JP4818898B2 - Radar image processing device - Google Patents

Description

  The present invention relates to a radar image processing apparatus capable of detecting changes occurring on the ground surface from between radar images taken at different times, classifying the detected changes, and detecting what kind of change has occurred. Is.

  A conventional radar image processing apparatus calculates coherence, which is a correlation between two radar images acquired at different times, and uses the fact that coherence decreases at a portion where a change occurs between images. A change occurring between images is detected by threshold processing based on strength (for example, see Non-Patent Document 1). The change occurring between the images indicates a change in the ground surface such as a vehicle dredging, a fire, or an explosion that occurred within the imaging time interval.

C.V.Jakowatz, Jr., D.E.Wahl, P.H.Eichel, D.C.Ghiglia, and P.A.Thompson, "Spotlight-Mode Synthetic Aperture Radar: A Signal Processing Approach", Kluwer academic publisher, p.330-340, 1999. JSLee, KWHoppel, SAMango, and ARMiller, "Intensity and Phase Statistics of Multilook Polarimetric and Interferometric SAR Imagery", IEEE Transactions on Geoscience and Remote Sensing, vol.32, no.5, pp.1017-pp.1028 , 1994. Mikio Takagi, Yoshihisa Shimoda, New Image Analysis Handbook, p.1551-p.1650, University of Tokyo Press, 2004.

  Since the conventional radar image processing apparatus uses the coherence whose relation to the state of the change occurring on the ground surface is not clear as the detection index value, it cannot be determined what kind of change is occurring at the point where the change is detected. There was a problem.

  In addition, the coherence decreases not only according to the change that occurred on the ground surface, but also changes in the observation conditions at the time of two radar image captures, for example, the irradiation angle of the radar wave and the trajectory of the platform on which the radar is mounted, Since the coherence value alone cannot distinguish between a decrease in coherence caused by changes other than those occurring on the ground surface and a decrease in coherence due to changes occurring on the ground surface, falsely detect ground surface changes. There was a problem.

  The present invention has been made to solve the above-described problems. The purpose of the present invention is to detect a change occurring on the ground surface with high accuracy and to a state of the change occurring on the ground surface. A radar image processing apparatus that can be categorized accordingly is obtained.

  A radar image processing apparatus according to the present invention calculates a coherence that is a complex correlation between a radar image storage unit that stores a plurality of radar images captured at different times and a radar image stored in the radar image storage unit. A coherence calculation unit, a phase difference calculation unit that calculates a phase difference between radar images stored in the radar image storage unit, and a predetermined pixel centered on a pixel at a position set by the coherence calculation unit to calculate coherence A statistical distribution calculating unit for calculating a statistical distribution of phase differences for each range of the above, a theoretical statistical distribution calculating unit for calculating a theoretical statistical distribution of phase differences based on the coherence calculated by the coherence calculating unit, and the statistical distribution Comparison between the statistical distribution calculated by the calculation unit and the statistical distribution calculated by the theoretical statistical distribution calculation unit, and the difference between the statistical distributions A test unit for calculating a statistical parameter, a threshold designating unit for inputting a threshold for detecting a change in the ground surface based on the statistical parameter calculated by the test unit, the statistical parameter calculated by the test unit, and the A detection processing unit that compares the threshold value input by the threshold value specifying unit and detects a location that exceeds the threshold value as a location on the ground surface is provided.

  The radar image processing apparatus according to the present invention can detect changes generated on the ground surface with high accuracy and can be classified according to the state of changes generated on the ground surface.

Embodiment 1 FIG.
A radar image processing apparatus according to Embodiment 1 of the present invention will be described with reference to FIGS. 1 is a block diagram showing a configuration of a radar image processing apparatus according to Embodiment 1 of the present invention. In the following, in each figure, the same reference numerals indicate the same or corresponding parts.

  In FIG. 1, a radar image processing apparatus according to Embodiment 1 includes a radar image storage unit 1 that stores a plurality of radar images captured at different times, an output storage unit 2 that stores the output of the present apparatus, A control unit 100 is provided that controls input / output to / from the radar image storage unit 1 and the output storage unit 2 and data exchange between parts constituting the apparatus described later.

  The control unit 100 includes a coherence calculation unit 200, a phase difference calculation unit 300, a statistical distribution calculation unit 400, a theoretical statistical distribution calculation unit 500, a test unit 600, a threshold specification unit 700, and a detection processing unit 800. Is provided.

  The coherence calculation unit 200 calculates the coherence between two complex radar images input via the control unit 100. The phase difference calculation unit 300 calculates a phase difference between complex radar images. The statistical distribution calculation unit 400 calculates the statistical distribution of the phase difference calculated by the phase difference calculation unit 300. The theoretical statistical distribution calculation unit 500 calculates a statistical distribution of theoretical phase differences based on the coherence calculated by the coherence calculation unit 200. The verification unit 600 compares the statistical distribution of the phase difference calculated by the statistical distribution calculation unit 400 with the statistical distribution calculated by the theoretical statistical distribution calculation unit 500, and tests the difference between these distributions. The detection processing unit 800 compares the distribution difference obtained by the verification unit 600 with a change detection threshold value input from the threshold value specifying unit 700, and changes on the ground surface are detected at locations where the distribution difference exceeds the threshold value. Detect as the place where it was.

  Next, the operation of the radar image processing apparatus according to the first embodiment will be described with reference to the drawings. FIG. 2 is a flowchart showing the operation of the radar image processing apparatus according to Embodiment 1 of the present invention.

  First, in step ST100, the control unit 100 reads two complex radar images captured at different times from the radar image storage unit 1.

  Next, in step ST200, the control unit 100 outputs (passes) two complex radar images to the coherence calculation unit 200. The coherence calculation unit 200 calculates coherence from two radar images at the position P (m, n) of each pixel in the radar image. The coherence ρ (m, n) at the position P (m, n) is expressed by the following equation (1).

Here, s _{1, k} in the first sheet of the radar image (m, n) to indicate the complex data values with the N pixels centered, s _{2, k} is the second piece of the radar image definitive complex data Indicates the value. The number N corresponds to a range for calculating coherence centered on (m, n). * Indicates a complex conjugate. ρ represents amplitude and θ represents declination. This coherence takes a value close to 1 when there is no change in the ground surface under the condition that two images are acquired in the same trajectory, and the ground surface has a larger area than the range N for calculating the coherence. It takes a value close to 0 at the place where the change occurs.

  Next, in step ST300, the control unit 100 outputs two complex radar images to the phase difference calculation unit 300. The phase difference calculation unit 300 calculates a phase difference from two radar images at the position P (m, n) of each pixel in the radar image. The phase difference Φ (m, n) at the position P (m, n) is expressed by the following equation (2).

Here, angle is a function indicating a declination of a complex number, and s ₁ (m, n) and s ₂ (m, n) are positions (m, n) on the first radar image and the second radar image, respectively. ) Indicates the complex data value.

  Next, in step ST400, the control unit 100 outputs the phase difference calculated by the phase difference calculation unit 300 to the statistical distribution calculation unit 400. The statistical distribution calculation unit 400 includes a frequency distribution of a phase difference Φ (m, n) for each predetermined range centered on a pixel at a position (m, n) set for the coherence calculation unit 200 to calculate coherence. That is, a statistical distribution is obtained.

  Next, in step ST500, the control unit 100 outputs the coherence calculated by the coherence calculation unit 200 to the theoretical statistical distribution calculation unit 500. The theoretical statistical distribution calculation unit 500 calculates the theoretical statistical distribution of the phase difference based on the coherence amplitude ρ and declination θ. This theoretical statistical distribution is expressed by the following equation (3) as derived in Non-Patent Document 2.

  Here, Φ is the phase difference, the range is from −π to π, n is the multilook number, and β = ρcos [Φ−θ]. Γ is a gamma function, and F is a Gaussian hypergeometric function.

  In addition, although formula (3) was used for calculation of theoretical statistical distribution, it is not restricted to this, You may use another formula as long as it shows statistical distribution of phase difference.

  Next, in step ST600, the control unit 100 outputs the statistical distribution calculated by the statistical distribution calculation unit 400 and the theoretical statistical distribution calculated by the theoretical statistical distribution calculation unit 500 to the test unit 600. The test unit 600 compares two input statistical distributions and calculates a difference between the statistical distributions. This statistical distribution difference is measured for statistical parameters such as mean, standard deviation, skewness, kurtosis, and higher order moments. Here, the calculation of the distribution difference may be any method that can calculate the statistical distribution difference, such as Student's t-test, F-test, and Kolmogorov-Smirnov test. Here, the calculated number of statistical parameters is Q. Further, the test unit 600 displays a statistical parameter that is a difference between the statistical distributions calculated here on a display or the like.

  Next, in step ST700, the analyst inputs the threshold value for the statistical parameter calculated by the test unit 600 and displayed on the display or the like as an index. The threshold designation unit 700 reads the threshold for change detection input by the analyst and outputs the read threshold to the detection processing unit 800. The number of threshold values read here corresponds to the number Q of statistical parameters calculated by the test unit 600.

  Finally, in step ST800, control unit 100 outputs the statistical distribution difference calculated by test unit 600 to detection processing unit 800. The detection processing unit 800 compares the statistical distribution difference with a threshold value for change detection, and detects a location where the statistical distribution difference exceeds all threshold values as a location on the ground surface. Then, the control unit 100 stores the detection result in the output storage unit 2.

  As described above, not only coherence but also a plurality of statistical parameters representing deviations from the theoretical value of the statistical distribution of the phase difference are obtained, and changes in the ground surface are detected using these as index values. It is possible to detect a minute change that cannot be detected by itself.

  In addition, since a plurality of statistical parameters are used as index values, a threshold value for detection can be weighted and determined, and a desired ground surface change, that is, a change in which only the statistical parameter specified by the threshold weighting changes. It is possible to detect only the portion where the occurrence occurs.

Embodiment 2. FIG.
In the first embodiment, the presence / absence of a change in the ground surface is detected using a deviation from the theoretical value of the statistical distribution as an index. In the second embodiment, the deviation of the ground surface is detected using the deviation as an index. It categorizes the type of change.

  A radar image processing apparatus according to Embodiment 2 of the present invention will be described with reference to FIGS. FIG. 3 is a block diagram showing the configuration of the radar image processing apparatus according to Embodiment 2 of the present invention.

  In FIG. 3, the same reference numerals as those in FIG. 1 are the same as or equivalent to those described above. A classification processing unit 900 provided in place of the detection processing unit 800 detects whether or not there is a change in the ground surface by using a deviation from the theoretical value of the statistical distribution of the phase difference as an index value, and classifies the change.

  Next, the operation of the radar image processing apparatus according to the second embodiment will be described with reference to the drawings. FIG. 4 is a flowchart showing the operation of the radar image processing apparatus according to the second embodiment of the present invention.

  In FIG. 4, the same reference numerals as those in FIG. 2 are the same or equivalent.

  In step ST900, the control unit 100 outputs the statistical distribution difference calculated by the test unit 600 to the classification processing unit 900. The classification processing unit 900 compares the statistical distribution difference with a threshold value for change detection, and detects a location where the statistical distribution difference exceeds the threshold value as a location on the ground surface. Next, using the statistical distribution difference at the location where the change is detected as an index value, the change at the location where the change is detected is classified. That is, those having similar statistical parameter values at locations where changes are detected are classified as having the same type of ground surface change. As this classification method, for example, a clustering method, a multi-level slice, a dension tree method, or the like as described in Non-Patent Document 3 is used. Then, the control unit 100 stores this classification result in the output storage unit 2.

  As described above, since changes in the ground surface are detected using a plurality of statistical parameters indicating deviations from the statistical distribution of phase differences as index values, and classification processing is performed based on the index values, changes that occur on the ground surface It is possible to classify the difference in the state of.

  In addition, since multiple statistical parameters are used as index values, statistical parameters that change due to factors other than changes that occur on the ground surface can be removed during detection and classification, and changes that occur on the ground surface can be detected and classified with high accuracy. can do.

Embodiment 3 FIG.
In the first and second embodiments, the statistical distribution of the phase difference is compared with the theoretical statistical distribution, and detection and classification are performed based on the difference. In the third embodiment, the phase difference in the image is detected. By comparing the statistical distributions, the types of changes that occurred on the ground surface are classified.

  A radar image processing apparatus according to Embodiment 3 of the present invention will be described with reference to FIGS. FIG. 5 is a block diagram showing a configuration of a radar image processing apparatus according to Embodiment 3 of the present invention.

  In FIG. 5, the same reference numerals as those in FIGS. 1 and 3 are the same or equivalent.

  Next, the operation of the radar image processing apparatus according to the third embodiment will be described with reference to the drawings. FIG. 6 is a flowchart showing the operation of the radar image processing apparatus according to Embodiment 3 of the present invention.

  In FIG. 6, the same reference numerals as those in FIG. 4 are the same or equivalent.

  In step ST610, the control unit 100 outputs the statistical distribution of the phase difference calculated in a predetermined range centered on each pixel calculated by the statistical distribution calculation unit 400 to the test unit 600. The test unit 600 calculates a statistical parameter of the statistical distribution of the phase difference. Further, the test unit 600 displays the difference between the distributions calculated here on a display or the like.

  In step ST910, control unit 100 outputs the statistical parameter of the statistical distribution calculated by test unit 600 to classification processing unit 900. This classification processing unit 900 compares the statistical parameter of the statistical distribution of the phase difference in a predetermined range centering on each pixel with the threshold value input by the analyst, and finds a place where the statistical parameter exceeds all the threshold values on the ground surface. It is detected as a place where there has been a change. Next, using the statistical parameter at the location where the change is detected as an index value, the change at the location where the change is detected is classified. Those with similar statistical parameter values are classified as having the same kind of ground surface change. Then, the control unit 100 stores the detection and classification results in the output storage unit 2.

  As described above, classification is performed using only the statistical distribution of the phase difference calculated between the input radar images without calculating the theoretical statistical distribution, so the calculation load required for calculating the theoretical statistical distribution is reduced. Reduction and high-speed processing can be realized.

  In addition, since multiple statistical parameters are used as index values, statistical parameters that change due to factors other than changes that occur on the ground surface can be removed during detection and classification, and changes that occur on the ground surface can be detected and classified with high accuracy. can do.

Embodiment 4 FIG.
In the first, second, and third embodiments, the statistical distribution of the phase difference is compared between the theoretical statistical distribution or the statistical distributions calculated from the input radar images. 4 categorizes the types of changes that have occurred on the ground surface by comparing with a database of statistical distributions of phase differences corresponding to the state of ground surface changes stored in advance.

  A radar image processing apparatus according to Embodiment 4 of the present invention will be described with reference to FIGS. FIG. 7 is a block diagram showing a configuration of a radar image processing apparatus according to Embodiment 4 of the present invention.

  In FIG. 7, the same reference numerals as those in FIG. 5 are the same or equivalent. In the statistical distribution database 3, the phase difference between the radar images before and after the change according to the state of the ground surface change, for example, the track of a large vehicle equipped with a caterpillar or a vehicle equipped with a tire, or a landslide. The statistical distribution is calculated and accumulated in advance by experimental or theoretical calculation. The database reference type test unit 620 replaced with the test unit 600 compares the statistical distribution of the phase difference calculated by the statistical distribution calculation unit 400 with the statistical distribution accumulated in the statistical distribution database 3, and tests the difference between the statistical distributions. To do. The database reference type classification processing unit 920 replaced with the classification processing unit 900 detects and classifies changes occurring on the ground surface based on the difference between the statistical distributions of the phase differences calculated by the database reference type test unit 620.

  Next, the operation of the radar image processing apparatus according to the fourth embodiment will be described with reference to the drawings. FIG. 8 is a flowchart showing the operation of the radar image processing apparatus according to Embodiment 4 of the present invention.

  In FIG. 8, the same reference numerals as those in FIG. 6 denote the same or corresponding parts.

  In step ST620, the control unit 100 outputs the statistical distribution of the phase difference in a predetermined range centered on each pixel calculated by the statistical distribution calculation unit 400 to the database reference type test unit 620. The database reference type test unit 620 compares the input statistical distribution of the phase difference with the statistical distribution of the phase difference accumulated in the statistical distribution database 3 to obtain a difference between the statistical distributions. Differences between statistical distributions (a set of statistical parameters) calculated here are calculated for each pixel by the number of statistical distributions accumulated in the statistical distribution database 3. Here, this number is indicated by R. Furthermore, the database reference type test unit 620 displays the difference between the distributions calculated here on a display or the like.

  Next, in step ST720, the analyst inputs the threshold value for the statistical parameter calculated by the database reference type test unit 620 and displayed on the display or the like. The threshold designation unit 700 reads a threshold for detecting a change input by the analyst and outputs the read threshold to the database reference type classification processing unit 920.

  Finally, in step ST920, the control unit 100 outputs the statistical distribution difference calculated by the database reference type test unit 620 to the database reference type classification processing unit 920. The database reference type classification processing unit 920 associates the statistical distribution of the phase difference calculated in each pixel with the statistical distribution stored in the statistical distribution database 3. Here, first, the database reference type classification processing unit 920 compares the difference between the R statistical distributions obtained in each pixel and accumulated in the statistical distribution database 3 with a threshold value. Of the statistical distributions in the statistical distribution database 3, a statistical distribution whose difference exceeds a threshold value is regarded as a non-corresponding statistical distribution. Here, when the difference with respect to all the statistical distributions in the statistical distribution database 3 exceeds a threshold value, there is no statistical distribution corresponding to the pixel. Next, it is assumed that the statistical distribution in the statistical distribution database 3 in which the difference is equal to or smaller than the threshold is the statistical distribution corresponding to the pixel, which has the smallest difference. Then, the control unit 100 stores the detection and classification results in the output storage unit 2.

  As described above, the statistical distribution of the phase difference according to the ground surface change state obtained in advance is referred to, and the difference from the distribution is calculated and classified. The status can be detected and classified.

  In addition, since multiple statistical parameters are used as index values, statistical parameters that change due to factors other than changes that occur on the ground surface can be detected and removed during classification, and changes that occur on the ground surface can be detected with high accuracy. And can be categorized.

It is a block diagram which shows the structure of the radar image processing apparatus which concerns on Embodiment 1 of this invention. It is a flowchart which shows operation | movement of the radar image processing apparatus which concerns on Embodiment 1 of this invention. It is a block diagram which shows the structure of the radar image processing apparatus which concerns on Embodiment 2 of this invention. It is a flowchart which shows operation | movement of the radar image processing apparatus which concerns on Embodiment 2 of this invention. It is a block diagram which shows the structure of the radar image processing apparatus which concerns on Embodiment 3 of this invention. It is a flowchart which shows operation | movement of the radar image processing apparatus which concerns on Embodiment 3 of this invention. It is a block diagram which shows the structure of the radar image processing apparatus which concerns on Embodiment 4 of this invention. It is a flowchart which shows operation | movement of the radar image processing apparatus which concerns on Embodiment 4 of this invention.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 1 Radar image storage part, 2 Output storage part, 3 Statistical distribution database, 100 Control part, 200 Coherence calculation part, 300 Phase difference calculation part, 400 Statistical distribution calculation part, 500 Theoretical statistical distribution calculation part, 600 Test part, 620 Database Reference type verification unit, 700 threshold specification unit, 800 detection processing unit, 900 classification processing unit, 920 database reference type classification processing unit.

Claims (5)

A radar image storage unit for storing a plurality of radar images captured at different times;
A coherence calculating unit that calculates coherence that is a complex correlation between radar images stored in the radar image storage unit;
A phase difference calculation unit for calculating a phase difference between radar images stored in the radar image storage unit;
A statistical distribution calculator that calculates a statistical distribution of phase differences for each predetermined range centered on a pixel at a position set by the coherence calculator to calculate coherence;
A theoretical statistical distribution calculator that calculates a theoretical statistical distribution of phase differences based on the coherence calculated by the coherence calculator;
A statistical unit calculated by the statistical distribution calculating unit and a statistical unit that compares the statistical distribution calculated by the theoretical statistical distribution calculating unit and calculates a statistical parameter indicating a difference between the statistical distributions;
Based on the statistical parameter calculated by the test unit, a threshold designating unit for inputting a threshold for detecting a change in the ground surface
A detection processing unit that compares the statistical parameter calculated by the verification unit with the threshold value input by the threshold value specifying unit and detects a location that exceeds the threshold value as a location on the ground surface. A radar image processing apparatus. Instead of the detection processing unit,
The statistical parameter calculated by the verification unit and the threshold value input by the threshold value specifying unit are compared, and a location exceeding the threshold value is detected as a location on the ground surface, and a change is detected. The radar image processing apparatus according to claim 1, further comprising: a classification processing unit that classifies the parts having similar statistical parameter values as having the same kind of change on the ground surface. A radar image storage unit for storing a plurality of radar images captured at different times;
A coherence calculating unit that calculates coherence that is a complex correlation between radar images stored in the radar image storage unit;
A phase difference calculation unit for calculating a phase difference between radar images stored in the radar image storage unit;
A statistical distribution calculator that calculates a statistical distribution of phase differences for each predetermined range centered on a pixel at a position set by the coherence calculator to calculate coherence;
A test unit that compares the statistical distributions calculated by the statistical distribution calculation unit and calculates statistical parameters indicating differences between the statistical distributions;
Based on the statistical parameter calculated by the test unit, a threshold designating unit for inputting a threshold for detecting a change in the ground surface
A classification processing unit that compares the statistical parameter calculated by the test unit with the threshold value input by the threshold value specifying unit and detects a location that exceeds the threshold value as a location on the ground surface. A radar image processing apparatus. The classification processing unit compares the statistical parameter calculated by the verification unit with the threshold value input by the threshold value specifying unit, and detects a location exceeding the threshold value as a location on the ground surface. 4. The radar image processing apparatus according to claim 3, wherein those having a similar statistical parameter value at a place where a change is detected are classified as having the same kind of change on the ground surface. A radar image storage unit for storing a plurality of radar images captured at different times;
A coherence calculating unit that calculates coherence that is a complex correlation between radar images stored in the radar image storage unit;
A phase difference calculation unit for calculating a phase difference between radar images stored in the radar image storage unit;
A statistical distribution calculator that calculates a statistical distribution of phase differences for each predetermined range centered on a pixel at a position set by the coherence calculator to calculate coherence;
A statistical distribution database in which statistical distributions of phase differences between radar images before and after the change according to the state of ground surface change are accumulated in advance,
A database reference type test unit that compares the statistical distribution calculated by the statistical distribution calculation unit with the statistical distribution stored in the statistical distribution database and calculates a statistical parameter indicating a difference between the statistical distributions;
Based on the statistical parameter calculated by the database reference type test unit, a threshold value specifying unit for inputting a threshold value for detecting a change in the ground surface;
The statistical distribution difference accumulated in the statistical distribution database is compared with the threshold input by the threshold designating unit, and the statistical distribution that does not exceed the threshold and has the smallest difference between the statistical distributions is the statistical distribution database. A radar image processing apparatus, comprising: a database reference type classification processing unit that detects and classifies changes in the ground surface by specifying in the data.

Images