JP2016057164A - Rader system, and target classification device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a radar system and a target classification device for extracting and classifying characteristics while specifying a whole shape of a target without integrating an image with a time axis even when the acquired ISAR image is not clear.SOLUTION: One ISAR image including a target ship is split into a plurality of regions, and a histogram of luminance gradient is prepared for each region. With a deviation of luminance gradient focused in these histograms, the region in which a bow side or a stern of the target ship is included is specified, and the entire shape is grasped. Characteristic information relating to the arrangement, the position and others of various kinds of structures provided in the ship is extracted based on a specified result of both end parts of these ships and collated with characteristic information for classification.SELECTED DRAWING: Figure 1

Description

  Embodiments described herein relate generally to a radar apparatus and a target classification apparatus.

  2. Description of the Related Art Conventionally, radar devices that classify targets such as ships using images acquired by an inverse synthetic aperture radar (ISAR) are known. In this type of apparatus, the target features are extracted from the acquired ISAR images, and the target features are classified by comparing them with features of a ship or the like that are stored in a database in advance. Examples of the target characteristics include the overall shape of the ship, various installed structures, and their installation positions. In the process of extracting these features from the ISAR image, both ends of the target, that is, the bow and stern are detected from the image, the overall image of the target is specified, and the position where each structure is installed is Identify in the image.

  Since the ISAR image is generated by synthesizing the radar echoes from the target, a clear image of the entire ship is not always acquired depending on the propagation state of the radar signal, and the bow or stern is low in brightness. It may be unclear images that only appear. For this reason, various methods for extracting a target feature even from a blurred ISAR image are disclosed.

  An ISAR image can be regarded as a two-dimensional image in which pixels each having a luminance level are two-dimensionally arranged. Then, paying attention to the luminance level of the two-dimensional image, it is possible to extract necessary features according to the above-mentioned target categories. At this time, if the target ISAR image is unclear, for example, there is a technique of performing extraction processing after integrating the ISAR image with time in order to compensate for the unclearness.

JP 2011-247777 A

Maki, A., "Automatic Ship Identification in ISAR imagery", Radar Conference, 2002.Proceedings of the IEEE

  However, when the ISAR image is integrated over time, a plurality of images are superimposed according to the integration time, so that the appearance blur is expected to be improved, but the overall shape is blurred. End up. For this reason, the position of a structure or the like is not accurately specified, and an error in extracting a target feature increases, resulting in a problem that target classification accuracy decreases. For this reason, it is desired to extract the target features while reliably detecting the target bow and stern using one image and specifying the overall shape without integrating the ISAR image. It was.

  The embodiment of the present invention has been made in consideration of the above-described circumstances, and is characterized by specifying the target overall shape without integrating the image on the time axis even when the acquired ISAR image is unclear. An object of the present invention is to provide a radar apparatus and target classification apparatus that extract and categorize the signals.

  In order to achieve the above object, the radar apparatus according to the present embodiment generates a plurality of characteristic information for classifying ships in advance in a radar apparatus that generates an ISAR (Inverse Synthetic Aperture Radar) image based on radar echoes. A storage unit for storing, an image selection unit for selecting one image including a ship to be classified from the generated ISAR image, and the selected ISAR image is divided into a plurality of regions of a predetermined size. In each area, the brightness levels of all the pixels in the area are compared with adjacent pixels in units of pixels, the direction of the brightness gradient vector calculated based on the difference in brightness levels is determined, and each brightness gradient vector direction is determined. Histograms created for each of these areas are created. A position specifying unit for specifying whether the region is a region including the bow or stern of the ship based on the deviation of the frequency value for each direction of the brightness gradient vector in the system, and a result of specifying the bow or stern region And extracting the feature information of the ship, comparing the extracted feature information with the stored contents of the storage unit, and classifying the ship, and displaying the classification result by the classification processor And a display unit.

  The target classification apparatus according to the present embodiment is a target classification apparatus used in a radar apparatus that generates an ISAR image based on radar echoes, and a storage unit that stores a plurality of pieces of feature information for classifying ships in advance. An image selection unit that selects one image including a ship to be classified from the generated ISAR image, and the selected ISAR image is divided into a plurality of regions of a predetermined size, and in each region , The brightness level of all pixels in the region is compared with neighboring pixels in pixel units, the direction of the brightness gradient vector calculated based on the difference in brightness level is determined, and the frequency determined for each direction of the brightness gradient vector And a histogram is created based on the frequency value bias for each direction of the luminance gradient vector in the histogram created for each region. And extracting the feature information of the ship based on the position specifying unit that specifies whether the area includes the bow or stern of the ship and the result of specifying the bow or stern area. And a classification processing unit that classifies the ship by comparing the characteristic information with the contents stored in the storage unit.

The block diagram which shows the structure of one Example of the radar apparatus which concerns on this embodiment. The 1st explanatory view for explaining the processing performed in each part in position specific part 62, and the result. The 2nd explanatory view for explaining the processing performed in each part in position specific part 62, and the result. The 3rd explanatory drawing for demonstrating the process performed by each part in the position specific | specification part 62, and its result. 2 is a flowchart for explaining the operation of the radar apparatus illustrated in FIG. 1.

  Hereinafter, the best mode for carrying out the radar device and the target classification device according to the present embodiment will be described with reference to FIGS.

  FIG. 1 is a block diagram showing a configuration of an example of a radar apparatus according to the present embodiment. As illustrated in FIG. 1, the radar apparatus 1 includes an antenna 10, a circulator 20, a transmitter 30, a receiver 40, an ISAR image generation unit 50, a target classification unit 60, a storage unit 70, a display processing unit 80, and a display. 90 and the operation unit 100.

  The antenna 10 radiates the radar signal from the transmitter 30, receives the reflected wave, and sends it to the receiver 40. The circulator 20 is a transmission line having directionality, and transmits the radar signal from the transmitter 30 to the antenna 10 and transmits the reflected wave of the radar signal received by the antenna 10 to the receiver 40. The transmitter 30 generates a radar signal having a predetermined period and sends it to the antenna 10.

  The receiver 40 performs reception processing such as low noise amplification, frequency conversion, and filtering on the reflected wave of the radar signal received by the antenna 10, and sends it to the ISAR image generation unit 50. The ISAR image generation unit 50 generates an ISAR image by performing synthesis processing or the like on the signal after reception processing.

  The target classification unit 60 selects one image including the classification target ship from the generated ISAR image, specifies the position of the bow or stern in the image, and based on the identification result, the target classification of the target ship. Extract feature information. And the target ship is classified by collating the extracted feature information and the memory content of the memory | storage part 70, and the result is output. Details of the target classification unit 60 will be described in detail later.

  A plurality of pieces of characteristic information for classifying various ships are stored in the storage unit 70 in advance, and are referred to when the target ships are classified by the target classification unit 60. The ship characteristic information includes, for example, characteristics related to the shape of the ship, such as the total length of the ship and position information of various structures such as a bridge and a mast.

  The display processing unit 80 edits display data for displaying the classification result in the target classification unit 60, displays an ISAR image generated by the ISAR image generation unit 50, and performs an input operation from the operation unit 100 described later. Edit the display data for display etc. associated with etc. The display 90 displays the edited display data on the screen. The operation unit 100 accepts an input operation by an operator including an ISAR image selection operation in the target classification unit 60, and transfers it to each unit in the apparatus. The target classification device can be configured using the target classification unit 60, the storage unit 70, and the operation unit 100 described above.

  Next, the target classification unit 60 will be described in detail. The target classification unit 60 includes an image selection unit 61, a position specifying unit 62, and a classification processing unit 63. The image selection unit 61 selects one ISAR image to be used for each target category from the images generated by the ISAR image generation unit 50. In this embodiment, an image can be selected by a selection operation by an operator or the like from the operation unit 100. The position specifying unit 62 specifies the bow or stern of the classification target ship in the selected ISAR image.

  Here, the position specifying unit 62 includes an image dividing unit 621, a histogram generating unit 622, and an evaluating unit 623. The image dividing unit 621 divides the ISAR image selected by the image selecting unit 61 into a plurality of regions having a predetermined size set in advance. The histogram generation unit 622 compares the luminance levels of all the pixels in each divided area with adjacent pixels in units of pixels, determines the direction of the luminance gradient vector calculated based on the difference in luminance level, and The frequency determined for each time is counted to create a histogram. The evaluation unit 623 evaluates the deviation of the frequency values for each direction in the histogram created for each divided area, and specifies the areas including the bow and stern.

  Each part of the position specifying part 62 will be further described in detail with reference to FIGS. FIGS. 2 to 4 are explanatory diagrams illustrating by modeling the processing performed by each unit in the position specifying unit 62 and the result of the selected ISAR image. FIG. 2A is an example of an ISAR image selected by the image selection unit 61. This image can be regarded as a two-dimensional image in which single pixels each having a luminance level are two-dimensionally arranged. FIG. 2B shows an example in which this ISAR image is divided into a plurality of areas having a size set in advance by the image dividing unit 621. In this example, an image is divided into four regions A to D evenly by a vertical dividing line.

  In the histogram generation unit 622, the luminance levels of all the pixels are compared with adjacent pixels for each divided region. In this embodiment, as illustrated in FIG. 3A, each pixel (circle in the figure) is compared with pixels adjacent to each other in four directions, up, down, left, and right on the paper surface, and the luminance level difference is The direction of the largest adjacent pixel (one of the top, bottom, left, and right directions indicated by a triangle in the figure) is determined. For example, when it is determined that the direction is upward, the frequency of the upward direction is counted by one count. The direction frequently counted by these determinations corresponds to the direction of the luminance gradient in the image. In the histogram generation unit 622, the above-described determination and counting are performed for all the pixels in each region, and a histogram of direction vs. frequency of adjacent pixels is created for each region. FIG. 3B shows an example of a histogram created for the region D among the histograms created in this way.

  FIG. 4A shows a direction (one of four directions, top, bottom, left, and right) that is frequently determined when the above-described determination is performed in each region illustrated in FIG. Only the position is modeled on the ISAR image. In a ship, in particular, the hull itself has many linear structures, but such a linear part is detected as a strong luminance gradient in the vertical direction on the paper surface in the case of the image as shown in FIG. In particular, this tendency becomes more prominent as shown in FIG. 4A in the vicinity of the ends of the bow, stern and the like where the number of conspicuous structures and the like is smaller than that of other parts. For this reason, in the histogram of the region corresponding to the end portion or the region including the end portion, a frequency peak occurs in the direction of the luminance gradient vector (in this example, the vertical direction).

  On the other hand, in the histogram of the region where many structures and the like are arranged, the frequency peak does not appear remarkably, and in the region where the target is not included, it is considered that there is basically no frequency peak. And the way the peaks appear in the histogram at this time is that the total number of pixels in the region is a constant number, so if the frequency in either direction increases relative to the average value of the frequency, the other direction decreases. And the frequency is biased.

  In the histogram created for each region, the evaluation unit 623 evaluates a significant frequency deviation due to such a peak by comparing with a threshold value, and extracts a region where an edge exists. . FIG. 4B illustrates a scene where evaluation is performed by applying a threshold value to the histogram of the region D illustrated in FIG. In this embodiment, the threshold value at this time is the value obtained by dividing the number of executions of luminance comparison by the number of directions of adjacent pixels (the number of directions in FIG. A predetermined constant k is added based on the average frequency value for, and if the frequency exceeds this value, it is determined that there is a significant bias. In the case of FIG. 4B, it is determined that there is a significant bias in the vertical direction, and it is determined that the region includes the end of the ship. Thereby, for example, the entire shape of the ship can be grasped in one ISAR image without using an integrated ISAR image or the like.

  In the description of the examples in FIG. 2 to FIG. 4, the area division is four areas by the vertical dividing lines. The part can be specified. In addition, although the direction of the luminance gradient is also four directions, up, down, left, and right, by adding a direction other than these (for example, an oblique direction) to the direction of the pixel to be compared, the luminance gradient can be biased more precisely. It can also be determined.

  The classification processing unit 63 further extracts feature information of the target ship based on the bow or stern identification result in the ISAR image identified by the position identifying unit 62. The feature information includes, for example, positions of various structures installed on the ship. Then, for example, the contents of the storage unit 70 are collated by a method such as pattern matching to classify the ships in the ISAR image, and the result is sent to the display processing unit 80.

  Next, the operation of the radar apparatus 1 according to the present embodiment configured as described above will be described with reference to the above-described flowcharts of FIGS. FIG. 5 is a flowchart for explaining the operation of the radar apparatus 1.

  First, a radar wave emitted from the transmitter 30 is radiated from the antenna 10, and the reflected wave is received by the antenna 10, subjected to reception processing by the receiver 40, and sent to the ISAR image generation unit 50. The ISAR image generation unit 50 further performs synthesis processing and the like on the signal after reception processing, and sequentially generates ISAR images. The generated images are sequentially sent to the target classification unit 60 and the display processing unit 80 (ST51).

  Next, in the image selection unit 61 in the target classification unit 60, one ISAR image including a target ship to be classified is selected. In this embodiment, an image is selected from among the sequentially generated images by a selection operation from the operation unit 100 by an operator or the like. An example of the selected ISAR image is shown in FIG. This ISAR image is sent to the position specifying unit 62 (ST52).

  Next, in the image dividing unit 621 in the position specifying unit 62, the selected one ISAR image is divided into a plurality of regions having a predetermined size. An example of the division is shown in FIG. FIG. 2B shows an example in which the image of FIG. 2A is divided into four regions A to D by vertical dividing lines (ST53).

  Next, in the histogram generation unit 622, a histogram of luminance gradient is created for each of the divided areas. That is, the luminance levels of all the pixels in each region are compared with the adjacent pixels on a pixel basis as shown in the explanatory diagram of FIG. 3A, and the direction of the adjacent pixel having the largest difference in luminance level is determined. Then, the frequency determined for each direction is counted, and a histogram is created. FIG. 3B is an example of a histogram created for the region D in FIG. 2B in this way (ST54).

  Next, in the evaluation unit 623, the generated histogram is evaluated, and a region including the bow and stern is specified. That is, in the region including the end of the hull having many linear structures such as the bow and stern in the image, as described above, the frequency gradient is biased in the histogram of the luminance gradient. Whether or not there is a significant frequency bias in the histogram of the brightness gradient of each region created in the operation step is determined by comparison with a threshold set based on the average frequency (ST55). .

  As a result of the comparison, if it is determined that there is a bias (Y in ST55), the region corresponding to the histogram is extracted as a region including the end of the ship. This makes it possible to grasp the overall shape of the target ship in one ISAR image. Then, the result is sent to the classification processing unit 63 (ST56). On the other hand, if there is no significant frequency bias in any of the histograms created for each region (N in ST55), for example, review of image division in the operation step of ST53, reselection of an ISAR image, and the like. After being made, the above operational steps are retried.

  Next, the classification processing unit 63 classifies the ships in the ISAR image. That is, in the classification processing unit 63, based on the identification result of the bow or stern in the ISAR image received from the position specifying unit 62, various features related to the target ship such as the positions of various structures installed on the ship. Information is acquired. Then, the acquired feature information is collated with the feature information for classifying various ships stored in the storage unit 70, and the ships in the ISAR image are classified. The classification result is sent to the display processing unit 80 (ST57).

  Next, display screen data for displaying the classification result is edited in the display processing unit 80. The display 90 displays, for example, an ISAR image, data associated with the operation of the operation unit 100, and the display processing unit 80 edits display screen data for performing various displays including classification results. Is done. The edited display screen data is sent to the display 90, and various information including the vessel classification result is displayed on the display 90 (ST58). Thereafter, the above operation steps are repeated until the operation of the apparatus is instructed (ST59).

  As described above, in this embodiment, one ISAR image including the target ship is divided into a plurality of regions, and a histogram of the brightness gradient is created for each region, and the brightness gradient in these histograms is also created. The area including the bow or stern of the target ship is specified by paying attention to the bias. Thus, for example, even if the ISAR image is unclear, both ends of the target ship can be obtained from one selected ISAR image without performing integration processing of the ISAR image that leads to an increase in ambiguity in the image. Can be identified and the overall shape can be grasped.

  In addition, since it is possible to more accurately identify the arrangement and position of various structures provided on the ship based on the identification results of both ends of the ship, the error in extracting the feature information is reduced. It is possible to improve accuracy when classifying targets by collating feature information.

  Therefore, even when the ISAR image is unclear, it is possible to obtain a radar device and a target classification device that ensure a good classification accuracy while specifying the entire target image from one image.

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

DESCRIPTION OF SYMBOLS 1 Radar apparatus 10 Antenna 20 Circulator 30 Transmitter 40 Receiver 50 ISAR image generation part 60 Target classification part 61 Image selection part 62 Position specification part 63 Classification processing part 70 Storage part 80 Display processing part 90 Display unit 100 Operation part 621 Image division Unit 622 histogram generation unit 623 evaluation unit

Claims (8)

In a radar apparatus that generates an ISAR (Inverse Synthetic Aperture Radar) image based on a radar echo,
A storage unit for storing a plurality of feature information for classifying ships in advance for each ship;
An image selection unit for selecting one image including a ship to be classified from the generated ISAR image;
The selected ISAR image is divided into a plurality of regions of a predetermined size, and in each region, the luminance levels of all the pixels in the region are compared with adjacent pixels in units of pixels, and based on the luminance level difference The direction of the calculated brightness gradient vector is determined, the frequency determined for each direction of the brightness gradient vector is counted to create a histogram, and the frequency for each direction of the brightness gradient vector in the histogram generated for each region A position identifying unit that identifies whether the region is a region including the bow or stern of the ship, based on the value bias;
Based on the identification result of the bow or stern area, the feature information of the ship is extracted, and a classification processing unit that classifies the ships by collating the extracted feature information with the storage content of the storage unit;
A radar apparatus comprising: a display unit for displaying a classification result by the classification processing unit.   The bias of the frequency value of the histogram created for each area divided by the position specifying unit is the presence or absence of a frequency value exceeding a preset threshold value for each histogram, and if there is a corresponding frequency value, The radar apparatus according to claim 1, wherein the determination is made based on the direction corresponding to a frequency value.   The preset threshold value for determining the bias of the frequency value is a value obtained by dividing the number of brightness comparisons with adjacent pixels performed in the divided area by the number of directions of adjacent pixels in the histogram. The radar apparatus according to claim 2, wherein the radar apparatus is set based on the setting.   The image processing apparatus further includes an operation unit that receives an operation for selecting one image from the generated ISAR image, and the image selection unit selects an image based on the operation received by the operation unit. The radar apparatus according to any one of claims 1 to 3. In a target classification apparatus used in a radar apparatus that generates an ISAR image based on a radar echo,
A storage unit for storing a plurality of feature information for classifying ships in advance for each ship;
An image selection unit for selecting one image including a ship to be classified from the generated ISAR image;
The selected ISAR image is divided into a plurality of regions of a predetermined size, and in each region, the luminance levels of all the pixels in the region are compared with adjacent pixels in units of pixels, and based on the luminance level difference The direction of the calculated brightness gradient vector is determined, the frequency determined for each direction of the brightness gradient vector is counted to create a histogram, and the frequency for each direction of the brightness gradient vector in the histogram generated for each region A position identifying unit that identifies whether the region is a region including the bow or stern of the ship, based on the value bias;
A classification processing unit that extracts feature information of the ship based on a result of specifying the bow or stern area, and classifies the ship by comparing the extracted feature information with the storage content of the storage unit; Target classification device characterized by that.   The bias of the frequency value of the histogram created for each area divided by the position specifying unit is the presence or absence of a frequency value exceeding a preset threshold value for each histogram, and if there is a corresponding frequency value, 6. The target classification apparatus according to claim 5, wherein the determination is made based on the direction corresponding to a frequency value.   The preset threshold value for determining the bias of the frequency value is a value obtained by dividing the number of brightness comparisons with adjacent pixels performed in the divided area by the number of directions of adjacent pixels in the histogram. 7. The target classification apparatus according to claim 6, wherein the target classification apparatus is set based on the target classification.   The image processing apparatus further includes an operation unit that receives an operation for selecting one image from the generated ISAR image, and the image selection unit selects an image based on the operation received by the operation unit. The target classification apparatus according to any one of claims 5 to 7.