JPH0252834B2 - - Google Patents

Description

[Detailed Description of the Invention] The present invention relates to a device for measuring magnetic anomalies in the earth's magnetic field using a highly sensitive magnetic anomaly detector mounted on an aircraft and automating the task of searching for a specific target. .

Conventionally, when searching for a shipwreck as a target, for example, when passing over the shipwreck, the magnetic anomaly of the earth's magnetism due to ferromagnetic materials such as iron, which is the constituent material, is generated as a guided waveform in the magnetic field that is the output of a magnetic anomaly detector. This was done by having an operator visually judge the magnetic field measurement signals that are continuously drawn on a recording paper by making use of the fact that they appear in the measurement signals. The reason for adopting this method is that the above-mentioned magnetic field measurement signal contains a lot of noise caused by distortions in the geomagnetic field caused by the topography and geology of the ocean floor, and it is difficult to detect guided waveforms based on magnetic anomalies caused by shipwrecks, etc. This is because advanced human recognition ability is required, but on the other hand, when humans are involved in detection, there is a drawback that it is not possible to maintain a uniform detection function over a long period of time due to individual differences, fatigue, etc.

In order to improve this situation, devices have been devised to automate the detection operations that were previously performed by humans. One of the methods is to perform Haar transformation on the magnetic field measurement signal, perform an appropriate linear combination of the resulting Haar spectra, and then obtain a signal obtained by squaring the Haar spectrum, thereby eliminating various noises contained in the magnetic field measurement signal. At the same time, by effectively extracting the characteristics of the guided waveform caused by a target object, such as a shipwreck, and comparing this with a preset threshold, the presence or absence of a guided waveform caused by a shipwreck, etc. can be automatically determined. There is an automatic magnetic anomaly detector that detects magnetic anomalies, but this type of conventional automatic magnetic anomaly detector has a fixed threshold, so the detection result changes accordingly when the level of the magnetic field measurement signal changes. .

This invention provides a threshold calculation section in a conventional automatic magnetic anomaly detector, calculates the dispersion of the magnetic field measurement signal over a certain period of time in the past, and then automatically calculates the threshold based on the dispersion according to the level of the magnetic field measurement signal. By using a calculation unit to automatically set the threshold, the above-mentioned drawbacks that were caused by the conventional fixed threshold can be removed, and the threshold can be adjusted according to the target object without being affected by the magnitude of the signal level. The purpose is to automatically detect the presence or absence of a guided waveform.

First, the principle of the automatic magnetic anomaly detector of the present invention will be explained. The magnetic field measurement signal measured by the magnetic anomaly detector can take the form of either an analog quantity or a digital quantity depending on the configuration of the magnetic anomaly detector, but here we will introduce magnetic anomaly automatic detection in which an analog quantity magnetic field measurement signal is input. This will be explained using a container as an example.

Figure 1 shows an example of the magnetic field measurement signal when a disturbance in the earth's magnetic field is measured with a high-sensitivity magnetic anomaly detector, and a part of the Haar function family used for Haar transform of the Haar transformer of this kind of conventional automatic magnetic anomaly detector. This is what I drew. In the figure, a shows a magnetic field measurement signal when the above-mentioned high-sensitivity magnetic anomaly detector is mounted on an aircraft to measure geomagnetic disturbances induced by a sunken ship or the like from above. (Hereafter, this signal will be referred to as the magnetic field measurement signal SG.) b is the magnetic field measurement signal SG with a sample period of T _s
The zero-order held signal after A _/ D conversion at
The symbol (γ) shall be used. ) is
The BX section shows waveforms induced by shipwrecks, etc. Note that the duration of the guided waveform is T. c depicts a part of the Haar function family used for Haar transformation. The Haar function family consists of the following sequence of functions defined in the interval 0≦τ≦1, and generally the interval 0
It is known that continuous functions of ≦τ≦1 and even step-like functions such as the magnetic field U can be uniquely expanded by the Haar function family.

X ₀ (τ)=1 0≦τ≦1 X ^(j) _k (τ)= 0 √2 ^k-1 −√2 ^k-1 00<τ<(2j−2)/2 ^k (2j−2) /2 ^k <τ<(2j−2)/2 ^k (2j−2)/2 ^k <τ<2j/2 ^k 2j/2 ^k <τ −(1) (k=1, 2,..., j= 1, 2, ..., 2 ^k-1 ) In other words, the sampling period T _s is set as (2) so that the number of samples N of the guided waveform in the part BX above becomes N = 2 ⁿ (N, n are positive constants). If you choose the value shown by the formula, the magnetic field at N points in the past from the current time t = γ_・T _s = T/(N-1) = T/(2 ⁿ -1) (2) U(γ-N+1) , U(γ-N+2), U(γ-N
+3), ..., U(γ) is in the interval [0,
1] can be uniquely expanded in the Haar function family as shown in equation (3) as a step function divided into 2 ⁿ equal parts.

U _i =U(γ-N+1+i)=h ₀ X ₀ (2i+1
/2 ⁿ⁺¹ )+ _o 〓 ^K=1 ₂ 〓 ^j=1k-1 h ^(j) _k X ^(j) _k (2i+1/2 ⁿ⁺¹ −(3) (i=0, 1, 2,... , 2 ⁿ −1, k=1, 2,
..., n, j=1, 2,...2 ^k-1 ) U (γ-N+1), U (γ-N+2), U in equation (3)
(γ−N+3), ...U(γ), the coefficient h ₀ and h ^(j) _k (k
= 1, 2, ..., n, j = 1, 2, ..., 2 ^k-1 ) is called Haar transformation, and the above coefficients h ₀ and h ^(j) _k (k = 1, 2,...n,
j = 1, 2, ..., 2 ^k-1 ) is called the Haar spectrum. In this case, the Haar spectrum h ₀ above,
h ^(j) _k (k=1, 2,...n, j=1, 2,..., 2 ^k-1 )
is U(γ-N+1+i) (i=0, 1, 2,...2n
-1), it is mathematically known that it can be obtained by the following calculation steps. ⁽⁰⁾ _i = U _i U (γ-N+1+i) -(4) (i=0, 1, 2,..., 2 ⁿ -1) ^(l) _n = ^(l-1) _2n + ^(l-1) _2n+1 (l=1, 2,..., n-1, m=0, 1, 2,
…, 2 ^nl −1) −(5) That is, if we define ⁽⁰⁾ _i and ^(l) _n by equations (4) and (5),
(6) and (7), the Haar spectrum h ₀ , h ^(j) _k (k=1,
2,..., n, j=1, 2,..., 2 ^n-1 ) can be calculated.

h ^(m+1) _o-k+1 = { ^(k-1) _2n − ^(k-1) _2n+1 }/2k√2 ^n-
k ) (k=1, 2,..., n, m=0, 1,..., 2 ^nk -
1) −(6) h ₀ = { ^(n-1) ₀ + ^(n-1) ₁ }2 ⁿ −(7) Figure 2 shows the above spectra h ₀ , h ^(j) _k (k=1, 2 ,
..., n, j = 1, 2, ..., 2 ^n-1 ), when n = 3, that is, the number of sample points is N = 2 ³ = 8 points.
h ₀ , h ^(2k-1) _k (k=1, 2, 3) are expressed on the same time axis as the magnetic field U. It can be seen from FIG. 2 that the Haar spectrum captures the characteristics of the above-mentioned guided waveform well.

The above Haar spectrum h ₀ , h ^(2k-1) _k (k=1, 2,
The features of the guided waveform extracted by ..., n) can be further emphasized by suitably linearly combining them and further using the squared function, that is, the feature extraction signal SS shown in equation (8). .

SS={a ₀・h ₀ +a ₁・h ⁽¹⁾ ₁ +a ₂・h ⁽²⁾ ₂ +a ₃・h ^(
4) ₃ +…+a _o・h ^(2n-1) _o } ² −(8) The coefficient a _k (k=0, 1,...n) is determined based on many actual magnetic field measurement signals SG. . FIG. 3 shows the feature extraction signal SS together with the magnetic field U, and it can be seen from this that the feature extraction signal SS effectively extracts the features of the induced waveform.

If a threshold TH is provided in the feature extraction signal SS, it is possible to automatically determine whether a waveform induced by a shipwreck or the like exists. However, in this method of setting the threshold in advance, the level of the magnetic field measurement signal changes depending on changes in the altitude of the aircraft, so for example, when the level of the magnetic field measurement signal changes by twice, the feature extraction signal SS becomes (3). (8), but since the threshold level is constant, the detection result may change if the feature extraction signal SS and the threshold level are close to each other. In this invention, the dispersion VR of the magnetic field U (γ) in the past fixed time shown in equation (9)
is calculated and multiplied by the coefficient K (positive constant) to determine the automatic setting threshold shown in equation (10).
By using it as an AH, the above-mentioned drawbacks that were caused by the conventional fixed threshold are removed. However, the value of distributed VR and automatically set threshold AH changes every cycle T _s , so when expressing the value at time γ・T _s , VR
(γ) and AH(γ) shall be used.

VR(γ)=1/L _L 〓 ⁱ⁼¹ {U(γ+1−i)−(k))} ² = _L 〓 ⁱ⁼¹ U(γ+1−i) ² −{1/L _L 〓 ⁱ⁼¹ U(γ+1-i)} ² (L≒3N) −(9) AH(γ)=K・VR(γ) −(10) However, (γ) is the magnetic field in the past L・T _s time shown by the following formula This is the average value of U.

(γ)=1/L _L 〓 ⁱ⁼¹ (γ+1-i) From equations (9) and (10), the automatically set threshold AH changes in proportion to the square of the magnetic field U. magnetic field measurement signal due to changes in altitude, etc.
When the level of SG changes, the feature extraction signal SS
Even if the change occurs by a factor of 2, the automatically set threshold AH also changes by a factor of 2, thereby making it possible to prevent the detection result from being influenced by the level of the magnetic field measurement signal SG.

FIG. 4 shows this situation. a in the diagram
is the magnetic field measurement signal SG, and b is the magnetic field measurement signal SG of a.
This figure shows the relationship between the feature extraction signal SS and the automatically set threshold AH corresponding to . c is the feature extraction signal SS and automatic setting threshold when the level of the magnetic field measurement signal of a becomes 1/2
This shows the AH relationship, and the parts BX1BX2 and BX3 are waveforms induced by shipwrecks, etc. If the level of the magnetic field measurement signal becomes 1/2 and the level of the feature extraction signal SS becomes 1/4 as shown in c, the conventional fixed threshold TH shown by the broken line
It can be seen that while the detection results have changed in the BX1 and BX2 sections, the detection results have not changed in the case of the automatic setting threshold AH because its own level is also 1/4. In this way, by calculating the automatically set threshold from the dispersion of the magnetic field measurement signal over a certain period of time in the past, the threshold level can be automatically changed according to the level of the magnetic field measurement signal. It is possible to detect guided waveforms from shipwrecks, etc., regardless of the level. The above is the principle according to this invention.

The automatic magnetic anomaly detector according to the present invention basically consists of five parts: an A/D converter, a Haar converter, a discriminator, a synchronous pulse generator, and a threshold calculator.

FIG. 5 is a block diagram showing the configuration of the automatic magnetic anomaly detector according to the present invention. In the figure, 1 is an A/D converter, 2 is a Haar converter, 3 is a synchronous pulse generator, and 4 is a 5 is a threshold calculation unit, 5 is a discrimination unit, SG is a magnetic field measurement signal (analog signal) from a magnetic anomaly detector, and U is a signal obtained by digitizing the magnetic field measurement signal SG by the A/D converter 1 and holding sampling. , P is the synchronization pulse output from the synchronization pulse generator 3, SS is the feature extraction signal calculated by the Haar transform unit 2, AH
is the automatically set threshold calculated by the threshold calculation section, and LT is the discrimination signal obtained by the discrimination section 5 described above.

The operation of each part will be explained below. FIG. 6 shows the synchronization pulse P generated by the synchronization pulse generator 3.
The pulse width T _c is set to be longer than the conversion time of the A/D converter 1, and the repetition period T _s is set to the value shown by equation (2). Figure 7 shows the operation of the A/D converter 1. When the synchronizing pulse P shown in a rises from 0 to 1, the A/D converter 1 converts the magnetic field measurement signal SG, which is an analog signal, into an analog quantity. The conversion starts, and after this, after a conversion time of T _c , A/
When the D conversion is completed and the synchronization pulse P falls from 1 to 0, the magnetic field measurement signal converted into a digital quantity is held in the zero order and output as the magnetic field U.

In this way, the magnetic field measurement signal, which is an analog signal,
As shown in b, SG is converted into a digital quantity into a magnetic field U every sampling period _Ts , and is sampled and held.

FIG. 8 shows the configuration of the Haar transform section 2 in the case where n=3, that is, the number of sample points N is 8 (=2 ³ ) points.

U is the magnetic field output from the A/D converter 1, P is the synchronization pulse generated from the synchronization pulse generator 3,
6 transforms the magnetic field U into Haar spectrum h ₀ , h ^(m+1) _o-k+1 (n=3, k=1, 2
,
3, m = 0, 1, ..., 2 ^nk -1), U ₀₄ , U ₁₄ , U ₃₃ , U ₇₂ are the outputs of the Haar transformer 6, and the Haar spectra are respectively
Signals corresponding to h ₀ , h ⁽¹⁾ ₁ , h ⁽²⁾ ₂ , and h ⁽⁴⁾ ₃ , and 7 input the above signals U ₀₄ , U ₁₄ , U ₃₃ , and U ₇₂ , and obtain the signals shown in equation (8). The Haar spectral feature extractor SS is the output of the Haar spectral feature extractor 7 that calculates the feature extraction signal.

FIG. 9 shows details of the Haar transformer 6 constituting the Haar transform section 2. As shown in FIG. (This is shown for the case where the number of sample points N is 2 ³ = 8 (n = 3).) In the figure, U is the magnetic field output from the A/D converter 1;
P is a synchronization pulse generated by the synchronization pulse generator 3, M ₀₁ , M ₁₁ ,..., M ₇₁ , M ₀₂ , M ₁₂ ,...,
M ₇₂ , M ₀₃ , M ₁₃ ,..., M ₃₃ and M ₀₄ , M ₁₄ are storage devices for storing intermediate calculation results, CM ₁ , CM ₂ ,
CM ₃ is a storage device storing constants C ₁ , C ₂ , C ₃ respectively, AD ₁₁ , AD ₂₁ , ..., AD ₄₁ , AD ₁₂ ,
AD ₂₂ AD ₁₃ is an adder, SD ₁₁ , SD ₂₁ ,..., SD ₄₁ ,
SD ₁₂ , SD ₂₂ , SD ₁₃ are subtracters, DV ₁₁ , DV ₂₁ ,...,
DV ₄₁ , DV ₁₂ , DV ₂₂ , DV ₀₃ , DV ₁₃ are dividers,
DL ₁ is a delay element that delays the synchronizing pulse P by a time T ₀ (positive constant) that is longer than the sum of the calculation times of the subtracter SD ₁₁ and the divider DV ₁₁ , and DL2 and DL3 each delay the synchronizing pulse P by 2・T ₀ , 3 ・Delay elements that delay T ₀ , P ₁ , P ₂ , P ₃ are the outputs of the above delay elements DL ₁ , DL ₂ , DL ₃ respectively, U ₀₁ , U ₁₁ ,
..., U ₇₁ are storage devices M ₀₁ , M ₁₁ , ..., M ₇₁ respectively
U ₀₂ , U ₁₂ ,..., U ₇₂ are the values stored in the storage devices M ₀₂ , M ₁₂ ,...M ₇₂ , respectively, and U ₀₃ , U 13 , U ₂₃ , U ₃₃ are the values stored in the storage devices M 02 , M ₁₂ ,...M 72, respectively. Storage device
The values stored in M ₀₃ , M ₁₃ , M ₂₃ , M ₃₃ ,
U ₀₄ and U ₁₄ are values stored in storage devices M ₀₄ and M ₁₄ , respectively; W ₀₂ , W ₂₂ , W ₄₂ , and W ₆₂ are outputs of adders AD ₁₁ , AD ₂₁ , AD ₃₁ , and AD ₄₁ , respectively; _W12 ,
W ₃₂ , W ₅₂ , W ₇₂ are subtractors SD ₁₁ , SD ₂₁ ,
The outputs of SD ₃₁ and SD ₄₁ , W ₀₃ and W ₂₃ are adders respectively.
The outputs of AD ₁₂ and AD ₂₂ , W ₁₃ and W ₃₃ are the outputs of the subtracters SD ₁₂ and SD ₂₂ , respectively, W ₀₄ is the output of the adder AD ₁₃ ,
W ₁₄ is the output of subtractor SD ₁₃ , AW ₁₂ , AW ₃₂ ,
AW ₅₂ and AW ₇₂ have dividers DV ₁₁ , DV ₂₁ , and
The outputs of DV ₃₁ and DV ₄₁ , AW ₁₃ and AW ₃₃ are respectively the outputs of dividers DV ₁₂ and DV ₂₂ , AW ₀₄ and AW ₁₄ are the outputs of dividers DV ₀₃ and DV ₁₃ , respectively.

In such a configuration, each time the synchronization pulse P is generated from the synchronization pulse generator 3, the storage device M ₁₁ ,
The contents of M ₂₁ , ..., M ₇₁ are stored in the adjacent storage device M ₀₁ ,
M ₁₁ ,..., M ₆₁ are sequentially transferred to the storage device in this order.
The magnetic field U is newly stored in _M71 . The values U ₀₁ , U ₁₁ , ..., U ₇₁ stored in the storage devices M ₀₁ , M ₁₁ , ..., M ₇₁ in this way are expressed as ⁽⁰⁾ _i = U _i (i=0,
1, 2, ..., 7) (however, n=3).

The values U ₀₁ , U ₂₁ , U ₄₁ , and U ₆₁ stored in the storage devices M ₀₁ , M ₂₁ , M ₄₁ , and M ₆₁ are each added by an adder.
The values U ₁₁ stored in the storage devices M ₁₁ , M ₃₁ , M 51 _, M ₇₁ by AD ₁₁ , AD ₂₁ , AD ₃₁ , AD ₄₁ ,
The signals W ₀₂ , W ₂₂ , W ₄₂ , and W ₆₂ shown in equation (11) are calculated as a result of addition with U ₃₁ , U ₅₁ , and U ₇₁ , but
⁽¹⁾ ₀ = ⁽⁰⁾ ₀ + ⁽⁰⁾ ₁ , ⁽¹⁾ ₁ as shown in equation (5), respectively.
= ⁽⁰⁾ ₂ +
⁽⁰⁾ ₃ , ⁽¹⁾ ₂ = ⁽⁰⁾ ₄ + ⁽⁰⁾ ₅ , ⁽¹⁾ ₃ = ⁽⁰⁾ ₆ +
⁽⁰⁾ Compatible with ₇ . (However, n=3, l=1) W ₀₂ = U ₀₁ + U ₁₁ W ₂₂ = U ₂₁ + U ₃₁ W ₄₂ = U ₄₁ + U ₅₁ W ₆₂ = U ₆₁ + U ₇₁ −(11) Also, the above storage devices M ₀₁ , M The values _{U 01 ,} U _{21 ,} U ₄₁ , U ₆₁ stored in 21 , M 41 , M ₆₁ are stored in the storage devices M ₁₁ , M ₃₁ , by subtracters SD ₁₁ , SD ₂₁ , SD ₃₁ , SD ₄₁ , respectively. The value U ₁₁ stored in M ₅₁ , M ₇₁ ,
U ₃₁ , U ₅₁ , and U ₇₁ are subtracted, and as a result, signals W ₁₂ , W ₃₂ , W ₅₂ , and W ₇₂ shown in equation (12) are calculated.

W ₁₂ = U ₀₁ − U ₁₁ W ₃₂ = U ₂₁ − U ₃₁ W ₅₂ = U ₄₁ − U ₅ ¹ W ₇₂ = U ₆₁ − U ₇₁ − (12) Furthermore, the above signals W ₁₂ , W ₃₂ , W ₅₂ , W ₇₂ is divided by the constant C ₁ (=4) by the dividers DV ₁₁ , DV ₂₁ , DV ₃₁ , and DV ₄₁ , respectively, and the results are the signals AW ₁₂ , AW ₃₂ , AW ₅₂ , and AW ₇₂ shown in equation (13). is calculated.

AW ₁₂ = W ₁₂ /4 = (U ₀₁ - U ₁₁ ) / 4 AW ₃₂ = W ₃₂ /4 = (U ₂₁ - U ₃₁ ) / 4 AW ₅₂ = W ₅₂ / 4 = (U ₄₁ - U ₅₁ ) / 4 AW ₇₂ = W ₇₂ /4 = (U _{61 -} U ₇₁ )/4 - (13) This is the Haar spectrum h shown in equation (6) ⁽¹⁾ ₃ = ( ⁽⁰⁾ ₀
− ⁽⁰⁾ ₁ )/4, h ⁽²⁾ ₃ = ( ⁽⁰⁾ ₂ − ⁽⁰⁾ ₃ )/4, h ^(3
) ₃ = ( ⁽⁰⁾ ₄
− ⁽⁰⁾ ₅ )/4, h ⁽⁴⁾ ₃ = ( ⁽⁰⁾ ₆ − ⁽⁰⁾ ₇ )/4 are effectively calculated into the signals AW ₁₂ , AW ₃₂ , AW ₅₂ , AW ₇₂ , respectively. It represents. (However, n=3, k
=1). The signal W ₀₂ calculated in this way,
W ₂₂ , W ₄₂ , W ₆₂ and AW ₁₂ , AW ₃₂ , AW ₅₂ ,
AW ₇₂ is a memory device each time a pulse delayed by T ₀ from pulse P, that is, pulse P ₁ becomes 1.
M ₀₂ , M ₂₂ , M ₄₂ , M ₆₂ and M ₁₂ , M ₃₂ , M ₅₂ ,
The value is written to _M72 . Next, storage device M ₀₂ ,
The values U ₀₂ and U ₄₂ stored in M ₄₂ are each added by an adder.
Added to the values U ₂₂ and U 62 stored in the storage devices M ₂₂ and M ₆₂ by AD ₁₂ and AD ₂₂ , the _result (14)
Signals W ₀₃ and W ₂₃ shown in the equations are calculated.

W ₀₃ = U ₀₂ + U ₂₂ = W ₀₂ + W ₂₂ W ₂₃ = U ₄₂ + U ₆₂ = W ₄₂ + W ₆₂ − (14) This is shown in equation (5) ⁽²⁾ ₀ = ⁽¹⁾ ₀ + ⁽¹⁾ ₁ , ^(twenty _one
= ⁽¹⁾ ₂ +
⁽¹⁾ Compatible with ₃ . (However, n=3, l=2) Also, the values U ₀₂ , U ₄₂ stored in the storage devices M ₀₂ , M ₄₂
The values U _{22 and U 62 stored in the storage devices M 22 and M 62 are subtracted by the subtracters SD 12 and SD 22 , respectively} , and the resulting signal W ₁₆ = U ₀₂ −U ₂₂ = W ₀₂ −W _{twenty two} ,
W ₃₃ = U ₄₂ − U ₆₂ = W ₄₂ − W ₆₂ is calculated, and the above signals W ₁₃ and W ₃₃ are input to the dividers DV 12 and DV ₁₂ , respectively.
It is divided by the constant C ₂ (=4√2) by DV ₂₂ , and as a result, the signals AW ₁₃ and AW ₃₃ shown in equation (15) are calculated.

AW ₁₃ = W ₁₃ /4√2 = (W ₀₂ −W ₂₂ ) /4√2 AW ₃₃ = W ₃₃ /4√2 = (W ₄₂ −W ₆₂ ) /4√2 − (15) This is (6 ) Haar spectrum h ⁽¹⁾ ₂ = ( ⁽¹⁾ ₀
− ⁽¹⁾ ₁ )/4√2, h ⁽²⁾ ₂ = ( ⁽¹⁾ ₂ − ⁽¹⁾ ₃ )/4
√2 is effectively calculated into the signals AW ₁₃ and AW ₃₃ , respectively. (However, n=3, k=2) The signals W ₀₃ , W ₂₃ and
Values of AW ₁₃ and AW ₃₃ are written into the memory devices M ₀₃ , M ₂₃ and M ₁₃ , M ₃₃ , respectively, every time the pulse delayed by 2·T ₀ from the pulse P, that is, the pulse P ₂ becomes 1. Further, the values U ₀₃ and U ₂₃ stored in the storage devices M ₀₃ and M ₂₃ are added by an adder AD ₁₃ , and the resultant signal W ₀₄ =U ₀₃ +U ₂₃ =W ₀₃ +W ₂₃ is calculated, and the above-mentioned signal W ₀₄ is divided by a constant C ₃ (=8) by a divider DV ₀₃ , and as a result, a signal AW ₀₄ shown in equation (16) is calculated.

AW ₀₄ = W ₀₄ /8 = (W ₀₃ + W ₂₃ ) / 8 = (W ₀₂ + W ₂₂
+W ₄₂ +W ₆₂ )/8 = (U ₀₁ +U ₁₁ +U ₂₁ +U ₃₁ +U ₄₁ +U ₅₁ +U ₆₁
+U ₇₁ )/8−(16) This indicates that the Haar spectrum h ₀ shown in equation (7), that is, shown in the following equation, has been effectively calculated as the signal AW ₀₄ . (However, n=3, k=3) h ₀ = { ^(3-1) ₀ + ^(3-1) ₁ }/2 ³ = ( ⁽²⁾ ₀ +
⁽²⁾ ₁ )/8={( ⁽¹⁾ ₀ + ⁽¹⁾ ₁ )+( ⁽¹⁾ ₂ + ⁽¹⁾ ₃
)}/8 = {( ⁽⁰⁾ ₀ + ⁽⁰⁾ ₁ ) + ( ⁽⁰⁾ ₂ + ^(0
) ₃ ) + ( ⁽⁰⁾ ₄ + ⁽⁰⁾ ₅ ) + ( ⁽⁰⁾ ₆ + ⁽⁰⁾ ₇ )}/8
Further, the value U ₀₃ stored in the storage device M ₀₃ is converted to the value U ₂₃ stored in the storage device M ₂₃ by a subtracter SD ₁₃ .
is drawn and the resulting signal W ₁₄ = U ₀₃ − U ₂₃ = W ₀₃ −
W ₂₃ is calculated and further the above signal W ₁₄ is divided by the divider
The signal AW ₁₄ shown in equation (17) is calculated by dividing by the constant C ₃ (=8) by DV ₁₃ .

AW ₁₄ = W ₁₄ /8 = (W _{03 -} W ₂₃ )/8 - (17) This is shown in equation (6) h ⁽¹⁾ ₁ = ( ⁽²⁾ ₀ - ⁽²⁾ ₁ )/8 is effective This indicates that the signal was calculated as AW ₁₄ . W ₁₄ and AW ₁₄ calculated in this way
is a pulse delayed by 3·T ₀ from the pulse P, that is, each time the pulse P ₃ becomes 1, the memory device M ₀₄ is stored.
and the value is written to _M14 . However, it is assumed that the pulse delay time 3·T ₀ is set to a value sufficiently smaller than the repetition period T _s of the pulse P. As a result, after the pulse P ₃ becomes 1, the signals U ₁₂ , U ₃₂ ,
Haar spectrum by U ₅₂ and U ₇₂ respectively
h ⁽¹⁾ ₃ , h ⁽²⁾ ₃ , h ⁽³⁾ ₃ , and h ⁽⁴⁾ ₃ are Haar spectra h ⁽¹⁾ ₂ and h ⁽²⁾ ₂ by signals U ₁₃ and U ₃₃ , respectively, The Haar spectrum h ⁽¹⁾ ₁ has been effectively determined by the signal U ₁₄ , and the Haar spectrum h ₀ has been effectively determined by the signal U ₀₄ , so that the Haar spectrum shown in equations (6) and (7) is finally obtained. It has been calculated.

FIG. 10 shows details of the Haar spectral feature extractor 7 constituting the Haar transform section 2. As shown in FIG. (This is shown for the case where the number of sample points N is 2 ³ = 8 (n = 3).) In the figure, U ₀₄ , U ₁₄ , U ₃₃ , and U ₇₂ are the storage devices M ₀₄ of the Haar transformer 6, respectively. , M ₁₄ , M ₃₃ ,
Signals output from M ₇₂ , MY ₁ , MY ₂ , MY ₃ ,
MY ₄ and MY ₅ are multipliers, CY ₁ , CY ₂ , CY ₃ , and CY ₄ are storage devices that store constants a ₀ , a ₁ , a ₂ , and a ₃ respectively, ADD is an adder, and ZA is the above adder. The output of
SS is the feature extraction signal obtained as the output of the multiplier _MY5 . In such a configuration, the signals U ₀₄ , U ₁₄ , U ₃₃ , and U ₇₂ are connected to the multipliers MY 1 and MY ₁ , respectively.
By MY ₂ , MY ₃ , MY ₄ the above storage device CY ₁ ,
Constants a ₀ , a ₁ , stored in CY ₂ , CY ₃ , CY ₄ ,
Multiplied by a ₂ and a ₃ respectively, the signals Z ₁ , Z ₂ , Z ₃ , Z ₄
Furthermore, the signals Z ₁ , Z ₂ , Z ₃ , and Z ₄ are added together by the adder ADD to form the signal ZA. In other words, the multipliers MY ₁ ,...MY ₄ , the storage devices CY ₁ ,...CY ₄ ,
and the adder ADD performs the calculation shown in equation (18).

ZA＝a ₀・U ₀₄ +a ₁・U ₁₄ +a ₂・U ₃₃ +a ₃・U ₇₂
-(18) Further, the signal ZA is multiplied by the square in the multiplier _MY5 , and as a result, the signal SS shown in equation (19) is calculated.

SS=ZA ² = {a ₀・U ₀₄ +a ₁・U ₁₄ +a ₂・U ₃₃ +a ₃・
U ₇₂ } ² − (19) This means that the above signals U ₇₄ , U ₁₄ , U ₃₃ , and U ₇₂ correspond to the Haar spectra h ₀ , h ⁽¹⁾ ₁ , h ⁽²⁾ ₂ , and h ⁽⁴⁾ ₃ , respectively. This means that the feature extraction signal SS shown in equation (8) is effectively calculated for the case where n=3.

After all, in the Haar transform unit 2 configured in this way, the magnetic field U inputted by the Haar transformer 6 is Haar-transformed into power-of-2 signal components, that is, the Haar spectrum h shown in equations (6) and (7). ₀ , h ^(m+1) _o-k+1 (n=3
,
k = 1, 2, 3, m = 0, 1, ..., 2 ^nk -1) are calculated, and among the Haar spectra, h ₀ , h ⁽¹⁾ ₁ ,
The signals U ₀ , U ₁₄ , U ₃₃ , and U ₇₂ corresponding to h ⁽²⁾ ₂ and h ⁽⁴⁾ ₃ are further sent to the Haar spectrum feature extractor 7, where they are extracted as feature extraction signals shown in equation (8). SS is calculated for the case where n=3.

In the end, in this way, in the Haar transformer 2, the magnetic field U
The Haar spectrum shown in equations (6) and (7) is calculated by applying the Haar transform, and then h ₀ , h ^(2k-1) _k (k=1, 2, 3) of the obtained Haar spectra are linearly They are combined and further multiplied by 2 to finally calculate the feature extraction signal SS shown in equation (8) for the case where n=3.
Although n=3 is used in this explanation to simplify the explanation, it is actually preferable to use an integer of 4 or more for n, and in this case, even if n increases, the calculation process can be achieved in exactly the same way.

FIG. 11 shows details of the threshold calculation section 4. In the figure, U is the magnetic field output from the A/D converter 1, P is the synchronization pulse generated from the synchronization pulse generator 3, MD ₁ , MD ₂ ,...,
MD _L-1 is a storage device that stores intermediate calculation results, CB,
CC is a storage device that stores constants L and K, respectively;
AX, AY are adders, SUB is subtracter, MP ₀ ,
MP ₁ ,...MP _L-1 , MB, MC are multipliers, DD, DE
are dividers, UU ₁ , UU ₂ , ..., UU _L-1 are values stored in storage devices MD ₁ , MD ₂ , ..., MD _L-1, respectively, V, V ₁ , V ₂ , ..., V _L-1 are the outputs of the multipliers MP ₀ , MP ₁ , ...MP _L-1 , respectively, SX and SY are the outputs of the adders AX and AY, respectively, SMX and AVE are the outputs of the dividers DD and DE, respectively, and SMY is the multiplier vessel
The output of MB, VR is the output of subtracter SUB, and AH is the output of multiplier MC.

In such a configuration, each time the synchronization pulse P is generated from the synchronization pulse generator 3, the storage device
The contents of MD _L-2 , MD _L-3 , ... MD ₁ are sequentially moved to the adjacent storage devices MD _L-1 , MD _L-2 , ..., MD ₂ in this order, and the storage device MD ₁ is exposed to the magnetic field U. is newly memorized. Signal U and the storage devices MD ₁ , MD ₂ ,
..., the values stored in MD _L-1 UU ₁ , UU ₂ , ...UU _L-1
are 2 by the multipliers MP ₀ , MP ₁ , ..., MP _L-1, respectively.
Multiplyed. The above squared signals V, V ₁ , V ₂
..., V _L-1 are added together in adder AX and signal SX
Further, the signal SX is divided by the value L stored in the storage device CB in the divider DD to become the signal SMX. In other words, the multipliers MP ₀ , MP ₁ ,...,
By MP _L-1 , adder AX and multiplier DD (20)
The calculations shown in the formula are performed.

SMX=SX/L=1/L (V ² +V ² ₁ +V ² ₂ +...+V ² _L-1
) = 1/L (U ² + UU ² ₁ +...UU ² _L-1 ) - (20) This is effectively the calculation formula for distributed VR. The first term of equation (9) is calculated as signal SMX. It represents that.

On the other hand, the above signals U, UU ₁ , ..., UU _L-1 are added to the adder
It is added at AY to become signal SY, and the above signal
SY is divided by the value L stored in the storage device CB in the divider DE to become the signal AVE, and the signal AVE is further multiplied by the square in the multiplier MB to become the signal.
Become SMY. In other words, adder AY, divider DE,
Multiplier MB performs the calculation shown in equation (21).

SMY=SY ² = {1/L((U+UU ₁ +UU ₂
+...+UU _L-1 )} ² - (21) This indicates that the second term of equation (9), which is effectively the calculation formula for distributed VR, is calculated as the signal SMY.

Further, the value of SMY is subtracted from the signal SMX by a subtracter SUB, and as a result, the variance VR shown by equation (22) is calculated.

VR=SMX−SMY=1/ _{L (} ^U2 + ^UU21 + ^UU22 +... ^UU2L _-1 )−
{1/L(U+UU ₁ +UU ₂ +...+UU _L-1 )} ² - (22) This means that the variance VR shown in equation (9) has been effectively calculated, and the above signal VR is multiplied by vessel
In MC, the automatic setting threshold AH shown in equation (10) is finally calculated by multiplying by K by the constant K stored in the storage device CC.

FIG. 12 shows the configuration of the discriminator 5. In the figure, SS is the feature extraction signal obtained by the Haar transform unit 2, AH is the automatically set threshold input from the threshold calculation unit, and SB is the automatic setting threshold input from the threshold calculation unit. Subtractor, EZ
is the output of the subtracter SB, RL is a relay that generates a logical output that is 1 if the input signal is positive and 0 if it is negative or 0, and LT is a discrimination signal output from relay RL. In such a configuration, first the subtractor
The value of the automatically set threshold AH is subtracted from the feature extraction signal SS by SB to calculate the signal EZ for comparing the two, that is, EZ = SS - AH. Next, the relay RL subtracts the value of the automatic threshold AH from the feature extraction signal The sign can be checked. In other words, if the feature extraction signal SS is greater than the automatic setting threshold AH, the signal EZ becomes positive, and as a result, the output of the relay RL is a logic 1 signal, and conversely, if the signal SS is less than or equal to AH, EZ becomes negative or 0, resulting in relay RL
A logic 0 signal is obtained at the output of . In other words, whether or not the feature extraction signal SS exceeds the automatically set threshold AH is expressed by a logic signal of 1 and 0.
LT signal is required.

In the end, the magnetic anomaly automatic detector configured as described above digitizes the magnetic field measurement signal output from the high-sensitivity magnetic anomaly detector mounted on the aircraft by the A/D converter 1, and converts the digitized magnetic field measurement signal into a digital signal. Based on this, the Haar transform unit 2 generates a feature extraction signal that extracts the characteristics of the waveform induced by a shipwreck, etc., and the threshold calculation unit 4 generates a threshold signal that automatically changes the magnetic hold level according to the level of the magnetic field measurement signal. Automatically set thresholds are required.

Next, the determination unit 5 compares the magnitude of the feature extraction signal and the automatically set threshold, and if the feature extraction signal exceeds the level of the automatically set threshold, it determines that the magnetic field measurement signal is a waveform induced by a shipwreck, etc. If the logic output 1 is not exceeded, the logic output 0 is the logic output that determines that the magnetic field measurement signal is noise.
is output as the discrimination signal LT.

By driving a buzzer, lamp, etc. using the determination signal LT obtained in this way, the presence of a shipwreck or the like can be automatically detected. In this explanation, a case has been described in which an analog magnetic field measurement signal is input, but when a digital magnetic field measurement signal is input, it may be input directly to the Haar converter without going through an A/D converter. The above description is one embodiment of the present invention, and various modifications may be made without departing from the gist of the present invention.

The above explanation is about automating the detection of magnetic anomalies for searching for underwater shipwrecks, etc. using a highly sensitive magnetic anomaly detector mounted on an aircraft.
The present invention is not limited to this, but can also be used to automate the measurement of the number of automobiles, etc. using a magnetometer.

As described above, the automatic magnetic anomaly detector according to the present invention can automate the process of detecting magnetic anomalies caused by a target object, such as a sunken ship, without being affected by the magnetic field measurement level, so the detection ability depends on individual differences in the operator. This method has the advantage that it eliminates the drawbacks of non-uniformity in detection ability due to variations in the magnetic field and fatigue, and also eliminates the troublesomeness of the operator having to reset the threshold each time the level of the magnetic field measurement signal changes.

[Brief explanation of the drawing]

Figure 1 shows an example of a magnetic field measurement signal and a part of the Haar function family used for Haar transformation, and Figure 2 shows how the Haar spectrum extracts the characteristics of waveforms induced by shipwrecks etc. according to the respective spectra. Figure 3 is a diagram showing the feature extraction signal together with the magnetic field U, and Figure 4 is a diagram showing how the level of the feature extraction signal and automatic setting threshold changes as the level of the magnetic field measurement signal changes. , FIG. 5 is a diagram showing the configuration block diagram of the automatic magnetic anomaly detector according to the present invention, FIG. 6 is a diagram showing the synchronization pulse P, and FIG.
The figure shows the holding of signals by the A/D converter, Figure 8 shows the configuration of the Haar converter, Figure 9 shows the Haar converter that constitutes the Haar converter, and Figure 10 shows the Haar converter. FIG. 11 is a diagram showing the Haar spectrum feature extractor constituting the conversion section, FIG. 11 is a diagram showing the threshold calculation section, and FIG. 12 is a diagram showing the discriminating section. In the figure, 1 is an A/D converter, 2 is a Haar converter, 3 is a synchronous pulse generator, 4 is a threshold calculation unit, 5 is a discrimination unit, 6 is a Haar converter, and 7 is a Haar spectrum feature extractor. , M ₀₁ , M ₁₁ ,…,
M ₇₁ , M ₀₂ , M ₁₂ ,..., M ₇₂ , M ₀₃ , M ₁₃ , M ₂₃ ,
M ₃₃ , M ₀₄ , M ₁₄ , MD ₁ , MD ₂ ,..., MD _L-1 is a storage device, CM ₁ , CM ₂ , CM ₃ , CY ₁ , CY ₂ , CY ₃ ,
CY ₄ , CB, CC are storage devices that store constants,
AD ₁₁ , AD ₂₁ ,…AD ₄₁ , AD ₁₂ , AD ₂₂ , AD ₁₃ ,
ADD, AX, AY are adders, SD ₁₁ , SD ₂₁ ,...
SD ₄₁ , SD ₁₂ , SD ₂₂ , SD ₁₃ , SUB, SB are subtractors,
DV ₁₁ , DV ₂₁ , DV ₃₁ , DV ₄₁ , DV ₁₂ , DV ₂₂ ,
DV ₀₃ , DV ₁₃ , DD, DE are dividers, MY ₁ ,
MY ₂ , MY ₃ , MY ₄ , MY ₅ , MP ₀ , MP ₁ ,…
MP _L-1 , MB, and MC are multipliers, and RL is a relay. Note that the same or corresponding parts in the figures are indicated by the same reference numerals.

Claims (1)

[Claims] 1. A magnetic field measurement signal measured by a magnetic anomaly detector is Haar-transformed to calculate a Haar spectrum, and the calculated Haar spectrum is appropriately linearly combined and squared. a Haar transform unit that obtains a feature extraction signal, and a threshold calculation that calculates the dispersion of the magnetic field measurement signal over a certain period in the past and multiplies the dispersion obtained by the above calculation by a real number to calculate an automatic setting threshold. a synchronous pulse generator for controlling the Haar transform section and the threshold calculation section; a feature extraction signal obtained from the Haar transform section; and an automatically set threshold obtained from the threshold calculation section. 1. An automatic magnetic anomaly detector comprising: a discriminator that compares the magnetic anomalies and determines whether there is a magnetic anomaly caused by a specific target object based on the comparison result. 2 The Haar transform unit is composed of a correlation unit that calculates the correlation between the magnetic field measurement signal and a sine cosine wave, and the correlation unit calculates the magnetic field caused by a specific target by correlating the magnetic field measurement signal with the constant sine cosine wave. An automatic magnetic anomaly detector according to claim 1, which is adapted to effectively extract anomalies.