JP6266882B2 - Tracking support device - Google Patents

Description

  The present invention relates to a tracking support apparatus that stably maintains a target to be tracked in a primary radar having a tracking function.

Some marine radars have a function of tracking a target of another ship or the like in order to ensure navigation safety.
FIG. 4 is a diagram illustrating a configuration example of a radar apparatus having a target tracking function.

  In the figure, the output of the transmission unit 41 is connected to the first opening of the circulator 42, and the second opening of the circulator is connected to the feeding point of the antenna system 44 via the feeding unit 43. The third opening of the circulator 42 is connected to the input of the instruction unit 47 via the reception unit 45 and the signal processing unit 46. The corresponding input / output ports of the control unit 48 are connected to the control terminals of the transmission unit 41, the reception unit 45, the signal processing unit 46 and the instruction unit 47.

  In the radar apparatus having such a configuration, the transmission unit 41 generates a transmission wave that is a pulse wave under the control of the control unit 48. Such a transmission wave is radiated from the aerial system 44 via the circulator 42 and the power feeding unit 43 toward the sea area where another ship can be located.

The reflected wave that has arrived at the aerial system 44 from a target such as another ship that has reflected such a transmitted wave is delivered to the receiving unit 45 via the power feeding unit 43 and the circulator 42.
The receiving unit 45 generates a baseband signal indicating the A scope by demodulating the reflected wave.

  The signal processing unit 46 generates an instruction image by mapping the A scope indicated by the baseband signal in synchronization with the sweep and scan performed by the antenna system 44. Note that such an instruction image appropriately includes the following images.

(1) An image showing the track of the target of another ship, etc., which is the source of the reflected wave (hereinafter referred to as “track image”)
(2) A symbol image that is generated under target tracking, which will be described later, and shows the target that is the target of tracking as visual information (meaning the risk of collision, etc.)

  Here, the target tracking is realized as a process of identifying the azimuth, the distance, the course, and the like for each target from the radar image obtained at a predetermined cycle and predicting the speed and the traveling direction based on these. In the course of such processing, the tracking performance is maintained by performing processing for reducing the possibility of transfer to a nearby target and the possibility of being lost for each tracking target.

The control unit 48 generates text and graphic information as a result of processing performed in accordance with the operating status of the radar apparatus and instructions given by the operator.
The signal processing unit 46 merges the instruction image (which may include a wake image and a symbol image), characters, and graphic information as an image having a predetermined format and layout, and delivers the image to the instruction unit 47.

The instruction unit 47 displays an instruction image based on a predetermined instruction method on the instruction screen.
Therefore, in a ship equipped with such a radar device having a tracking function, a ship operator can perform an appropriate ship operation while easily grasping the movement of another ship.

In addition, there existed patent document 1 thru | or patent document 4 mentioned later as a prior art relevant to this invention.
(1) “In a radar device that obtains a target radar image by improving the resolution in the range direction by pulse compression and further improving the resolution in the cross range direction by utilizing the Doppler effect caused by the rotation or movement of the target. Image reproduction means for reproducing the target radar image from the received signal output from the machine, target tracking means for measuring the position, moving direction, speed, etc. of the target from the same received signal; Point image response estimation means for estimating the image response function, target aspect angle estimation means for estimating the angle formed by the line of sight and the target traveling direction, and three-dimensional shape data of the target to be recognized and identified in advance are stored Target shape data storage means to be read out, and the target shape data are sequentially read out from the target shape data storage means, RCS calculation means for calculating an RCS (Radar Cross Section) distribution on the target from the target shape data based on the target aspect angle estimated by the spect angle estimation means, the RCS distribution on the target calculated thereby, and the point image response Convolution integration means for generating a dictionary image for recognizing and identifying a target by performing convolution integration with a function, and radar image display means for simultaneously displaying a dictionary image sequentially output and a radar image reproduced from the received signal The radar apparatus is characterized in that “the operator can easily recognize and identify the target from the radar image or can be automatically identified by the radar”.

(2) “A two-dimensional radar image consisting of a range-Doppler plane created from radar echoes is divided into regions of a specified size, and an image cell having the maximum amplitude in each region including the target portion is divided for each region. The image cell selection means for selecting the image cell, and each image cell selected by this means has the maximum amplitude within the area centered on the point of the image cell selected for the radar image one unit time ago. Image cell tracking means for outputting the position and complex value of a point in the image cell as a new selected image cell, and calculating the complex value phase of the selected image cell point sequence and calculating the second derivative with respect to the phase time Then, the image cell acceleration calculating means for converting the calculation result into acceleration and outputting it, the two-dimensional radar image, and the acceleration are input, and the two-dimensional position (record) on the range-Doppler plane of the image cell tracking point is input. 3D quantity (range, Doppler frequency, acceleration) added to acceleration and Doppler frequency) and the intensity of the image cell tracking point, the amount of (range position, Doppler frequency, acceleration, intensity) is the amount of the image cell 3D radar image creation means for outputting as a 3D radar image of the tracking point, and (range position, Doppler frequency, acceleration) of the 3D radar image are regarded as coordinate points and are represented by a range axis, a Doppler axis, and an acceleration axis. 3D radar image display means for displaying a point indicating the intensity on a three-dimensional coordinate system, without collecting data in various aspect directions of the radar as viewed from the radar target. 3D shape information of the target can be obtained using the radar image obtained on the established range-Doppler plane. Da image processing apparatus ... Patent Document 2

(3) “The target motion based on the tracking data, which is the target position / velocity information, including the tracking data of the candidate target observed by the radar and the tracking data generated by the simulation, and the radar observation data. Alternatively, a target parameter estimator for estimating a target parameter estimation value related to a posture, target-specific specification data that is three-dimensional shape data of a candidate target, the radar observation specification data, and target parameter estimation from the target parameter estimator And a dictionary time-series image sequence generator for generating dictionary time-series image sequence data for each candidate target, observation time-series image sequence data obtained by observing an unknown target with a radar, and Based on the dictionary time-series image sequence data from the dictionary time-series image sequence generator, the type identification of the unknown target included in the observed time-series image sequence data A target discriminating device characterized by "realizing class discrimination between targets with similar shapes and motions for image data obtained by radar" ... Patent Reference 3

(4) In the radar device that detects the target of interest from the image data generated based on the input signal including the echo signal that is reflected by the transmitted signal, the echo signal of the target of interest A storage unit that stores template image data generated based on characteristics, a unit that performs pattern matching between the image data and the template image data, and a target of interest detected from the image data as a result of pattern matching , Including a specifying unit that specifies a region in the image data including the target of interest related to the template image data ”, thereby enabling a specific target of interest to be accurately detected from the image data. "Radar device characterized by point ... Patent Document 4

Japanese Patent Application Laid-Open No. 06-174838 JP 08-271612 A Japanese Patent No. 4772341 JP 2012-108050 A

  By the way, in the above-described conventional example, to ensure the accuracy and stability of target tracking, in general, the sea area where a ship or the like targeted for target tracking can be located is geographically wide and the weather conditions are diverse. Therefore, it has been difficult to achieve with radar signal processing alone.

  Note that the accuracy and stability of such target tracking are generally secured by applying the techniques listed below, and for example, obstruction factors caused by noise, sea surface reflection, rain and snow reflection, etc. are reduced. obtain.

(1) Automatic Identification System (AIS: Automatic Identification)
System)
(2) Cooperation with gyro mounted on own ship along with radar equipment
(3) Combined correction of true motion by azimuth angle sensors such as logs

  However, these techniques have been difficult to apply to radar devices mounted on small ships due to cost constraints.

  An object of the present invention is to provide a tracking support apparatus that can flexibly adapt to a target shape, size, relative distance, relative position, course substance, and course, and can track the target stably and inexpensively.

  According to the first aspect of the present invention, the feature identifying means obtains a feature amount of a specific part where a target to be tracked is located in an image obtained from a reflected wave of a radar signal. The tracking target estimation means estimates that the target is located in a part of the image that has the maximum correlation with the feature amount or exceeds a predetermined threshold.

  That is, the position of the tracking target is a conventional radar signal in which the direction, distance, course, etc. of the tracking target is identified based on the intensity and frequency component of the reflected wave, and the speed and traveling direction are predicted based on these. Unlike the processing, the accuracy is high based on the feature of the image indicated by the reflection, and it is easily estimated by utilizing the surplus processing amount.

  According to a second aspect of the present invention, in the tracking support device according to the first aspect, the feature amount is a feature amount suitable for a target that can be a target of the tracking.

  That is, the position of the tracking target is estimated based on the feature amount suitable for the tracking target.

  According to a third aspect of the present invention, in the tracking support device according to the first or second aspect, the feature amount includes all or part of the shape, size, relative distance, relative position, and course of the tracking target. Is a combination of a plurality of feature amounts obtained by learning either or both of the entity and the course.

  That is, the tracking target position is estimated accurately and stably based on a combination of a plurality of feature amounts that are not necessarily high in accuracy even when only one feature amount cannot achieve high accuracy.

  According to a fourth aspect of the present invention, in the tracking support device according to any one of the first to third aspects, the feature amount includes all or one of a relative distance, a relative position, and a course of the tracking target. Selected based on the history of the part.

  That is, the estimation of the position of the tracking target is realized while flexibly adapting to the state and behavior of the target.

  According to the present invention, the accuracy and stability of target tracking can be improved even when the area where the tracking target is located and the weather conditions in the area can be variously changed.

  In the present invention, even if there are various targets that can be tracked, if the feature quantity suitable for the target can be specified (selected) and calculated, the accuracy and stability of the target tracking can be maintained high. Is done.

Furthermore, target tracking can be realized with high accuracy by flexibly adapting to target target attributes.
In addition, tracking in accordance with the target substance is realized with high accuracy and stability.

  Therefore, in a radar site or a mobile body equipped with the radar apparatus according to the present invention, the reliability of target tracking is improved and stable maintenance is achieved, thereby improving navigation safety.

It is a figure which shows one Embodiment of this invention. It is an operation | movement flowchart of the control part in this embodiment. It is a figure explaining operation | movement of this embodiment. It is a figure which shows the structural example of the radar apparatus which has the function of target tracking.

Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.
FIG. 1 is a diagram showing an embodiment of the present invention.
In the figure, components having the same functions and configurations as those shown in FIG. 4 are given the same reference numerals, and descriptions thereof are omitted here.

The difference between the present embodiment and the conventional example shown in FIG. 4 is that a signal processing unit 11 is provided instead of the signal processing unit 46.
FIG. 2 is an operation flowchart of the control unit in the present embodiment.
FIG. 3 is a diagram for explaining the operation of the present embodiment.

The operation of this embodiment will be described below with reference to FIGS.
In the present embodiment, the feature of the present invention resides in the procedure of target tracking processing performed by the signal processing unit 11 as will be described later.

  The functions and operations of elements other than the signal processing unit 11 are the same as those in the conventional example shown in FIG.

  In the present embodiment, as shown in FIG. 3, the main memory of the signal processing unit 11 is pre-arranged with registers for storing the following items that are referred to in the course of the target tracking process performed in the present embodiment. The

(1) Binary information (logical values “1” and “0”) corresponding to each of the feature amounts a) to v) listed below and indicating whether or not the corresponding feature amount is applied to target tracking Corresponds to “appropriate” and “no”, respectively, and the flag register 11 _FR is stored packed.

a) Haar-like feature value: given as the brightness difference between two adjacent rectangular areas.
b) EOH (Edge of Orientation Histograms) feature quantity: The feature quantities of many local regions are given as a combination of low-level feature quantities.
c) HOG (Histogram of Oriented Gradient) feature value: given as an edge histogram.

d) Edgelet feature: Means the connection of edges.
e) Joint Haar-like feature: given as the correlation of luminance features.
f) Shapelet feature amount: means the correlation of edges that can fall under a feature.

g) Joint HOG feature: Means the correlation of edge gradients in two regions.
h) SIFT (Scale-Invariant Feature Transform) feature
i) SURF (Speeded Up Robust Features) features

j) RIFF feature
k) BRIFF feature
l) ORB (Oruented FAST and Rotated BRIEF) features

m) BRISK (Binary Robust Invariant Scalable Keypoint) feature
n) CARD (Compact And Real-time Descriptors) feature
o) Center of gravity moment

p) Zernike moment
q) Higher order local autocorrelation features: found by Log-Polar transformation, universal for both magnitude, rotation, and tilt of the arrhythmia plane relative to the optical axis of the optical system.
r) LBP (Local Binary Pattern) features

s) Center-Symmetric LBP feature
t) Ferns feature
u) HoF feature
v) FREAK (Fast retina Keypoint descriptor) feature

(2) A feature value register 11 _FVR which is a set of storage areas in which the values of the feature amounts a) to v) are individually stored.

  The signal processing unit 11 performs the following processing when the radar apparatus according to the present embodiment is started and when the operation mode (range switching, etc.) of the radar apparatus is changed.

(1) The logical value of each binary information on the flag register 11 _FR is initialized to an initial value suitable for the state of the radar apparatus at that time, or updated to a value suitable for a new operation form of the radar apparatus. (FIG. 2, step S1).
(2) Feature value register 11 All the storage areas of the _FVR are initialized (step S2 in FIG. 2).

  In the steady state, the signal processing unit 11 maps the A scope indicated by the above-described baseband signal in synchronization with the sweep and the scan in the same manner as the conventional example, so that it is combined with the above-described “wake image”. Then, an instruction image including the “symbol image” generated as a result of the target tracking is generated (step S3 in FIG. 2).

Such target tracking is performed in the following procedure.
(1) For the first time after starting, identify the direction, distance, course, etc. of the target that is the target of tracking (hereinafter referred to as “tracking target”) and predict the speed and direction of travel based on these. (Step S4 in FIG. 2).

(2) In the instruction image at this time point, the part where the tracking target with the predicted speed and traveling direction is located is specified as described above. (FIG. 2, step S5)

(3) Among the above-described feature values a) to v), a specific feature value Vf whose binary information value stored in the flag register 11 _FR is “1” is calculated for the part. The quantity register 11 is stored in the corresponding storage area of the _FVR (step S6 in FIG. 2, FIG. 3 (1)).

(4) In the subsequent sweep process, the specific feature value Vf ′ is calculated for the tracking target on the updated instruction image and its neighboring region (FIG. 3 (2)) (step S7 in FIG. 2), The degree of correlation between the feature quantity Vf ′ and the feature quantity Vf is compared with a predetermined threshold value (step S8 in FIG. 2).

(5) When the degree of correlation is below the threshold, the feature value Vf is updated with the average value of feature values Vf and Vf ′ (may be a moving average or an exponential smooth value) (step S9 in FIG. 2, FIG. 2). 3 (3)), the speed and traveling direction of the tracking target are predicted by identifying the azimuth, distance, and course of the tracking target based on the same procedure as in the conventional example (step S10 in FIG. 2).

(6) Conversely, when the correlation degree exceeds the threshold value, the following processing is performed.
(6-1) The feature value Vf is updated to the latest feature value Vf ′ (step S11 in FIG. 2).
(6-2) The direction, distance, course, etc. of the tracking target are identified based on the same procedure as in the conventional example (step S12 in FIG. 2).

(6-3) By correcting the predicted result of the tracking target's speed and traveling method by regarding the direction, distance, and course as the tracking target position, the position where the corresponding correlation is maximized on the instruction image is considered ( FIG. 2 step S13).

(7) Characters and graphic information generated in the same manner as in the conventional example based on the result of such target tracking are merged with the above-described instruction image (a wake image or a symbol image may be included). Delivered to the unit 47 (step S14 in FIG. 2).

  That is, according to the present embodiment, it is determined with high certainty that the tracking target has not changed or lost based on the feature amount of the image of the tracking target and its surrounding parts on the instruction image. Target tracking is realized with higher accuracy than the example.

  Therefore, in a ship equipped with the radar apparatus according to the present embodiment, the reliability of target tracking is improved, so that the convenience of maneuvering is improved and thus the safety of navigation is enhanced.

  In the present embodiment, the change of the tracking target and the avoidance of the loss are attempted based on only one feature amount determined by the form in which the radar apparatus is operated.

However, in the present invention, the target determination may be performed based on “Bags-of-Word expression” in which a plurality of local feature amounts are expressed as a histogram, instead of only one feature amount.

In the present invention, the combination and weighting of local feature amounts may be appropriately set or updated based on any of the following machine learning methods.
(1)
Boosting
(2)
SVM (Support Vector Machine)
(3)
Decision tree learning
(Four)
neural network
(Five)
Clustering

  Further, in such a learning process, for example, by selecting a suitable feature amount selected based on the history of all or part of the relative distance, relative position, and course of the target, the state and behavior of the target are selected. A flexible adaptation to may be achieved.

Furthermore, in the present embodiment, an instruction image to be delivered to the instruction unit 47 is directly referred to for characteristic processing performed for target tracking.
However, if the image referred to for target tracking is an image having a content and format suitable for the characteristic processing described above, the image is generated under a process separately performed in parallel with the generation of the instruction image. Any image may be substituted.

  Further, in the present embodiment, the feature amount that can be applied is flexibly adapted to the target shape, size, relative distance, relative position, course substance and course without impairing desired accuracy and real-time characteristics. If so, any feature amount may be used.

Furthermore, in the present embodiment, the above characteristic processing is realized by the signal processing unit 11 utilizing the surplus processing amount.
However, such processing may be executed by a module or package that is separate from or linked with an existing tracking processing module or package.

Further, in the present embodiment, the position where the aforementioned correlation is maximized on the instruction image is regarded as the position of the tracking target, so that the speed of the tracking target and the prediction result of the traveling method are corrected.
However, the present invention is not limited to such a configuration, and for example, the position where the aforementioned correlation is maximized may be applied as the update value of the position of the tracking target obtained in advance.

  Further, when such a configuration is adopted, the area where the correlation is taken on the instruction image is not limited to the current value of the position of the tracking target and the vicinity thereof, for example, all on the instruction image or Any of partial images obtained by dividing the instruction image may be used.

  Further, in the present embodiment, the tracking target speed and the traveling direction prediction result may not be corrected in a state where the maximum correlation point on the instruction image exists reliably and stably.

  Further, the present invention is not limited to the above-described embodiments, and various configurations can be made within the scope of the present invention, and any improvement may be applied to all or some of the components.

DESCRIPTION OF SYMBOLS 11, 46 Signal processing part 11 _FR flag register 11 _FVR feature value register 41 Transmission part 42 Circulator 43 Power supply part 44 Antenna system 45 Reception part 47 Instruction part

Claims (1)

Among the images obtained from the reflected wave of the radar signal transmitted by the radar device, among the plurality of feature values of the specific part where the target to be tracked is located , “appropriate” for each of the plurality of feature values. ”And“ flag ”in which the binary information that means“ not ”is described, and at least one feature amount of the value that means“ suitable ” is selected from the binary information in the flag register , A feature identifying means for obtaining a feature value Vf of a specific part and obtaining a feature value Vf ′ of the specific part and a neighborhood region of the specific part after a predetermined time;
Tracking target estimation means for estimating that the target is located in a part of the feature quantity Vf ′ that has the maximum correlation with the feature quantity Vf or exceeds a predetermined threshold;
The plurality of feature quantities are
(A) haar-like feature value,
(B) EOH (Edge of Orientation Histograms) feature,
(C) HOG (Histogram of Oriented Gradient) feature quantity,
(D) Edgelet feature amount,
(E) Joint Haar-like feature,
(F) Shapelet features,
(G) Joint HOG feature,
(H) SIFT (Scale-Invariant Feature Transform) feature,
(I) SURF (Speeded Up Robust Features) features
(J) RIFF feature amount,
(K) BRIFF feature,
(L) ORB (Oriented FAST and Rotated BREF) features,
(M) BRISK (Binary Robust Innovative Scalable Keypoint) feature,
(N) CARD (Compact And Real-Time Descriptors) feature quantity,
(O) Center of gravity moment,
(P) Zernike moment,
(Q) higher order local autocorrelation features;
(R) LBP (Local Binary Pattern) feature,
(S) Center-Symmetric LBP feature,
(T) Ferns feature,
A tracking support apparatus, characterized in that (u) a HoF feature amount and (v) a FREEAK (Fast retina keypoint descriptor) feature amount.