JPH09159744A - System for automatically separating multitarget signal - Google Patents

Abstract

PROBLEM TO BE SOLVED: To automatically separate line-detecting signals by targets by using azimuth information after analyzing acoustic signals for frequency. SOLUTION: A signal processing section 11 analyzes acoustic signals for frequency and a line detecting section 12 and an azimuth extracting section 13 detects lines from the analyzed results of the section 11 and extract the azimuths of the lines. A line number searching section 21, an excessive azimuth group deleting section 22, and an azimuth group integrating section 23 divide the lines at prescribed time into groups by using azimuth information. A continuity discriminating section 31 discriminates the continuity between azimuth groups at two time and a time series discriminating section 32 separates the lines by targets by time sequentially discriminating the lines. Therefore, the possibility of including noise in the same target group is reduced and a target existing in close azimuths can also be separated.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an automatic multi-target signal separation system, and more particularly to a multi-target separation of underwater acoustic signals.

[0002]

2. Description of the Related Art Conventionally, as a method of separating a frequency spectrum included in an acoustic signal for each target, for example, Japanese Patent Laid-Open No.
Japanese Unexamined Patent Publication No. 223374/1992, "Spectrum target display method for frequency spectrum" document (1), and Japanese Unexamined Patent Publication No. H4-65689, "2D echo pattern identification method" document (2).

In the above document (1), a display method is shown in which only frequency components coming from a predetermined azimuth range are displayed on a screen. In order to finally separate the goals, the person judged by looking at the displayed result. In (2), a method of separating a target from a received signal obtained by transmitting an acoustic wave using azimuth information and distance information is shown.

[0004]

In the conventional method of Document (1), an acoustic signal is frequency-analyzed, the intensity of frequency components is represented by shading, and displayed in time series (hereinafter referred to as LO).
FAR gram), a person visually detects the line and determines the separation, which takes time and effort.
There is a problem that there is an individual difference in separation performance. Further, when the number of input audio signals becomes plural and the amount of processing to be judged by a person increases, there is a problem that information on a target may be overlooked. Further, when the arrival direction of the signal changes due to the movement of the target, a predetermined azimuth range must be widened, and the target cannot be separated in a situation where the plurality of targets move azimuthally. was there.

[0005] Next, according to the conventional method of Document (2), separation can be performed according to the target position, but there is a problem that it is impossible to separate what kind of sound is generated from each target.
Further, there is a problem that it is necessary to transmit an acoustic wave in order to know the position of the target, and separation is impossible under a situation where only the received signal is processed without being known to the target.

[0006]

According to the automatic multi-target signal separation method of the present invention, a preprocessing unit for detecting a line included in an input audio signal and an azimuth of the line at a predetermined time is provided. From the result at a predetermined time of the processing unit,
A separation processing unit that groups lines by direction; a determination processing unit that determines a result of the separation processing unit in time series and separates a line for each target; and a storage that stores necessary processing results in each processing unit. Device.

The preprocessing unit analyzes the frequency of the input acoustic signal and outputs the intensity and direction of the frequency component. And a direction extracting unit that extracts a direction of the line at a predetermined time from the frequency analysis result and the line detection result.

Next, the separation processing section sets a window in a predetermined azimuth angle range from the line azimuth information, and changes the line number existing in the window at a predetermined time and the center azimuth of the window. While searching in all directions, the line number search unit that searches for the center direction and line number of the window, and compares the set of line numbers from the search results for each azimuth window, is included in the combination of the line numbers of other windows A surplus bearing group deletion unit that deletes a surplus bearing group from the search results,
The azimuth group integration unit integrates azimuth groups in which combinations of line numbers completely match from the deletion result into one candidate by averaging the center azimuths.

The determination processing section uses a continuity determination section for determining the temporal continuity of the azimuth group from the result of the separation processing section at two successive analysis times, and a continuity determination result in time series. A time series determination unit that determines line numbers belonging to the same azimuth group as the same target group for a predetermined time or more.

Further, the storage device includes a frequency analysis result storage unit for storing an output result of the signal processing unit, a line detection unit,
A line information storage unit for storing an output result of the direction extraction unit;
The azimuth group storage unit stores the output results of the line number search unit, the redundant azimuth group deletion unit, the azimuth group integration unit, and the azimuth group continuity storage unit that stores the output results of the continuity determination unit.

[0011]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Next, the present invention will be described with reference to the drawings. FIG. 1 shows a block diagram of one embodiment of the present invention. The automatic multi-target signal separating apparatus according to the present embodiment includes a pre-processing unit 1
0, a separation processing unit 20, a determination processing unit 30, and a storage device 40. The pre-processing unit 10 includes a signal processing unit 11
, A line detecting unit 12 and an azimuth extracting unit 13.
The separation processing unit 20 includes a line number search unit 21, a surplus orientation group deletion unit 22, and an orientation group integration unit 23. The determination processing unit 30 includes a continuity determination unit 31 and a time-series determination unit 32. The storage device 40 includes a frequency analysis result storage unit 41, a line information storage unit 42, an azimuth group storage unit 43, and an azimuth group continuity storage unit 44.

The signal processing unit 11 gives a directivity to the passive sonobuoy and performs frequency analysis on a signal received by a sonobuoy which can detect a direction of arrival of the sound, hereinafter referred to as a diphasonobuoy, and outputs a LOFAR gram. The frequency analysis result storage unit 41 stores signal strength information and azimuth information of each frequency component from the result of the signal processing unit 11. The line detection unit 12 fetches the frequency analysis result from the frequency analysis result storage unit 41, detects a signal component, assigns a line number, and outputs a line number and a frequency at each analysis time. The line information storage unit 42 stores the line detection unit 12
The line number and the frequency at each analysis time are stored from the result of (1). Here, the signal line detection method is disclosed in Japanese Patent Application No. Hei.
The method proposed in US Pat. No. 3,324,432 can be used. Briefly, this method detects signal strength peaks at each analysis time of a LOFAR gram. Then, these peaks are tracked within a preset frequency fluctuation width to form a signal line. The azimuth extracting unit 13 reads the time and frequency information of the detected line from the line information storage unit 42 and fetches the azimuth information corresponding to the time and frequency information from the frequency analysis result storage unit 41. The line information storage unit 42 stores the line number and the frequency information stored in the result of the direction extraction unit 13 in association with the direction.

The line number search unit 21 fetches the line number and the azimuth at a predetermined time from the line information storage unit 42, sets a window of a predetermined azimuth angle range, and determines the line number existing in the window by the window. The search is performed in all directions while changing the center direction, and a set of the center direction of the window and the line number is output as a direction group candidate. The azimuth group storage unit 43 stores a set of the center azimuth of the window and the line number from the result of the line number search unit 21. The surplus azimuth group deletion unit 22 reads the center azimuth and the line number of the azimuth group candidate from the azimuth group storage unit 43, compares the line number for each azimuth group candidate, and has a smaller number of lines than the number of lines in the other windows. And the combination of line numbers
The surplus orientation group that is included in the combination of the line numbers of any one of the windows is deleted. The azimuth group storage unit 43 deletes the surplus azimuth group from the storage device, reflecting the result of the surplus azimuth group deletion unit 22. The azimuth group integration unit 23 includes the azimuth group storage unit 4
Then, the direction group candidates are read from No. 3 and the direction group candidates in which the combinations of the line numbers completely match are averaged for the center direction to be grouped into one direction group.
The orientation group storage unit 43 records the result of the orientation group integration unit 23 as an orientation group, and deletes the stored orientation group candidate.

The continuity determining unit 31 reads the information of the azimuth group at two predetermined times from the azimuth group storage unit 43, determines the temporal continuity of the azimuth group by comparing the line numbers, and determines the determination result. It is stored in the azimuth group continuity storage unit 44. The time-series determination unit 32 reads the determination result from the azimuth group continuity storage unit 44 in chronological order, determines the line numbers belonging to the same azimuth group as the same target group for a predetermined time or more, and determines the line numbers of the same target group. Is output.

As described above, since the signal line is separated for each target when the duration of the line exceeds a predetermined time using the time-series information, the possibility that noise is included in the same target group is reduced. Also, close targets can be separated due to differences in movement.

Next, a first embodiment of the present invention will be described with reference to FIG.
~ It demonstrates using FIG. FIG. 2A shows two goals A,
4 shows the positional relationship between B and Daifa Sonobui. At time t _N , the two targets are close to each other, and it is difficult to separate the frequency components of the signal only from the azimuth information. FIG. 2B shows frequencies at which the targets A and B occur. Goal A is 1
0, 15, 20, 30 Hz signals, target B is 40, 5
It is assumed that a signal of 5, 70 Hz is emitted.

FIG. 3A shows the result of inputting the signal received by the diphasonobuoy to the signal processing unit 11 and writing a LOFAR gram. Noise and signal with strong signal strength are displayed. The frequency analysis result storage unit 41 stores the time t
_{From 0} to T _N , each frequency and its direction at each time are stored. FIG. 3B shows an output in which the output result of the signal processing unit 11 is sequentially input to the line detection unit 12. Here, eight lines are detected. One, two,
Lines 3 and 4 are sounds generated from target A, 5, 6, 7
The line No. 8 is a sound generated from the target B, and the line No. 8 is noise. The line information storage unit 42 stores the time t
_{From 0} to time t _N , the line number existing at each time and its frequency are stored.

FIG. 4 shows the result of inputting the output of the line detecting unit 12 to the azimuth extracting unit 13. Using the line number and the frequency, the direction corresponding to the frequency is read from the frequency analysis result storage unit 41, and the line number is associated with the direction.
Here, the direction of the noise is the same as that of the target B. This output is stored in the line information storage unit 42.

FIG. 5 shows a processing result obtained by inputting the output of the azimuth extracting unit 13 at time t ₀ to the line number searching unit 21. Here, the width of the window is set to 10 degrees, and the change in the center orientation of the window is set to 5 degrees. The center direction is changed, and a line number existing in the range of the window is searched using the direction information.
By performing the above processing, 6 from F21 to F26
Two orientation group candidates are created. This is stored in the direction group storage unit 43.

FIG. 6A shows a processing result obtained by inputting the output of the line number searching unit 21 to the surplus direction group deleting unit 22. The azimuth group candidates F24 and F26 both have a smaller number of lines than the azimuth group candidate F25 and completely include a combination of line numbers. Therefore, the azimuth group candidates F24 and F26 are deleted, and as a result, four of G21 to G24 are deleted. Two orientation group candidates are stored in the orientation group storage unit 43. FIG. 6B shows a processing result obtained by inputting the result of the surplus orientation group deletion unit 22 to the orientation group integration unit 23. Orientation group candidates G21 to G23
Since the combinations of the line numbers are the same, the azimuth groups are combined into one, and the central azimuth is the average value of the central azimuths of the three azimuth group candidates. As a result, the two orientation groups H21 and H22 are stored in the orientation group storage unit 43. Similarly, the above processing is performed at time t ₁ , and the output result of the azimuth group integration unit 23 is as shown in FIG.
Let's say FIG. 6D shows the times t ₀ and t _1.
This is an output when the integration result is input to the continuity determination unit 31. The set of line numbers at time t ₀ and time t ₁ is compared, and it is considered that the azimuth group in which even one line number matches is continued. Here, H21 and I21,
It is determined that H22 and I22 continue, and stored in the azimuth group continuity storage unit 44.

FIG. 7A shows the output of the azimuth group integrating unit 23 at time t _N-1 . It is assumed that two azimuth groups have been created so far. FIG. 7B shows the time t _N.
Is the output of the azimuth group integration unit 23 in FIG. Time t
_{In N} , since the azimuths of the two targets are close to each other, they form one azimuth group. Therefore, the continuity determination unit 31
Are determined to belong to the same orientation group as L21 for both K21 and K22. (FIG. 7 (C)).

FIG. 7 (D) shows an output result of the time series determination unit 32 when the output of the continuity determination unit 31 from time t ₀ to time t _N is input. If the azimuth group continues beyond a predetermined time length, the line numbers belonging to the azimuth group for a predetermined time or more are determined to be the same target group. The noise of line number 8 is not included in the same target group because the azimuth does not continue temporally. Also, line number 1
Are determined to be one azimuth group at time t _N , but are separated by a predetermined time or more,
It is separated into two identical goal groups.

Next, a second embodiment of the present invention will be described. The operation of this embodiment will be described with reference to FIG. FIG. 8A shows the positional relationship between the two targets A and B and the Daifaso Nobuy, as in FIG. 2A. Time t _N
From time t to time t _V , the two targets are close to each other, and it is difficult to separate the frequency components of the signal only from the azimuth information. The frequencies at which the two targets are generated are the same as in FIG. 2B, and the correspondence between the line numbers and the generated frequencies is also the same as in the example of FIG. FIG.
(B) is an output result of the azimuth group integration unit 23 at time t _V. From the time t _N to time t _V, the line number 1 7 and was regarded as one azimuth groups. The processing up to the azimuth group integration unit 23 is the same as that of the first embodiment, and the description is omitted. FIG. 8C shows time t.
It is an output result of the azimuth group integration unit 23 at _{V + 1} . Here, two orientation groups have been created. Time t
FIG. 8D shows the output of the continuity determining unit 31 at _{V + 1} . FIG. 8E shows the result of inputting the output of the continuity determining unit 31 from time t _N to time t _{V + 1} to the time-series determining unit 32. Since the time separated into two is less than the predetermined time, it is determined as one and the same target group. FIG. 8F shows the result of inputting the output result of the continuity determining unit 31 from time t _N to time t _X to the time-series determining unit 32. It is assumed that two azimuth groups have been separated from time t _{V + 1} to time t _x . Since they were separated for more than a predetermined time, they were separated into two identical target groups.

FIG. 9 shows a third embodiment. FIG. 9 (A) shows the positional relationship between the two targets A and B and the diphasone knob as in the first and second embodiments. Here, a specific description will be given using an example in which the first embodiment and the second embodiment are continuous, that is, an example in which two targets cross each other. Time t _N
The steps up to this are the same as in the first embodiment. FIG. 9B shows the time t.
_This is the result of inputting the output result of the continuity determination unit 31 from ₀ to time t _V to the time series determination unit 32. Since the two azimuth groups have been separated from time t ₀ to time t _N−1, they are also separated into two identical target groups at time t _V. FIG. 9C shows the result of inputting the output result of the continuity determining unit 31 from time t ₀ to time t _{V + 1} to the time-series determining unit 32. When continuity is determined in time series, C
31 direction groups are B31,..., L1, K2
2, I22 and H22. L1 is determined to be the same as the two azimuth groups K21 and K22, but the combination of the line number of C31 and K21, K22
The combination of 22 line numbers is compared, and connection is made with the azimuth group in which even one line number matches. Thus, C31 is determined to be in the same orientation group as K22. Similarly, the direction group of C32 is B32,.
, L1, K21, I21, and H21 are in the same azimuth group, and the two azimuth groups C31 and C32 are separated by a predetermined time or more, so they are separated as two same target groups.

In the first, second and third embodiments, the number of targets has been described as two, but any number of targets can be used. Also, the target movement is described as being linear, but any movement can be handled.

Since the movement of the target is tracked by using the time-series information as described above, it is possible to separate the targets existing in close directions, and it is possible to reduce the risk that noise is erroneously included in the same target group.

[0027]

As described above, since the multi-target signal automatic separation system according to the present invention is configured so that the target separation can be performed automatically, the time and labor can be greatly reduced as compared with the prior art, and the manual separation performance can be reduced. This has the effect of eliminating individual differences and oversights.

In addition, since the determination is made using the time-series target motion information, there is an effect that separation is possible even in a situation where a plurality of targets move azimuthally.

Furthermore, since only information of the received signal is used, there is an effect that the sound generated by each target can be separated without noticing the target.

[Brief description of the drawings]

FIG. 1 is a block diagram showing an embodiment of an automatic multi-target signal separation system according to the present invention.

FIG. 2 is a diagram for explaining the operation of the first embodiment;
(A) is a diagram showing the positional relationship between Daifaso Nobuy and two targets,
(B) is a frequency at which the targets A and B occur.

FIG. 3 is a diagram for explaining the operation of the embodiment of FIG. 1;
(A) is an output of the signal processing unit 11 and (B) is an output of the line detection unit 12.

FIG. 4 is a diagram for explaining the operation of the embodiment in FIG. 1, and is a diagram showing an output of a direction extracting unit 13;

[Figure 5] is a view for explaining the operation of the embodiment of FIG. 1, shows the output of the line number search unit 21 at time t _0.

FIG. 6 is a diagram for explaining the operation of the embodiment of FIG. 1;
(A) shows the surplus orientation group deletion unit 22 at time t ₀ .
Output, (B) the output of the azimuth group integrating unit 23 at time t _0, (C) the output of the azimuth group integrating unit 23 at time t _1, (D) the output of the continuity judgment section 31 at time t ₁ .

FIG. 7 is a diagram for explaining the operation of the embodiment of FIG. 1;
(A) is the output of the azimuth group integration unit 23 at time t _N−1 , and (B) is the azimuth group integration unit 2 at time t _N.
3, (C) is a continuity determination unit 31 at time t _N.
Output, (D) the output of the time-series determination unit 32 at time t _N.

FIG. 8 is a diagram showing the operation of the second embodiment of the present invention;
(A) is a diagram of the positional relationship between Daiso Sonobuoy and two targets,
(B) the output of the azimuth group integrating unit 23 at time t _V, (C) the time t _{V + 1} at an azimuth group integrating section 2
3 output, (D) is the continuity determination unit 3 at time t _{V + 1} .
1 output, (E) is the time series determination unit 3 at time t _{V + 1}
2 (F) is a time series determination unit 32 at time t _x
Output.

FIGS. 9A and 9B are diagrams for explaining the operation of the third embodiment of the present invention, wherein FIG.
(B) is the output of the time series determination unit 32 at time t _v ,
(C) is the output of the time series determination unit 32 at the time _{tv + 1} .

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 10 Pre-processing part 11 Signal processing part 12 Line detection part 13 Direction extraction part 20 Separation processing part 21 Line number search part 22 Excessive direction group deletion part 23 Direction group integration part 30 Judgment processing part 31 Continuity judgment part 32 Time series judgment part 40 storage device 41 frequency analysis result storage unit 42 line information storage unit 43 azimuth group storage unit 44 azimuth group continuity storage unit

Claims (5)

[Claims] 1. A line for each azimuth based on a line included in the input acoustic signal and a preprocessing unit for detecting the azimuth of the line at a predetermined time, and a result of the preprocessing unit at the predetermined time. And a determination processing unit that determines the results of the separation processing unit in time series and separates the lines for each target, and a storage device that stores the necessary processing result in each processing unit. An automatic multi-target signal separation method comprising: 2. The preprocessing unit frequency-analyzes the input acoustic signal, and outputs a signal processing unit that outputs the intensity and azimuth of the frequency component; and a result of arranging the frequency analysis results in time series,
A line detecting unit that detects a signal component represented on a line and gives a line number, and an azimuth extracting unit that extracts the azimuth of the line at a predetermined time from the frequency analysis result and the line detection result. The automatic multi-target signal separation method according to claim 1. 3. The separation processing unit sets a window in a predetermined azimuth angle range from the azimuth information of the line, changes a line number existing in the window at a predetermined time, and changes a central azimuth of the window. While searching in all directions, a line number search unit that searches the center direction and the line number of the window, and a set of line numbers from the search result is compared for each direction window. Few lines,
And the surplus azimuth group that is included in the combination of line numbers, the surplus azimuth group deletion unit that deletes it from the search result, and the azimuth group in which the combination of line numbers completely matches from the deletion result is averaged of the central azimuths. 2. The automatic multi-target signal separation method according to claim 1, further comprising an azimuth group integration unit that combines the candidates into one candidate. 4. The determination processing unit determines the temporal continuity of the azimuth group from the results of the separation processing unit at two consecutive analysis times, and the continuity determination result in time series. 2. The automatic multi-target signal separation method according to claim 1, further comprising a time-series determination unit that determines line numbers belonging to the same direction group as the same target group for a predetermined time or longer. 5. The storage device includes a frequency analysis result storage unit that stores an output result of the signal processing unit, a line information storage unit that stores output results of the line detection unit and an azimuth extraction unit, and the continuity. The automatic multi-target signal separation method according to claim 1, further comprising an azimuth group continuity storage unit that stores an output result of the determination unit.