JP2006343290A - Imaging radar device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To reduce estimation error of compensation amount increasing due to the influence of interference among a plurality of reflection points existing in the same range, when unnecessary range variations are estimated. <P>SOLUTION: The imaging radar device is equipped with a two-dimensional Fourier transformation compensating circuit 20, which includes a time series data retrieving means 201 for retrieving a time series data from range history, a Doppler history generating means 202 for generating a Doppler history by applying short-time Fourier transformation, a history clipping means 203 for clipping the short-time Doppler history, Doppler variability estimating means 204 to 206 for estimating variability of Doppler frequency from the short-time Doppler history, a range variation calculating means 207 for estimating range variations between objects and the radar, based on the variability of Doppler frequency, a compensation calculating means 208 for calculating compensation amount based on the range variation, and a motion compensator 21 for compensating variations of unnecessary range and phase of the range history, based on the compensation amount. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention receives a reflected wave from a target by transmitting a radio wave to a moving target, generates a range history that is a time history of the range profile, and creates a target image by cross-range compression of the range history. The present invention relates to an obtained image radar apparatus.

  Image radars such as SAR (Synthetic Aperture Radar) and ISAR (Inverse SAR) generate reflection points on the observation target (hereinafter referred to as the target) by the range that is the distance from the radar and the relative motion between the radar and the target. The radio wave image related to the target is generated by separating at the Doppler frequency.

  However, the range of each reflection point and the Doppler change on the image due to the influence of unnecessary components included in the relative motion, resulting in a blurred image. In contrast, the following measures are taken in the conventional image radar apparatus.

  A conventional image radar apparatus receives a high-frequency pulse modulated by a transmitter and irradiated to a target via a transmission / reception antenna, and a time history of a range profile obtained by pulse compression (hereinafter referred to as a range history). The Doppler profile time history (hereinafter referred to as the Doppler history) obtained by first short-time Fourier transforming the received signal to the above-mentioned range and Doppler changes caused by unnecessary relative motion between the target and the radar. (See, for example, Patent Document 1).

  On the Doppler history, there are a plurality of reflection points on the target. In the conventional method, these trajectories are regarded as straight lines having the same inclination, and a plurality of integration paths having different inclinations passing through the origin are set on a spectrum obtained by two-dimensional Fourier transform of the amplitude distribution of the Doppler history. In the conventional method, a straight line on the spectrum is specified based on the maximum value of the spectrum integration result along each integration path, and a primary change with respect to time of the Doppler frequency of the reflection point on the target is determined based on the information. Estimated.

  Further, the change in the secondary range and phase is estimated from the information (the secondary change in phase corresponds to the primary change in Doppler). Finally, based on this estimation result, it is possible to perform compensation so as to eliminate the change in the range of each reflection point and the Doppler frequency, and to remove image blur. In addition, after using this process as a pre-process, another method can be applied to further improve the compensation accuracy.

JP 2001-74832 A (first page, FIG. 1)

  However, the prior art has the following problems. In order to perform high-precision motion compensation, the conventional radar apparatus estimates the distance change based on the phase change of each reflection point on the received signal, and the interference between a plurality of reflection points existing in the same range. There is a problem that the estimation error of the compensation amount increases due to the influence.

  Further, when estimating the compensation amount based on the position change of each reflection point on the received signal, there is a problem that the compensation amount estimation accuracy is insufficient.

  The present invention has been made to solve the above-described problems. When estimating an unnecessary distance change, an estimation error of a compensation amount that increases due to the influence of interference between a plurality of reflection points existing in the same range. An object of the present invention is to obtain an image radar apparatus capable of reducing the above.

  The image radar apparatus according to the present invention transmits a radio wave to a moving target, receives a reflected wave from the target, generates a range history that is a time history of the range profile, and performs cross range compression of the range history. In an image radar apparatus for obtaining a target image, time-series data extracting means for extracting time-series data from a range history, and a Doppler history that is a time history of a Doppler profile by applying a short-time Fourier transform to the extracted time-series data Doppler history generating means for generating a history, a history extracting means for extracting a short-time Doppler history that is a Doppler history in a segmented time region from the generated Doppler history, and a Doppler of a locus of each reflection point on the extracted short-time Doppler history Doppler change rate estimation means for estimating the frequency change rate; A range that estimates the Doppler frequency change over time based on the Doppler frequency change rate estimated for a number of short-time Doppler histories and estimates the range change between the target and the radar based on the estimation results Compensation amount to calculate the range compensation amount and phase compensation amount normalized by the range resolution to compensate for the influence of the range history and the unnecessary Doppler change based on the estimated range change A two-dimensional Fourier transform compensation circuit having calculation means and a motion compensator that compensates for an unnecessary range and phase change in the range history based on the calculated range compensation amount and phase compensation amount is provided.

  According to the present invention, an unnecessary distance change is obtained by dividing a history into divided regions and estimating a compensation amount based on a history change rate obtained by applying a two-dimensional Fourier transform to each divided region. It is possible to obtain an image radar apparatus that can reduce an estimation error of a compensation amount that increases due to the influence of interference between a plurality of reflection points existing in the same range when estimating.

A preferred embodiment of an image radar apparatus of the present invention will be described below with reference to the drawings.
The image radar apparatus of the present invention is characterized in that a compensation amount with a reduced estimation error can be estimated by performing Fourier transform on the history divided for each divided region.

Embodiment 1 FIG.
FIG. 1 is a configuration diagram of an image radar apparatus according to Embodiment 1 of the present invention. In FIG. 1, the transmitter 1 generates and amplifies a broadband pulse that irradiates a target. The transmission / reception switch 2 switches between transmission and reception. The transmission / reception antenna 3 is an antenna that transmits the broadband pulse generated by the transmitter 1 to the target and receives the reflected broadband pulse.

  The receiver 4 receives the signal received by the transmission / reception antenna via the transmission / reception switch 2. Further, the pulse compressor 5 performs pulse compression on the received signal. The pulse-compressed signal undergoes compensation processing by the coarse compensation circuit 10, the 2D-FT compensation circuit 20, and the fine compensation circuit 30 and is input to the cross range compressor 40.

  The coarse compensation circuit 10 includes a coarse compensation amount estimator 100 and a motion compensator 11 and performs coarse compensation of motion. The rough compensation amount estimator 100 estimates an unnecessary motion component that causes blurring and image quality degradation at the time of image reproduction with rough accuracy, and sets a compensation amount for compensating for the motion component. Then, the motion compensator 11 compensates the motion using the estimated motion compensation amount.

  The 2D-FT compensation circuit 20 is a two-dimensional Fourier transform (2D-FT) compensation circuit, and includes a 2D-FT compensation amount estimator 200 and a motion compensator 21 and compensates for motion. The 2D-FT compensation amount estimator 200 estimates an unnecessary motion component based on the two-dimensional Fourier transform, and sets a compensation amount for compensating for the motion component. The motion compensator 21 compensates for the motion using the estimated motion compensation amount.

  The fine compensation circuit 30 includes a fine compensation amount estimator 300 and a motion compensator 31, and performs fine motion compensation. The fine compensation amount estimator 300 finely estimates unnecessary motion components and sets the compensation amount. The motion compensator 31 compensates for the motion using the estimated motion compensation amount.

  The cross-range compressor 40 is a cross-range compression circuit that obtains a radar image after aperture synthesis by Fourier-transforming a signal after motion compensation is sequentially performed by the coarse compensation circuit 10, the 2D-FT compensation circuit 20, and the fine compensation circuit 30. It is.

  The image radar apparatus according to the first embodiment uses a 2D-FT compensation amount estimator 200 in the 2D-FT compensation circuit 20 to estimate an unnecessary distance change between a plurality of reflection points existing in the same range. It is characterized in that the estimation error of the compensation amount that increases due to the influence of the interference is reduced. The detailed configuration of the 2D-FT compensation amount estimator 200 will be described next.

  FIG. 2 is an internal configuration diagram of the 2D-FT compensation amount estimator 200 in the image radar apparatus according to Embodiment 1 of the present invention. The 2D-FT compensation amount estimator 200 in FIG. 2 includes a time series data extraction unit 201, a Doppler history generation unit 202, a history extraction unit 203, a 2D-FT unit 204, a spectral image inclination estimation unit 205, and a Doppler change rate calculation unit 206. A range change calculating unit 207 and a compensation amount calculating unit 208, and further includes an STFT score storage unit 291, a history extraction score storage unit 292, and a spectral image inclination storage unit 293 as storage units.

  Further, FIG. 3 is an internal configuration diagram of the spectral image inclination estimation means 205 in the first embodiment of the present invention, and shows a detailed configuration of the spectral image inclination estimation means 205 in FIG.

  3 includes spectral image inclination estimation main means 211a and 211b, zero inclination switching means 212, known inclination adding means 213, 2D-FT means 214, known inclination removing means 215, and spectral image inclination output. Consists of means 216.

  Here, the spectral image inclination estimation main means 211a and 211b have the same function, the spectral image inclination estimation main means 211a corresponds to the first spectral image inclination estimation main means, and the spectral image inclination estimation main means 211b. This corresponds to the second spectral image inclination estimation main means. The 2D-FT means 214 has the same function as the 2D-FT means 204 and corresponds to a second 2D-FT means.

  Further, FIG. 4 is an internal configuration diagram of the spectral image inclination estimation main unit 211 in Embodiment 1 of the present invention, and shows a detailed configuration of the spectral image inclination estimation main unit 211a and 211b in FIG.

  The spectral image inclination estimation main means 211 in FIG. 4 includes a Y value setting means 221, an inclination candidate setting means 222, an X continuous value calculation means 223, an X cutoff value setting means 224, an X difference value calculation means 225, an adjacent X value integration. Means 227, weighted normalization vector generation means 226, and peak slope detection means 228 are provided, and an integration result storage unit 294 is provided as a storage unit.

  Hereinafter, specific processing in the image radar apparatus according to the first embodiment will be described based on the configuration diagrams of FIGS. The transmitter 1 generates a high-frequency pulse modulated into a signal (chirp signal) whose frequency changes with time, and supplies it to the transmission / reception antenna 3 via the transmission / reception switch 2.

  The transmitting / receiving antenna 3 transmits the supplied high frequency pulse to the target 6. The transmitted high-frequency pulse is reflected by the target 6, and the reflected signal (echo) enters the transmission / reception antenna 3 and is demodulated by the receiver 4 via the transmission / reception switch 2.

  The signal demodulated by the receiver 4 is obtained by delaying the time required for the round trip of radio wave propagation between the image radar apparatus and the target 6 with respect to the instantaneous frequency of the transmission signal. Therefore, the pulse compressor 5 can obtain an impulse at a time corresponding to the delay by applying a matched filter to the received signal using the transmission signal. This improves the range resolution. Hereinafter, this compressed waveform is referred to as a range profile.

  The above-described series of processes of irradiating the target 6 with high frequency pulses and pulse-compressing the received signal is repeated while changing the relative positional relationship between the target 6 and the radar, and the obtained range profile is accumulated. In the following description, the number of the irradiated high frequency pulse is referred to as a hit h (h = 0, 1, 2,..., H−1, where H is the number of hits).

  If the sampling frequency of the range is Bs, the sampling period is 1 / Bs. The received signal whose delay from the sampling start time is t [s] is sampled at the point t / (1 / Bs) = tBs.

  Hereinafter, this sampling number is expressed as a range cell m (m = 0, 1, 2,..., M−1: where M is the number of range cells). Note that the delay t [s] corresponds to a distance difference tC / 2 [m] where the light speed is C [m / s]. Therefore, one range resolution cell is C / (2Bs) [m].

  Below, the range profile for the accumulated H hits is given as a two-dimensional array S_0 (m, h). Since S_0 (m, h) gives a time history of the range profile, this is called a range history.

  Next, the compression principle in the cross range direction orthogonal to the range will be described. FIG. 5 is an explanatory diagram relating to the compression principle in the cross-range direction according to Embodiment 1 of the present invention. When the target moving in the xy plane shown in FIG. 5 is appropriately compensated so as to fix the distance to the virtual center of gravity G on the target, the same target is fixed in space for the received signal from the target. Therefore, it can be considered that the received signal is equal to the received signal when the equivalent rotation is performed around G.

  The unit vector in the centroid position vector direction with respect to the radar is ii_0, the angular velocity vector giving the equivalent rotational motion (or the rotational motion in the case of a target that performs only the rotational motion) is ww, and the center frequency corresponds. When the wavelength is λ, the Doppler frequency f_p of the point P where the position vector with G as a reference is given by rr_p is approximately given by the following equation (1).

  That is, it can be seen that the Doppler frequency f_p is a value proportional to the position of the point P in the (ii_0 × ww) direction.

  Here, since ii_0 and (ii_0 × ww) are orthogonal to each other, the two-dimensional distribution of the range and the Doppler frequency related to the received signal from the target defines the reflection intensity distribution on the target as ii_0 and (ii_0 × ww). It corresponds to the image projected on the two-dimensional plane.

  FIG. 6 is a view showing an example of an image obtained by projecting the target on the two-dimensional plane in the first embodiment of the present invention. The ISAR obtains a range and Doppler distribution of a received signal, that is, a two-dimensional radio image related to a target, by performing Fourier transform on the range history in the hit direction for each range.

  Here, FIG. 7 is a diagram showing a trajectory in which a curve in an image represents a range at each time of a plurality of reflection points in the first embodiment of the present invention. FIG. 8 is a diagram showing a locus in which a curve in an image represents a Doppler position at each time of a plurality of reflection points in Embodiment 1 of the present invention.

  As described above, since each reflection point is separated by the range and the Doppler frequency in the ISAR, for example, as shown in FIGS. 7 and 8, the range and the Doppler frequency of each reflection point change in the received signal. Then the image will be blurred.

  Therefore, it is necessary to compensate so as to remove changes in the range and Doppler frequency. First, the coarse compensation amount estimator 100 in the coarse compensation circuit 10 uses a target tracking, a result of estimation processing based on a conventional received signal, or the like to change the distance between the radar and the center of gravity at each hit h. Estimate r_raw (h).

  Based on the estimated distance change r_raw (h), a phase change φ_raw (h) and a range change (hereinafter referred to as a normalized range change) m_raw (h) normalized by a distance width corresponding to the sampling interval, It calculates with following Formula (2) and (3).

  Next, based on these values, the motion compensator 11 compensates the range and phase of S_0 (m, h) by the following equation (4), and obtains the range history s_1 (m, h) after rough compensation. obtain.

  However, Align which is a process in the equation (4) is a process for compensating the range, and is given by the following equations (5) to (7).

  Here, assuming that the compensation amount of the range is given with an accuracy of one range cell or less, the range compensation is performed based on the correction of the phase of the spectrum of the range as shown in the equations (5) to (7). ing.

  After compensation by the coarse compensation circuit 10, the 2D-FT compensation circuit 20 is obtained by performing a short time Fourier transform (STFT) on the received signal in the internal 2D-FT compensation amount estimator 200. The compensation amount is estimated based on the Doppler history.

  The time-series data extracting unit 201 in FIG. 2 extracts a data string d (h) for generating a Doppler history from the range history s_1 (m, h). Here, FIG. 9 is a schematic diagram of the range history in the first embodiment of the present invention, and the thick lines in the drawing represent the loci of several reflection points.

  The time-series data extraction unit 201 sets a range extraction range (range range minimum value m_st, range maximum value m_end) indicated by a horizontal line in FIG. 9 so that the locus of the reflection point of interest is included, The cutout range is summed by the following equation (8) to obtain d (h).

  Here, d (h) is extended to the range of h <0 and h> H−1 for the preparation after the next stage, and the value is set to 0. Note that m_stt and m_end may be set to m_st = 0 and m_end = M−1 so that the data in the entire range is summed.

  Next, the Doppler history generation means 202 first generates a Doppler history generation array d_cut (h ′, h_c) given by the following equation (9) from d (h).

  Here, D_cut is the number of cut-out points from which data is cut out from d (h), and a value stored in advance in the STFT point storage unit 291 (hereinafter, this value is referred to as STFT point number) is read out. D_hokan is a zero-padded score at the time of Doppler history generation, h_c is a cut center hit, and h ′ is a hit at the time of Doppler history generation.

  Here, by setting D_cut ≦ D_hokan, the Doppler history is interpolated to zero by D_hokan / D_cut times to refine the Doppler history.

  Next, as shown in the following equation (10), the Doppler history generation unit 202 performs Fourier transform in the h ′ direction on d_cut (h ′, h_c) to obtain Doppler history dop_hist (k ′, h_c).

  However, k 'is a Doppler cell number in Doppler history. Here, considering the processing in the next stage, dop_hist (k ′, h_c) is expanded to the range of h_c <0 and h_c> H−1, and 0 is set.

  On the Doppler history, the locus of a plurality of reflection points appears. When each reflection point is not blurred in the Doppler direction on the ISAR image, the locus of each reflection point on the Doppler history is an equal Doppler straight line.

  However, when each reflection point on the ISAR image is blurred in the Doppler direction, the locus of each reflection point changes in the Doppler direction as the hit changes (that is, the Doppler frequency of each reflection point changes as the hit changes). Change).

  However, since the change in the Doppler frequency is caused by the change in the position of the same target, the change amount itself is expected to be substantially equal at the total reflection point. In the following, the amount of change is estimated.

  The history cutout means 203 cuts out the input history (here, Doppler history) based on the history cutout points D2_cut accumulated in the history cutout point storage unit 292.

  Hereinafter, the input history is generally expressed as img00 (y, x0). Here, the y axis in img00 (y, x0) corresponds to the k ′ axis in dop_hist (k ′, h_c), and the x0 axis corresponds to the h_c axis, respectively.

  As shown in the following equation (11), the history cutout unit 203 cuts out a short-time history having a width in the x direction D2_cut from img00 (y, x0) while changing x_c which is the central x value, (y, x, x_c) is stored in the history cut-out point storage unit 292.

  FIG. 10 is a view showing an example of short-term history extraction in the first embodiment of the present invention. In FIG. 10, a horizontal line area is cut out when x_c = x_c1, and a hatched area is cut out when x_c = x_c2. As the cutout width D2_cut becomes shorter, the locus on the image can be regarded as a straight line.

  That is, when D2_cut is shortened, it is expected that the problem of estimating the change of the trajectory in each segmented area can be simplified to the problem of estimating the inclination of the straight line having the same inclination on the image.

  Next, the 2D-FT means 204 performs a two-dimensional Fourier transform on the two-dimensional image obtained for each x_c to obtain a spectrum.

  Here, in img0 (y, x, x_c), a two-dimensional array related to y and x determined when x_c is given is generally referred to as img (y, x) again. Here, the number of pixels in the y direction of img (y, x) is N_y, and the number of pixels in the x direction is N_x.

  After detecting the amplitude of img (y, x), a two-dimensional Fourier transform is performed to obtain a spectrum img (f_y, f_x) of img (y, x) as shown in the following equation (12).

  Here, f_x and f_y are spatial frequencies in the x and y directions, respectively. Next, the spectral image inclination estimation means 205 estimates the inclination of the locus on the complex image img (y, x) based on the input spectral image img (f_y, f_x).

  FIG. 11 and FIG. 12 are explanatory diagrams when estimating the inclination in the first embodiment of the present invention. Arbitrary straight lines (FIG. 11) having the same slope and the same y-intercept (the number of pixels that advance in the y direction while moving one pixel in the x direction) on the image | img (y, x) | As described above, the inclination a is converted into a straight line given by the following equation (13) (see FIG. 12).

  As shown in FIG. 3, the spectrum image inclination estimation main means 211a inside the spectrum image inclination estimation means 205 is a spectrum image fimg (f_y, f_x) input from the 2D-FT means 204, passes through the origin, and has an inclination. A plurality of different integration paths are set, and a straight line in the spectrum image is detected based on the maximum value of the integration values along each integration path.

  As shown in FIG. 4, the pY value setting unit 221 inside the spectral image inclination estimation main unit 211a sets the value in the f_y direction of the spectrum image at the point 2N_x in order to set the integration path. First, fy0 (i_ny) of the following equation (14) is set.

  Next, assuming that the pixel number in the f_y direction of the spectrum image is given by 0 to N_y−1, f_y (i_ny) is given by the modulo operation of the following equation (14) ′.

  The inclination candidate setting means 222 sets N_A types of inclinations a_cand (i_nA) (i_nA = 0, 1,..., N_A−1), and sequentially outputs them. Next, the X continuous value calculation unit 223 obtains the value of f_x (i_ny) for each f_y0 (i_ny) by the following equation (15) for each A_cand (i_nA) output from the inclination candidate setting unit 222.

  The f_x (i_ny) obtained here is a continuous value. However, when integration is performed, it is necessary to discretize and express the pixel in order to specify the pixel.

  Here, the integral value is calculated by linear interpolation of two adjacent pixels in the vicinity of the continuous value. For this purpose, the X cut-off value setting means 224 obtains a value f_xom0 (i_ny) obtained by cutting off the fractional part of f_x (i_ny).

  Next, consider a case where the value of f_xom0 (i_ny) is outside the spectrum image, that is, f_xom0 (i_ny) <0 or f_xom0 (i_ny)> N_x-1.

  On the spectrum image, the trajectory that exits from one end of the image appears from the opposite end due to folding, and here, the value of fx after folding f_xom (i_ny) is obtained by the modulo operation of the following equation (16). Get.

  Similarly, a pixel number f_xomp1 (i_ny) that is one greater than f_xom (i_ny) is obtained by the following equation (17).

  FIG. 13 is an explanatory diagram of the integration process on the spectrum image in the first embodiment of the present invention. In FIG. 13, for the continuous value f_x (i_ny) indicated by a black circle, cells corresponding to f_xom (i_ny) are indicated by cross hatching, and cells corresponding to f_xomp1 (i_ny) are indicated by vertical hatching. Show.

  The X difference value calculating means 225 calculates the difference value delta_fx (i_ny) between f_x (i_ny) and fxom0 (i_ny) by the following equation (18).

  The weighted normalization vector generation means 226 sets the weight vector W (i_ny) for each f_y at the time of integration by the following equation (19).

  Here, Win_data (fimg (f_y, f_x)) is a weight determined from the input spectrum image fimg (f_y, f_x). In general, the signal level of the signal fimg (f_y, f_x) decreases as the distance from the direct current increases. Therefore, it is useful to raise the signal on the high frequency side far from the direct current in order to improve the estimation accuracy of the slope.

  Based on this, for example, as Win_data (fmg (f_y, f_x)), a weight that makes the maximum amplitude value of the fmg (f_y, f_x) equal in each f_y, and the total of the fimg (f_y, f_x) in each f_y Normalization weights that make power equal can be considered.

  Win (i_ny) is a weight for correcting the weight of the normalization system by Win_data (fimg (f_y, f_x)), and an appropriate weight is used according to the purpose.

  For example, in order to mitigate excessively raising the level on the high frequency side, a weight that decreases the level on the high frequency side, i.e., i_ny = 0 or i_ny = 2N_y−1 in terms of i_ny (for example, Hanning weight) ) Is used.

  The adjacent X value integration means 227 performs line integration by the following equation (20), and stores the integration result s_int in the integration result storage unit 294.

  The spectral image inclination estimation main unit 211a repeats the above processing while changing the inclination A_cand (i_nA).

  Then, after the integration with respect to all the integration paths is completed, the peak inclination detecting means 228 searches for the maximum value of the integrated value, and obtains the inclination at that time as A_est. Then, the inclination a obtained by substituting A_est for A in Expression (13) is output as the estimation result a_est of the inclination of the locus on the input image img (x, y). The inclination estimation result a_est estimated by the spectral image inclination estimation main unit 211a corresponds to the first inclination amount.

  Next, the zero slope switching means 212 in FIG. 3 checks the slope a_est and switches the subsequent processing depending on whether or not this value is zero. First, when the inclination is not zero, the inclination a_est is sent to the spectral image inclination output means 216.

  In addition, when the slope is zero, the slope estimation based on the two-dimensional Fourier transform has a low accuracy near the slope zero and is often erroneously estimated as zero. The process proceeds to the adding unit 213.

  As shown in the following equation (21), the known inclination adding means 213 is arranged so that the inclination of the locus deviates from near zero with low accuracy with respect to img (y, x) input from the history extraction means 203. Compensation of the known slope a_0 is performed.

  The 2D-FT means 214 applies a two-dimensional Fourier transform to this img_2 (y, x) to obtain fimg_2 (f_y, f_x). The spectral image inclination estimation main unit 211b detects a straight line passing through the above-described origin with respect to fimg_2 (f_y, f_x), and obtains a_est2. The inclination estimation result a_est2 estimated by the spectral image inclination estimation main unit 211b corresponds to the second inclination amount.

  The known inclination removing means 215 subtracts the known inclination a_0 added by the known inclination adding means 213 from the inclination a_est2 obtained by the previous spectral image inclination estimation main means 211b, and obtains the obtained a_est as the spectral image inclination output means. To 216. The a_est obtained as the difference corresponds to the third inclination amount.

  The spectral image tilt output unit 216 outputs a_est which is the first tilt amount obtained from the zero tilt switching unit 212 or a_est which is the third tilt amount obtained from the known tilt removing unit 215. Then, the inclination a_est that is the output of the spectral image inclination estimation means 205 is accumulated in the spectral image inclination storage unit 293.

  The 2D-FT compensation amount estimator 200 in FIG. 2 repeats the process from the history extraction by the history extraction unit 203 to the accumulation of the inclination in the spectrum image inclination storage unit 293 while changing x_c that gives the center of the extraction position. . Here, x_c is expressed as h_c in the sense of the center of the hit, and the accumulated gradient is represented as a_est (h_c) as a gradient with respect to h_c.

  When the input history is Doppler history, a_est (h_c) is a change rate of the Doppler frequency cell per hit. Therefore, the Doppler change rate calculation means 206 first converts a_est (h_c) into the change rate α_est [Hz / s] of the Doppler frequency [Hz] per 1 [s] by the following equation (22).

  Here, PRI is a pulse repetition period. Next, the time t (h_c) at each center hit h_c is given by the following equation (23).

  Here, FIG. 14 is an explanatory diagram of a process for estimating the Doppler change rate in the first embodiment of the present invention. As shown in FIG. 14, the P-th order least squares method is applied to each time t (h_c) and α_est (h_c), and the Doppler frequency change rate α_fit (t) is expressed by the following equation (24). This is expressed by a P-th order polynomial with respect to time t.

  The range change calculation means 207 first obtains the Doppler frequency change f_fit (t) at each time t by the following equation (25).

  Next, based on the fact that the relationship between the Doppler frequency change f_fit (t) at each time t and the corresponding distance change r_fit (t) is given by Expression (26), r_fit (t) is expressed by Expression (27). Get like.

  The phase φ_2DFT (t) corresponding to r_fit (t) is given by the following equation (28).

  Also, a normalized range change m_fit (t), which is a range change based on the range sampling period Δ_r, is given by the following equation (29).

  The compensation amount calculation means 208 calculates the phase φ_2DFT (h_c) and the normalized range change m_2DFT (h_c) at each hit time t (h_c) based on the above φ_fit (t) and m_fit (t) and outputs this. To do.

  The motion compensator 21 at the subsequent stage of the 2D-FT compensation amount estimator 200 calculates φ_2DFT (h_c) and m_2DFT (h_c) output from the compensation amount calculation unit 208 as φ_raw (h) and m_raw (h) in Expression (4). Further, the range history s_1 (m, h) obtained from the coarse compensation circuit 10 is substituted for S_0 (m, h) to compensate for unnecessary motion remaining in the range history after the coarse compensation.

  The compensation result by the 2D-FT compensation circuit 20 is represented as s_2 (m, h). In s_2 (m, h), since compensation based on 2D-FT is performed as described above, unnecessary range and Doppler frequency changes are reduced.

  Therefore, there is a possibility that an ISAR image is formed only by applying the below-described cross-range compression to s_2 (m, h), but here, an unnecessary motion component remains in s_2 (m, h). Therefore, the fine compensation circuit 30 is used to further compensate for the remaining unnecessary motion components.

  In the fine compensation circuit 30, the data string d_2 (h) used when estimating the compensation amount from the range history s_2 (m, h) that is the output of the 2D-FT compensation circuit 20 is expressed as s_1 (m H) and substituting s_2 (m, h).

  However, based on the fact that the compensation error has been reduced by the processing of the 2D-FT compensation circuit 20 in the previous stage, the interval between m_stt and m_end for setting the extraction range is set narrower than the interval in the 2D-FT compensation amount estimator 200.

  By narrowing this interval, it is possible to reduce the problem that the compensation accuracy of the compensation amount deteriorates due to interference with reflection points existing in different ranges. Further, the fine compensation amount estimator 300 applies the FFT processing of the following equation (30) to d_2 (h).

  Next, k = k_max that maximizes the absolute value of the signal D_2 (k) after the FFT is searched as in the following equation (31).

  Next, according to the following equation (32), D_2 (k) is circularly shifted by −k_max.

  Here, shift (A (x), xshift) is an operator that circularly shifts the array A (x) by xshift. By this process, the signal component having the maximum power among the signals included in D_2 (k), that is, d_2 (h), becomes the DC component.

  Next, with respect to D_3 (k), as shown in the following equation (33), only a component near the DC component is extracted, and a filter process for suppressing other frequency components is performed.

  Here, Win (k) is an appropriate weighting factor for extracting only signal components near DC and suppressing other frequency components. For example, when this weight is configured as a rectangular filter having a pass bandwidth 2W_k + 1 Doppler cell, Win (k) is given by the following equation (34).

  With the above processing, only signals near the DC component can be extracted. That is, in the estimation of the compensation amount to be performed next, it is possible to reduce the influence of interference between signals at a plurality of reflection points with different Doppler frequencies included in the extracted signal d_2 (h).

  In order to reduce the influence of interference, it is necessary to narrow the passband width of the filter. However, in order to pass the Doppler spread component generated based on unnecessary motion, the signal at the reflection point of interest is passed. Needs to increase the bandwidth of the filter.

  Here, the spread of the Doppler generated based on unnecessary motion generally increases as the unnecessary motion component increases, so the smaller the influence of the unnecessary motion component, the narrower the passband width of the filter and the interference. It becomes easy to reduce the influence of.

  As for s_2 (m, h), it is expected that unnecessary motion components are greatly reduced by the 2D-FT compensation circuit 20 in the previous stage. Therefore, the influence of interference is further reduced by narrowing the passband width. It is expected to be possible.

  Next, the fine compensation amount estimator 300 obtains d_4 (h) by performing inverse FFT on D_4 (k) using the following equation (35).

  Then, for the phase of d_4 (h), applying the removal of phase wrapping every 2π (unwrapping) and the least square method of an appropriate order for the phase after unwrapping are unnecessary as in the conventional method. A phase change based on a simple motion component is obtained.

  Further, this phase change is converted into a distance change based on the relationship of Expression (28). Furthermore, after converting the obtained distance change into a normalized range change, this value and the phase change after applying the least square method are output.

  The motion compensator 31 subsequent to the fine compensation amount estimator 300 performs motion compensation of s_2 (m, h) based on the obtained normalized range change and phase change, and the range history s_3 (m, h after motion compensation). )

  The cross range compressor 40 applies a Fourier transform in the hit direction of the following equation (36) to the obtained range history s_3 (m, h), and finally obtains an ISAR image img (m, k). .

  As described above, according to the first embodiment, the Doppler history is divided into divided regions, and the Doppler change rate of each reflection point is estimated by applying the two-dimensional Fourier transform for each divided region. . By setting the width of the divided area to be narrow, the locus of each reflection point on the Doppler history can be regarded as a straight line. Furthermore, since each reflection point exists on the same target, the rate of change of Doppler is expected to be substantially equal.

  Furthermore, since straight lines with the same inclination on the image are generally converted into straight lines with the same inclination passing through the origin by two-dimensional Fourier transform, even when there are signals of a plurality of reflection points on the Doppler history, for example. By performing a search for a straight line passing through the origin in the spectrum of the Doppler history section area, it is possible to reduce the influence of interference between signals and estimate the compensation amount.

  Furthermore, by performing motion compensation based on the two-dimensional Fourier transform in the 2D-FT compensation circuit 20, the fine compensation circuit can estimate the compensation amount while narrowing the range and the Doppler width, thereby reducing the influence of interference. The compensation amount can be estimated with high accuracy.

  Further, in the 2D-FT compensation circuit 20, since a high-order least-squares method is applied to the rate of change of the Doppler frequency at each reflection point on each piecewise Doppler history to estimate the change in Doppler frequency, interference occurs. The amount of compensation can be estimated up to higher-order components while reducing the effect of.

  Furthermore, since the width of the Doppler cell is reduced by performing zero padding interpolation when generating the Doppler history, there is an advantage that the compensation amount can be estimated with high accuracy.

  Furthermore, since the data string for generating the Doppler history is extracted by summing up the signals of a plurality of adjacent ranges, the signal at the reflection point can be used even if the target reflection point moves in the range. Thus, there is an advantage that the compensation amount can be estimated with high accuracy.

  Furthermore, the rate of change of the Doppler frequency at each reflection point on each piecewise Doppler history is simplified to the problem of detecting a straight line passing through the origin on the spectral image obtained using the two-dimensional Fourier transform, Integration of spectral images along multiple integration paths with different inclinations through the origin and peak detection of the integration results reduce the influence of interference from multiple reflection points and estimate the compensation amount with high accuracy There are advantages you can do.

  Furthermore, when performing integration along the integration path, a high frequency component is lifted or weights that suppress excessive lifting are added, so that there is an advantage that the estimation accuracy of the change rate is improved. Furthermore, when integrating along the integration path, the integration is performed based on linear interpolation of the amplitude of the cells in the vicinity of the integration path, so that there is an advantage that the estimation accuracy of the change rate is improved.

  Furthermore, after estimating the rate of change in the piecewise Doppler history, it is determined whether or not the rate of change is zero. If it is determined that the rate of change is not zero, it is output as it is, and the rate of change is assumed to be zero. If it is determined, after compensating so that a known Doppler rate of change is added to the piecewise Doppler history, a spectral image is generated again to estimate the rate of change. Since the subtraction result is output as the final change rate estimation result, the change rate can be correctly estimated even when the change rate is close to zero and the value is easily dragged to zero.

Embodiment 2. FIG.
In the second embodiment, another configuration example of the 2D-FT compensation amount estimator 200 will be described. FIG. 15 is an internal configuration diagram of the 2D-FT compensation amount estimator 200 according to Embodiment 2 of the present invention. Compared to the 2D-FT compensation amount estimator 200 of FIG. 2 in the first embodiment, the 2D-FT compensation amount estimator 200 of FIG. 15 includes a time-series data compensation unit 231, an STFT point number change unit 232, and a corrected compensation amount. It is characterized in that a storage unit 295 is further provided.

  Hereinafter, specific processing according to the second embodiment will be described with reference to FIGS. 1 and 15. After performing high-frequency pulses in the transmitter 1 and processing until coarse compensation is performed in the coarse compensation circuit 10 and motion compensation based on the compensation amount obtained by the 2D-FT compensation amount estimator 200, The processing until the cross range compression is performed by the cross range compressor 40 is the same as that of the first embodiment.

  The 2D-FT compensation amount estimator 200 in FIG. 15 uses the compensation amount estimated with respect to the time series data d (h) extracted by the time series data extraction unit 201 to repeat the compensation process and sequentially. The compensation amount is obtained. Therefore, the compensation amount calculation unit 208 stores the calculated compensation amount in the correction compensation amount storage unit 295 sequentially.

  The time series data compensation unit 231 uses the phase change φ_comp (h) accumulated in the correction compensation amount storage unit 295 for the time series data d (h) obtained by the time series data extraction unit 201 to Compensation of equation (37) is performed.

  However, in the initial state, φ_comp (h) = 0 (h = 0, 1, 2,..., H−1). The correction compensation amount storage unit 295 also stores the initial value of the normalized range change as n_comp (h) = 0 (h = 0, 1,..., H−1).

  Next, the Doppler history generation unit 202 generates a Doppler history based on the STFT score D_cut accumulated in the STFT score storage unit 291 as in the first embodiment. However, the value of the score D_cut for the STFT accumulated in the STFT score storage unit 291 is changed by the STFT score changing means 232. The operation of this STFT point number changing means 232 will be described later.

  The processing from the history extraction unit 203 to the compensation amount calculation unit 208 is the same as that in the first embodiment, and the description thereof is omitted.

  Assuming that the phase change and the normalized range change calculated by the compensation amount calculation unit 208 are φ_comp0 (h) and n_comp0 (h), the compensation amount calculation unit 208 is already stored in the corrected compensation amount storage unit 295. The result of adding the above-described φ_comp0 (h) and n_comp0 (h) to φ_comp (h) and n_comp (h) as in the following equations (38) and (39) is the new update data φ_comp (h) And n_comp (h) are stored in the correction compensation amount storage unit 295.

  The time series data d (h) that is the output of the time series data extraction unit 201 is compensated with the new compensation amount obtained in this way, and thereafter, a series of processing up to the compensation amount calculation unit 208 is performed a predetermined number of times. Or, it is repeated until the change in accumulated compensation amount becomes sufficiently small. Then, the compensation amount calculation means 208 outputs φ_comp (h) and n_comp (h) that are finally updated.

  Next, the operation of the STFT point number changing means 232 will be described. When performing STFT, the larger the number of cut points D_cut, the higher the resolution of the Doppler history. However, since the time resolution at the time of Doppler history generation deteriorates as D_cut increases, the Doppler frequency width of each reflection point on the Doppler history widens when the Doppler frequency at the reflection point varies greatly. In other words, the locus of each reflection point on the Doppler history becomes thick.

  When the trajectory of each reflection point on the Doppler history becomes thick, the STFT point number changing unit 232 initially sets D_cut to a small value in consideration that the estimation accuracy of the Doppler change rate also deteriorates.

  Then, the STFT score changing means 232 performs D_cut every time d (h) is compensated with φ_comp (h) and n_comp (h) stored in the correction compensation amount storage unit 295 in order to improve the Doppler resolution. Make it bigger. Since the change width of the Doppler frequency is narrowed by performing the compensation, it is expected that the width of the locus on the Doppler history does not become wide even if D_cut is increased.

  According to the second embodiment, even for a target with a large Doppler change width, the compensation of motion can be stably performed with high Doppler resolution by gradually increasing the number of STFT points and repeating the calculation of the compensation amount. .

Embodiment 3 FIG.
In the third embodiment, another configuration example of the 2D-FT compensation amount estimator 200 for realizing the repeated calculation of the compensation amount described in the second embodiment will be described. FIG. 16 is an internal configuration diagram of the 2D-FT compensation amount estimator 200 according to Embodiment 3 of the present invention.

  Compared to the 2D-FT compensation amount estimator 200 of FIG. 15 in the second embodiment, the 2D-FT compensation amount estimator 200 of FIG. 16 includes a Doppler history cut-out point changing unit 241 instead of the STFT point changing unit 232. Is different. In the second embodiment, the STFT score is gradually increased with the compensation amount repeatedly calculated. In the third embodiment, the history cut-out score is changed with the repeatedly calculated compensation amount. It is characterized by doing.

  Hereinafter, specific processing according to the third embodiment will be described with reference to FIGS. 1 and 16. As in the second embodiment, in the initial state, the correction compensation amount storage unit 295 stores the phase change φ_comp (h) = 0 (h = 0, 1, 2,..., H−1) and the normalized range change n_comp. (H) = 0 (h = 0, 1,..., H−1) is accumulated.

  In each loop, compensation of time series data is performed using the compensation amount stored in the correction compensation amount storage unit 295, and Doppler history generation means 202, history extraction means 203, 2D-FT means 204, Spectral image inclination estimation means 205, Doppler change rate calculation means 206, and range change calculation means 207 are applied to estimate the normalized range and phase compensation amount.

  Then, the estimated compensation amount is added to each compensation amount accumulated in the corrected compensation amount storage unit 295, and a certain number of loops is reached, or a change between each compensation amount loop is less than a certain value. When it becomes, the loop is terminated. Then, the compensation amount calculation means 208 outputs the finally updated compensation amounts, and a series of these processes is almost the same as in the second embodiment.

  However, in each loop, the D_cut accumulated in the STFT score storage unit 291 was changed in the second embodiment. However, as described above, in the third embodiment, the D_cut accumulated in the history cutout score storage unit 292 is changed. The cutout point D2_cut of the history is changed.

  As D2_cut increases, the estimated resolution of the rate of change improves, but when the time resolution deteriorates and the rate of change changes significantly, the value of the Doppler rate of change in the extracted history changes. The accuracy is degraded.

  By taking account of these properties, adjusting the cutout width of the history (for example, increasing the Doppler history cutout width D2_cut as the change in the “Doppler change rate” becomes smaller, improving the estimated resolution of the Doppler change rate, etc. ), The estimation accuracy of the Doppler change rate can be improved. Therefore, it is estimated that the estimation accuracy of the compensation amount is finally improved.

  In the above description, the process has been described in which D_cut accumulated in the STFT score storage unit 291 is made constant. However, as in the process of the second embodiment, the process is accumulated in the STFT score storage unit 291 for each loop. Both D_cut and D2_cut stored in the history cut-out point storage unit 292 may be changed at the same time.

  FIG. 17 is an internal configuration diagram of another 2D-FT compensation amount estimator 200 according to Embodiment 3 of the present invention. FIG. 17 is characterized in that D_cut and D2_cut are simultaneously changed by using the Doppler history cutting point changing means 241 and the STFT score changing means 232 in each loop, which corresponds to the combination of the configurations of FIGS. 15 and 16. To do. It is expected that the accuracy of compensation amount estimation will be further improved by appropriately adjusting both parameters at the same time.

  According to the third embodiment, it is possible to improve the estimation accuracy of the Doppler change rate by adjusting the cutout width of the history and repeating the calculation of the compensation amount, and finally improve the estimation accuracy of the compensation amount. Can do. Further, as described in the second embodiment, the STFT point number is also appropriately adjusted and calculation of the compensation amount is repeated at the same time, whereby the compensation amount estimation accuracy can be further improved, and the target having a large Doppler variation range is obtained. However, the motion can be compensated stably with high Doppler resolution.

Embodiment 4 FIG.
In the fourth embodiment, another configuration example of the 2D-FT compensation amount estimator 200 will be described. FIG. 18 is an internal configuration diagram of the 2D-FT compensation amount estimator 200 according to Embodiment 4 of the present invention. Compared with the 2D-FT compensation amount estimator 200 of FIG. 2 in the first embodiment, the 2D-FT compensation amount estimator 200 of FIG. 18 is stable in width instead of the Doppler history generation means 202 and the STFT point storage unit 291. It is characterized in that a type Doppler history generation means 251 and a reference Doppler history width storage unit 296 are provided.

  Hereinafter, specific processing according to the fourth embodiment will be described with reference to FIGS. 1 and 18. As described in the second embodiment, when the number of STFT points for generating Doppler history is increased, the time resolution is degraded and the width of the locus on the Doppler history is increased. Conversely, if the STFT score is too small, the Doppler resolution deteriorates, and the width of the locus on the Doppler history also increases. That is, the number of STFT points has an optimum value for minimizing the width of the locus.

  In the fourth embodiment, when the Doppler history is generated from the time series data d (h), the Doppler frequency width of the reflection point on the Doppler profile at each center hit h_c is the minimum value or the minimum value for all the center hits. The number of STFT points is controlled so as to be approximately equal to.

  For this purpose, the width stable Doppler history generation unit 251 reads the reference Doppler frequency width W_st stored in the reference Doppler history width storage unit 296, and determines whether the Doppler frequency width of the reflection point is closest to the above W_st in each h_c. Alternatively, the STFT cut-off point D_cut (h_c) is adjusted so as to be substantially equal. Here, W_st is appropriately set based on the predicted motion of the target, or simply set to zero.

  According to the fourth embodiment, for each center hit h_c, the number of cut points D_cut (h_c) based on the change in the target Doppler frequency and the Doppler resolution is set, and as a result, the estimation accuracy of the compensation amount can be improved. .

Embodiment 5. FIG.
In the fifth embodiment, another configuration example of the 2D-FT compensation amount estimator 200 will be described. FIG. 19 is an internal configuration diagram of 2D-FT compensation amount estimator 200 in Embodiment 5 of the present invention. Compared with the 2D-FT compensation amount estimator 200 of FIG. 2 in the first embodiment, the 2D-FT compensation amount estimator 200 of FIG. 19 calculates an FFT history instead of the history extraction unit 203 and the 2D-FT unit 204. It is characterized by comprising means 261, FFT history cutout means 262, and 1D-FT means 263.

  Hereinafter, specific processing of the fifth embodiment will be described with reference to FIGS. 1 and 19. The purpose of the fifth embodiment is to improve the efficiency of the processing of the 2D-FT means 204 in the first to fourth embodiments. The processing until the Doppler history generation unit 202 generates the Doppler history is the same as in the first to fourth embodiments.

  Hereinafter, the input history is generally expressed as img00 (y, x0). The FFT history calculation means 261 obtains fimg00 (f_y, x0) by Fourier transforming the absolute value of the input history (here, Doppler history) in the range or Doppler direction according to the following equation (40).

  Next, as shown in the following equation (41), the FFT history cutout unit 262 cuts out the section history having the width in the x direction D2_cut from the fimg00 (f_y, x0) while changing the central x value x_c. , Fimg0 (f_y, x, x_c) are stored in the history cut-out point storage unit 292.

  Next, the 1D-FT means 263 obtains a spectrum by performing a one-dimensional Fourier transform on the two-dimensional image obtained for each x_c in the x direction. Here, in FIGmg0 (f_y, x, x_c), a two-dimensional array relating to f_y and x determined when x_c is given is generally referred to as fig1 (f_y, x). Here, the number of pixels in the f_y direction of FIGmg1 (f_y, x) is N_y, and the number of pixels in the x direction is N_x.

  The 1D-FT means 263 performs one-dimensional Fourier transform on the fimg1 (f_y, x) in the x direction, and finally obtains fimg (f_y, f_x) by taking this absolute value. This fimg (f_y, f_x) is equal to the fimg (f_y, f_x) in the equation (12).

  However, in Expression (12), since the two-dimensional Fourier transform is performed every time on the extracted piecewise history, the same x0 profile is Fourier-transformed many times in the y direction, and the calculation efficiency is poor. It was.

  On the other hand, in the method described in the fifth embodiment, the FFT history calculation unit 261 first performs a one-dimensional Fourier transform in the y direction, and then cuts out the 1D-FT unit 263. Since the x-direction Fourier transform is used, there is no problem that the y-direction Fourier transform overlaps.

  According to the fifth embodiment, by combining the FFT history calculation means, the FTT history extraction means, and the 1D-FT means, the same effect as that of the 2D-FT means can be obtained, and duplication of Fourier transform can be avoided. The calculation load can be reduced.

Embodiment 6 FIG.
In the first to fifth embodiments, the compensation amount is estimated by applying a piecewise two-dimensional Fourier transform process to the amplitude distribution of the Doppler history. In the sixth embodiment, a case where the compensation amount is estimated by applying a piecewise two-dimensional Fourier transform process to the amplitude distribution of the range history will be described.

  FIG. 20 is an internal configuration diagram of the rough compensation amount estimator 100 according to Embodiment 6 of the present invention. The coarse compensation amount estimator 100 in FIG. 20 includes history cutout means 103, 2D-FT means 104, spectral image inclination estimation means 105, range history change calculation means 171 and compensation amount calculation means 108, and a history as a storage unit. A cut-out point storage unit 192 and a spectral image inclination storage unit 193 are provided.

  Components other than the range history change calculation unit 171 basically have the same functions as those provided in the 2D-FT compensation amount estimator 200. Hereinafter, specific processing according to the sixth embodiment will be described with reference to FIGS. 1 and 20.

  The history cut-out means 103 follows the divided range history rhc (m, h_rhc, h_c) from the range history S_0 (m, h) while changing the center hit h_c based on the cut-out point G accumulated in the history cut-out point storage unit 192. Cut out as in equation (42).

  However, it is assumed that S_0 (m, h) is expanded to be 0 when h <0 and h> H−1. Next, the 2D-FT means 104 performs a two-dimensional Fourier transform on the absolute value of each extracted range history. The spectral image inclination estimation means 105 estimates the inclination of the locus on the input two-dimensional image (in this case, the range history) as in the first to fifth embodiments.

  The same processing is repeated for the range history cut out for each center hit h_c, and the obtained inclination is stored as a_est (h_c) in the spectrum image inclination storage unit 193.

  a_est (h_c) is the rate of change of the range cell per hit. Therefore, the range history change calculating means 171 first converts a_est (h_c) into the change rate v_est [m / s] of the range [m] per 1 [s] by the following equation (43).

  Next, the P-order least squares method is applied to the set of times t (h_c) and v_est (h_c) at each center hit h_c, and the range change rate v_fit (t) is expressed by the following equation (44). Is represented by a polynomial at time t.

  Therefore, the target range change r_fit (t) caused by unnecessary movement is given by the following equation (45).

  The compensation amount calculation means 108 estimates the compensation amount based on this range change. Thereafter, the process of compensating for the range history S_0 (m, h) by the motion compensator 11 subsequent to the coarse compensation amount estimator 100 is the same as in any one of the first to fifth embodiments.

  According to the sixth embodiment, by applying the piecewise two-dimensional Fourier transform process to the amplitude distribution of the range history, it is possible to improve the accuracy of compensation in the rough compensation circuit 10 as well.

  As described above, the image radar apparatus of the present invention is characterized in that a compensation amount with reduced estimation error can be estimated by performing Fourier transform on the history divided for each divided region. The case where the estimation of the compensation amount is applied to the 2D-FT compensation circuit 20 is described in the first to fifth embodiments, and the case where the compensation amount is applied to the coarse compensation circuit 10 is described in the sixth embodiment.

  Based on these, not only is the estimation error reduced by the three compensation circuits of the coarse compensation circuit 10, the 2D-FT compensation circuit 20, and the fine compensation circuit 30 shown in FIG. 1, but a combination of any two compensation circuits. Or the rough compensation circuit 10 or the 2D-FT compensation circuit 20 alone can reduce the estimation error.

It is a block diagram of the image radar apparatus in Embodiment 1 of this invention. It is an internal block diagram of the 2D-FT compensation amount estimator of the radar apparatus in Embodiment 1 of the present invention. It is an internal block diagram of the spectrum image inclination estimation means in Embodiment 1 of this invention. It is an internal block diagram of the spectrum image inclination estimation main means in Embodiment 1 of this invention. It is the geometry of observation of the radar apparatus in Embodiment 1 of this invention. It is an example of the target ISAR image in Embodiment 1 of this invention. It is an example of the range history in Embodiment 1 of this invention. It is an example of the Doppler history in Embodiment 1 of this invention. It is a schematic diagram explaining the process which cuts out a signal from the range history in Embodiment 1 of this invention. It is a schematic diagram explaining the process which cuts out a short time Doppler history from the Doppler history in Embodiment 1 of this invention. It is an example of the two-dimensional image containing the several straight line of the same inclination in Embodiment 1 of this invention. In Embodiment 1 of this invention, it is an example of the spectrum image obtained by carrying out the two-dimensional Fourier transform of the two-dimensional image of FIG. It is a figure explaining the integration process on the spectrum image in Embodiment 1 of this invention. It is a figure explaining the process which estimates the change rate of Doppler in Embodiment 1 of this invention. It is an internal block diagram of the 2D-FT compensation amount estimator in Embodiment 2 of this invention. It is an internal block diagram of the 2D-FT compensation amount estimator in Embodiment 3 of this invention. It is an internal block diagram of another 2D-FT compensation amount estimator in Embodiment 3 of this invention. It is an internal block diagram of the 2D-FT compensation amount estimator in Embodiment 4 of this invention. It is an internal block diagram of the 2D-FT compensation amount estimator in Embodiment 5 of this invention. It is an internal block diagram of the rough compensation amount estimator in Embodiment 6 of this invention.

Explanation of symbols

    DESCRIPTION OF SYMBOLS 1 Transmitter, 2 Transmission / reception switching device, 3 Transmitting / receiving antenna, 4 Receiver, 5 Pulse compressor, 6 Target, 10 Coarse compensation circuit, 11 Motion compensator, 20 2D-FT compensation circuit, 21 Motion compensator, 30 Fine compensation Circuit, 31 motion compensator, 40 cross-range compressor, 100 coarse compensation amount estimator, 103 history extraction means, 104 2D-FT means, 105 spectral image inclination estimation means, 108 compensation amount calculation means, 171 range history change calculation means 192 History cut-out point storage unit, 193 Spectral image inclination storage unit, 2002 2D-FT compensation amount estimator, 201 Time series data extraction unit, 202 Doppler history generation unit, 203 History cut-out unit, 204 2D-FT unit, 205 spectrum Image tilt estimation means, 206 Doppler change rate calculation hand , 207 Range change calculating means, 208 Compensation amount calculating means, 211, 211a, 211b Spectral image inclination estimation main means, 212 Zero inclination switching means, 213 Known inclination adding means, 214 2D-FT means, 215 Known inclination removing means, 216 Spectral image inclination output means, 221 Y value setting means, 222 inclination candidate setting means, 223 X continuous value calculation means, 224 X truncation value setting means, 225 X difference value calculation means, 226 weighted normalized vector generation means, 227 adjacent X Value integration means, 228 peak slope detection means, 231 time series data compensation means, 232 STFT point change means, 241 Doppler history extraction point change means, 251 width stable Doppler history generation means, 261 FFT history calculation means, 262 FFT history extraction hand H.263 1D-FT means, 291 STFT point storage unit, 292 history cut-out point storage unit, 293 spectral image inclination storage unit, 294 integration result storage unit, 295 correction compensation amount storage unit, 296 reference Doppler history width storage unit, 300 fine Compensation amount estimator.

Claims (17)

An image that transmits a radio wave to a moving target, receives a reflected wave from the target, generates a range history that is a time history of the range profile, and obtains an image of the target by cross-range compression of the range history In radar equipment
Time-series data extracting means for extracting time-series data from the range history;
Doppler history generating means for generating a Doppler history that is a time history of a Doppler profile by applying a short-time Fourier transform to the extracted time series data;
A history cutout means for cutting out a short-time Doppler history that is a Doppler history of a segment time region from the generated Doppler history;
Doppler change rate estimation means for estimating the change rate of the Doppler frequency of the locus of each reflection point on the cut out short-time Doppler history;
Based on the Doppler frequency change rate estimated for a plurality of short-time Doppler histories, the Doppler frequency change over time is estimated, and the range change between the target and the radar is estimated based on the estimation result. Range change calculating means;
Compensation amount calculating means for calculating a range compensation amount and a phase compensation amount normalized with a range resolution for compensating for an influence of an unnecessary change in the range and the Doppler in the range history based on the estimated change in the range; ,
An image radar comprising: a two-dimensional Fourier transform compensation circuit comprising: a motion compensator that compensates for an unnecessary range and phase change of the range history based on the calculated range compensation amount and the phase compensation amount. apparatus. The image radar device according to claim 1,
The Doppler change rate estimating means is:
Two-dimensional Fourier transform means for generating a spectral image by applying a two-dimensional Fourier transform to the amplitude distribution of the cut-out short-time Doppler history;
A spectral image inclination estimation means for estimating an inclination of a locus on the complex image based on the generated spectral image;
An image radar apparatus comprising: a Doppler change rate calculating unit that estimates a change rate of a Doppler frequency based on the estimated inclination. The image radar device according to claim 2,
The spectral image inclination estimation means includes:
First spectral image inclination estimation main means for estimating the inclination of a straight line passing through the origin on the generated spectral image as a first inclination amount;
Zero inclination switching means for determining whether or not the first inclination amount is zero, and for switching processing between the case where the first inclination amount is zero and the case where the first inclination amount is not zero;
A known slope that compensates for a known slope amount with respect to the short-time Doppler history cut out by the history cutout means when the zero slope switching means determines that the first slope amount is zero. Additional means;
A second two-dimensional Fourier transform means for generating a spectral image by applying a two-dimensional Fourier transform to the amplitude distribution of the short-time Doppler history after compensation by the known tilt amount;
Second spectral image inclination estimation main means for estimating the inclination of a straight line passing through the origin on the spectral image generated by the second two-dimensional Fourier transform means as a second inclination amount;
The known inclination removal for estimating the third inclination amount by subtracting the known inclination amount added by the known inclination adding means from the second inclination amount estimated by the second spectral image inclination estimation main means. Means,
If it is determined by the zero inclination switching means that the first inclination amount is not zero, the first inclination amount is output, and the zero inclination switching means indicates that the first inclination amount is zero. An image radar apparatus comprising: a spectral image tilt output unit that outputs the third tilt amount when it is determined. The image radar apparatus according to claim 3,
The first spectral image inclination estimation main means and the second spectral image inclination estimation main means are:
Y value setting means for setting a Y value of an integration path for a spectrum image composed of a two-dimensional plane of XY,
An inclination candidate setting means for setting an inclination candidate on the spectrum image of the integration path;
X continuous value calculation means for setting a continuous value of X based on the Y value set by the Y value setting means and the inclination set by the inclination candidate setting means;
X rounded value setting means for rounding off the decimal point of the X continuous value set by the X continuous value calculating means to obtain an X rounded value discretized by rounding down;
X difference value calculation means for calculating a difference value between the continuous value of X and the X cutoff value;
Based on the calculated difference value, the cell on the two-dimensional plane set with the X cut-off value, and the cell adjacent to the cell and set with a value obtained by adding 1 to the X cut-off value After linearly interpolating the amplitude values, the adjacent X value integrating means for performing integration, and normalizing the spectrum image with the maximum amplitude value for all X values for each Y value, and further weighting each Y value Weighted normalization vector generation means for generating a weight vector for performing;
A peak inclination detecting means for searching for a peak of an integration result by the adjacent X value integrating means and calculating an inclination of a locus on the short-time Doppler history inputted based on an inclination of an integration path where the peak appears. An image radar device characterized by that. The image radar device according to any one of claims 2 to 4,
The Doppler change rate calculating means applies a fitting method such as a high-order least squares method to a pair of calculation results of the change rate of the Doppler frequency in each divided region and the center time of each divided region. Obtain the rate of frequency change as a polynomial over time,
The range change calculating means calculates a range change over time using a time change in Doppler frequency obtained by integrating the polynomial. The image radar apparatus according to any one of claims 1 to 5,
The time-series data extracting means sets an extraction range of the range on the assumption that the reflection point has moved beyond the range, and generates time-series data by summing the extraction range. Radar device. The image radar device according to any one of claims 1 to 6,
The Doppler history generating means performs zero padding on a segmented data string cut out from time series data, and interpolates the Doppler history in the Doppler frequency direction. The image radar apparatus according to any one of claims 1 to 7,
A correction compensation amount storage unit for storing the range compensation amount and the phase compensation amount calculated by the compensation amount calculation means;
Time-series data compensation means for compensating the phase of the time-series data extracted by the time-series data extraction means based on the phase compensation amount stored in the correction compensation amount storage unit;
STFT score changing means for changing the time series data score used for the short-time Fourier transform executed by the Doppler history generating means based on the range compensation quantity and the phase compensation quantity calculated by the compensation quantity calculating means; In addition,
The Doppler history generation means performs a new Doppler by performing a short-time Fourier transform on the time series data compensated by the time series data compensation means using the points of the time series data changed by the STFT point change means. Generate history,
The compensation amount calculation means includes a range compensation amount and a phase compensation amount calculated corresponding to the new Doppler history, and a previous range compensation amount and a phase compensation amount stored in the correction compensation amount storage unit, respectively. The updated range compensation amount and phase compensation amount are calculated by adding and stored in the correction compensation amount storage unit, and the updated range compensation amount and phase compensation amount change amount are within a predetermined value, Alternatively, the image radar device is characterized in that the compensation amount calculation process is repeatedly performed until the predetermined number of times is calculated. The image radar apparatus according to any one of claims 1 to 7,
A correction compensation amount storage unit for storing the range compensation amount and the phase compensation amount calculated by the compensation amount calculation means;
Time-series data compensation means for compensating the phase of the time-series data extracted by the time-series data extraction means based on the phase compensation amount stored in the correction compensation amount storage unit;
Further comprising: a Doppler history cut-out point changing means for changing a cut-out point for cutting out the short-time Doppler history by the history cut-out means based on the range compensation amount and the phase compensation amount calculated by the compensation amount calculating means,
The Doppler history generation means generates a new Doppler history by performing a short-time Fourier transform on the time series data compensated by the time series data compensation means,
The history cutout means cuts out a new short-time Doppler history using the cutout points changed by the Doppler history cutout point changing means,
The compensation amount calculating means includes a range compensation amount and a phase compensation amount calculated corresponding to the new short-time Doppler history, a previous range compensation amount and a phase compensation amount stored in the correction compensation amount storage unit, Are added to each other to calculate the updated range compensation amount and phase compensation amount and store them in the correction compensation amount storage unit, and the updated range compensation amount and phase compensation amount change amount are within a predetermined value. Alternatively, the compensation amount calculation process is repeatedly performed until a predetermined number of times is calculated. The image radar apparatus according to any one of claims 1 to 7,
A correction compensation amount storage unit for storing the range compensation amount and the phase compensation amount calculated by the compensation amount calculation means;
Time-series data compensation means for compensating the phase of the time-series data extracted by the time-series data extraction means based on the phase compensation amount stored in the correction compensation amount storage unit;
STFT score changing means for changing the score of time-series data used for the short-time Fourier transform executed by the Doppler history generating means based on the range compensation quantity and the phase compensation quantity calculated by the compensation quantity calculating means; A Doppler history cut-out number changing means for changing a cut-out point for cutting out the short-time Doppler history by the history cut-out means based on the range compensation amount and the phase compensation amount calculated by a compensation amount calculation means, and
The Doppler history generation means performs a new Doppler by performing a short-time Fourier transform on the time series data compensated by the time series data compensation means using the points of the time series data changed by the STFT point change means. Generate history,
The history cutout means cuts out a new short-time Doppler history using the cutout points changed by the Doppler history cutout point changing means,
The compensation amount calculating means includes a range compensation amount and a phase compensation amount calculated corresponding to the new short-time Doppler history, a previous range compensation amount and a phase compensation amount stored in the correction compensation amount storage unit, Are added to each other to calculate the updated range compensation amount and phase compensation amount and store them in the correction compensation amount storage unit, and the updated range compensation amount and phase compensation amount change amount are within a predetermined value. Alternatively, the compensation amount calculation process is repeatedly performed until a predetermined number of times is calculated. The image radar device according to claim 1,
The history cut-out means is
FFT history calculation means for performing a one-dimensional Fourier transform on the input amplitude distribution of the Doppler history to calculate an FFT history in order to estimate the inclination of the trajectory on the section history;
FFT history cutting means for cutting out the FFT history in the hit direction based on the number of cut points when performing a short-time two-dimensional Fourier transform,
The Doppler conversion rate estimation means is:
1D-FT means for generating a spectral image of the input Doppler history by Fourier-transforming the FFT history cut by the FFT history cutting means in the hit direction;
A spectral image inclination estimation means for estimating an inclination of a locus on the complex image based on the generated spectral image;
An image radar apparatus comprising: a Doppler change rate calculating unit that estimates a change rate of a Doppler frequency based on the estimated inclination. The image radar device according to claim 1,
In the short-time Fourier transform for generating the Doppler history, the Doppler history generation means keeps the number of data points after zero padding constant, and the reflection point when the time-series data is subjected to the short-time Fourier transform is a predetermined reference Doppler. An image radar apparatus, wherein the Doppler history is generated while changing the number of cut-out points of the time series data so as to be within a history width. The image radar device according to any one of claims 1 to 12,
A time series signal is extracted from the range history after motion compensation, the time series signal is filtered, and an unnecessary range remaining in the range history after motion compensation is based on the phase change of the time series signal after filtering processing. Precise compensation amount estimation that estimates the change and calculates the range compensation amount and phase compensation amount normalized by the range resolution to compensate for the influence of the unnecessary range change of the range history and the Doppler based on the unnecessary range change And
A fine compensation circuit comprising: a fine motion compensator that compensates for an unnecessary range change remaining in the range history after motion compensation using the range compensation amount and the phase compensation amount that are outputs of the fine compensation amount estimator. In addition,
The fine radar compensation amount estimator takes in a range history compensated by the motion compensator in the two-dimensional Fourier transform compensation circuit as a range history after motion compensation. An image that transmits a radio wave to a moving target, receives a reflected wave from the target, generates a range history that is a time history of the range profile, and obtains an image of the target by cross-range compression of the range history In radar equipment
A history cutout means for cutting out a short time range history that is a range history of each divided time region from the range history,
Range change rate estimation means for estimating the range change rate of the locus of each reflection point on the short-time range history cut out by the history cutout means;
Range history change calculation means for integrating the range change rate and estimating a change in range over the entire time;
Based on the range change estimated by the range history change calculation means, a range compensation amount and a phase compensation amount normalized by range resolution for compensating for the influence of the range history and unnecessary Doppler change are calculated. Compensation amount calculating means;
An image radar apparatus comprising: a coarse compensation circuit including: a coarse motion compensator that compensates for an unnecessary range and phase change in a range history based on the calculated range compensation amount and the phase compensation amount. The image radar device according to claim 14, wherein
Time-series data extracting means for extracting time-series data from the range history;
Doppler history generating means for generating a Doppler history that is a time history of a Doppler profile by applying a short-time Fourier transform to the extracted time series data;
A history cutout means for cutting out a short-time Doppler history that is a Doppler history of a segment time region from the generated Doppler history;
Doppler change rate estimation means for estimating the change rate of the Doppler frequency of the locus of each reflection point on the cut out short-time Doppler history;
Based on the Doppler frequency change rate estimated for a plurality of short-time Doppler histories, the Doppler frequency change over time is estimated, and the range change between the target and the radar is estimated based on the estimation result. Range change calculating means;
Compensation amount calculating means for calculating a range compensation amount and a phase compensation amount normalized with a range resolution for compensating for an influence of an unnecessary change in the range and the Doppler in the range history based on the estimated change in the range; ,
A time-series data extraction means further comprising: a motion compensator that compensates for an unnecessary range and phase change in the range history based on the calculated range compensation amount and the phase compensation amount. The image radar apparatus extracts time series data from a range history compensated by the coarse motion compensator in the coarse compensation circuit. The image radar device according to claim 15,
A time series signal is extracted from the range history after motion compensation, the time series signal is filtered, and an unnecessary range remaining in the range history after motion compensation is based on the phase change of the time series signal after filtering processing. Precise compensation amount estimation that estimates the change and calculates the range compensation amount and phase compensation amount normalized by the range resolution to compensate for the influence of the unnecessary range change of the range history and the Doppler based on the unnecessary range change And
A fine compensation circuit comprising: a fine motion compensator that compensates for an unnecessary range change remaining in the range history after motion compensation using the range compensation amount and the phase compensation amount that are outputs of the fine compensation amount estimator. In addition,
The fine radar compensation amount estimator takes in a range history compensated by the motion compensator in the two-dimensional Fourier transform compensation circuit as a range history after motion compensation. The image radar device according to claim 14, wherein
A time series signal is extracted from the range history after motion compensation, the time series signal is filtered, and an unnecessary range remaining in the range history after motion compensation is based on the phase change of the time series signal after filtering processing. Precise compensation amount estimation that estimates the change and calculates the range compensation amount and phase compensation amount normalized by the range resolution to compensate for the influence of the unnecessary range change of the range history and the Doppler based on the unnecessary range change And
A fine compensation circuit comprising: a fine motion compensator that compensates for an unnecessary range change remaining in the range history after motion compensation using the range compensation amount and the phase compensation amount that are outputs of the fine compensation amount estimator. In addition,
The image radar apparatus according to claim 1, wherein the fine compensation amount estimator takes in a range history compensated by the coarse motion compensator in the coarse compensation circuit as a range history after motion compensation.

Images