JP5448757B2 - Radar image processing device - Google Patents

Description

  The present invention relates to a radar image processing apparatus that is used in a synthetic aperture radar, a reverse synthetic aperture radar, a real aperture radar, and the like and performs image processing of these radars.

  In general, a synthetic aperture radar, a reverse synthetic aperture radar, and a real aperture radar obtain a radar image by irradiating an observation target with a radio wave, receiving the reflected wave, and performing signal processing. In these types of radar, when a radio wave is irradiated to the observation target, a part of the radio wave is also irradiated outside the observation target range, so that a strong reflected wave may be generated by an object outside the observation target range. . In such a case, when a radar image in the observation target range is reproduced, there is a problem that an image based on a strong reflected wave from outside the observation target range is displayed as an image in the observation target range. An image reflected in a radar image of an object that does not exist within the observation range is called a virtual image.

  Here, for example, in a conventional radar image virtual image detection method as shown in Non-Patent Document 1, an image is reproduced for each data on the right side of the beam, the center of the beam, and the left side of the beam, and only the image on the right side or left side of the beam is reproduced. The captured image can be identified as a virtual image. The virtual image that can be detected by this conventional virtual image detection method is azimuth ambiguity, which is a virtual image generated by a strong reflected wave entering from the side of the radar beam. Note that a virtual image detection method other than azimuth ambiguity is not generally known, and in many cases, a virtual image is detected from a reproduced radar image by visual observation of a radar image user.

G. Hajduch, P. Leilde, and V. Kerbaol, “Ship detection on ENVISAT ASAR data: results, limitations and perspectives,” Proceedings of SEASAR 2006.

  The conventional virtual image detection method as described above uses the feature that the azimuth ambiguity is generated by the reflected wave from the side. For this reason, the conventional virtual image detection method as described above is used for a virtual image due to a strong reflected wave from above or below the radar beam called range ambiguity, broadcasting, communication or other radar called an interference wave. There has been a problem that it is impossible to detect a virtual image or the like caused by receiving a radio wave. Therefore, the conventional virtual image detection method as described above cannot detect a virtual image of a type other than azimuth ambiguity.

  The present invention has been made to solve the above-described problems, and an object of the present invention is to obtain a radar image processing apparatus that can also detect virtual images of types other than azimuth ambiguity.

The radar image processing apparatus according to the present invention is configured to process the radar image based on a preset shift amount having a resolution as a basic unit with respect to a processing target image that is one of all or part of the acquired radar image. A shift processing unit that shifts in the axial direction of at least one of two axes for representing, the processing target image after the shift by the shift processing unit, and the processing target image before the shift; A coherence map calculation unit that calculates a coherence map that is a coherence distribution for each pixel in the pixel and pixels around each pixel, and when a virtual image exists in the processing target image, based on the coherence map And a virtual image position specifying unit for specifying the position of the virtual image existing in the processing target area .

According to the radar image processing device of the present invention, the coherence map calculation unit calculates each of the processing target images from the processing target image after the shift with the shift amount having the resolution as a basic unit and the processing target image before the shift. A coherence map, which is a distribution of coherence for pixels and surrounding pixels of each pixel, is calculated, and when a virtual image exists in the processing target image, the virtual image position specifying unit performs processing based on the coherence map. By specifying the position of the virtual image existing inside, it is possible to detect a virtual image of a type other than azimuth ambiguity.

1 is a block diagram illustrating a radar image processing apparatus according to Embodiment 1 of the present invention. It is explanatory drawing for demonstrating the image formation state of a radar image. 2 is a flowchart showing an operation of the radar image processing apparatus of FIG. 1. It is a block diagram which shows the radar image processing apparatus by Embodiment 2 of this invention. It is a flowchart which shows operation | movement of the radar image processing apparatus of FIG. It is a block diagram which shows the radar image processing apparatus by Embodiment 3 of this invention. It is a graph which shows an example of the function used for the analysis process of the image blur analysis part of FIG. It is a flowchart which shows operation | movement of the radar image processing apparatus of FIG. It is a block diagram which shows the radar image processing apparatus by Embodiment 4 of this invention. It is a graph which shows an example of the function used for the estimation process of the resolution estimation part of FIG. 10 is a flowchart showing the operation of the radar image processing apparatus of FIG.

Hereinafter, embodiments for carrying out the present invention will be described with reference to the drawings.
Embodiment 1 FIG.
FIG. 1 is a block diagram showing a radar image processing apparatus according to Embodiment 1 of the present invention.
In FIG. 1, a radar image processing apparatus 100 processes a radar image obtained by irradiating an observation target with radio waves from an antenna (not shown). The radar image processing apparatus 100 includes a radar image storage unit 101, a parameter storage unit 102, an image shift processing unit 103, a coherence calculation unit 104, a virtual image determination unit 105, and a determination result storage unit 106.

  In the radar image storage unit 101, a part or all of a radar image (SAR image: Synthetic Aperture Radar) subjected to reception processing by a reception processing device (not shown) in the previous stage is stored as a processing target image. Note that the processing target image stored in the radar image storage unit 101 may be data after the image reproduction processing of the synthetic aperture radar, or data during the image reproduction processing.

  The parameter storage unit 102 stores in advance a shift parameter that indicates the amount by which the processing target image is shifted, the shift direction of the processing target image, and the like. The shift amount of this shift parameter is basically one resolution or more. The image shift processing unit 103 shifts (displaces) the processing target image stored in the radar image storage unit 101 based on the shift amount of the shift parameter stored in the parameter storage unit 102. The coherence calculation unit 104 has a coherence (correlation) indicating a degree of correlation between the pre-shift processing target image stored in the radar image storage unit 101 and the post-shift processing target image shifted by the image shift processing unit 103. Number).

  Based on the coherence calculated by the coherence calculation unit 104, the virtual image determination unit 105 determines whether or not a blur region exists in the processing target image. Further, when the virtual image determination unit 105 determines that a blurred region exists in the processing target image, the virtual image determination unit 105 determines that a virtual image exists in the processing target image. The determination result storage unit 106 stores a determination result regarding the presence / absence of a blur region or the presence / absence of a virtual image by the virtual image determination unit 105.

  Next, the imaging state of the radar image will be described. The virtual image to be detected by the radar image processing apparatus 100 is an image (that is, ambiguity) obtained by projecting an object existing outside the observation target region assumed by the radar onto the observation target region, a broadcast wave, a communication wave, It is an image based on interference waves such as transmission waves and interference waves from other radar devices. The observation target region assumed by the radar is a region to be imaged by the radar, and is generally a region irradiated with the main beam of the antenna. Furthermore, the radar image has a feature that the information of the observation target at each point in the image falls within one resolution except when the image is blurred.

  FIG. 2 is an explanatory diagram for explaining the imaging state of the radar image. Points A and B in FIG. 2 are images formed without being blurred, and a region C in FIG. 2 is a blurred image, that is, a blurred region. Further, the radar image of FIG. 2 is represented by two-axis coordinates of an x axis (for example, a range direction: a distance direction) and a y axis (for example, an azimuth direction: an angle direction). Also, the width of the grid in FIG. 2 is the resolution of each of the x-axis and y-axis of the image.

  In FIG. 2, when coherence is calculated between points separated by 1 resolution or more from each other in the radar image, the coherence is almost 0 when the image is not blurred, and the coherence is 0 when the image is blurred. Is also a large value. In the radar image reproduction process, only the reflected wave from the observation target region is imaged as an image (that is, an image that is not blurred). For this reason, the ambiguity based on radio waves outside the observation region appears as a blurred image in the radar image.

  Further, the transmission wave of the radar is modulated with a fixed method and frequency, and when the image is reproduced, the received wave is demodulated and processed. On the other hand, the interference wave that is the basis of the virtual image is modulated with a method and frequency different from that of its own transmission wave. For this reason, since the interference wave cannot be demodulated by itself, the virtual image due to the interference wave has a characteristic that it is blurred in the radar image like the ambiguity. Note that, in the reception processing device in the preceding stage of the radar image processing device 100, even when the reception processing for the reflected wave in the observation region is incomplete, a blur region occurs in the radar image.

  Next, a coherence calculation method by the coherence calculation unit 104 and a virtual image or blur region detection method by the virtual image determination unit 105 will be described more specifically. When the pixel value of the coordinates (x, y) on the image is p (x, y) and the shift amounts in the x-axis and y-axis directions are Δx and Δy, respectively, the coherence calculation unit 104 calculates the following equation: The coherence γ (Δx, Δy) is calculated by executing the calculation using (1).

ただし、Ａはコヒーレンスの算出範囲（処理対象範囲）であり、処理対象画像の一部の範囲又は全範囲である。＊は複素共役である。

However, A is a coherence calculation range (processing target range), which is a partial range or the entire range of the processing target image. * Is a complex conjugate.

  Then, the virtual image determination unit 105 determines the presence / absence of a blurred region in the processing target image based on the coherence calculated by the coherence calculation unit 104. Specifically, when the virtual image determination unit 105 confirms that the coherence value is approximately 0, the virtual image determination unit 105 determines that the blurred region does not exist in the processing target image. On the other hand, if the virtual image determination unit 105 confirms that the coherence value is greater than 0, the virtual image determination unit 105 determines that a blur area exists in the processing target image, and determines that the blur area is a virtual image. That is, the virtual image determination unit 105 detects a blurred area and a virtual image by confirming that the coherence value is greater than zero.

  Therefore, the radar image processing apparatus 100 detects not only a virtual image due to azimuth ambiguity but also a virtual image due to range ambiguity and a virtual image due to interference waves by detecting a blur region in the radar image using coherence. Can be targeted.

  Here, the reference for determining that the blur region does not exist in the processing target image (determining that the coherence is approximately 0) may be determined by the radar image user based on experience or in advance. A set threshold value may be used. This threshold value may be set by a radar image user based on experience. Further, the coherence when a virtual image or a blurred region does not exist may be calculated by simulation, and the threshold value may be set using a statistic such as the maximum value, average value, or standard deviation of the coherence as an index.

  Next, the operation will be described. FIG. 3 is a flowchart showing the operation of the radar image processing apparatus 100 of FIG. First, in step S101, the image shift processing unit 103 shifts the processing target image according to the shift amount. Thereafter, in step S102, the coherence calculation unit 104 calculates the coherence between the processing target image before the shift and the processing target image after the shift.

  Subsequently, in step S <b> 103, the virtual image determination unit 105 determines the presence or absence of a virtual image or a blurred region in the processing target image from the coherence value calculated by the coherence calculation unit 104. In step S <b> 104, the virtual image determination unit 105 stores (saves) the determination result regarding the presence or absence of the virtual image or the blurred area in the processing target image in the determination result storage unit 106. Thereby, the operation of the radar image processing apparatus 100 ends.

  According to the radar image processing apparatus 100 of the first embodiment as described above, the coherence calculation unit 104 performs processing between the processing target image after the shift with the shift amount having the resolution as a basic unit and the processing target image before the shift. Calculate coherence. Then, when the virtual image determination unit 105 determines whether or not a blur region exists in the processing target image based on the coherence, and determines that the blur region exists in the processing target image, the virtual image in the processing target image Is determined to exist. With this configuration, when a virtual image exists in the processing target image, the pixel where the blurred image about the virtual image exists correlates with the surrounding pixels in resolution units, and therefore, other types than azimuth ambiguity A virtual image, that is, a range ambiguity, a virtual image due to an interference wave, or the like can also be a detection target.

  In addition, a determination result about the presence or absence of a virtual image or a blurred area in the processing target image stored in the determination result storage unit 106 is obtained from, for example, the processing target image in a processing device (not shown) at the subsequent stage of the radar image processing device 100. It can be used for the removal process of the virtual image or the blur area.

Here, the conventional virtual image detection method as shown in Non-Patent Document 1 has a problem that an observation target having a characteristic of returning a strong reflected wave from the side is erroneously detected as a virtual image.
On the other hand, in the radar image processing apparatus 100 according to the first embodiment, the virtual image determination unit 105 determines the presence / absence of a virtual image in the processing target image based on the determination result of the presence / absence of the blur region in the processing target image. . With this configuration, it is possible to avoid erroneous detection as a virtual image of an observation target having a characteristic of returning a strong reflected wave from the side.

Further, in the conventional virtual image detection method as shown in Non-Patent Document 1, images on the right side of the beam, the center of the beam, and the left side of the beam are respectively reproduced, or all radio waves received on the right side of the beam, the center of the beam, and the left side of the beam are used. After reproducing one image, it is necessary to process the reproduced one image into images on the right side, the center of the beam, and the left side of the beam.
In addition, in the conventional virtual image detection method as shown in Non-Patent Document 1, it is necessary to create images on the right side of the beam, the center of the beam, and the left side of the beam, and then superimpose them with an error of one pixel or less. For this reason, if only the images on the right side of the beam, the center of the beam, and the left side of the beam are created, a deviation of one pixel or more may occur due to, for example, the aircraft trajectory not being a straight line. A separate process was required.
On the other hand, the radar image processing apparatus 100 according to the first embodiment does not require image decomposition processing or superimposition processing, so that the processing load can be reduced as compared with the conventional virtual image detection method. The calculation time can be shortened.

Furthermore, in the method of detecting a virtual image by the visual observation of the radar image user, the virtual image may appear only on the image with a relatively weak signal strength. It was difficult to do.
On the other hand, since the radar image processing apparatus 100 according to the first embodiment detects a virtual image by signal processing, a virtual image having a relatively weak signal intensity that is difficult to visually observe can be detected.

  Instead of the coherence of the expression (1) in the first embodiment, a method for calculating the correlation between images, for example, a method for calculating a correlation coefficient after performing a correlation function or variable conversion may be used.

  Further, the coherence between the pixel at coordinates (x, y) and the pixel at coordinates (x + Δx, y + Δy) may be directly calculated without going through step S102 in FIG. 3 of the first embodiment. However, in order to detect a virtual image or a blurred region of an image, at least one of the shift amounts Δx and Δy needs to be a value equal to or higher than the resolution.

Embodiment 2. FIG.
In the radar image processing apparatus 100 according to the first embodiment, the presence or absence of a virtual image in the processing target image can be determined by using coherence before and after the entire processing target image is shifted. However, specifying the position of the virtual image or the blur region is not possible. could not. On the other hand, in the radar image processing apparatus 200 according to the second embodiment, instead of the coherence of the processing target image, the coherence map is a coherence of each pixel of the processing target image, that is, a local coherence distribution of the processing target image. And the position of the virtual image or blurred region can be specified based on the coherence map.

  FIG. 4 is a block diagram showing a radar image processing apparatus according to Embodiment 2 of the present invention. 4, the outline of the configuration of the radar image processing apparatus 200 according to the second embodiment is the same as the block configuration according to the first embodiment, but the coherence calculation unit 104, the virtual image determination unit 105, and the determination result according to the first embodiment. Instead of the storage unit 106, the coherence map calculation unit 204, the virtual image position specifying unit 205, and the virtual image distribution storage unit 206 of FIG.

  The coherence map calculation unit 204 calculates (generates) a coherence map between the processing target image stored in the radar image storage unit 101 and the processing target image shifted by the image shift processing unit 103. The virtual image position specifying unit 205 specifies the position of the virtual image based on the coherence map calculated by the coherence map calculating unit 204. The virtual image distribution storage unit 206 stores information on the position of the virtual image specified by the virtual image position specifying unit 205. Other configurations are the same as those in the first embodiment.

  Next, the calculation method of the coherence map by the coherence map calculation unit 204 will be described more specifically. The coherence map is a local coherence distribution map created based on the coherence with the surrounding pixels of each pixel in the processing target image. The coherence map calculation unit 204 uses the following equation () for local coherence γ (X, Y, Δx, Δy) when the coordinates of the calculation target pixel are (X, Y) and the shift amounts are Δx and Δy. It can be calculated by executing the calculation using 2).

ただし、Ａは座標（Ｘ，Ｙ）の周囲の画素の領域を示す。

However, A shows the area | region of the pixel around a coordinate (X, Y).

  Here, there are various methods for setting the pixels around the coordinates (X, Y). For example, there are eight neighboring regions that are one pixel at a time in the vertical and horizontal directions and the diagonal direction around the pixel to be calculated, A region including only a plurality of pixels in the vertical direction, a region including only a plurality of pixels in the horizontal direction of the calculation target pixel, a region set in a rectangular shape centering on the calculation target pixel, and the like can be considered. Further, the setting method of the pixels around the coordinates (X, Y) does not have to be uniform in the processing target image, and the setting method may be appropriately changed for each position in the processing target image.

  Next, a method for specifying the position of the virtual image or the blurred area by the virtual image position specifying unit 205 will be described more specifically. The virtual image position specifying unit 205 confirms the pixel value (local coherence value) of each pixel of the coherence map calculated by the coherence map calculation unit 204. Then, when the virtual image position specifying unit 205 confirms that the pixel value of each pixel of the coherence map is approximately 0, the virtual image position specifying unit 205 determines that no virtual image or blur area exists within the range of the coherence map.

  On the other hand, when the virtual image position specifying unit 205 confirms that the pixel value of each pixel of the coherence map is greater than 0, the virtual image position specifying unit 205 determines that a virtual image or a blur region exists within the range of the coherence map. Further, a threshold value for the pixel value of each pixel of the coherence map may be set in the virtual image position specifying unit 205 in advance. In this case, the virtual image position specifying unit 205 can specify an area where the pixel value of each pixel of the coherence map is a value equal to or larger than the threshold as a position where the virtual image or the blur area exists.

  This threshold value may be set by a radar image user based on experience. Further, the threshold may be set based on the appearance probability distribution (histogram) of the pixel values of the coherence map. In addition, if a virtual image or an area where there is no blurred area of the image is known as prior information, the maximum value, average value, or standard deviation of the pixel value of the coherence map of the known area is added to the statistics. A threshold may be set based on this.

  Next, the operation will be described. FIG. 5 is a flowchart showing the operation of the radar image processing apparatus 200 of FIG. In FIG. 5, in step S <b> 201, the image shift processing unit 103 shifts the processing target image according to the shift amount. Thereafter, in step S202, the coherence map calculation unit 204 calculates a coherence map between the processing target image before the shift and the processing target image after the shift.

  Subsequently, in step S203, the virtual image position specifying unit 205 specifies the position of the virtual image or the blurred area from the coherence value of each pixel in the coherence map. In step S204, the virtual image position specifying unit 205 stores the virtual image specified in step S203 or the virtual image distribution that is the position of the blurred area in the virtual image distribution storage unit 206. Thereby, the operation of the radar image processing apparatus 200 according to the second embodiment ends.

  According to the radar image processing apparatus 200 of the second embodiment as described above, the coherence map calculation unit 204 includes the processing target image after the shift with the shift amount having the resolution as a basic unit, the processing target image before the shift, From this, a coherence map is calculated. With this configuration, a virtual image of a type other than the azimuth ambiguity can be set as a detection target, and the position of the virtual image or the blurred region of the image existing in the processing target image can be specified.

  Further, the virtual image distribution information stored in the virtual image distribution storage unit 206 is, for example, a virtual image from the processing target image or the like in the processing apparatus (not shown) at the subsequent stage of the radar image processing apparatus 200, as in the first embodiment. It can be used for blur region removal processing or the like. Furthermore, it can also be used to display the position of a virtual image or blurred area. In addition, the type of virtual image can be estimated from the virtual image distribution information. For example, if there is a linear virtual image along the axis in the range direction (distance direction) or azimuth direction (angular direction) of the radar image, it can be estimated that the virtual image is caused by an interference wave.

  Furthermore, when this coherence map is displayed on a display device, a virtual image or a blur region existing in the processing target image can be specified by visual observation of a radar image user.

  Note that, instead of the coherence of the expression (2) in the second embodiment, a method for calculating correlation between images, for example, a correlation coefficient or a correlation coefficient after variable conversion is calculated, as in the first embodiment. A method or the like may be used.

  Further, the coherence map between the pixel at coordinates (x, y) and the pixel at coordinates (x + Δx, y + Δy) may be directly calculated without going through step S202 in FIG. 5 of the second embodiment. However, in order to detect a virtual image or a blurred region of an image, at least one of the shift amounts Δx and Δy needs to be a value equal to or higher than the resolution.

  Furthermore, the coherence map calculation unit 204 in the second embodiment may be added to the radar image processing apparatus 100 in the first embodiment. In this case, a threshold value is set based on the appearance probability distribution (histogram) of the coherence in the coherence map calculated by the coherence map calculation unit 204, and the virtual image determination unit 105 uses the set threshold value in the processing target image. It may be determined whether or not a virtual image exists.

Embodiment 3 FIG.
The radar image processing apparatus 100 according to the first embodiment determines the presence or absence of a virtual image or blur area in the processing target image, and the radar image processing apparatus 200 according to the second embodiment has a virtual image or blur area in the processing target image. Identified the area that is. On the other hand, the radar image processing apparatus 300 according to the third embodiment analyzes the spreading direction and size (blurring degree) of a virtual image or a blur area existing in the processing target image.

  FIG. 6 is a block diagram showing a radar image processing apparatus according to Embodiment 3 of the present invention. In FIG. 6, a radar image processing apparatus 300 according to the third embodiment includes a radar image storage unit 301, a parameter storage unit 302, an image shift processing unit 303, a coherence calculation unit 304, a shift amount versus correlation correspondence unit 305, and an image blur analysis unit. 306 and an analysis result storage unit 307. The configurations of the radar image storage unit 301, the parameter storage unit 302, the image shift processing unit 303, and the coherence calculation unit 304 are the same as the configurations of the functional blocks 101 to 104 in the first embodiment.

  The shift amount versus correlation correspondence unit 305 receives from the coherence calculation unit 304 and the parameter storage unit 302 the coherence value calculated by the coherence calculation unit 304 and the shift parameter used to calculate the coherence. Also, the shift amount / correlation correspondence unit 305 stores the coherence calculated by the coherence calculation unit 304 and the shift amount used in calculating the coherence in association with each other and stored as shift amount / correlation information.

  Furthermore, the shift amount-correlation correspondence unit 305 stores the shift amount-correlation information, and then stores the shift parameter stored in the parameter storage unit 302 in a shift parameter for which coherence has not yet been calculated (that is, the shift amount and the shift direction). To change the shift amount and the shift direction for the processing target image. The range in which the shift amount is changed is a preset range, and is a range in which the resolution of each of the x-axis and the y-axis is the minimum unit. The shift amount correlation correlation unit 305 causes the image shift processing unit 303 and the coherence calculation unit 304 to repeatedly execute the respective processes after changing the shift parameter.

  The image blur analysis unit 306 receives shift amount versus correlation information from the shift amount versus correlation correspondence unit 305. In addition, the image blur analysis unit 306 analyzes the spread direction and the degree of blur of a virtual image or a blur region existing in the processing target image based on the shift amount versus correlation information. The analysis result storage unit 307 stores the analysis result by the image blur analysis unit 306.

  Next, the analysis method performed by the image blur analysis unit 306 will be described more specifically. The image blur analysis unit 306 includes, for example, a one-dimensional function as illustrated in FIG. 7, a two-dimensional function having two shift axes, a function created by variable conversion of the shift amount or coherence, or the shift amount Using the statistic calculated from the coherence calculation result or the like, the direction in which the virtual image or the blur region exists in the processing target image and the degree of blur are analyzed. Here, a description will be given using the example of FIG.

  When analyzing the spreading direction of the blur region, the image blur analysis unit 306 calculates the coherence function of FIG. 7 for the shift amount Δx of the axis x of the image and the coherence of the shift amount Δy of the axis y. Create the result of calculating the function. However, the basic unit of the shift amount at this time is the resolution of the image in each axis direction (for example, the horizontal axis of the graph is marked as twice the resolution). The image blur analysis unit 306 can determine that the blur region is widened in the y-axis direction when it is confirmed that Δy has a lower degree of reduction in coherence than Δx.

  The image blur analysis unit 306 uses a threshold as shown in FIG. 7 when analyzing the size of the blur region. In other words, the image blur analysis unit 306 can analyze the shift amount when the coherence is below the threshold as the size of the blur region. This threshold value may be set based on the experience of the process operator. Alternatively, a region where a virtual image does not exist may be obtained from the virtual image position specifying result of the second embodiment, and may be set based on a coherence average value or a statistic such as variance.

  Next, the operation will be described. FIG. 8 is a flowchart showing the operation of the radar image processing apparatus 300 of FIG. In FIG. 8, in step S301, the image shift processing unit 303 shifts the processing target image according to the shift amount. Thereafter, in step S302, the coherence calculation unit 304 calculates the coherence between the processing target image before the shift and the processing target image after the shift.

  Subsequently, in step S303, the shift amount correlation correlation unit 305 associates the coherence calculated by the coherence calculation unit 304 with the shift amount used in the calculation of the coherence, and shift amount pair correlation information. Save as. In step S304, the shift amount correlation correlation unit 305 updates the shift amount and the shift direction. Here, the arithmetic processing of steps S301 to S304 is repeatedly executed for each shift amount and shift direction within a predetermined range.

  After the calculation process of steps S301 to S304 is completed, the process proceeds to the calculation process of step S305. In step S305, the image blur analysis unit 306 analyzes the spread direction and size of the blur region based on the shift amount versus correlation information. In step S306, the image blur analysis unit 306 stores information on the analysis result in step S305 in the analysis result storage unit 307. Thereby, the operation of the radar image processing apparatus 300 according to the third embodiment is completed.

  According to the radar image processing apparatus 300 of the third embodiment as described above, the image blur analysis unit 306 exists in the processing target image based on the shift amount versus correlation information from the shift amount versus correlation correspondence unit 305. Analyze the direction and size of the blur area. With this configuration, it is possible to classify (categorize) virtual images using information on the analysis result of the image blur analysis unit 306. For example, when a blur area that is uniform in the range direction or the azimuth direction exists in the processing target image, it can be estimated that the blur area is a virtual image generated by the interference wave.

  The analysis result information stored in the analysis result storage unit 307 is, for example, processed in a processing device (not shown) subsequent to the radar image processing device 300 to remove a virtual image or a blur region from the processing target image, It can be used for correction processing and the like. Further, out of the analysis result information stored in the analysis result storage unit 307, the information on the size of the blur region is obtained by the autofocus processing device (not shown) at the subsequent stage of the radar image processing device 300. Can be used for the radar auto-focus processing with the operation parameter as.

Embodiment 4 FIG.
In the first to third embodiments, the processing condition is that the resolution of the radar image is known information. On the other hand, in the fourth embodiment, when the resolution of the radar image is unknown information, the resolution estimation unit 406 estimates the resolution.

  FIG. 9 is a block diagram showing a radar image processing apparatus according to Embodiment 4 of the present invention. In FIG. 9, the configuration of the radar image processing apparatus 400 according to the fourth embodiment is such that the basic unit of the shift amount is a distance, and the image blur analysis unit 306 and the analysis result storage unit 307 of the radar image processing apparatus 300 are replaced. Except for having a resolution estimation unit 406 and a resolution estimation value storage unit 407, the configuration is the same as that of the radar image processing apparatus 300 of the third embodiment. That is, in the radar image processing apparatus 400 of the fourth embodiment, the functional blocks 301 to 305 of the radar image processing apparatus 300 of the third embodiment are function blocks for resolution estimation.

  The resolution estimation unit 406 receives the shift amount versus correlation information from the shift amount versus correlation correspondence unit 305. Further, the resolution estimation unit 406 estimates the resolution of the radar image based on the shift amount versus correlation information. Specifically, the resolution estimation unit 406 creates a graph representing the relationship between the shift amount and coherence as shown in FIG. The difference between the graph of FIG. 10 and the graph of FIG. 7 in the third embodiment is that the unit of the shift amount is not a resolution but a distance (for example, a meter unit).

  The resolution estimation unit 406 can estimate the shift amount when the coherence falls below a preset threshold as the resolution. This threshold setting method is the same as in the third embodiment. The estimated resolution value storage unit 407 stores the estimated resolution value by the resolution estimation unit 406. FIG. 11 is a flowchart showing the operation of the radar image processing apparatus 400 of FIG. 9. The outline of the operation of the radar image processing apparatus 400 is the same as the operation of the radar image processing apparatus 300 of the third embodiment. .

  According to the radar image processing apparatus 400 of the fourth embodiment as described above, the resolution estimation unit 406 estimates the resolution of the radar image based on the shift amount versus correlation information from the shift amount versus correlation correspondence unit 305. With this configuration, when the resolution of the radar image is unknown information, the resolution can be estimated from the processing target image.

  The configuration of the radar image processing apparatus 400 according to the fourth embodiment may be combined with the configuration of each of the radar image processing apparatuses 100 to 300 according to the first to third embodiments. In such a combined configuration, even when the resolution is unknown information, the radar image processing apparatuses 100 to 300 use the shift amount having the resolution calculated by the radar image processing apparatus 400 as a basic unit, Detection, position specification, or analysis of a blurred area can be performed.

  100 to 400 Radar image processing apparatus, 101, 301 Radar image storage unit, 102, 302 Parameter storage unit, 103, 303 Image shift processing unit, 104, 304 Coherence calculation unit, 105 Virtual image determination unit, 106 Determination result storage unit, 204 Coherence map calculation unit, 205 virtual image position specifying unit, 206 virtual image distribution storage unit, 305 shift amount correlation correlation unit, 306 image blur analysis unit, 307 analysis result storage unit, 406 resolution estimation unit, 407 resolution estimation value storage unit.

Claims (6)

At least one of two axes for representing the radar image based on a preset shift amount having a resolution as a basic unit for a processing target image that is one of all or part of the acquired radar image A shift processing unit for shifting in the axial direction of
Coherence that is a coherence distribution of each pixel in the processing target image and pixels around each pixel from the processing target image after the shift by the shift processing unit and the processing target image before the shift. A coherence map calculation unit for calculating a map;
A radar image processing apparatus comprising: a virtual image position specifying unit that specifies a position of a virtual image existing in the processing target area based on the coherence map when a virtual image exists in the processing target image. . The virtual image position specifying unit, the pixel in which the pixel values of the coherence map exceeds a predetermined threshold value, determines that the virtual image, according to claim 1, characterized in that locating the virtual image of the processing target area The radar image processing apparatus described. The virtual image position specifying unit determines a threshold based on the appearance probability distribution of the coherence in the coherence map calculated by the coherence map calculation unit, and uses the determined threshold instead of the predetermined threshold in the coherence map. The radar image processing apparatus according to claim 2 , wherein a position of a virtual image in the processing target area is specified. For a processing target image that is a part or all of the acquired radar image, at least one of two axes for representing the radar image based on a preset shift amount with resolution as a basic unit A shift processing unit for shifting in the axial direction;
A coherence calculation unit that calculates coherence between the processing target image after being shifted by the shift processing unit and the processing target image before the shift;
The coherence calculated by the coherence calculation unit and the shift amount used in the calculation of the coherence are associated with each other as shift amount-correlation information, and the shift amount and the shift direction for the processing target image are set. A shift amount-correlation correspondence unit that causes the shift processing unit and the coherence calculation unit to repeatedly execute the respective processes,
An image blur analysis unit that analyzes a spread direction and a size of a blur region existing in the processing target image based on the shift amount pair correlation information from the shift amount pair correlation correspondence unit. Radar image processing device. The image blur analysis unit calculates a function indicating a relationship between a shift amount and coherence in the shift amount-correlation information for each axial direction of the two axes, and based on the calculated function set, the processing target image The radar image processing apparatus according to claim 4, wherein a direction and a size of a blur area existing in the area is analyzed. The shift processing unit is a shift processing unit for image processing,
A shift processing unit for resolution estimation that shifts in the axial direction of at least one of the two axes based on a shift amount set in advance with the distance as a basic unit for the processing target image;
A resolution estimation coherence calculation unit that calculates coherence between the processing target image after the shift by the resolution estimation shift processing unit and the processing target image before the shift;
The coherence calculated by the coherence calculating unit for resolution estimation and the shift amount used in the calculation of the coherence are associated with each other as shift amount-correlation information for resolution estimation, and the processing target image Changing the shift amount and the shift direction for the resolution estimation, the resolution estimation shift processing unit, and the resolution estimation coherence calculation unit repeatedly executing the respective processing, a resolution estimation shift amount versus correlation correspondence unit,
A resolution estimation unit that estimates the resolution of the radar image based on the resolution estimation shift amount pair correlation information from the resolution estimation shift amount pair correlation correspondence unit;
The shift processing unit for image processing shifts the processing target image in at least one of the two axes based on a shift amount having the resolution estimated by the resolution estimation unit as a basic unit. The radar image processing apparatus according to any one of claims 1 to 5, wherein

Images