JP5619544B2 - Speed estimation apparatus, computer program, and speed estimation method - Google Patents

Description

  The present invention relates to a speed estimation device that estimates the speed of an observation target observed by a synthetic aperture radar.

In the total aperture radar, the resolution in the azimuth direction is increased by synthetic aperture processing (azimuth compression processing).
If the observation target observed by the synthetic aperture radar is moving, the image of the observation target may be blurred or the position in the azimuth direction may be shifted.

JP 2010-14700 A JP 2006-29979 A

If the reference signal used for the azimuth compression matches the moving speed of the observation target, the image of the observation target can be focused and the correct position in the azimuth direction can be observed.
However, in order to use a reference signal that matches the movement speed of the observation target, it is necessary to correctly estimate the movement speed of the observation target.
In particular, in a situation where a plurality of observation objects with different movement speeds are densely packed within a relatively narrow range, it is necessary to correctly estimate the movement speeds of the plurality of observation objects.
The present invention has been made, for example, in order to solve the above-described problems, and an object thereof is to correctly estimate the speed of an observation target.

The speed estimation device according to the present invention is:
Based on the received signal received by the synthetic aperture radar, in the speed estimation device that estimates the speed of the observation target,
A compression observation calculation unit, a peak extraction unit, and a velocity estimation unit;
The compressed observation calculation unit obtains a correlation between a series of observation values having a predetermined sampling interval generated from the received signal and a reference signal corresponding to the estimated speed for each of a plurality of estimated speeds. Calculating a series of compressed observations having a sampling interval shorter than the predetermined sampling interval,
The peak extraction unit extracts a peak of a series of compression observation values calculated by the compression observation calculation unit for each of the plurality of estimated speeds,
The speed estimation unit obtains an estimated speed at which the peak extracted by the peak extraction unit is a maximum from the plurality of estimated speeds, and sets the estimated speed as the observation target.

According to the speed estimation device of the present invention, the compression observation calculation unit generates a series of compressed observation values having a sampling interval shorter than the sampling interval of the series of observation values, so that the peak of the compression observation values can be more accurately determined. And can accurately estimate the speed of the observation target.
In particular, it is possible to accurately capture the peak of the compressed observation value even if it is difficult to capture the peak of the compressed observation value where multiple observation objects with different moving speeds are densely packed within a relatively narrow range, It is possible to correctly estimate the speed of each of a plurality of observation objects.

1 is a system configuration diagram showing an example of the overall configuration of a synthetic aperture radar system 800 in Embodiment 1. FIG. FIG. 2 is a hardware configuration diagram illustrating an example of a hardware configuration of an observation result analysis apparatus 100 according to the first embodiment. FIG. 3 is a block configuration diagram illustrating an example of functional blocks of an observation result analysis apparatus 100 according to Embodiment 1. FIG. 4 is a flowchart showing an example of a flow of speed estimation processing S610 in the first embodiment. FIG. 6 is a diagram for explaining the operation of the azimuth compression unit 120 according to the first embodiment. The figure for demonstrating the peak of a series of compression observation values. The figure for demonstrating an azimuth shift. The figure which shows the change of the distance between the observation apparatus 810 and the observation object 850. FIG. The figure for demonstrating the subject and its cause which arise when the sampling interval of a compression observation value is wide. The figure for demonstrating the effect by narrowing the sampling interval of a compression observation value.

Embodiment 1 FIG.
The first embodiment will be described with reference to FIGS.

FIG. 1 is a system configuration diagram showing an example of the overall configuration of a synthetic aperture radar system 800 in this embodiment.
The synthetic aperture radar system 800 includes an observation device 810 and an observation result analysis device 100.
The observation device 810 is mounted on a moving body such as an artificial satellite or an aircraft, and moves relatively at a moving speed 711 in a predetermined direction, and relatively around a predetermined direction (for example, a direction perpendicular to the moving direction). Radio waves 701 are emitted toward a wide range. The radio wave 701 emitted by the observation device 810 is, for example, a pulsed chirp signal emitted at a predetermined time interval (for example, 1 microsecond). The observation device 810 receives the reflected wave 702 (received signal) reflected by the radiated radio wave hitting the observation target 850. The observation object 850 is stationary, or is moving at a movement speed 721 that is relatively smaller than the movement speed 711 of the observation apparatus 810. Of the movement speed 721 of the observation object 850, the component in the direction parallel to the movement direction of the observation apparatus 810 is the “azimuth direction movement speed” (722), and the component in the direction perpendicular to the movement direction of the observation apparatus 810 is “range”. This is referred to as “direction moving speed” (723).
The observation result analysis apparatus 100 (speed estimation apparatus) analyzes the reflected wave 702 received by the observation apparatus 810 and generates a synthetic aperture radar image. The observation result analyzing apparatus 100 obtains the position and moving speed 721 of the observation target 850. The observation result analysis apparatus 100 may be mounted on the same mobile body as the observation apparatus 810, or may be installed at a location different from the observation result analysis apparatus 100 such as the ground.

FIG. 2 is a hardware configuration diagram illustrating an example of a hardware configuration of the observation result analysis apparatus 100 in this embodiment.
The observation result analysis apparatus 100 is a computer, for example. The observation result analysis apparatus 100 includes, for example, a storage device 914, a processing device 911, an input device 912, and an output device 913.
The storage device 914 stores a program executed by the processing device 911, data processed by the processing device 911, and the like. The storage device 914 is, for example, a primary storage device such as a volatile memory or a nonvolatile memory, or a secondary storage device that stores a program or data using a storage medium such as a magnetic disk or an optical disk.
The processing device 911 processes data by executing the program stored in the storage device 914 and controls the entire observation result analysis device 100.
The input device 912 inputs information from outside the observation result analysis device 100 and converts it into data in a format that can be processed by the processing device 911. The data converted by the input device 912 may be processed directly by the processing device 911 or may be temporarily stored by the storage device 914. The input device 912 is, for example, an operation input device that inputs a user's operation such as a keyboard or a mouse, an analog-digital conversion device that inputs an analog signal and converts it into digital data, and receives and demodulates a signal transmitted by another device Receiving device.
The output device 913 outputs the data processed by the processing device 911 and the data stored by the storage device 914 to the outside of the observation result analysis device 100. The output device 913 is, for example, an information output device that outputs information in a form that can be perceived by humans, such as a display device or a speaker, a digital-analog conversion device that converts digital data into an analog signal, and generates a modulation signal. For example, a transmission device that transmits data to a device.

  The functional blocks of the observation result analysis device 100 described below are realized by the processing device 911 executing the computer program stored in the storage device 914. Note that these functional blocks may be realized not by the computer program being executed by the processing device 911 but by an electrical configuration such as a digital circuit or an analog circuit, or other configurations.

FIG. 3 is a block configuration diagram illustrating an example of functional blocks of the observation result analysis apparatus 100 according to this embodiment.
The observation result analysis apparatus 100 includes, for example, an azimuth compression unit 120, a peak extraction unit 141, and a speed estimation unit 151.

The azimuth compression unit 120 (compression observation calculation unit) uses the processing device 911 to azimuth-compress the observation value data sequence generated from the reflected wave 702 received by the observation device 810, thereby increasing the resolution in the azimuth direction. The observation value data string input by the azimuth compression unit 120 is data representing a series of observation values in the time domain sampled at a predetermined sampling frequency. The signal that is the basis of sampling is, for example, a signal in which the resolution in the range direction is improved by performing range compression on the reflected wave 702 received by the observation apparatus 810. Alternatively, the azimuth compression unit 120 may be configured to input a series of data strings in the time domain generated as a result of the range compression as observation data strings.
The azimuth compression unit 120 uses the processing device 911 to obtain a correlation between the observation value data string and the reference signal, thereby generating a compressed observation value data string having an improved resolution in the azimuth direction. The compression observation value data string generated by the azimuth compression unit 120 is data representing a series of compression observation values in the time domain.

  The reference signal used for azimuth compression needs to use a signal that matches the moving speed 721 of the observation target 850. When azimuth compression is performed using a reference signal that does not match the moving speed 721 of the observation target 850, the focus is blurred. In other words, the moving speed 721 of the observation target 850 can be estimated by performing azimuth compression using reference signals corresponding to a plurality of estimated speeds and finding the most focused one. In this way, the observation result analyzing apparatus 100 estimates the moving speed 721 of the observation target 850.

  In addition, the sampling frequency of the compressed observation value data sequence generated by the azimuth compression unit 120 is higher than the sampling frequency of the observation value data sequence input by the azimuth compression unit 120. That is, the time interval between compressed observations is shorter than the time interval between observations, and the number of compressed observations is greater than the number of observations. The azimuth compression unit 120 generates a compressed observation value data sequence having a shorter time interval than the input observation value data sequence, for example, by performing an interpolation process.

  The azimuth compression unit 120 includes, for example, a range direction estimation speed selection unit 121, an azimuth direction estimation speed selection unit 122, a frequency domain reference signal generation unit 123, a Fourier transform unit 131, a multiplication unit 132, and a zero padding unit 133. And an inverse Fourier transform unit 134.

The range direction estimated speed selection unit 121 uses the processing device 911 to select the range direction estimated speed one by one from a plurality of possible range direction estimated speeds as the range direction moving speed 723 of the observation target 850. .
The azimuth direction estimated speed selection unit 122 uses the processing device 911 to select the estimated azimuth direction speed one by one from a plurality of possible azimuth direction estimated speeds as the azimuth direction moving speed 722 of the observation target 850. .

  Using the processing device 911, the frequency domain reference signal generation unit 123 is based on the range direction estimated speed selected by the range direction estimated speed selection unit 121 and the azimuth direction estimated speed selected by the azimuth direction estimated speed selection unit 122. Then, a reference signal corresponding to the set of estimated speeds is generated. The correlation process with the observation value data string is performed in the frequency domain. Therefore, the frequency domain reference signal generation unit 123 generates data (frequency domain reference signal value data string) representing a series of reference signal values (frequency domain reference signal values) in the frequency domain obtained by Fourier transforming the reference signal. For example, the frequency domain reference signal generation unit 123 stores a frequency domain reference signal value data sequence corresponding to each set of estimated speeds in advance using the storage device 914. The frequency domain reference signal generation unit 123 uses the processing device 911 to select the range direction estimated speed and the azimuth direction estimated speed selected by the range direction estimated speed selection unit 121 from the stored frequency domain reference signal value data sequence. The frequency domain reference signal value data string corresponding to the set with the estimated azimuth direction speed selected by the unit 122 is selected.

  The Fourier transform unit 131 uses the processing device 911 to Fourier-transform a series of observation values in the time domain based on the observation value data string, thereby calculating a series of observation values (frequency domain observation values) in the frequency domain. Then, data representing the calculated series of frequency domain observation values (frequency domain observation value data string) is generated. For example, the Fourier transform unit 131 extracts a predetermined number of observation values from the observation value data string by window processing, and performs a discrete Fourier transform on the extracted series of observation values using a fast Fourier transform algorithm. The number of frequency domain observation values calculated by the Fourier transform unit 131 is equal to the number of observation values taken out by the Fourier transform unit 131.

The multiplying unit 132 uses the processing device 911 to generate the frequency domain observation value data sequence generated by the Fourier transform unit 131 and the frequency domain reference signal data sequence generated by the frequency domain reference signal generating unit 123 for each frequency. The product (frequency domain compressed observation value) obtained by multiplying the frequency domain observation value and the frequency domain reference signal value is calculated. Multiplier 132 generates data (frequency domain compressed observation value data string) representing the calculated series of frequency domain compressed observation values.
According to the convolution theorem, calculating a product for each frequency in the frequency domain is equivalent to performing a convolution integral in the time domain. That is, the multiplier 132 calculates a product for each frequency in the frequency domain, thereby obtaining a correlation between the series of observation values and the reference signal in the time domain.

Using the processing device 911, the zero padding unit 133 inserts and adds data representing one or more zero values in the middle of the frequency domain compressed observation value data string generated by the multiplication unit 132, and performs frequency domain compressed observation. Increase the number of values.
The middle of the series of values obtained by the discrete Fourier transform corresponds to half the sampling frequency. The half of the sampling frequency is the highest frequency that can be reproduced by sampling at the sampling frequency. If a zero value is inserted and added here, it is assumed that the component having a frequency higher than half the sampling frequency is 0, and the sampling frequency is increased.

  Using the processing device 911, the inverse Fourier transform unit 134 performs inverse Fourier transform on the frequency domain compressed observation value data sequence to which the zero padding unit 133 has added the data representing the zero value, thereby generating a series of compressed observation values in the time domain. Calculate and generate data (compressed observation value data string) representing the calculated series of compressed observation values. The inverse Fourier transform unit 134 performs discrete Fourier inverse transform on a series of frequency domain compressed observation values represented by the frequency domain compressed observation value data string using, for example, a fast Fourier transform algorithm. The number of compressed observation values calculated by the inverse Fourier transform unit 134 is equal to the number of frequency domain compression observation values to which the zero padding unit 133 has added 0 values. That is, the number of series of compressed observation values calculated by the inverse Fourier transform unit 134 is larger than the number of series of observation values taken out by the Fourier transform unit 131 by the number of zero values added by the zero padding unit 133. On the other hand, the time corresponding to the series of compressed observation values calculated by the inverse Fourier transform unit 134 is the same as the time corresponding to the series of observation values taken out by the Fourier transform unit 131. That is, the time interval between the compressed observation values is shorter than the time interval between the observation values.

  For example, the azimuth compression unit 120 generates a compressed observation value data string having a sampling frequency higher than that of the input observation value data string.

The peak extraction unit 141 uses the processing device 911 to extract a series of compression observation value peaks based on the compression observation value data string generated by the azimuth compression unit 120. The peak of the series of compressed observation values corresponds to the observation object 850, and the peak becomes larger as the focus is adjusted. Further, when there are a plurality of observation objects 850, there are a plurality of compressed observation value peaks correspondingly. The peak extraction unit 141 extracts, for example, all peaks that are larger than a predetermined threshold from a series of compressed observation values. Alternatively, the peak extraction unit 141 may be configured to extract only one maximum peak from a series of compressed observation values.
The peak extraction unit 141 is a series calculated by the azimuth compression unit 120 for each set of the range direction estimated speed selected by the range direction estimated speed selection unit 121 and the azimuth direction estimated speed selected by the azimuth direction estimated speed selection unit 122. Extract the peak of the observed compression.

  The speed estimation unit 151 uses the processing device 911 to select the range direction estimated speed selected by the range direction estimated speed selection unit 121 and the azimuth direction estimated speed selected by the azimuth direction estimated speed selection unit 122 from the set. A set of estimated speeds at which the peak extracted by the peak extraction unit 141 is maximized is obtained. The speed estimation unit 151 uses the processing device 911 to estimate that the determined set of estimated speeds is the moving speed 721 (the azimuth direction moving speed 722 and the range direction moving speed 723) of the observation target 850. The speed estimation unit 151 uses the output device 913 to output the estimated speed (the azimuth direction estimated speed and the range direction estimated speed).

  As described above, the peak of the series of compressed observation values extracted by the peak extraction unit 141 becomes larger as the focus is adjusted. Therefore, the set of estimated speeds at which the peak is maximized is the moving speed 721 of the observation target 850. Fits best. That is, the set of estimated speeds at which the peak is maximum is closest to the azimuth direction moving speed 722 and the range direction moving speed 723 of the observation target 850. For this reason, the speed estimation unit 151 estimates that the set of estimated speeds at which the peak extracted by the peak extraction unit 141 is maximized is the moving speed 721 of the observation target 850.

In addition, when there are a plurality of observation objects 850 and the movement speeds 721 of the observation objects 850 are different, the set of estimated speeds to be focused on is different for each observation object 850. Since a series of compressed observation peak values appear corresponding to each observation target 850, the speed estimation unit 151 obtains a set of estimated speeds that maximize the peak for each peak when a plurality of peaks appear. Thus, the moving speed 721 of each observation target 850 can be estimated.
In addition, even when the peak extraction unit 141 is configured to extract only one maximum peak, the peak extracted by the peak extraction unit 141 is maximized when focusing on any one of the plurality of observation targets 850. . For this reason, a plurality of sets of estimated speeds at which the peaks extracted by the peak extraction unit 141 are maximized appear. Therefore, the speed estimation unit 151 can estimate the moving speed 721 of each observation target 850 by obtaining all the sets of estimated speeds at which the peaks extracted by the peak extraction unit 141 are maximum.

FIG. 4 is a flowchart showing an example of the flow of speed estimation processing S610 in this embodiment.
In the speed estimation process S610, the observation result analysis apparatus 100 estimates the moving speed 721 of the observation target 850. The speed estimation process S610 includes, for example, a Fourier transform step S611, a range direction estimated speed selection step S612, an azimuth direction estimated speed selection step S613, a frequency domain reference signal generation step S614, a multiplication step S615, and a zero padding step S616. And an inverse Fourier transform step S617, a peak extraction step S618, and a speed estimation step S619.

  In the Fourier transform step S611, the Fourier transform unit 131 uses the processing device 911 to Fourier transform a series of observation values represented by the observation value data string to calculate a series of frequency domain observation values.

  In the range direction estimated speed selection step S612, the range direction estimated speed selection unit 121 uses the processing device 911 to select one unprocessed range direction estimated speed from among a plurality of predetermined range direction estimated speeds. When all the range direction estimated speeds have been processed and there is no unprocessed range direction estimated speed, the range direction estimated speed selection unit 121 uses the processing device 911 to advance the process to the speed estimation step S619. When there is an unprocessed range direction estimated speed, the range direction estimated speed selection unit 121 uses the processing device 911 to select one range direction estimated speed from the unprocessed range direction estimated speeds, and the azimuth direction The process proceeds to estimated speed selection step S613.

  In the azimuth direction estimated speed selection step S613, the azimuth direction estimated speed selection unit 122 uses the processing device 911 to select the range direction estimated speed selection in the range direction estimated speed selection step S612 from among a plurality of predetermined azimuth direction estimated speeds. A set with the range direction estimated speed selected by the unit 121 selects one unprocessed azimuth direction estimated speed. When all the estimated azimuth direction speeds have been processed and there is no unprocessed azimuth direction estimated speed, the azimuth direction estimated speed selection unit 122 uses the processing device 911 to return the process to the range direction estimated speed selection step S612. The range direction estimated speed selection unit 121 selects the next range direction estimated speed. When there is an unprocessed azimuth direction estimated speed, the azimuth direction estimated speed selection unit 122 selects one azimuth direction estimated speed from the unprocessed azimuth direction estimated speeds using the processing device 911, and the frequency domain The process proceeds to reference signal generation step S614.

  In the frequency domain reference signal generation step S614, the frequency domain reference signal generation unit 123 uses the processing device 911, the range direction estimated speed selected by the range direction estimated speed selection unit 121 in the range direction estimated speed selection step S612, and the azimuth. A frequency domain reference signal value data sequence corresponding to a set with the azimuth direction estimated speed selected by the azimuth direction estimated speed selection unit 122 in the direction estimated speed selection step S613 is generated.

  In the multiplication step S615, the multiplication unit 132 uses the processing device 911 to generate a series of frequency domain observation values calculated by the Fourier transformation unit 131 in the Fourier transformation step S611 and frequency domain reference signal generation in the frequency domain reference signal generation step S614. A series of frequency domain reference signal values represented by the frequency domain reference signal value data sequence generated by unit 123 is multiplied for each frequency to calculate a series of frequency domain compressed observation values.

  In the zero padding step S616, the zero padding unit 133 uses the processing device 911 to insert one or more zero values into the middle of the series of frequency domain compressed observation values calculated by the multiplier unit 132 in the multiplication step S615. Append.

  In the inverse Fourier transform step S617, the inverse Fourier transform unit 134 performs inverse Fourier transform on the series of frequency domain compressed observation values to which the zero padding unit 133 has added zero values in the zero padding step S616 using the processing device 911. Calculate a series of compressed observations.

In the peak extraction step S618, the peak extraction unit 141 uses the processing device 911 to extract a series of compressed observation values calculated by the inverse Fourier transform unit 134 in the inverse Fourier transform step S617. The peak extraction unit 141 uses the storage device 914 to store data representing the extracted peak.
Using the processing device 911, the peak extraction unit 141 returns the process to the azimuth direction estimated speed selection step S613, and the azimuth direction estimated speed selection unit 122 selects the next azimuth direction estimated speed.

In the speed estimation step S619, the speed estimation unit 151 uses the processing device 911 to select the peak represented by the data stored in the peak extraction step S618 from the set of the plurality of range direction estimated speeds and the plurality of azimuth direction estimated speeds. Is determined as a set of estimated speeds, and an estimated value of the moving speed 721 of the observation target 850 is determined.
The speed estimation unit 151 uses the output device 913 to output an estimated value of the estimated moving speed 721 of the observation target 850.

  FIG. 5 is a diagram for explaining the operation of the azimuth compressing unit 120 in this embodiment.

  The Fourier transform unit 131 performs a Fourier transform on the series of observation values 510 to calculate a series of frequency domain observation values 520. A series of observed values 510 Fourier-transformed by the Fourier transform unit 131 is composed of N complex numbers. N is preferably a power of 2 due to the nature of the fast Fourier transform algorithm, for example, 65536 (2 to the 16th power). The series of frequency domain observation values 520 calculated by the Fourier transform unit 131 is composed of N complex numbers, like the series of observation values 510.

  The frequency domain reference signal generation unit 123 generates a series of frequency domain reference signal values 530 corresponding to the estimated speed. A series of frequency domain reference signal values 530 generated by the frequency domain reference signal generation unit 123 is composed of N complex numbers, like the series of frequency domain observation values 520.

  The multiplier 132 multiplies the series of frequency domain observation values 520 and the series of frequency domain reference signal values 530 for each frequency to calculate a series of frequency domain compression observation values 540. The series of frequency domain compressed observation values 540 calculated by the multiplier 132 is composed of N complex numbers, like the series of frequency domain observation values 520 and the series of frequency domain reference signal values 530.

  The zero padding unit 133 adds M 0 values to the middle of the series of frequency domain compressed observations 540 to form a series of frequency domain compression observations 550. The series of frequency domain compressed observations 550 to which the zero padding unit 133 has added a zero value is followed by the first half [N / 2] complex numbers of the series of frequency domain compression observations 540 followed by M zero values. This is followed by [N / 2] complex numbers in the second half of the series of frequency domain compressed observations 540, consisting of [N + M] complex numbers in all.

The inverse Fourier transform unit 134 performs inverse Fourier transform on the series of frequency domain compressed observation values 550 to calculate a series of compressed observation values 560. The series of compressed observation values 560 calculated by the inverse Fourier transform unit 134 is composed of [N + M] complex numbers, like the series of frequency domain compression observation values 550. [N + M], like N, is preferably a power of 2 due to the nature of the fast Fourier transform algorithm, for example, 262144 (2 to the 18th power).
Since a series of observations 510 consisting of N complex numbers and a series of compressed observations 560 consisting of [N + M] complex numbers correspond to the same length of time, the sampling period T of the series of observations 510 and The relationship of T ′ = T · N / (N + M) is established between the sampling period T ′ of the series of compressed observation values 560. For example, if N is 65536 and [N + M] is 262144, T ′ is a quarter of T.

FIG. 6 is a diagram for explaining a peak of a series of compressed observation values.
An envelope 570a indicated by a solid line and an envelope 570b indicated by a broken line are envelopes of a series of compressed observation values calculated at different range direction estimation speeds. The envelope 570a is narrower and has a larger peak than the envelope 570b. This means that the range direction estimated speed of the envelope 570a is closer to the range direction moving speed 723 of the observation target 850 than the range direction estimated speed of the envelope 570b.
In addition, the envelope 570a and the envelope 570b differ not only in the peak size but also in the peak time (azimuth time). This is due to a phenomenon known as azimuth shift.

FIG. 7 is a diagram for explaining the azimuth shift.
Since the observation device 810 is moving at the moving speed 711, when the observation target 850 is stationary, the observation target 850 is viewed in the direction opposite to the movement direction of the observation device 810 when viewed from the observation device 810. It seems to move at a moving speed 731. In addition, when the observation target 850 is moving at the moving speed 721, the observation target 850 seems to be moving at the moving speed 732 obtained by combining the moving speed 721 and the moving speed 731 when the observation device 810 is used as a reference. Looks like.
When the observation object 850 is stationary, when the observation object 850 comes in a direction perpendicular to the moving direction of the observation apparatus 810, the distance between the observation apparatus 810 and the observation object 850 becomes the shortest distance 741. .
On the other hand, when the observation object 850 is moving at the moving speed 721, the observation object 850 comes to the position 751 slightly before the observation object 850 comes in a direction perpendicular to the movement direction of the observation device 810. The distance between the observation device 810 and the observation object 850 becomes the shortest distance 742. This is because the range direction moving speed 723 of the observation target 850 is directed in the direction away from the observation apparatus 810, and when the range direction movement speed 723 of the observation target 850 is directed in the direction approaching the observation apparatus 810. On the contrary, the distance between the observation device 810 and the observation object 850 becomes the shortest after the observation object 850 passes through the direction perpendicular to the moving direction of the observation device 810. The distance between the observation apparatus 810 and the observation object 850 is the shortest because the observation object 850 comes to a position in a direction perpendicular to the apparent moving speed 732 of the observation object 850 as viewed from the observation apparatus 810. Is the time.

FIG. 8 is a diagram illustrating a change in the distance between the observation apparatus 810 and the observation target 850.
A distance 761 indicated by a broken line indicates a case where the observation target 850 is stationary. A distance 762 indicated by a solid line indicates a case where the observation target 850 is moving at the moving speed 721 (when the azimuth direction moving speed 723 is a direction away from the observation apparatus 810).
As described above, when the observation object 850 is stationary, the distance 761 between the observation apparatus 810 and the observation object 850 is the shortest in the direction perpendicular to the moving direction of the observation apparatus 810. It is time 771 when 850 has come.
On the other hand, when the observation target 850 is moving at the movement speed 721, the distance 762 between the observation device 810 and the observation target 850 is the shortest because the observation target 850 is viewed from the observation device 810. This is a time 772 when the observation object 850 comes in a direction perpendicular to the moving speed 732.

In azimuth compression, a reference signal configured so that the peak of the compressed observation value appears at the time when the observation device 810 comes in a direction perpendicular to the moving direction of the observation device 810. For example, when azimuth compression is performed using a reference signal that assumes that the observation target 850 is stationary, the peak of the compressed observation value appears at the time when the distance between the observation device 810 and the observation target 850 becomes the shortest. .
For this reason, when the azimuth compression is performed using the reference signal that the observation target 850 is assumed to be stationary even though the observation target 850 is actually moving at the moving speed 721, the actual measurement of the observation target 850 is performed. The peak of the compressed observation value appears not at the time 771 corresponding to the azimuth direction position but at the time 772 when the distance between the observation device 810 and the observation target 850 is the shortest. In contrast, if azimuth compression is performed using a reference signal that matches the range direction moving speed 723 of the observation target 850, a peak of the compressed observation value appears at time 771 corresponding to the actual position of the observation target 850 in the azimuth direction. .

  Thus, when the range direction estimated speed is changed, not only the peak size of the compressed observation value but also the time (azimuth time) at which the peak of the compressed observation value appears changes.

FIG. 9 is a diagram for explaining a problem and its cause that occur when the sampling interval of the compressed observation values is wide.
A series of compressed observation values 580a are compression observation values calculated at a certain range direction estimated speed. The series of compression observation values 580b are compression observation values calculated at a range direction estimated speed slightly different from the compression observation value 580a.
In this example, the range direction estimated speed used to calculate the series of compressed observation values 580b is higher than the range direction estimated speed used to calculate the series of compressed observation values 580a. It is close to 723. For this reason, the peak of the series of compressed observation values 580b should be larger than the peak of the series of compression observation values 580a. However, the series of compressed observations 580a includes compression observations that capture the times when the compression observations peak, whereas the series of compression observations 580b captures the times when the compression observations peak. There are no compressed observations. This is because the time at which the compressed observation value reaches its peak shifts due to the azimuth shift. As a result, the series of compression observations 580a has a larger peak of the compression observations than the series of compression observations 580b.
The peak value 590A represents a peak of a series of compressed observation values calculated by gradually changing the range direction estimated speed. Due to the phenomenon described above, since the true peak of the compressed observation value may not be captured, the peak value 590A changes to jagged as shown in this figure. For this reason, it becomes difficult to determine the estimated speed in the range direction where the peak is maximum.

  When a large number of observation objects 850 having different moving speeds 721 are clustered within a relatively narrow range, the situation is further deteriorated. Since a plurality of peaks corresponding to a plurality of observation objects 850 cannot be distinguished by position, a plurality of peaks corresponding to a plurality of observation objects 850 are overlapped, and determination of a plurality of range direction estimation speeds where the peaks are maximum is as follows: Almost impossible.

FIG. 10 is a diagram for explaining the effect of narrowing the sampling interval of the compressed observation values.
A series of compressed observation values 580c are compression observation values calculated at a certain range direction estimated speed. The series of compression observation values 580d are compression observation values calculated at a range direction estimated speed slightly different from the compression observation value 580c.
The series of compressed observations 580c includes compression observations that capture the time when the compression observations peak, whereas the series of compression observations 580d includes compression that captures the time when the compression observations peak. There are no observations. However, since the sampling interval is narrow, the peak of the series of compressed observations 580d is almost close to the true peak.
The peak value 590B represents a peak of a series of compressed observation values calculated by gradually changing the range direction estimated speed. Since the sampling interval is narrow, even if the true peak of the compressed observation value cannot be captured, a value almost close to the true peak can be obtained, and the jagged peak value 590B is smaller than the peak value 590A. For this reason, it is relatively easy to determine the range direction estimated speed at which the peak is maximum.

  When a large number of observation objects 850 having different moving speeds 721 are dense within a relatively narrow range, a plurality of peaks corresponding to the plurality of observation objects 850 cannot be distinguished from each other by position. Corresponding peaks overlap. However, since the waveform of the peak corresponding to one observation object 850 has a gentle shape as shown in this figure, even if a plurality of these overlap, a plurality of range direction estimation speeds at which the peaks are maximum It can be easily distinguished.

  As the sampling frequency of the series of observation values 510 is higher, the amount of calculation required for the Fourier transform unit 131 to perform the Fourier transform increases. For this reason, the sampling frequency of the series of observation values 510 is preferably as low as possible. In consideration of this point, the sampling frequency of the series of observation values 510 is set as low as possible based on the azimuth direction resolution required for the synthetic aperture radar system 800.

If the sampling frequency of the series of compressed observations 560 is the same as the sampling frequency of the series of observations 510, the designed azimuth resolution can be achieved, but as described above, the true peak of the compression observations Therefore, it is insufficient to correctly estimate the moving speed 723 of the observation target 850 in the range direction.
By making the sampling frequency of the series of compressed observations 560 higher than the sampling frequency of the series of observations 510, the true peak of the compression observations can be captured, and the range direction moving speed 723 of the observation target 850 is correctly estimated. can do.

  As the sampling frequency of the series of compressed observation values 560 increases, the true peak of the compressed observation values can be captured more reliably, but the amount of calculation required for the inverse Fourier transform unit 134 to perform the inverse Fourier transform increases. However, since it is known in advance that the zero value added by the zero padding unit 133 is 0, the inverse Fourier transform unit 134 omits the calculation for the zero value added by the zero padding unit 133. can do. If the configuration of the inverse Fourier transform unit 134 is optimized in this way, an increase in the amount of calculation required for the inverse Fourier transform unit 134 to perform the inverse Fourier transform can be suppressed to some extent.

  Note that, in order to make the sampling frequency of the series of compressed observation values 560 higher than the sampling frequency of the series of observation values 510 (resampling), as described above, 0 value is added in the frequency domain and inverse Fourier transform is performed. In addition to the method of performing inverse Fourier transform and returning to the time domain, interpolation processing (eg, nearest-neighbor interpolation, linear interpolation, cubic convolution (cubic-) (convolution) and the like. The azimuth compression unit 120 may be configured such that the sampling frequency of the series of compressed observation values 560 is higher than the sampling frequency of the series of observation values 510 by using such other interpolation method. However, when performing interpolation by nearest neighbor interpolation or linear interpolation, the azimuth compression unit 120 increases the number of compressed observation values by interpolation, and then lowers the original sampling frequency by a low-pass filter. However, it is necessary to remove a high frequency component. In the case of performing interpolation by the cubic convolution interpolation method, it is not necessary to remove the high frequency component by the low pass filter.

Compared to the method of performing interpolation processing in the time domain, the method of adding a zero value in the frequency domain is superior in the following points.
First, the method of adding a zero value in the frequency domain has the highest reproducibility of the true peak of the compressed observation value. This is because the true peak of the compressed observation is very sharp.
Second, the method of adding a zero value in the frequency domain requires a smaller amount of calculation than other interpolation methods. Generally speaking, a method of adding a zero value in the frequency domain requires a Fourier transform to operate in the frequency domain, and an inverse Fourier transform to return to the time domain. The amount of calculation is larger than the method. However, in azimuth compression, Fourier transformation and inverse Fourier transformation are performed to obtain a correlation with the reference signal. Therefore, it is not necessary to separately perform Fourier transform / inverse Fourier transform, and an increase in calculation amount can be suppressed.
Because of such advantages, a method in which the azimuth compression unit 120 makes the sampling frequency of the series of compressed observation values 560 higher than the sampling frequency of the series of observation values 510 is preferably a method of adding a zero value in the frequency domain. .

  The number of 0 values added by the zero padding unit 133 may be a predetermined constant or a number specified by the user of the observation result analysis apparatus 100. For example, the zero padding unit 133 inputs the number of zero values designated by the user using the input device 912 and stores it using the storage device 914. The zero padding unit 133 uses the processing device 911 to add a zero value to the series of frequency domain compressed signal values calculated by the multiplication unit 132 based on the stored number of zero values.

The observation result analysis apparatus 100 in this embodiment is a speed estimation apparatus that estimates the speed of the observation target (850) based on the received signal (reflected wave 702) received by the synthetic aperture radar (observation apparatus 810).
The velocity estimation device (observation result analysis device 100) includes a compression observation calculation unit (azimuth compression unit 120), a peak extraction unit (141), and a velocity estimation unit (151).
The compression observation calculation unit (120), for each of a plurality of estimated speeds, between a series of observed values (510) having a predetermined sampling interval generated from the received signal and a reference signal corresponding to the estimated speed By taking the correlation, a series of compressed observation values (560) having a sampling interval shorter than the predetermined sampling interval are calculated.
The peak extraction unit (141) extracts a peak of a series of compressed observation values (560) calculated by the compression observation calculation unit (120) for each of the plurality of estimated speeds.
The speed estimation unit (151) obtains an estimated speed at which the peak extracted by the peak extraction unit (141) is a maximum from the plurality of estimated speeds, and sets it as the estimated speed of the observation target (850).

The compressed observation calculation unit generates a series of compressed observations that have a sampling interval shorter than the sampling interval of the series of observations, so that the peak of the compressed observations can be captured more accurately and the speed of the observation target is correctly set. Can be estimated.
In particular, it is possible to accurately capture the peak of the compressed observation value even if it is difficult to capture the peak of the compressed observation value where multiple observation objects with different moving speeds are densely packed within a relatively narrow range, It is possible to correctly estimate the speed of each of a plurality of observation objects.

The compression observation calculation unit (120) includes a Fourier transform unit (131), a frequency domain reference signal generation unit (123), a multiplication unit (132), a zero padding unit (133), and an inverse Fourier transform unit (134). And have.
The Fourier transform unit (131) performs a Fourier transform on the series of observation values (510) to calculate a series of frequency domain observation values (520).
For each of the plurality of estimated speeds, the frequency domain reference signal generation unit (123) generates a series of frequency domain reference signal values (530) obtained by Fourier-transforming the reference signal corresponding to the estimated speed.
The multiplication unit (132), for each of the plurality of estimated speeds, a series of frequency domain observation values (520) calculated by the Fourier transform unit (131) and a series generated by the frequency domain reference signal generation unit (123). Are respectively calculated with a frequency domain reference signal value (530) to obtain a series of frequency domain compressed observation values (540).
The zero padding unit (133) adds one or more 0 values to the series of frequency domain compressed observation values (540) calculated by the multiplication unit (132) for each of the plurality of estimated speeds.
For each of the plurality of estimated speeds, the inverse Fourier transform unit (134) performs inverse Fourier transform on the series of frequency domain compressed observation values (550) to which the zero padding unit (133) has added a zero value, and the series of the above-described series of estimated speeds. The compression observation value (560) is calculated.

  The zero padding part adds a zero value in the frequency domain to make the sampling interval of a series of compressed observations shorter than the sampling interval of a series of observations, thus reproducing the peak of the compressed observations with a small amount of calculation can do.

The compression observation calculation unit (120) calculates the series of compression observation values (560) for each of the plurality of range direction estimated velocities.
The peak extraction unit (141) extracts a peak of a series of compressed observation values (560) calculated by the compression observation calculation unit (120) for each of the plurality of range direction estimated velocities.
The speed estimation unit (151) obtains a range direction speed at which the peak extracted by the peak extraction unit (141) is a maximum from the plurality of range direction estimation speeds, and the range direction of the observation target (850) Estimated speed.

  Changing the range direction estimation speed makes it difficult to capture the peak of the compressed observation due to the azimuth shift, but the compression observation calculation unit generates a series of compressed observations with a sampling interval shorter than the sampling interval of the series of observations. The peak of the compressed observation value can be grasped more accurately, and the moving speed in the range direction of the observation object can be correctly estimated.

The compression observation calculation unit (120) performs the above series of compression observation values (for each set of the range direction estimated speed selected from the plurality of range direction estimated speeds and the azimuth direction estimated speed selected from the plurality of azimuth direction estimated speeds). 560) is calculated.
The peak extraction unit (141) extracts a peak of a series of compressed observation values (560) calculated by the compression observation calculation unit (120) for each set of the range direction estimated speed and the azimuth direction estimated speed.
The speed estimation unit (151) includes a range direction estimation speed and an azimuth direction estimation speed at which the peak extracted by the peak extraction unit (141) is maximized from the set of the range direction estimation speed and the azimuth direction estimation speed. The range direction estimated speed and the azimuth direction estimated speed of the observation object (850) are obtained.

For each set of range direction estimated speed and azimuth direction estimated speed, the compression observation calculation unit calculates a series of compression observation values, and the peak extraction unit extracts a peak of the calculated series of compression observation values and extracts the extracted peaks. Since the speed estimator obtains a set of the range direction estimated speed and the azimuth direction estimated speed at which is maximized, the range direction speed and the azimuth direction speed to be observed can be estimated simultaneously.
In particular, even in situations where multiple observation targets with different moving speeds are densely packed within a relatively narrow range, it is difficult to capture the peak of the compressed observation value. By estimating, peaks corresponding to a plurality of observation targets can be captured, and the moving speed of each of the plurality of observation targets can be accurately estimated.

  The observation result analysis apparatus 100 in this embodiment can be realized by causing a computer to execute a computer program that causes the computer to function as the speed estimation apparatus.

  100 observation result analyzer, 120 azimuth compression unit, 121 range direction estimated speed selection unit, 122 azimuth direction estimated speed selection unit, 123 frequency domain reference signal generation unit, 131 Fourier transform unit, 132 multiplication unit, 133 zero padding unit, 134 Inverse Fourier transform unit, 141 peak extraction unit, 151 velocity estimation unit, 510 observation value, 520 frequency domain observation value, 530 frequency domain reference signal value, 540, 550 frequency domain compression observation value, 560, 580 compression observation value, 570 envelope Line, 590 peak value, 701 radio wave, 702 reflected wave, 711, 721, 722, 723, 731, 732 travel speed, 741, 742, 761, 762 distance, 751 position, 771, 772 time, 800 synthetic aperture radar system, 810 observation equipment, 850 observation object, 11 processing unit, 912 input unit, 913 output unit, 914 storage device.

Claims (5)

Based on the received signal received by the synthetic aperture radar, in the speed estimation device that estimates the speed of the observation target ,
For each estimated speed of the multiple, by taking a series of observations with a predetermined sampling interval generated from the received signal, a correlation between the reference signal corresponding to the estimated speed, the predetermined sampling interval A compression observation calculator that calculates a series of compression observations having a shorter sampling interval ,
For each top Symbol plurality of estimated speed, a peak extractor for extracting peaks of a series of compression observations that the compression observing calculating unit has calculated,
From among the above SL plurality of estimated speed, seeking the estimated velocity peaks the peak extraction unit has extracted is maximized, and the speed estimating section to estimate the speed of the observed object
Have
The compression observation calculation unit
A Fourier transform unit for Fourier transforming the series of observations to calculate a series of frequency domain observations;
For each of the plurality of estimated speeds, a frequency domain reference signal generation unit that generates a series of frequency domain reference signal values obtained by Fourier transforming a reference signal corresponding to the estimated speed;
For each of the plurality of estimated velocities, a product of a series of frequency domain observation values calculated by the Fourier transform unit and a series of frequency domain reference signal values generated by the frequency domain reference signal generation unit is calculated, respectively. A multiplier for the frequency domain compression observation of
For each of the plurality of estimated speeds, a zero padding unit that inserts and adds one or more zero values in the middle of the frequency domain of the series of frequency domain compression observation values calculated by the multiplication unit;
For each of the plurality of estimated speeds, an inverse Fourier transform unit that calculates a series of compressed observation values by performing an inverse Fourier transform on the series of frequency domain compression observation values to which the zero padding unit adds a zero value;
Speed estimation device according to claim <br/> to have. The compression observation calculation unit calculates the series of compression observation values for each of a plurality of range direction estimated speeds,
The peak extraction unit extracts a peak of a series of compression observation values calculated by the compression observation calculation unit for each of the plurality of range direction estimation speeds,
The speed estimation unit obtains a range direction speed at which the peak extracted by the peak extraction unit is a maximum from the plurality of range direction estimation speeds, and sets the range direction speed of the observation target, The speed estimation device according to claim 1 . The compression observation calculation unit calculates the series of compression observation values for each set of the range direction estimated speed selected from the plurality of range direction estimated speeds and the azimuth direction estimated speed selected from the plurality of azimuth direction estimated speeds. ,
The peak extraction unit extracts a peak of a series of compression observation values calculated by the compression observation calculation unit for each set of the range direction estimation speed and the azimuth direction estimation speed,
The speed estimation unit obtains a set of the range direction estimated speed and the azimuth direction estimated speed at which the peak extracted by the peak extraction unit is maximized from the set of the range direction estimated speed and the azimuth direction estimated speed. , the speed estimation device of claim 1 or 2, characterized in that the range direction estimated velocity of the observation target and the azimuth direction estimated speed. A computer program that, when executed by a computer, functions as the speed estimation device according to any one of claims 1 to 3 . In a speed estimation method in which a speed estimation device having a compression observation calculation unit, a peak extraction unit, and a speed estimation unit estimates the speed of an observation target based on a received signal received by a synthetic aperture radar,
The compression observation calculating unit, for each of the plurality of estimated speeds, using a series of observations obtained by sampling the received signal at predetermined sampling intervals, and the received signal, between the reference signal corresponding to the estimated speed By taking a correlation, a series of compressed observations having a time interval shorter than the predetermined sampling interval is calculated,
The peak extraction unit extracts a peak of a series of compression observation values calculated by the compression observation calculation unit for each of the plurality of estimated speeds,
The speed estimator obtains a speed at which the peak extracted by the peak extractor is maximized from the plurality of estimated speeds, and determines the estimated speed of the observation target ,
When the compression observation calculation unit calculates the series of compression observation values for each of the plurality of estimated speeds,
The Fourier transform unit of the compressed observation calculation unit Fourier-transforms the series of observation values to calculate a series of frequency domain observation values,
The frequency domain reference signal generation unit of the compression observation calculation unit generates a series of frequency domain reference signal values obtained by Fourier transforming the reference signal corresponding to the estimated speed for each of the plurality of estimated speeds,
For each of the plurality of estimated velocities, the multiplication unit of the compression observation calculation unit performs a series of frequency domain observation values calculated by the Fourier transform unit, and a series of frequency domain reference signal values generated by the frequency domain reference signal generation unit. To calculate a series of frequency domain compression observations,
The zero padding unit of the compression observation calculation unit inserts one or more zero values in the middle of the frequency domain of the series of frequency domain compression observation values calculated by the multiplication unit for each of the plurality of estimated speeds. ,
The inverse Fourier transform unit of the compressed observation calculation unit performs inverse Fourier transform on a series of frequency domain compressed observation values to which the zero padding unit adds a zero value for each of the plurality of estimated speeds, and the series of compressed observation values. A speed estimation method characterized by calculating .

Images