation unit 123, and a relative spread calculation unit 124.

  The main reflection position extraction unit 121 extracts a main reflection position of a target that is a detection target. That is, if a peak with significant power is detected almost continuously from a point that seems to be the same position in the target for a certain period of time, that peak is used as the main reflection position of the target, and the position in the range direction within that target is To detect.

  The power recording unit 122 records the power for each of the extracted main reflection positions during a preset processing period, that is, a plurality of CPI (Coherent Processing Intervals).

  When there is a significant power fluctuation within the processing period for each main reflection position, the frequency calculation unit 123 calculates the frequency of the power fluctuation, that is, the peak or bottom frequency in the power fluctuation within the period.

  The relative spread calculation unit 124 outputs the calculated frequency in association with the position in the range direction within the target of each main reflection position.

Next, the operation of this embodiment will be described. The electric field of the reflected wave from the object having a spread from a to b as shown in FIG. 2 can be described as follows.

ξ = 2πfθ / c, c is the speed of light, f is the frequency, and θ is the target tilt angle. i is an imaginary unit. ρ _f is the integration result in the y direction of the reflected wave at each x point, and is equivalent to the reflection coefficient at each x point. In the case of FIG. 2, a is a negative value. Equation (1) uses Weierstrass infinite product display,

Can be transformed. It is w _n is zero point. It should be noted that the range of θ is limited to a range where the target looks almost the same shape. Dividing the reflected wave into phase and amplitude,

The angle g detected by phase monopulse angle measurement is the slope of the wavefront,

Can be written. Where the denominator of the third term approaches zero, i.e., xi] is g is increased at approaching w _n, glint noise is generated. At this time, the expression (2) is close to 0, and the reception power is decreasing. That is, glint noise occurs at an angle at which the received power decreases. ξ is a real number, the conditions under which the glint noise occurs, becomes a condition that ξ is closer to the real part of w _n. ρ _f

As described above, when divided into amplitude and phase, w _n is approximately

It becomes. n is an integer, and ξ approaches the real part of w _n corresponding to several n in the range where the target appears to be approximately the same shape. From ξ = 2πfθ / c and the equation (6), under the condition that ξ, that is, the frequency is fixed, when θ changes, glint noise is periodically generated, and after that period, that is, glint noise is generated, The angle δθ until it occurs in

It becomes. However, L = b−a, which is the target spread. λ is a wavelength and is c / f.

  From the above, if you want to know L, you need to know δθ. In order to know δθ correctly, with the target stationary, the radar moves and changes θ, or the radar movement (including stationary) is known, and the target moves in a known manner, It must be a condition to change θ. In order to know δθ, it is necessary to detect glint noise.

  However, the condition premised on the present embodiment is a condition in which δθ cannot be known because the target has an unknown fluctuation. In addition, when the target is far away, or when the reflection coefficient of the reflection point is small, it is not always possible to obtain a sufficient SNR that can be detected by distinguishing the occurrence of glint noise from thermal noise.

  First, consider SNR. When the target angle is measured with a monopulse, the phase difference between the two antennas is detected. The angle calculation in the monopulse angle measurement is performed by dividing two signals generated from two antenna outputs. When multiplying two systems of signals in which noise and a desired wave are mixed, the noise in the SNR of the multiplication result is not the thermal noise itself, but is a multiplication of one thermal noise and the other desired wave, and the SNR is each before multiplication. 3 dB worse than the signal. In the case of division, the result is similar if approximation is performed using Taylor expansion, and the SNR deteriorates by approximately 3 dB. On the other hand, range detection and power detection in that range can be performed with SNR equivalent to coherent addition of signals from two antennas by combining the two systems with maximum ratio so that the highest SNR can be obtained. As a result, the SNR is improved by 6 dB over the case of one antenna. Therefore, the SNR of the detected power is 9 dB higher than the SNR at the time of angle measurement, and the influence of thermal noise is small.

(2) Looking at the expression, in the condition where ξ is a glint noise generation condition approaches w _n, the reception power decreases. In other words, glint noise always occurs with a significant decrease in power, so that it is possible to perform measurement with higher sensitivity by detecting the frequency of power increase / decrease with a higher detection SNR than glint noise.

  In addition, glint noise often appears at several times the angular spread of the target part, but when the target is far away, the spread is often less than 0.1 degree, and several times as much. Even if glint noise occurs, it is often indistinguishable from thermal noise. On the other hand, if the target can be sufficiently resolved by the high resolution radar, the number of reflection points included in a certain range is reduced. In such a case, the power increase / decrease accompanying the occurrence of glint noise has a very large fluctuation range of 10 dB to 20 dB. Therefore, high detection sensitivity can be obtained by detecting power fluctuations instead of detecting glint noise that cannot be distinguished from thermal noise.

Next, detection of δθ will be considered. In many cases, the target fluctuation is mechanical vibration, and the angle of the posture often changes so as to move back and forth at a relatively slow speed. When the range of the angle is Φ and the spread at a certain position in the target is L ₁ , the number of times m ₁ when the power decreases while the target posture angle changes Φ is approximately m ₁ ≈Φ / δθ. = 2L ₁ Φ / λ. Similarly, placing the spread of different locations in the target and L _2, while the angle of the target posture changes [Phi, the number m ₂ of power is decreased _{approximately, m 2 ≒ Φ / δθ =} 2L 2 Φ / λ. The important point is that these positions with the spread of L ₁ and L ₂ are different positions within the same target, and if there is no significant deformation of the target within the measurement period, the two points are similarly only Φ It is to change the posture. As a result, Φ can be treated as common for both L ₁ and L ₂ although it is unknown. Therefore, when m ₁ / m ₂ is calculated, m ₁ / m ₂ = (2L ₁ Φ / λ) / (2L ₂ Φ / λ) = L ₁ / L ₂ . That is, when the state of interference at each position within the target changes due to the fluctuation of the target, the temporal change in the received power of the radar wave at each position is detected, and the frequency at which the power decreases is compared. , You can know the relative spread of each point of the target.

  It should be noted that the target shape that can be known in this way is a shape whose vertical / horizontal scale ratio is unknown.

  From the above discussion, the frequency that should be measured originally is the frequency of the bottom of the power, but the bottom and the peak naturally occur alternately, so any of them may be measured.

  Next, details of the present embodiment will be described.

  FIG. 3 is a block diagram illustrating an example of a functional configuration of the radar processing unit 112 according to the first embodiment. FIG. 3 is a diagram mainly showing reception baseband processing of the synthetic band radar. In FIG. 3, the transmission unit is omitted. A pulse train transmitted from a transmission unit (not shown) is reflected by the target and received by the antenna 111. The received reflected wave is processed by the radar processing unit 112 in FIG.

  In describing FIG. 3, the pulse configuration of the synthetic band radar will be described. A schematic diagram of the pulse configuration is shown in FIG.

One square is one pulse. The horizontal axis is time, and the vertical axis is frequency. The pulse repetition interval (PRI) is T ₂ and N _p pulses are transmitted in one frequency step. After transmitting the N _p pulses at a certain frequency step, the frequency by the frequency step interval f _stp transmits a similarly N _p pulses at different next frequency step. This is repeated N _f number of frequency step minute. One CPI consists of N _p × N _f pulses in total. A bandwidth of f _stp × N _f can be roughly obtained, but since the bandwidth of one pulse is sufficiently smaller than this, sufficient resolution can be obtained while suppressing the bandwidth and sampling rate of the baseband processing unit. .

As this variation, after sweeping with N _f frequency step component 1 frequency step 1 pulse, construction is also possible repeatedly to 1CPI with this N _p times. Since there is no big difference in performance and operation, explanation is omitted. Here, the description is continued with the pulse configuration of FIG.

  The pulse signal received by the antenna 111 is input to the radar processing unit 112 in FIG. The reception RF unit 1121 appropriately amplifies and filters the received pulse signal, and down-converts it to baseband at the transmission frequency of the pulse signal. As a result, the pulse signal is converted into a form with the Doppler frequency suffered during the transmission / reception process as the center frequency.

  Further, the reception RF unit 1121 converts the analog format pulse signal into a digital format at an appropriate sampling rate. Further, the reception RF unit 1121 demodulates the pulse signal by a method suitable for the pulse signal, and outputs a pulse representative value of one pulse per point to the Doppler frequency detection unit 1122 for one range bin or one gate.

  The Doppler frequency detection unit 1122 detects the Doppler frequency based on the pulse representative value from the reception RF unit. The Doppler frequency detection unit 1122 outputs the pulse representative value to the frequency step representative value extraction unit 1124 and outputs the detected Doppler frequency to the movement speed calculation unit 1123. When a plurality of targets having different Doppler frequencies to the extent that separation is possible are included, the Doppler frequency detection unit 1122 separates the Doppler frequencies and outputs the separated Doppler frequencies to the subsequent stage. In the present embodiment, the target detection device 10 detects the relative spread of different positions in the same target, but the Doppler frequencies at different positions in the target that are not so large do not differ greatly. Different positions within the target cannot be separated.

  The moving speed calculator 1123 converts the Doppler frequency from the Doppler frequency detector 1122 into a moving speed.

  The frequency step representative value extracting unit 1124 extracts one frequency step representative value from one frequency step based on the moving speed from the moving speed calculating unit 1123.

  The correcting unit 1125 corrects the frequency step representative value from the frequency step representative value extracting unit 1124 based on the moving speed from the moving speed calculating unit 1123. At this time, the phase corresponding to the change in the range of each pulse in the CPI is corrected in the frequency step representative value. The correction unit 1125 outputs the corrected frequency step representative value to the range calculation unit 1126.

  The range calculator 1126 estimates a range waveform based on the frequency step representative value from the corrector 1125 and extracts a peak range having significant power. For example, the range calculation unit 1126 performs IFFT (Inverse Fast Fourier Transform) on the frequency step representative value, and detects a peak range whose amplitude is equal to or greater than a threshold value. Alternatively, the range calculation unit 1126 may detect the peak range by applying MUSIC (Multiple Signal Classification), which is one type of super-resolution method, to the frequency step representative value. Further, the range calculation unit 1126 performs linear prediction called the Prony method described in Y. Zheng et al. “Prony's method for monopulse 3D imaging using stepped frequency waveform” (Proceedings of SPIE, Vol. 3461, pp. 308-314). Peak detection may be performed by a modification of the method.

  The amplitude extraction unit 1127 refers to the range, the number of peaks, the frequency step representative value, the waveform after IFFT, and the like from the range calculation unit 1126, and detects the amplitude of each peak, that is, information about power.

  For example, the amplitude extraction unit 1127 refers to the waveform after IFFT and detects the amplitude of the detected peak. Alternatively, the amplitude extraction unit 1127 extracts the complex amplitude in the detected peak, and as described in Zheng's literature, the number of significant peaks in the waveform, its range, and the inverse matrix from the single peak waveform May be used to estimate the amplitude of each peak alone. In such a method using a single peak waveform and an inverse matrix, even if adjacent peaks overlap and the waveform after IFFT leaks into each other, the height of the peak before overlapping can be estimated. . In the method of Zheng's document, a general inverse matrix is used for estimation. However, it is not always necessary to use the general inverse matrix, and the range of each peak, the complex amplitude before separation, and the single peak waveform are used. You may estimate with a normal inverse matrix. In the case of using the Prony method or MUSIC, a plurality of peaks approaching the resolution of the waveform after IFFT can be separated and detected, so it is better to estimate the amplitude before overlapping using an inverse matrix.

  As described above, the high resolution radar unit 11 extracts a range and amplitude having significant power in each CPI, and outputs them to the main reflection position extraction unit 121 of the relative spread detection unit 12.

  Note that a high-resolution radar that can resolve multiple points at different positions in the target may not be a synthetic band radar, such as a short pulse radar with a very short pulse width, a chirped pulse radar with a very wide bandwidth. But you can. In this case, the radar processing unit performs transmission / reception and modulation / demodulation suitable for each radar type, and outputs a significant peak range and amplitude of the resulting range waveform. In addition to the range and amplitude, the phase is output.

  The main reflection position extraction unit 121 extracts a main reflection position within the target. Specific processing is as follows, for example. First, for several consecutive CPIs, ranges and amplitudes having significant power are recorded side by side as shown in the table shown in FIG. In FIG. 5, the horizontal axis represents the position in the range direction within the target, and the vertical axis represents the CPI number. In FIG. 5, the position in the target is divided into 16 boxes, and when a peak having significant power is detected in each CPI in a range corresponding to the position in the target, the amplitude (or power) is expressed as Record in that box. In FIG. 5, the box in which the peak having significant power is recorded is shaded.

  As the target moves, its range changes from moment to moment. Therefore, in order to associate the position in the target with the detection range, information on how the range of the part representing the target, such as the head or center of the target, changes with time is necessary. is there. When the target moves, the range is normally tracked, and the gate for demodulating the pulse is determined based on the tracking information. When clutter separation, etc. is completed and the target to be detected in the gate is narrowed down to the target to be detected, for example, the leading significant reflection point in the gate is the same part in the target in any CPI Assuming that the point is a reflection point from the back, a significant reflection point on the back side may be put in the box with the head as a reference. Alternatively, the target start range may be determined based on the range tracking information, and the in-target position may be determined based on the difference between the range and the detected range. Note that the size of the box, that is, the classification interval of the in-target position may be determined based on the resolution at the time of range detection. However, when the super-resolution method or the like is used at the time of range detection, the resolution changes in conjunction with the SNR. In such a case, the interval may be determined based on the expected SNR at the stage of extracting the main reflection position.

  When the main reflection position extraction unit 121 creates a table, the ratio of the number of times data in a range having significant power is entered in a period of 22 CPIs in that period, as shown in FIG. The position within the target exceeding the ratio is extracted. For example, if 60% is set as the threshold value, in the case of FIG. 5, the in-target positions 2, 7, and 12 exceed the threshold value. The main reflection position extraction unit 121 has a significant power detected by each CPI with respect to the number of the position within the target, the relative range of the center of the box with respect to the head of the position within the target, or the head of the position within the target. The average value of the relative position of the peak range is defined as the main reflection position.

  If the distance between the boxes that define the position in the target is small compared to the resolution, the measurement error is larger than the resolution, or the tracking prediction accuracy is not very high, the same within the target There are cases where the position within the target where the reflected wave from the part is recorded varies. In such a case, the main reflection position extraction unit 121 may count the number of times the box has been filled, including a certain range such as front and back. For example, there is a CPI in which nothing is recorded in the in-target position 7 but recorded in 6 and 8. When such variations occur, 6, 7 and 8 may be evaluated together and the center of 7 may be set as the main reflection position, or the average value of these relative ranges may be set as the main reflection position.

  The main reflection position extraction unit 121 outputs the main reflection position extracted in this way to the power recording unit 122.

  The power recording unit 122 selects a reflection point output from the high resolution radar unit 11 and having a significant power, which is estimated to be a reflection from the main reflection position, and records the power for a predetermined period. I will do it. The power may be dB, linear, or amplitude. Further, complex amplitude may be used as long as power information can be correctly transmitted. The power recording unit 122 mainly determines whether or not the reflection is from the main reflection position based on whether or not it is within a certain allowable error range with respect to the target position output from the main reflection position extraction unit 121. The allowable error range may be determined based on the resolution. That is, it may be determined based on the performance of the measurement as in the case of determining the interval between the boxes at the target position in FIG. 5 and the main reflection position.

  When the high-resolution radar unit 11 is a synthetic band radar, if there is a speed detection error, an error that exceeds the detection range due to resolution or thermal noise may occur. In the synthetic band radar, the speed detection error and the range error have a fixed constant proportional relationship determined by the specifications of the radar. Therefore, the detection speed is output to the high resolution radar unit 11 and the detection speed is tracked. Then, the speed error may be calculated and the range error generated thereby may be corrected before recording. Alternatively, the speed detection accuracy may be predicted in advance, and the allowable error range may be determined based on the prediction. Note that the correction of the range error and the determination of the error range due to the speed error may be performed by processing of the main reflection position extraction unit 121.

  The period recorded by the power recording unit 122 is a period longer than the period used for detection by the main reflection position extraction unit 121. Further, when determining whether or not the reflection is from the main reflection position, tracking information or the like may be used in the same manner as performed by the main reflection position extraction unit 121.

  When the high-resolution radar unit 11 uses the super-resolution method or the like for range detection, the SNR increases as the target approaches, and a plurality of peaks having significant power are observed within the allowable error range. is there. In such a case, basically, it is sufficient to select a peak closer to the main reflection position. However, if angle measurement is performed as described later, angle measurement tracking data is created. A peak close to the angle predicted within the allowable error range may be selected.

  Next, details of the frequency calculation unit 123 of the present embodiment will be described. FIG. 6 is a block diagram illustrating a functional configuration of the frequency calculation unit 123 according to the first embodiment.

  The power correction unit 1231 corrects the power of each main position in the stored target so as to remove the influence of the range change. An example is shown in FIG. FIG. 7A is a graph in which changes in power at a target position are plotted in dB. As the range approaches, the received power increases in inverse proportion to the fourth power of the range according to the radar equation. Accordingly, FIG. 7A is a plot that rises to the left as a whole. Even in this state, the number of bottoms and peaks can be determined, but it is easier to determine the fluctuation range if the influence of the range change is removed. FIG. 7B is obtained by correcting FIG. 7A by multiplying the fourth power of the range. According to FIG. 7B, it can be seen that the change in the power due to the change in the range is removed, and only the change in the power is pure, so that it is easier to determine the unevenness.

  The low-pass filter unit 1232 applies a low-pass filter to the power change corrected by the power correction unit 1231 to remove high-speed fluctuation due to thermal noise or the like. An example is shown in FIG. FIG. 8A shows the range-corrected power at a certain position in the target recorded over a certain period, and the vertical axis is not dB but linear. The reason why the plots are discontinuous, especially in the vicinity of the vertical axis 0, is that these points were not detected because the power was too small, and they were missing. Since the waveform is disturbed due to the influence of the thermal noise, the peak and bottom of the unevenness due to the thermal noise will be detected as shown in FIG. Therefore, FIG. 8B shows a result of replacing the missing measurement point in FIG. 8A with 0 and applying a low-pass filter. It can be seen that fine irregularities due to thermal noise are removed and the peaks and bottoms are clear.

  Note that, as shown in FIG. 8A, if the cause of the missing measurement can be almost limited to the significant decrease in power, the missing measurement may be replaced with 0 as described above. However, for example, when the super-resolution method is used as a method for extracting the main reflection points from the range waveform of the radar output, and there are a plurality of reflection points with a separation distance that can be resolved by the super-resolution method. However, it is not always possible to separate these points. If separation is not possible, data for any of the primary reflection positions may be missing. In such cases, missing measurements do not necessarily occur when power is reduced. In such a case, filling missing data with zero will intentionally create a bottom that does not exist. In such a case, after missing data is interpolated from surrounding points by applying an interpolation method such as polynomial interpolation, a low pass filter is similarly applied. The low pass filter unit 1232 outputs the power fluctuation waveform to which the low pass filter has been applied to the fluctuation range determination unit 1233.

  The fluctuation range determination unit 1233 determines whether or not the main reflection position has a spread depending on whether the ratio (dB difference) between the maximum value and the minimum value of the power fluctuation waveform from the low-pass filter unit 1232 exceeds a predetermined threshold. Determine. As a result of the determination, the fluctuation range determination unit 1233 outputs only the data related to the main reflection position determined to have a spread to the frequency count unit 1234.

  Even if there is a main reflection position consisting of almost a single reflection point, if there is a main reflection position with a power fluctuation in the near range and a stronger average power, the power from there will leak. Even though it is a single reflection point, it may be observed that there is a power fluctuation. Therefore, the maximum value of the range correction power recorded for the main reflection position is compared with the maximum value of the range correction power recorded for other main reflection positions adjacent within a certain range. If there is a large main reflection position, it is preferable to increase the threshold of the width for judging the presence or absence of fluctuation. In particular, when the high resolution radar unit 11 detects the amplitude of each peak, such processing is necessary when the peak amplitude of the waveform is used as it is.

  Although the threshold value may be simply relaxed in this way, whether the power leaks from the adjacent main reflection position can also be detected by taking a correlation of the recorded range correction power. Even if there is a fluctuation in the power of a certain primary reflection position that exceeds the threshold, it is only slightly over the threshold (below the second threshold), and the power of the adjacent primary reflection position is more constant within a predetermined range. If the absolute value of the correlation is above a certain value and a strong correlation is observed, the power fluctuation at the main reflection position is mainly due to leakage. Therefore, it may be determined that the main reflection position has no spread.

  Since leakage from the adjacent main reflection position actually leaks with a complex amplitude, the power does not necessarily increase or decrease with the same sign with respect to increase or decrease in power at the adjacent main reflection position. Therefore, when determining the correlation, it is preferable to use the absolute value of the correlation value instead of the correlation value itself. In addition, since the phase may fluctuate due to changes in conditions, leakage from adjacent main reflection positions occurred at one time so that the power increased, but at another time the power decreased. May occur. For this reason, when determining the correlation, it is better not to collectively determine the correlation for a very long period. In other words, if the period for detecting the power fluctuation frequency is relatively long, it is preferable to divide it into several ranges, judge the correlation, and judge by the total value thereof.

  Based on the determination result from the fluctuation range determination unit 1233, the frequency counting unit 1234 calculates the number of irregularities of the power fluctuation frequency at each main reflection position within the period for calculating the power fluctuation frequency, that is, the number of peaks or bottom. Count the number of times.

  The frequency count unit 1234 outputs each main reflection position and frequency as a set. At this time, it is also preferable to output data indicating that the main reflection position does not spread even for the main reflection position determined to have no fluctuation by the fluctuation range determination unit 1233. In this way, it is possible to output the relative relationship between the angular spreads of the main reflection positions in the target.

  If the target fluctuation is irregular or the multipath leaks and the power waveform is greatly disturbed, the number of irregularities in the measurement period can be simply counted as described above. However, when the measurement conditions are relatively good, the target fluctuation method may appear in the power fluctuation waveform, and this can be used to detect with higher accuracy.

  The operation of the frequency counting unit 1234 at this time will be described with reference to FIG. When the posture of a target fluctuates as indicated by ◇ in FIG. 9, the fluctuations in the range correction power of two main reflection positions with different spreads within the target are indicated by Δ and □. . The power is a measured parameter, but the target posture angle change is a parameter that cannot be measured. The target posture angle fluctuates at a constant cycle, and moves back and forth with the portion indicated by the arrow as the apex. In such a case, the observed power fluctuation waveform is nearly symmetrical with respect to the point of the arrow. Around the point of the arrow, the power waveform shows a peak or a bottom, but in reality, these are not true peaks or bottoms, as if the target reversed the rotation direction of the posture at that timing. It just looks like a peak or bottom. If the timing of the arrow can be detected, we want to avoid counting the peaks and bottoms detected here. Therefore, for example, the following algorithm may be applied to the frequency counting unit 25.

  First, with respect to the power fluctuation waveform from the fluctuation width determination unit 1233, the square error of a point that is symmetrical with respect to each time is calculated in the range of about two peaks of the waveform, and the average error is calculated. Let the reciprocal be the cost function. The point where the cost function value is large is a point where the left and right waveforms are symmetrical with respect to the point, and when the target is shaken, it can be determined that the time is closer to the apex. FIG. 10 shows the result of calculating such a cost function for the waveform indicated by □ in FIG. Although the horizontal axis in FIG. 9 is the range, the horizontal axis in FIG. 10 is the CPI number, and the direction of the horizontal axis is reversed in FIGS. 9 and 10. In FIG. 10, the cost function value has a clear peak at the point indicated by the arrow, and it is estimated that these CPI numbers are vertices as indicated by the arrow in FIG.

  In such a case, it is desirable to calculate the frequency without crossing the point of the arrow. However, as can be seen from FIG. 9, the number of peaks and bottoms included between the arrows is not very large. Therefore, the frequency is not directly counted, but the intervals between a plurality of peaks or the intervals between a plurality of bottoms included between an arrow and an adjacent arrow are detected as a “period” of occurrence of interference. In addition, when there are not a plurality of peaks and bottoms, twice the interval between the bottoms adjacent to the peaks is detected as the “cycle” of occurrence of interference. When three or more points corresponding to the vertices (arrow times) are detected, by observing all periods in the interval between the vertices, averaging them, and taking the inverse , Calculate the frequency.

  Next, the operation of the target detection apparatus configured as described above will be described together with the result of simulation. FIG. 11 is a diagram illustrating an outline of a flowchart regarding the processing operation of the target detection apparatus 10 according to the first embodiment.

  First, the target detection apparatus 10 receives a radar wave by the high resolution radar unit 11 (step S111). The high resolution radar unit 11 generates an amplitude waveform in which the amplitude of the range corresponding to the main reflection point shows a peak, and outputs the amplitude waveform to the relative spread detection unit 12. FIG. 12 shows a waveform after IFFT in the range calculation unit 1126, and shows a target used in this simulation. In FIG. 12, the main reflection positions are the top, center, and back three points. The top includes only one main reflection point, but the center and the back include a plurality of reflection points, and the glint noise and power increase / decrease occur due to the change in posture angle caused by the fluctuation of the target. In the input value, the spread in the center cross range direction is 4.3 times the spread in the rear cross range direction.

  The target detection apparatus 10 extracts the main reflection position by the main reflection position extraction unit 121 (step S112), and records the power about the extracted main reflection position during the processing period preset by the power recording unit 122. (Step S113). FIG. 13 shows the result of recording the power at each main reflection position for a certain period under the condition that the target approaches the radar, and converting the power into the range correction power by the power correction unit 1231. FIG. 13A shows the power at the first main reflection position, FIG. 13B shows the power at the center main reflection position, and FIG. 13C shows the power at the rear main reflection position. Not only the power, but also the angle measurement error when measuring the angle is plotted. It can be seen that there is a spike-like angle measurement error that seems to be glint noise, but it is difficult to distinguish it from thermal noise. In particular, since the top is one reflection point, no glint noise is generated, but in the range where the SNR is low, the influence of thermal noise on the angle measurement is strong, and it is difficult to determine whether glint noise is generated. I understand. On the other hand, looking at the power waveform at each main reflection position, it can be seen that stable power can be observed even in the range where the angle measurement error is large at the head shown in FIG. Further, power fluctuations are clearly seen in the center shown in FIG. 13B and the back shown in FIG. 13C, and the advantage of using power instead of angle can be understood.

  Subsequently, the target detection device 10 calculates the frequency of power fluctuations by the frequency calculation unit 123 when there is significant power fluctuations within the processing period for each main reflection position. At this time, the target detection apparatus 10 uses the fluctuation range determination unit 1233 to determine whether or not the power fluctuation range within the processing period exceeds a preset threshold value (step S114). For example, in the fluctuation range determination unit 1233, if the threshold value of the fluctuation range for determining that there is a power fluctuation is set to 10 dB, it can be determined that the head has no spread because the width of the power fluctuation is lower than this. .

  When the power fluctuation width exceeds the threshold value (Yes in step S114), the target detection apparatus 10 calculates the number of peaks or bottoms within the processing period by the frequency counting unit 1234 (step S115). For example, when the number of peaks in the period is simply calculated with respect to the determination result of the fluctuation range determination unit 1233 for the power waveform at the center and the back, the center is 10 and the back is 4. When the power fluctuation width does not exceed the threshold value (No in step S114), the target detection apparatus 10 proceeds to step S117.

  The target detection apparatus 10 outputs the frequency calculated by the relative spread calculation unit 124 in association with the position of each main reflection position in the range direction within the target (step S116). According to the frequency calculation, since the center is 10 and the back is 4, it can be calculated that the center has a spread of 2.5 times relative to the back.

  When the algorithm for detecting symmetry and avoiding vertices described with reference to FIG. 9 is applied, the center frequency is 12.5 and the back frequency is 4.8. About 2.6 times wider.

  In the actual model, since the relative spread ratio is 4.3 times, the estimation accuracy is not high. The reason for this is that the speed of fluctuation between the vertices is not always equal, and when three or more reflection points contribute to the power fluctuation, the angular intervals at which interferences cancel each other are unequal, This is because the error increases when the number of samples is small. In the above examples, only a few cycles could be detected, and the error increased.

  The relationship between the top, center, and rear range is as shown in FIG. 12. As a result of the evaluation as described above, the top is an unexpanded point, and the center is located at a position about 2.7 m away from the front, It can be seen that it has a spread of about 5 to 2.6 times, and the back has a spread at a position about 5.2 m away from the head. Thereby, although the aspect ratio is unknown, the approximate shape of the target can be known as shown in FIG.

  Subsequently, the target detection apparatus 10 determines whether or not processing has been performed for all of the extracted main reflection positions (step S117). When processing has not been performed for all of the main reflection positions (No in step S117), The process proceeds to step S113, and the processes in steps S113 to S116 are repeated until the processes for all the main reflection positions are completed. When the process is completed for all the main reflection positions (Yes in step S117), the target detection apparatus 10 ends the process.

  As described above, in the first embodiment, the portion where the amplitude of the reflected radar wave is large is extracted as the main reflection position. And the target detection apparatus 10 calculates the frequency of the power fluctuation | variation in a main reflection position, and calculates | requires the relative spread in a target based on the frequency of the power fluctuation | variation.

  When there are a plurality of reflection points with high reflectivity at a certain position in the range direction within the target, the interference state of the reflected wave from them changes when the target fluctuates. The speed of the change is approximately proportional to the spread of those reflection points in the cross range direction. Multiple positions within the same target usually have the same sway, so even if the target is swayed uncertainly, the relative positions of the different positions are relatively different due to the difference in power fluctuation speed. Difference in general spread can be calculated.

  Further, power fluctuation due to a change in the interference state fluctuates with a very large fluctuation width exceeding 10 dB, so that detection with a low SNR is possible, and even a long-distance target can be detected. In the case of a system that measures angles using two antennas, range detection can be performed with an SNR that is 6 dB higher than the case of one antenna by combining two antennas, but the SNR during angle measurement is equivalent to two antennas. Since the data is divided, the SNR is deteriorated by about 3 dB from the state of one antenna. As described above, by using the power fluctuation, it is possible to perform detection at a longer distance and a lower SNR as compared with the conventional method based on the angle measurement data.

  Therefore, according to the target detection apparatus 10 according to the first embodiment, the absolute amount of spread in the cross range direction of each point of the target cannot be detected, but there is a small number of antennas and the target has uncertain fluctuations. Even in this case, the relative spread of each position of the target can be detected with a low SNR, and the approximate shape of the target can be estimated.

(Second Embodiment)
FIG. 15 is a block diagram illustrating a functional configuration of the target detection device 20 according to the second embodiment. In the first embodiment, angle measurement is not essential, but most radars have an angle measurement function. Since the performance can be improved by using the result of the angle measurement, or a worse condition can be dealt with, an embodiment in which the angle measurement is also performed is shown here.

  The high resolution radar unit 21 shown in FIG. 15 has two antennas per direction. Two antennas are shown together as an antenna 211. In addition, the antenna 211 normally consists of four antennas in order to measure the two directions of the azimuth and the elevation angle. However, in order to simplify the explanation, only two of one direction are shown here and one antenna is shown. Only the process will be described.

  The radar processing unit 212 correctly demodulates the received signal from the antenna 211 according to the radar format, detects a range having significant power by peak detection or the like, and detects the power. At that time, the two systems for angle measurement are processed, the angle is measured for the range corresponding to each peak, and the amplitude and angle are output for the range having significant power.

  FIG. 16 is a block diagram illustrating an example of a functional configuration of the radar processing unit 212 according to the present embodiment. FIG. 16 is a diagram mainly showing reception baseband processing of the synthetic band radar. Two signals from the first system and the second system are input to the radar processing unit 212. These two systems may directly correspond to two antennas, respectively. However, when performing monopulse angle measurement, it is often the case that the reception RF units 2121-1 and 211-2 are processed after generating the sum signal and difference signal of the two antennas at the antenna end. Therefore, here, it is assumed that a sum signal and a difference signal are generated at the antenna end, the first system is a sum signal, and the second system is a difference signal. Each system is subjected to RF processing in the reception RF units 2121-1 and 211-2. The content of the processing is the same as that of the reception RF unit 1121 shown in FIG. However, it is assumed that the parameters used for extracting one pulse representative value per one gate or one range bin at the time of demodulation are always the same in the two systems. In other words, the two systems are processed with exactly the same parameters without changing the gate position for each system, changing the coefficient of the pulse compression filter for each system, or extracting the pulse representative value from different range bins. . Hereinafter, in other cases until the angle measurement is completed, in most cases, the two systems are processed with the same parameters.

  The Doppler frequency detection unit 2122 detects the Doppler frequency from the extracted pulse representative value. The two systems may be combined and processed with a coefficient that maximizes the SNR, such as maximum ratio combining. If deterioration in detection performance is allowed, simpler processing such as detection from only one system or detection by equal gain synthesis may be employed. If the Doppler can be detected, the moving speed calculation unit 2123 calculates the moving speed. This is the same as the moving speed calculator 1123 shown in FIG.

  The frequency step representative value extraction unit 2124 extracts a frequency step representative value based on the same moving speed for each system. The rest is the same as the processing of the frequency step representative value extraction unit 1124 shown in FIG.

  The correction unit 2125 also performs correction based on the same moving speed of the two systems. The rest is the same as the correction unit 1125 shown in FIG.

  The range calculation unit 2126 performs range detection in the same manner as the range calculation unit 1126 shown in FIG. The two systems may be combined at the maximum ratio if possible, but when there are a plurality of reflection points with different ranges, the combination ratios are not the same and cannot be determined. . In the case of the synthetic band radar, the range detection accuracy is almost determined by the speed detection accuracy, and the angle detection accuracy is substantially determined by the SNR in the first place. Therefore, the synthesis method at the time of range calculation does not greatly affect the performance.

  However, when a correlation matrix such as MUSIC is used in the range calculation unit 2126, processing such as front-rear space averaging is performed in order to fully restore the rank of the correlation matrix. At that time, in the configuration of FIG. 3, it is necessary to recover from rank 1 to full rank, while in the configuration of FIG. 16, since recovery is from rank 2, when the number of antennas is 2, the number of subarrays is approximately halved. it can. Therefore, the subarray size can be increased, and the range detection with higher resolution than in the case of 1 antenna can be performed.

  When the range calculator 2126 detects several peaks and detects each range, the amplitude / phase extractor 2127 extracts the amplitude and phase, that is, the complex amplitude for each of the two systems. Again, for one peak, two systems are always processed with the same parameters. That is, the complex amplitudes of the completely same range are extracted by the two systems. The method of extraction is the same as that of the range calculator 2126 shown in FIG. 3 except that the amplitude is complex.

  The angle calculation unit 2128 calculates an angle from the extracted complex amplitude. The angle calculation method is, for example, a phase monopulse method. The phase monopulse method is a method of estimating the angle of arrival from the inclination of the wavefront input to the two antennas. Detailed processing of the phase monopulse method is not directly related to the present application, and is therefore omitted.

  As described above, the radar processing unit 212 detects the range, amplitude, and angle of reflected waves from a plurality of positions within the target.

  The relative spread detector 22 shown in FIG. 15 receives the range, amplitude and angle of the reflected wave from the high resolution radar unit 21 and first extracts the main reflection position by the main reflection position extraction unit 221. The basic operation of extracting the main reflection position is the same as that of the main reflection position extracting unit 121 shown in FIG. However, the power for the two systems is not processed individually, but is processed by making the total power for the two systems.

  In the main reflection position extraction unit 221, a table is similarly created for the angles. Even if significant power is continuously detected at the same target position, tracking to that angle of arrival clearly arrives from a different direction from the target being detected. This position is not considered the main reflection position. This occurs, for example, when a transmitted radar wave is reflected by a target, and once an interference wave called a multipath is received that is reflected from another place such as the ground and returned to the radar. Since the multipath arrives from the ground direction, the arrival angle thereof is usually different from the arrival angle of the detection target. In multipath, since the Doppler frequency is not significantly different from the target to be detected, separation by pulse Doppler radar is difficult, and separation in an angular region is desirable.

  The power recording unit 222 records the power of each main reflection position detected as described above for a predetermined period. The operation is the same as that of the power recording unit 122 shown in FIG. Similarly to the power, the angle recording unit 225 records the angle of each main reflection position for the same period as the period during which the power is recorded.

  The frequency calculation unit 223 performs detection using not only power but also angle. FIG. 17 is a block diagram illustrating a functional configuration of the frequency calculation unit 223 according to the second embodiment.

  The power correction unit 2231 corrects the power of each main position in the stored target so as to remove the influence of the range change.

  The low-pass filter unit 2232 receives the corrected power waveform from the power correction unit 2231 and the angle measurement result of the main reflection position recorded in the angle recording unit 225. The low-pass filter unit 2232 selects a portion where the error of the angle measurement result is small, and determines the characteristics of the low-pass filter based on the portion. This is because the portion where the error in the angle measurement result is small can be estimated as a portion where the influence of the interference wave is small, such as multipath.

  Here, the operation of the low-pass filter unit 2232 will be described with reference to FIG. FIG. 18 shows the result of recording the power of each main reflection position for a certain period under the condition that the target approaches the radar, and converting the power into range correction power by the power correction unit 2231.

  In the example of FIG. 13, the power was detected relatively stably even at a low SNR where the influence of thermal noise appears on the angle measurement error. On the other hand, when an interference wave such as a multipath is generated, noise may occur in the angle measurement and power detection results even if the influence of thermal noise is small. An example is shown in FIG. At this time, the target is the same target as in the example of FIG. 12, and the main reflection positions are the three points shown in FIG. FIGS. 18A, 18B, and 18C show changes in reception power and angle measurement error at the head, center, and rear, respectively. In FIGS. 18B and 18C, the angle error fluctuates greatly, and the power is also disturbed as compared with FIG. 13 particularly in the portion where the range is far and the influence of multipath is strong.

  Multipath usually arrives later than the direct reflected wave that is the desired wave, but the delay time is often longer as the range approaches, and when the range approaches to some extent, the multipath is separated, No effect. The multipath is sequentially separated from the peak near the head. At the head shown in FIG. 18A, the multipath is already separated at a point of 2000 m, so there is no influence. In the center shown in FIG. 18B, multipaths overlap up to a point of 1000 m, and in the back shown in FIG. 18C, overlaps up to a point of 900 m. Since FIG. 18 shows the result detected by the high-resolution radar, when the multipath has a peak between the head and the center in FIG. 12, for example, the peak can be separated to some extent. For this reason, the power waveform is roughly similar to that shown in FIG. However, the influence on the angle measurement, particularly when the reception power is reduced, is very large. This is because the power of the direct reflected wave is reduced, and the error is increased by detecting the leaked multipath angle, and the glint noise is generated due to the interference between the multipath and the direct reflected wave. There are two reasons for this, but these two are continuation and cannot be clearly distinguished in principle. The angle error when glint noise occurs is roughly proportional to the spread of the target, but when there is multipath, the spread of the target appears to be a spread including the multipath. Multipath can be treated as a wave from a target virtual image at a symmetrical position with respect to the ground, so the target spread that causes glint noise is very wide, resulting in a very large angular error. . The right vertical scale in FIG. 18 is five times the scale in FIG.

  The low-pass filter unit 2232 first extracts a range having a small angle measurement result error from the angle measurement result from the angle recording unit 225. In the entire evaluation period for evaluating the angular spread, about 130 consecutive CPIs are detected in FIG. 18. Among them, for example, a plurality of small periods of about 40 CPI are defined. The small periods may be divided over the whole period, but may overlap each other. For example, three small periods shifted by 40 CPI may be defined, but 18 shifted by 5 CPI may be defined. Here, the latter method that can better extract a small range of error in the angle measurement result will be described as an example.

  The low-pass filter unit 2232 calculates an error average value of the angle measurement results within each small period, for example, RMSE (square root of square error average). The angle recording unit 225 or other average angle calculation unit (not shown) tracks the angle measurement result of each main reflection position, calculates a reference value for error calculation, and outputs the reference value to the low-pass filter unit 2232 at the same time. To do. When the reference value changes every moment within the entire evaluation period, all the changed values are input to the low-pass filter unit 2232. The low-pass filter unit 2232 calculates RMSE from the difference between the reference value and the unprocessed angle measurement result.

  Subsequently, the low-pass filter unit 2232 selects the small period with the smallest RMSE. In the example of FIG. 18B, for example, a small period of 40 CPI corresponding to the range 700 m to 1100 m is selected, and in the example of FIG. 18C, a period of 40 CPI corresponding to 1100 m to 1500 m is selected. .

  At that time, the angle measurement RMSE within a short period may be simply evaluated, but it may be evaluated so as to exclude a portion where normal glint noise is generated as follows. As shown in the example of FIG. 13, glint noise may occur regardless of multipath. In such a case, even if glint noise occurs and the angle measurement RMSE is large, the interference wave is generated. I have not joined. Therefore, for example, first, the maximum value of the power in the entire period is detected, and the angle of the point where the power is a certain value, for example, 10 dB or less, is excluded from the RMSE calculation. Calculate Next, a small period is selected in which the ratio of the points that are excluded from the target within the small period is equal to or less than a certain value and the calculated angle measurement RMSE is the minimum.

  This eliminates the influence of normal glint noise and makes it possible to evaluate the angle measurement error due to interference. Therefore, the length of a small period can be made longer than when RMSE is simply calculated. As a result, the speed of variation for determining the filter coefficient can be detected with high accuracy.

  Next, the speed of power fluctuation within these small periods is detected. Evaluation of the speed of power fluctuation may be performed by, for example, Fourier-changing the power waveform during the period to obtain a spectrum and evaluating an approximate bandwidth. Alternatively, a very high-speed change may be removed in advance with a loose low-pass filter (for example, a moving average filter with a small number of taps) and then the root mean square of the differential coefficient may be evaluated. For example, in FIG. 18B, the vibration speed is extracted about three times within 40 CPI, and in FIG. 18C, the speed of fluctuation corresponding to about one vibration is extracted.

  Subsequently, the characteristics of the low-pass filter corresponding to the extracted fluctuation speed are determined. The low-pass filter may be either IIR (infinite impulse response) or FIR (finite impulse response). If it is IIR, a time constant corresponding to the extracted fluctuation speed is determined, and if it is FIR, a tap coefficient corresponding to the extracted fluctuation speed is set. In either case, the characteristics of the filter are determined so as to have a transfer function that allows passage of fluctuations about the speed of the extracted fluctuations and removes fluctuations faster than that. In the case of a specific filter type, for example, IIR, the type such as Chebyshev and Butterworth and the normalized shape determined by the order may be determined in advance. At this stage, only the bandwidth is determined. Change it.

  FIG. 19 shows a power fluctuation waveform after filtering in this way. 19A is after the filter of FIG. 18B, and FIG. 19B is after the filter of FIG. 18C. When the filter for obtaining FIG. 19 (b) is applied to FIG. 18 (b), the irregularities that should be detected are removed, and the filter for obtaining FIG. 19 (a) is applied to FIG. 18 (c). Then, the unevenness to be removed cannot be removed. However, using the angle measurement results as described above, select a part with relatively little interference, that is, a part where the original characteristic of the main reflection position is good, and determine the filter coefficient from there. Thus, a filter having an appropriate characteristic can be applied.

  The fluctuation range determination unit 2233 shown in FIG. 17 determines the fluctuation range of the power waveform obtained in this way. The details regarding the fluctuation range determination are the same as those of the fluctuation determination unit 1233 shown in FIG. In addition, the fluctuation range determination unit 2233 uses the low fluctuation point selection unit to specify information for specifying the main reflection position, such as the position or processing identification number, of the main reflection position whose fluctuation range is equal to or less than the threshold. To 226. When there are a plurality of main reflection positions that are equal to or less than the threshold value, the fluctuation range determination unit 2233 outputs each of the fluctuation ranges with the respective fluctuation ranges attached, or ranks the output according to the magnitude of the fluctuation range and outputs the result.

  The frequency counting unit 2234 detects the relative spread for the main reflection position whose fluctuation range is equal to or greater than the threshold value. Details are the same as described above. However, when the original waveform is largely disturbed as shown in FIG. 18, it is often difficult to detect symmetry. In such a case, since a point with high symmetry cannot be detected, it is only necessary to detect the peak or the bottom with the entire range as one range without forcibly detecting it. When a peak is detected based on the waveform of FIG. 19, there are 10 peaks at the center shown in FIG. 19 (a) and 4 peaks at the back shown in FIG. 19 (b). Using the back as a reference, it can be calculated that the center has a 2.5-fold spread. The frequency counting unit 2234 outputs the detection result to the relative spread detection unit 224.

  The relative spread detection unit 224 performs the same processing as the relative spread calculation unit 124 of FIG. The relative spread detection unit 224 outputs information indicating that there is no relative spread and no spread corresponding to each main reflection position.

  The low fluctuation point selection unit 226 receives the main reflection position whose fluctuation range is equal to or less than the threshold value from the fluctuation range determination unit 2233. The low fluctuation point selection unit 226 is supplied with the result recorded in the angle recording unit 225, and selects the notified angle recording result of the main reflection position with a small fluctuation range from the result supplied from the angle recording unit 225. Output. Since the main reflection position with a small fluctuation width output in this way often indicates the target angle relatively accurately, this is used for generating a guidance signal for following the target. When there are multiple main reflection positions with a small fluctuation range, in the subsequent stage, when using the angle information to follow the target, select the main reflection position having the tendency of angle fluctuation most suitable for the usage method and Outputs angle measurement results. For example, the angle of the main reflection position with the smallest power fluctuation may be output. Alternatively, the main reflection position with the smallest angle variation may be selected, or the main reflection position with the smallest missing measurement may be selected.

  Next, the operation of the target detection apparatus configured as described above will be described. FIG. 20 is a diagram illustrating an outline of a flowchart regarding the processing operation of the target detection device 20 according to the second embodiment.

  First, the target detection apparatus 20 receives a radar wave by the high resolution radar unit 21 (step S201). The high resolution radar unit 21 generates an amplitude waveform in which the amplitude of the range corresponding to the main reflection point has a peak, and outputs the amplitude waveform to the relative spread detection unit 22.

  The target detection apparatus 20 extracts the main reflection position by the main reflection position extraction unit 221 (step S202), and records the power about the extracted main reflection position during the processing period preset by the power recording unit 222. (Step S203). Moreover, the target detection apparatus 20 records the arrival angle of the extracted main reflection position for the same period as the period during which the power is recorded by the angle recording unit 225 (step S204).

  Subsequently, the target detection device 20 calculates the frequency of power fluctuation when the frequency calculation unit 223 causes significant power fluctuation within the processing period for each main reflection position. At this time, the target detection apparatus 20 determines whether or not the power fluctuation width within the processing period exceeds a preset threshold by the fluctuation width determination unit 2233 (step S205). When the power fluctuation width exceeds the threshold (Yes in step S205), the target detection apparatus 20 calculates the number of peaks or bottoms within the processing period by the frequency counting unit 2234 (step S206). When the power fluctuation width does not exceed the threshold value (No in step S205), the target detection apparatus 20 outputs the angle recording result of the main reflection position where the power fluctuation width does not exceed the threshold value by the low fluctuation point selection unit 226, and performs processing. The process proceeds to step S209.

  After calculating the number of peaks or bottoms in step S206, the target detection apparatus 20 outputs the calculated frequency in association with the position in the range direction within the target of each main reflection position by the relative spread calculation unit 224 ( Step S208).

  Subsequently, the target detection device 20 determines whether or not processing has been performed for all of the extracted main reflection positions (step S209), and when processing has not been performed for all of the main reflection positions (No in step S209), The process proceeds to step S203, and the processes in steps S203 to S208 are repeated until the processes for all the main reflection positions are completed. When the process is completed for all the main reflection positions (Yes in step S209), the target detection apparatus 20 ends the process.

  As a very simplified modification of the above, if a small period with a small error in the angle measurement result is selected and the speed of fluctuation is detected there, the speed may be used as the frequency of relative spread as it is. Alternatively, a portion where the error in the angle measurement result is small may be selected in a wider range, and the frequency may be calculated only from the portion where the error is small. For example, remove the range where the error of the angle measurement result is more than a certain value, calculate the peak period, bottom period, or twice the length between the peak and bottom from the remaining range, and average it It may be converted into frequency.

  As described above, in the second embodiment, a portion where the amplitude of the reflected radar wave is high is extracted as the main reflection position. And the target detection apparatus 20 calculates the frequency of the power fluctuation | variation in a main reflection position, and calculates | requires the relative spread in a target based on the frequency of the power fluctuation | variation. Thereby, even if the target is uncertain, the difference in relative spread can be calculated for different positions from the difference in the speed of power fluctuation. Further, power fluctuation due to a change in the interference state fluctuates with a very large fluctuation width exceeding 10 dB, so that detection with a low SNR is possible, and even a long-distance target can be detected.

  In the second embodiment, the low-pass filter unit 2232 determines that a portion where the error in the angle measurement result is small is a portion where the influence of the interference wave is small, such as a multipath, and uses the low-pass filter based on that portion. The characteristics are determined. As a result, the low-pass filter unit 2232 can eliminate the influence on the detection of power fluctuations caused by interference of multipath and direct reflected waves.

  In the second embodiment, the low fluctuation point selection unit 226 outputs an angle recording result for a main reflection position whose power fluctuation width is equal to or smaller than a threshold value. This output is used to generate a guidance signal for following the target. As described above, the target detection device 20 can effectively generate a signal effective for generating the induction signal in addition to detecting the relative spread of the target.

  Therefore, according to the target detection apparatus 20 according to the present embodiment, it is possible to estimate the shape including the target cross-range direction even when there is target fluctuation with a small number of antennas and a low SNR.

  In the second embodiment, the relative spread detection unit 22 having the functional configuration of FIG. 15 has been described as an example, but the configuration of the relative spread detection unit is not limited to this. For example, as shown in FIG. 21, the low variation point selection unit 226 may not be provided. Further, as shown in FIG. 22, the low-pass filter unit 2232 of the frequency calculation unit 223 may determine the filter coefficient without receiving the angle recording result from the angle recording unit 225.

(Third embodiment)
FIG. 23 is an embodiment related to the guidance device according to the third embodiment. FIG. 23 is a block diagram illustrating a functional configuration of the guidance device according to the third embodiment. The guidance device illustrated in FIG. 23 includes the target detection device 10 or 20 described in the first or second embodiment, and a guidance signal generation unit 31.

  The target detection devices 10 and 20 obtain the target relative shape, the target range, the target angle, and, as necessary, the target moving speed, as shown in the first and second embodiments, as a guidance signal generation unit 31. Output to. The target range, the target angle, and the target moving speed are parameters used in a normal guidance device, and thus description thereof is omitted.

  The induction signal generation unit 31 outputs a control signal for controlling the traveling direction, the moving speed, and the like to a flying unit driving unit (not shown).

  The relative shape of the target generated by the target detection devices 10 and 20 is used to detect whether or not the following target is a correct target. This occurs, for example, when the target is composed of a plurality of separate objects and cannot be distinguished at a distance but can be resolved by approaching. Or it occurs when the target detects the approach of the flying object carrying the guidance device and releases decoy or chaff.

  The guide signal generator 31 stores in advance a rough shape of the target, for example, vertical and horizontal lengths, a part having a characteristic spread, and the detected relative shape of the target is collated with it. , Determine whether it fits. If it is compatible, the tracking of the target being followed is continued, and if it is not compatible, the target detection devices 10 and 20 are instructed to search for other reflected waves in the vicinity, and the target to be tracked is searched. ,decide.

  Although several embodiments have been described, these embodiments have been presented by way of example and are not intended to limit the scope of the invention. These embodiments can be implemented in various other forms, and various omissions, replacements, and changes can be made without departing from the spirit of the invention. These embodiments and their modifications are included in the scope and gist of the invention, and are also included in the invention described in the claims and the equivalents thereof.

DESCRIPTION OF SYMBOLS 10, 20 ... Target detection apparatus 11, 21 ... High-resolution radar part 111, 211 ... Antenna 112, 212 ... Radar processing part 1121, 1212-1, 1211-2 ... Reception RF part 1122, 2122 ... Doppler frequency detection part 1123 2123 ... Movement speed calculation unit 1124, 2124 ... Frequency step representative value extraction unit 1125, 2125 ... Correction unit 1126, 2126 ... Range calculation unit 1127 ... Amplitude extraction unit 2127 ... Amplitude / phase extraction unit 2128 ... Angle calculation unit 12, 22 ... Relative spread detection units 121, 221... Main reflection position extraction units 122, 222 .. power recording units 123, 223... Frequency calculation units 123 1, 2231, power correction units 1232, 2232, low-pass filter units 1233, 2233. , 2234... Frequency counting section 124, 224. Spread calculator 225 ... angle recording unit 226 ... low variation point selection section 30 ... guide device 31 ... induction signal generator

Claims (12)

A plurality of ranges in which a measurement value peaks within a range measurement range; and a radar unit that acquires an amplitude value of the peak ; and
A main reflection position extraction unit that selects a main reflection position in the range direction within the target based on the reception status of the range and the amplitude value;
A power recording unit that records an amplitude value for each main reflection position during a preset processing period;
It is determined for each main reflection position whether or not the fluctuation range of the amplitude data recorded in the processing period exceeds a preset threshold value, and amplitude data of the main reflection position whose fluctuation range exceeds the threshold value is output. A fluctuation range determination unit;
A frequency counting unit that counts the variation frequency of the amplitude data of the main reflection position from the variation range determination unit;
A target detection apparatus comprising: a relative spread calculation unit that associates the main reflection position and the count result with a position in a range direction and a relative spread in an angle direction within the target. A power correction unit that corrects the amplitude data recorded in the processing period so as to remove the influence of a change in the range within the processing period;
The target detection apparatus according to claim 1, wherein the fluctuation range determination unit performs the determination process on the corrected amplitude data. A low-pass filter unit that applies low-pass filtering to the corrected amplitude data;
The target detection apparatus according to claim 2, wherein the fluctuation range determination unit performs the determination processing on the low-pass filtered amplitude data. The radar unit performs angle measurement for a plurality of ranges where the measurement value reaches a peak, and further obtains an arrival angle,
The target detection apparatus according to claim 1, further comprising an angle recording unit that records an arrival angle for each main reflection position during the processing period. The fluctuation range determination unit outputs a main reflection position where the fluctuation range does not exceed the threshold value,
Of the arrival angle data for each of the main reflection positions recorded in the angle recording unit, a low variation point selection unit that outputs arrival angle data for the main reflection position from the variation range determination unit is further provided. The target detection apparatus according to claim 4. A power correction unit that corrects the amplitude data recorded in the processing period so as to remove the influence of the range change in the processing period;
Variation in arrival angle data for each of the main reflection positions recorded in the angle recording unit is calculated for each of a plurality of shorter sub-periods within the processing period, and the main reflection in a sub-period where the variation is minimized. A low-pass filter that determines a low-pass filtering characteristic based on a change in position amplitude data, and further applies the low-pass filtering of the characteristic to the corrected amplitude data;
The target detection apparatus according to claim 4, wherein the fluctuation range determination unit performs the determination processing on the low-pass filtered amplitude data. The fluctuation range determination unit outputs a main reflection position where the fluctuation range does not exceed the threshold value,
Of the arrival angle data for each of the main reflection positions recorded in the angle recording unit, a low variation point selection unit that outputs arrival angle data for the main reflection position from the variation range determination unit is further provided. The target detection apparatus according to claim 6.   The low-pass filter unit excludes an arrival angle corresponding to a range lower than a preset value from the maximum value of the corrected amplitude data among the corrected amplitude data from the calculation target of the variation. The target detection apparatus according to claim 6.   The frequency counting unit counts the variation frequency of the amplitude data so as to detect the symmetry of the amplitude data of the main reflection position from the variation range determination unit and avoid the vertex of the target posture angle. The target detection apparatus according to claim 1.   The target detection apparatus according to claim 1, wherein the radar unit is a synthetic band radar. A target detection device according to at least one of claims 1 to 10;
A guidance signal generation unit configured to generate a signal for guiding the flying object based on the relationship between the position in the range direction within the target and the relative spread in the angle direction from the target detection device; A guidance device characterized by that. A plurality of ranges in which the measurement value becomes a peak within the range measurement range, and the amplitude value of the peak are acquired,
Based on the reception status of the range and the amplitude value, select a main reflection position in the range direction within the target,
The amplitude value for each main reflection position is recorded for a preset processing period,
It is determined for each main reflection position whether or not the fluctuation range of the amplitude data recorded in the processing period exceeds a preset threshold value, and amplitude data of the main reflection position whose fluctuation range exceeds the threshold value is output. ,
Count the fluctuation frequency of the amplitude data of the output main reflection position,
A target detection method comprising associating the main reflection position and the count result with a position in a range direction and a relative spread in an angular direction within the target.

Images