FR2778746A1 - Detection, location and identification of magnetic objects, using transducer in relative motion - Google Patents

Abstract

An observation plane W contains M locations of the magnetometric sensor and N object positions. Vector v characterizes instantaneous relative speed at instant t, between sensor and object. Coordinates Ei, Dj are defined in W. The signal SM received at each point is projected mathematically onto the basis phi . Energy identified iota i(ij), from the signal SM, is projected onto this basis for each coordinate Ei,Dj. A criterion of detection Cij is calculated for each point Ei,Dj. This is defined as the ratio of energy identified at each point iota i(ij)(ij) with the total energy iota t. The maximum value Ckl of Cij is determined, such that k and l represent coordinates Ek and Dl of the point Ei,Dj corresponding to the position of the object sought. An Independent claim is included for a computer system handling the calculations. It is defined in terms of preferred calculation precision, rate and data capacity. Preferred Features: The exact analytical representation is a system of orthonormal and orthogonal bases. Mathematical expressions are provided for dipolar and quadripolar magnetic objects; relevant parameters include terrestrial magnetic field vector, object magnetic moment and an approach ratio defined in terms of the path of relative motion.

Description

The present invention relates to a passive device enabling the

  detecting, locating and identifying objects having magnetic properties. These results are obtained by measuring the magnetic field created by these objects and the processing of the time signal (s) coming from one or two magnetometers. Current and future applications of this device are very numerous. All objects with magnetic characteristics are detectable; for example: Manufactured or natural objects of any size, containing ferrous substances, deposits of magnetic minerals or containing magnetic impurities,

  15. Clay materials cooked ...

  Objects with direct current or DC current circuits.

  The fields of application of this device are also very numerous and varied, for example: Archaeological and mining research ...

  The surveillance of the passage of vehicles, of ships ... The porters of airports, the research of wrecks, submarines, lost or buried objects on the ground or at sea, pipelines,

drilling heads ...

  The present invention relates to the detection of objects in relative motion with respect to the measuring magnetometer (s) and behaving magnetically in a substantially dipole or quadrupole manner, which is the case with sufficient precision in all cases where the distance

  between the object to be detected and the sensor or sensors is greater than 2 or 3 times the largest dimension of the object.

  Numerous magnetic object detection systems are used. For passive systems, ie not proceeding by emission of known electromagnetic field and analysis of the response of the object to this emission are prominently the systems measuring and analyzing the magnetic field

  static (natural or not) generated by wanted objects.

  Two major categories of such systems exist: - Those using directional sensor information - Those using module sensor information

magnetic field.

  The invention described in US Pat. No. 4,309,659 of January 5, 1982 is very representative of the first

category of systems.

  The author uses a set of several direct magnetometers

  ionels, that is to say sensitive to the components of the magnetic field (parasitic fields + field to be measured) on each of the measurement axes of said magnetometers and thus obtains measured values on the one hand from the components according to an XYZ reference point of the magnetic field of the object and values

  directional field gradients along the same axes.

  In the case where the object is magnetically dipolar the

  unknown to detect and identify

  The three XYZ coordinates of the object with respect to the measurement sensors and the three components Mx, My, Mz of

dipole moment M of the object.

  The author thus makes 6 measurements: 3 measurements of field Mx, My, Mz with one of the magnetometers C) è- H y ___ 30. 3 measures of guard of field - -x -èy, - r With three or four spacing magnetometers x, 6yz

known relative.

  3 - From the classical equations of the field of a dipole, lt

  6 unknowns are calculated from the 6 measurements made.

  This device is suitable for applications where the field of the object is much higher than parasitic fields of various origins. For certain applications where the search for objects takes place in the superimposed presence of the natural magnetic field of the earth, this device suffers from very important limitations due to the use of directional magnetometers, to noise, ie to temporal fluctuations. and the Earth's magnetic field and the defects

  unavoidable relative positioning of magnetometers.

  For example, the gradient of the Earth's natural field due to geological structures is usually much higher than the gradient value of a dipole object even at low

distance.

  In this case, the three measured gradients are unrelated to those created by the object to be detected and therefore lead to

  to erroneous solutions for solving equations.

  In this first category of systems it is not assumed a priori that the object to be detected is driven by a relative speed with respect to the sensors and the effect of such a possible speed is secondary for the detection and therefore, well that the principle of calculation is based on the equations of the field of dipolar objects, there is no possibility to verify that the received signals are actually those of a dipole object and not spurious signals that can be of varied origin and who are

exploited erroneously.

  In the second category the detection method is based on the measurement by module sensor (thus insensitive to its orientation) of the field of the object superimposed on the different

parasitic fields.

  -4- In this case although a set of three directional magnetometers

  oriented along the axes of a theoretically

  allow the calculation of the magnetic field modulus, it is most often called upon magnetometers to

  resonances because they are not directive.

  In this second category of systems, the magnetometer for measuring the field's module value is moved relative to the object and the detection system analyzes the evolution of the measured signal as a function of time. This time signal is the sum of the signal due to

  the object and various magnetic noises.

  Most often the detection is carried out when the signal to be measured exceeds a certain value the signals due to noise. In some cases it is also taken into account the duration of this overrun taking into account a certain

number of thresholds.

  By way of example, French Patent No. 2,106,657 discloses a

  detection method elaborated on the previous principle.

  According to this method, the measured signal is filtered in a filter battery passes contiguous strips performing a

summary spectral analysis.

  In each of these frequency bands is performed a

  signal envelope detection and a measurement of the ampli-

  ridge study. The comparison of the peak amplitudes between the envelopes of the filtered signals coming from the filters

  adjacent areas allows, through a decision logic to

construct a detection signal.

  Although its performance is far superior to the methods of the first category, the method is effective only when the target useful signal is far superior to the signals created by the parasitic magnetic noise and thus has a limited detection range -5 As before, this method does not distinguish the signals of a dipole object parasitic signals exceeding the

decision threshold.

  The objective of the invention is to significantly improve the performance of prior systems in the detection of magnetic objects at distances

  such that their magnetic appearance is dipolar.

  We will see that the invention also applies to objects of more complex magnetic form and in particular having a quadrupole behavior in practical conditions.

detection.

  By performance is meant at first order the detection range. This range is a function of the signal (object) to noise ratio (various magnetic noise). In the methods of the prior art, detection is only possible when the signal is much greater than the noise in

  a ratio of at least 2 or 3.

  The first advantage of the invention is to be able to detect dipole signals having levels much lower than the noise and thus significantly increase the detection range compared to the systems of the prior art. A second object of the invention is to allow the detection of magnetic objects using only the measurement of the

  or signals provided by one or more magneto-

  metric measuring the module of the field of the object and parasitic magnetic fields and therefore greatly facilitates the use compared to methods using directional sensors whose stability of the relative orientations of the axes of sensitivity is particularly

  difficult to insure and maintain over time.

  This is obtained, inter alia, by a new analytical formulation of the signals to be detected (dipole or quadri-polar) and having the advantages, on the one hand, of allowing the detection, but also the calculation of the location of the magnetic object without ambiguity; neither multiple solutions

  and mathematically exact.

  The invention in addition to a new adapted analytical formulation, also proposes a new method of effective resolution

  equations leading to detection and localization.

  Indeed, these equations are not likely to be solved directly given the number of unknowns and

their non-linearity.

  In particular, in the case of the use of two magnetometric measurement sensors, one of the advantages of the invention, besides the detection and the location of the magnetic object, is to allow the precise calculation of the magnetic moments of the object, allowing thus its identification in the magnetic aspect which offers the advantage of only detecting the objects whose magnetic appearance corresponds to the objects to be searched and contributes largely to the safety of detection by increasing the performances, and minimizing

false alarms.

  Another advantage of the invention is to propose a method for performing these operations in real time, it is

  to say without loss of time, nor delay.

  The invention will be better understood from the description

  next with reference to the accompanying drawings: FIG. 1A is a representation of the magnetic field created

  by a dipolar object in the space that surrounds it.

  Figure 19 shows the 3 typical shapes of the signals observed when a magnetic field module sensor

  is moving relative to the magnetic object.

  FIG. 1C represents one of the preceding signals exchanging

  tilted, ie when the measurement is made at

time intervals at t.

  FIG. ZA is the representation of the basic functions of the prior art f 0, f 1, f 2, making it possible to represent

any dipolar signal.

  FIG. 22 is the representation of the basic functions' iy / LIz according to the invention allowing not only to represent any dipole signal but also to perform

  the detection of the object and its location.

  FIG. 3 is the representation according to the invention of the choice of the selected geometric parameters, defined by the object to be detected N, and the relative velocity vector v between

  the magnetometric sensor and said object to be detected.

  FIG. 4, similar to the previous one, shows the geometrical parameters chosen in the case of the use of two

  magnetometers M1 and M2.

  FIG. 5 is the diagram explaining the method used according to the invention for obtaining the detection and location parameters of the magnetic object from the signal

  SM of a single magnetometric sensor.

  FIG. 6, analogous to the previous one, shows the diagram explaining one of the methods for obtaining the detection and localization parameters of the magnetic object at

  signals SM1 and SM2 from two magneto-

  metrics. FIG. 7 graphically represents the shape of the detection criteria obtained according to the method at times

  different before and after passage in the plane of the object.

  FIG. 8A is a representation of a simplified embodiment using only a limited number (8 for example)

  observation points in the observation plan.

  FIG. 8B shows, at two moments before and after passing through the plane of the object, the values of the detection criteria, -8- Conventionally the field of a magnetic dipole of moment

  _4 placed in O created at the point t D is carried out the measurement is -

  given by: At o (see Figure 1 A) U is the unit vector carried by

  the vector De and R is the distance OM.

  Due to the use for the measurement in M of a magnetometer

  meter measuring the module of the total magnetic field and because the field Hd in practice is always very small, in front of Ht, the signal present at the output of the magnetometer is the image of AH, projection of Hd on the earth's magnetic field Ht. When the magnetic object and the magnetic sensor are moved in relative motion, the output information of the sensor representing AtH is variable with time and is a function of the temporal variations of the parameters and relations expressed by the sensor. equation (1) in which the

  coefficient k is a function of the units chosen.

  Explicitly the AH signal measured at a given instant is a function of the following parameters: - Orientation of the terrestrial field Ht (2 angles) Orientation of the unit vector U (2 angles) - Orientation of the magnetic moment> (2 angles) - Magnetic moment module / t (1 parameter) Distance from the object to the R sensor (1 parameter)

a total of 8 parameters.

  The knowledge, a priori, of these 8 parameters makes it possible to calculate the signal AH, but conversely the measurement of ZH starting from a single equation (1) does not allow in any case to calculate the 8 unknown parameters, or even the 6 in the case where one assumes the orientation of the known terrestrial field by means for example of auxiliary directional magnetometers. Moreover, the evolution as a function of the time of aH involves two additional parameters which are the relative trajectory of the object and the sensor and their

relative speed.

  In current practice we can assimilate the relative trajectory of the sensor and the object to a straight line (trajectory

  rectilinear) and traveled at a constant speed v constant.

  This assumption is verified for almost all

  applications of the invention, furthermore the analysis and ex-

  practical experience have shown us that mistakes made using the method of invention for objects and

  current search vehicles remained of negligible impact.

  on the performance of the system.

  In this context, experience has shown that the pace of the signals measured as a function of time by the magnetometric sensor could have three typical shapes.

  represented by the curves 1,2 and 3 of figure [1B].

  Still in this hypothesis, MR. J.E. ANDERSON showed in 1949 that all magnetic signals, whatever the configuration of the object and the relative trajectory, could be expressed mathematically according to a

  linear combination of three signals or basic functions.

  Various magnetic signals are represented by changing the contribution of each of these basic functions. The mathematical representation of these basic functions is

  commonly referred to as ANOERSON functions.

  By using the dimensionless parameter u defined as the ratio of the distance E traveled along the relative trajectory and the shortest distance D of passage between the object and the sensor, the magnetometric signal SM seen by the sensor in function time is expressed by: sM = t (a) (z + uS1> + ^ utr t CT46UdC.S [- 5 SI + air, a) S ,; The three functions (3) are the basic functions of ANDERSON, the coefficients AO, A1, A2, coefficients whose value, which varies according to the case, represent the contribution. to the total SM signal of each of the basic signals WD, Figure 2-A is a graphical representation of these functions, as in Equation (1), equation (2) can represent any signal magnetic SM as a function of time once chosen all the parameters already mentioned, but does not allow more than the previous one to determine the said parameters, that is to say to perform

  localization detection from the sole know-

  of the magnetic signal SM as a function of time.

f

_ 11 _

  The analysis shows that there are in fact an infinity of possible representations of the signal SM and that we can find as many corresponding sets of three basic functions that can represent exactly SM. A first characteristic of the invention is to propose a set of three basic functions, obviously allowing to represent the signal SM from the knowledge a priori of the different parameters connecting the object, the sensor, and their respective trajectories, but especially allowing from the sole knowledge of a or two SM signals to have access to all the mentioned parameters and thus to perform the detection, the location and the determination of the characteristics

magnetic objects detected.

  The second characteristic of the invention is a new method which effectively makes it possible to start from an SM signal lk. (or 2 SM signals, and SM,) to calculate the different

  above-mentioned parameters and to perform locational

E0 tion of the object.

  The basic functions according to the invention were therefore determined

  mined and selected according to a number of criteria and characteristics necessary to be

  able to use the new detection method.

  These criteria according to the invention are as follows: The bases must be orthogonal and standardized to 1. That is to say that no basic signal is projected on the other two bases and in particular the part of the signal Sm which can be projected on each of the bases is unique which leads to the determination of unique projection coefficients (Ai) for a given signal SM. What is not the case of the ANDERSON functions for which the bases are not orthogonal there is an infinity of solutions of projection of the signal S, on these bases which leads to the impossibility of determination of the unknown parameters

who are not separated.

I.

_1 2 -

  -12-, The basic functions should be as differentiated as possible which in practice o the useful signal SM is embedded in various noises allows the precise calculation

  projections and therefore projection coefficients Ai.

  We have therefore sought to establish a set of bases called fall off decreasing, that is to say asymptotic behavior also

  as different from each other as possible.

  A third fundamental characteristic of the invention is the choice of the geometrical parameters by which the equations are expressed so as to minimize the number of unknown parameters, which made it possible to find a method making it possible in a first step to solve the problem by not having that a reduced number of unknown parameters to find. The choice of these parameters is explained in FIG. 3 in which: M and N are respectively the positions of the sensor M and and of the object N at a given instant, v is the relative velocity vector at this instant between the object and the sensor, V is the object plane as it passes through the object N and is orthogonal to the vector 7, W is the observation plane such that this plane contains

M, N and v.

  This choice of parameters is valid in all cases of respective movements of M and N and in particular applies both to the detection of a mobile object N by a fixed sensor M, a fixed object by a sensor

  mobile or a mobile object and a mobile sensor.

  In the practical case o during the observation period

  signals the relative trajectory of M and N is relative

  straight and traveled at substantially constant speed v, the straight line passing through M and parallel to v is

constant V, -

  The relative trajectory of M with respect to N and intersects the plane V at the point O. O is chosen as the origin of the coordinates, OX is parallel to v, OY is orthogonal to OX and passes through N, D is the shortest distance between M and N, E is the distance from the sensor M to the object plane V, to a

  given moment, H is the terrestrial magnetic field vector

  be a priori any in the OXY mark.

  Any temporal signal Sm coming from the magnetometric sensor and created by the magnetic / dipolar moment of the object N in its relative displacement with respect to the sensor can be written: S t. Either: SM = Ao. + Pf, + Az expression (E) E- in which according to the invention and with u =: L fl.,. e4, L) S / L

C _____

  C! = V. z_ The graphical representation of these new functions

  basic, both orthonormal and asymptomatic

  different tick is given in Figure 2-B. It can be seen that these three new functions are very well differentiated throughout the practical domain of variation of the parameter u unlike the functions of ANDERSON for which L

Yes ----

  -14- f0 and 2 are merged for u 1-1 1 and 2 are merged for u> -1 i i1 and 2 are merged for u> 0? O ,? 1 and 2 are merged for u>, 1 In the case where the minimum distance D of passage between the sensor M and the object N is no longer very large in front of the dimensions of the object, the dipole model can no longer

  be considered to represent exactly the behavior

  of the object, these new bases may be supplemented by two additional bases, representing

  the quadrupole behavior of the object.

  In this case, according to the invention, equation (4) is written: o, t ,, C, are given by 5) and, j and f, by: (4 'Ba ^ ^^^^ L ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

  different from each other and with respect to the first three.

  The set Y to C describes completely and exactly the

  signal from a quadrupole object.

  The detection method which is the subject of the invention applies equally to objects considered to be dipolar (equations 4 and 5) and to those with quadrupole behavior

(Equations 6, 5 and 7).

  In equations 4 and 6: the coefficients Ai are nonlinear functions of the parameters Ht and / t: the new basic functions 5 and 7 depend on the parameters E, D and Ht also in a non-linear manner. As a result, and having fewer equations than unknowns in the dipolar case as in the quadripolar case, the problem of the detection and localization of the object is not soluble by mathematical methods.

  because the equations are not inversely

  The first essential characteristic of the invention is the choice we make to search for the unknown object N in the plane W called the plane of observation, although it is not so linear in terms of E, D, H and. this plane is a priori unknown except that it contains the relative velocity vector v and the sensor M, this is made possible by the choice made of the parameters E and D and the reduced parameter u = E / D making it possible to express the functions of base t only according to E, and D, in the plane W. A second essential characteristic of the invention is to search for the unknown object N in certain positions chosen a priori in the plane W. For this, we define a priori in the plan W a

  number of points each defined by their coordi-

born and DJ.

  We thus establish a table of the coordinates of these points of dimension m X n such that the following: Ed Di EL D - - E D - - E. Dj

- D-ED E- D

E D EDDE E D.E Eu D

2778746 1

-16- -

  This table comprises in principle a number m X n of common coordinate points E i Det for each of which we know the numerical values of the basic functions t o1 SZ 3 from equations 5 in dipole model or functions go a1 C2 q3 q, 4 from equations 5 and '

7 in quadrupole model.

  Furthermore, according to the invention, the signal SM coming from the magnetometric sensor is sampled at equal time intervals Δt, which gives rise to the measurement (during a defined time (T) called the measuring horizon of P successive values of SM noted S, S'M .. SM and corresponding to the evolution of the signal created by the unknown object and measured by the magnetometric sensor in its displacement

  relative to the object sought.

  According to the invention, for each point EL D chosen a priori in the plane W, the following equation (8) is derived from the equation (4) of the expressions of the functions of J base C (i, of the measurement of P samples SPM MS of the SM signal of the magnetometer: 2 5 i SL x + ia q the solution of this equation is Ar. e in which 9t is the transposed matrix of -ij and the term [gT.SM] represents the correlation vector between the actual signal measured S. and the signal which would be generated by an object N arbitrarily placed at the coordinate point ECDj of the plane W, and IBI is the noise matrix. , to a close translation for all points E; D- of the table corresponding to a constant Didonne distance Equation (9) represents the solution that minimizes the error between the measured real signal and the signal of a virtual object placed in Ej, ie the error between the real signal and the signal coming from the model for an object placed at the chosen point Ej represented by the equations

(4) and (5).

  If the object sought has a four-fold

  polar, equations similar to (8) and (9) are established for each point Ej3 from the equations (6 and (5s) and (71

  the proposed method remains the same elsewhere.

  Equation (9) is called "projection operation" since it performs the mathematical projection of the real signal S on the bases tp for each of the points EL.Dj of the planes W, ie performs the identification in the signal Real SK and in sounds superimposed on this signal of the part identified as coming from a magnetic source

  dipolar (or quadrupole) placed at the point of analysis ED].

  Another essential characteristic of the invention is to call upon the detection of the location of the object N to calculate the energies of the noise and the identified signal and to calculate the detection criterion defined as the ratio of the energy identified and the total energy equal to the sum of said identified energy and the noise energy. It is the use of this energy criterion which is mathematically exact together

  with the exact analytical description of the possible signals

  from each of the ELD points] which is superior to the method which is the subject of the invention with respect to the previously known methods based on amplitude measurements, since the signal embedded in the noise i is extracted from the signal. all the information it contains and in a strictly exact way in the mathematical sense of that term. According to the invention, the detection method consists of: i - calculating the energy úE (idu real signal S. projectable on the bases of each of the ELDj points.

  & (at is the identified energy of the SM signal - Calculate the total energy of the actual signal and noise

  e on the time horizon T measure defined previously.

  - Calculate the detection criterion CLj corresponding to each ELDi point and defined according to the invention by the ratio Cj = j) Ct - Evaluate for the set of points E.Dj the maximum value] of the criterion C j which is the value Cke (k and 1 representing the coordinates in Ek and Du of the point E-Dj for which CL "is maximum) -Comparer the value of C; at a threshold v - Si CKL is greater than Extract C = CFu / ≥ confidence level of the detection 25. Extract E = And

D = DL

  which are the coordinates in the axes previously defined

  detected object in relation to the magneto-

  metric M. From the notation of the equations: 8> and 9 the identified energy C of the signal S Mcorresponding to each point EgDj is obtained by applying the following equation (10): qi -19- i (ij) = A " (10)

  o A "is the transposed matrix of A matrix of coefficients

  A. Ad Az in dipolar (or A, AA AI, A, A4 quadrupole). J5 8 (/ could also be calculated by: i (ij) = Ac + AA + AD or i (ij) = Ad + A + An A & + A4 since the bases used are orthonormal, but with a fundamental error due to The effect of filtering and the fact that the noise superimposed on the signal in the measurement frequency band may not be perfectly white for certain application cases. The total energy and sum L [of the energy of the real signal possible and the energy of the noises is obtained by (12) l_ & t = .M

201.:(12)

S

  + (s f + .... (SM) Each value of Sm being obtained by sampling the magnetometer signal at the rate nt over the measurement horizon T and corresponding to P successive samples

with P = T / at.

  As indicated above, the information provided by the method that is the subject of the invention when a single measurement sensor is used is: parameter C, level of

  confidence of the detection, the parameters E and D of

  Magnetic object with respect to the sensor M. E is the distance (at a given moment) between the sensor

  magnetometric and object plane V which is thus complete

  defined, D is in this plane the distance between the point 0 (known as the intersection of the plane V and the relative trajectory object sensor defined by the axis of the

  v. velocity vector) and the detected object.

  The position of the object thus detected is known only by a location of position which is the circle of center 0 and radius D, or at least a fraction of a circle for all applications. In practice the use of a single sensor does not allow, given the physical phenomena involved, a complete location of the objects to be detected, or a determination of their magnetic characteristics.

  On the other hand, this explains why, given the particular choice

  parameters used for the equation

  signals and basic functions, a choice that eliminates

  Some localization parameters allow the partial resolution of the localization problem by

applying the proposed method.

  Despite an incomplete localization of the magnetic object detected, the method described above is sufficient especially in applications where a position of the desired object (s) is known a priori; this is for example the case when looking for objects

  resting on the ground or at the bottom of the sea o slightly buried.

  In these cases the search is carried out by moving the magnetometer aboard a submerged fish towed by a boat, a land vehicle or an airplane. Since the trajectory of the magnetometer is substantially horizontal, it is known first of all the altitude of the magnetometer with respect to the ground or with respect to the seabed and the position of the magnetometer.

  the detected object is determined by

  section of the ground plane or the bottom of the sea and the circle

  position determined by the proposed method.

I 2778746

  In most cases this intersection is composed of two points symmetrical with respect to the trajectory of the sensor. The raising of doubt between these 2 positions

  possible of the object is, if necessary, obtained by

  killing two successive passes for substantially orthogonal velocity vector orientations v. We thus obtain 4 possible position points (2 per passage) of which two (1 per passage) are combined, thus giving

  the unambiguous position of the detected object.

  In other cases, it is necessary: either to be able to carry out a detection and a complete localization of the object without knowledge a priori of a first place of position (horizontal plane of the ground) it is the case for example

  search for objects that can be immersed in deep

  various objects, or to identify the objects detected by the evaluation of their magnetic characteristics. This second case is for example that of the search for specific objects in a port, whose bottom is congested with

  various and uninteresting magnetic debris and among

  what it is necessary to identify a particular object through the identification of particular magnetic features. In all cases, the method described above makes it possible to respond to the problems posed by the use of a pair of magnetometers MA and ML in place of the magnetometer.

  unique M used so far.

  Figure (4) is a representation of the geo-

  metrics chosen for the use of two magnetometric sensors MD and M. M4 and ML are for example placed at the ends of the wings in the

case of search by plane.

  In this example, the relative velocity v between the sensors and the object sought is substantially the speed of the aircraft and the direction MA, Mt is orthogonal to the vector v -22-, as previously called object plane (unknown a priori)

  the plane (V) passing through the object N and orthogonal to the vector v.

  In a manner similar to the preceding description and in accordance

  In accordance with the invention, we choose two observation plans

* Wation corresponding to each of the signals of the sensors

magnetometric MA and MI.

  The plane W ,,, observation plane of the signals of the sensor MH.

  is such that it contains the point M4, the vector v and the object N. Similarly the plane Wi, signal observation plane of the sensor Ml contains the point MI, the vector vet the object N. The points M 'A and M'I are the intersection with the object plane V of the trajectory relative to the object of the

  Magnetometric sensors MA and Mt.

  Another essential originality of the invention consists in this case of using two sensors to be applied simultaneously and independently to each of the signals issued J

  MA and MI sensors the method described above.

  The process applied to the Mo sensor signal leads to - a CA detection criterion - an EA distance estimate - an DA distance estimate

  likewise for the signal coming from M, we obtain the para-

  meters C, E; D. As the principle distance MA MA = d separating the magnetometric sensors is known, the triangle M, M'b N located in the plane object is completely determined by the length of its three sides Dl Db and d, therefore the position of the object N in this plane with respect to MA M '; in the same way, the distance Er = Es between this plane and the sensors M., and M. is known at each computation cycle. The complete position of the object N is thus obtained by

  compared to magnetometric sensors.

  In the case already cited of an airborne search of objects N, the sensors M4 and Mt are placed at the end of the wings so as to have a distance d not insignificant compared to the detection distances D. and D. In this case, the trajectory of the sensors Mo and M, is parallel to that of the aircraft and substantially horizontal, as well as the line joining MA and ML as well as MH and M 'The object plane V - orthogonal to v speed of the plane is

a vertical plane.

  We can thus use a cylindrical coordinate system to express the position of the object N with respect to the center M of the plane (M middle of M, Mt) and to the vertical Z. X = Eg = Ez = E is the horizontal distance between N and

the plane at a given moment.

  D = oblique distance between N and the plane (shorter

distance of passage).

  = elevation angle of N with respect to the vertical.

  These cylindrical coordinates are obtained on the one hand by X = EA = Ez, on the other hand by the resolution of the triangle M '> M' N from the calculated parameters DA and D_ and the prior knowledge of the distance d separating the magnetometers MA and Mt That is to say: X = Es = Ez D = V 2 (D + D) - dP (13) = 2 Arctg | (P-DA) (P-Ds) P (P-d) with P = (DA + D: + d) -24- Another feature of the invention is to allow the determination of the magnetic characteristics of the detected object, when its complete position has been determined according to the previously described method using two Magnetometric sensors MA and Mt. According to the invention, the position of the object in the plane W (or the planes W4 and Wt) is obtained for the point of the plane W 1 (W.1, W) for which the criterion of Ckl detection as

previously defined is maximum.

  For this point, we have already calculated according to equation (9) the matrix Akl = A * A A tk1 so we know the value of the three coefficients Am A A; for projecting the signal measured on the bases, which are known as corresponding to the best estimate of the position in W of the detected object N whose moment J

Dipolar magnetic is / t.

  The coefficients Ao As Aiconnus now are functions

  only parameters \ 3 and or parameters Ht and -

  according to the analytical representation chosen.

  In particular, the analytical expression of the coefficients Ao AA A, is linear as a function of / e - which allows the computation of / û., All the other parameters being now

determined.

  If we call x, y, z a so-called square torsorrequestangle of measurement defined as follows: x is the direction of the vector velocity relative v y is the orthogonal direction along the axis of the sensors MA M_

  z is orthogonal to the previous ones in the forward direction.

  It will be noted that in the case of airborne research with Mo M sensors; at the end of wings, the provenre xyz is then the aircraft with: x longitudinal axis y with transverse (to the right) z normal axis (towards the "bottom")] one is able to determine the dipolar magnetic moment of the object N by its three components / x, y,, z such that and all calculations made, from equations (1) and (8) according to equation (14): tx = D3t h _ $; h At td 4.1 = - 3 h to 11) * - tt -k2xhh - zfr2-i.t1. Ad (14)

-

  3.x T, an expression in which: 20. D = DA or D, depending on whether we use the coefficients Ao AA At corresponding to Cllobtained for the observation plane WA or Wl h, h3 are the components according to the temperature of measurement of the local terrestrial field known a priori or measured for example by a directional fluxgate magnetometer triaxial along said axes of the measurement test. We now describe examples of embodiments of devices for detecting, locating and identifying magnetic objects using processes.

new described above.

  Given the volume and complexity of the calculations according to the method, any application of this method uses a digital calculator more or less fast and powerful according to the objectives sought for each application, so that, as examples , for the search of objects made from a mobile (aircraft, boat ...) a suitable calculator is constituted by the association of a calculator of the mark "DIGITAL" type PDP 11-34 and a Array Processor type MAP 200 of the brand

C.S.P.I.

  In the case of fixed sensors detecting moving objects, a processor of the brand "DIGITAL" type VAX 8530 with

  floating co-processor is very suitable.

  Moreover, each type of calculator uses programs and tools for program development that are specific to it and therefore lead to application algorithms that will be specific to it, and that will also depend on habits or the more or less

great skill of the programmer.

  We will not develop here a program that would only be usable on a type of calculator, but we will give all the necessary information allowing the person skilled in the art, the choice of calculator adapted to a particular application of the detection method, as well as the organization and the realization

  different stages of the calculation and the value of the para-

  process-specific meters for different applications allowing it to achieve its

practical application.

  We will describe the choice of the basic parameters, the organization and sequencing of the successive operations for two applications by way of example: - search of object starting from an aircraft, - search of object starting from of an underwater fish

towed by a boat.

  We will show how to apply in the two previous examples the method to the use of one or two magnetometric sensors and the influence of the choice of the number of points ELi in the observation plane on the necessary computing capacity and its influence. on the accuracy of localization performance of the detected object. The first parameter to be defined is the observation time horizon T. Since the signal SM is a temporal signal linked to the

  relative displacement sensor / object, the energy of this signal r "" increases with the observation time.

  t Since the object is dipolar, an observation horizon is chosen such that the ratio of the following observation distance E to the minimum distance D of passage between the object and the target is equal to 5 makes it possible to acquire a duration signal such that the acquired energy is about 95% of the total energy of the signal which is more than enough

in practice.

  For an airborne application for which the optimal detection distance chosen is 500 meters, the observation horizon will be 500 X 5 = 2500 meters following the trajectory of the aircraft, for a typical speed of 100 m / s

  this horizon is therefore T = 25 seconds.

  For this same speed of 100 m / s if we choose a

  no sampling of 20 meters of the corresponding SM signal

  to the accuracy of the location calculation according to E L

-2 8 -

  the sampling rate of Sm will be 1 / t = ú-- = 5 per second, ie a time step

  Sampling At = 0.2 seconds.

  The number of samples p on which the signal will be examined will be p = 1 /, t x T = 5 x 25 = 125 In the same way from the selected ratio E / D = 5,

  the detection of objects placed at the bottom of the water by a magneto-

  meter aboard a fish towed by a boat at a speed of 10 knots or about 5 m / s for an optimal range of 25 meters with a resolution of 1 meter leads to the identical choice of these parameters is: ft = 0.2 s T = 25 sp = 125 samples By optimal scope is meant here the range for which the probability of detection will be 100 ', which is obviously not the limit of scope of the process which we will see later is limited only by the threshold t

  defined above which is usable in practice.

  The diagram of Figure 5 explains the succession and the organization of the steps of application of the method

  proposed in the case of the use of a magneto-

single metric.

  H is the clock defining the complete cycle of calculation and

  is derived from the base clock of the calculator.

  For both examples, its period t is

0.2 seconds.

  (1) is the input element of the calculator performing

  the acquisition of the SM signal from the magnetometer.

  In the case where this signal is analog, (1) is an encoder sampler which delivers a measurement of SM in digital form at the rate of the clock H is every 0.2 s. This signal is transferred into the buffer M -29- which contains in our examples p = 125 successive samples of the signal St.A each clock cycle H, a new sample is acquired and the oldest sample

is eliminated.

  F1 is a digital filter designed to improve the signal-to-noise ratio of the detector by eliminating noise in frequency bands outside the

  band containing the desired signals.

  In our examples FA is a bandpass filter of order 4, whose cutoff frequencies are respectively 0.01 Hz for the high pass filter and 0.3 Hz for the filter

pass low.

  Sç is the digital signal Se filtered in the above-mentioned frequency band. S is calculated at each clock cycle H, ie every 0.2 seconds. According to the method, this SF signal is "compared" every 0.2 seconds to the theoretical signals from a certain number of EcDi points.

the observation plan.

  For this, the function (2) generates the basic functions corresponding to each of the chosen points EjDi. These signals are calculated from equations 5 in the case where it is limited to an identification of the only dipole signals, which proves to be sufficient in the case of the chosen application examples. Practically, this calculation of the values of the three basic functions ,, ,, is made in real time by the computer from :, E / Dj with D = D every 0.2 seconds for E E values or well is from a precalculated table and in this case the function (2) is reduced to an array of numerical values of the three basic functions stored for each

  point E, Dj chosen in the observation plan.

  The basic signals are generated by reading the numerical values of the functions. t successively in each line Dj 'and for the successive values of

EA

-30-

- 30 -

  This reading in Table 2 is done at the rate

  of the clock, and thus in synchronism with the

  the SM magnetometer signal every 0.2 seconds.

  We therefore have, every 0.2 seconds, basic functions @, C ^ each represented by a number p

  of samples (p = 125) for each of the chosen points E Dj.

  As for the SM signal, each of the basic functions t.tqEt of each point EL% is according to the invention filtered by a digital filter Fo identical to the filter FA described above. In practice, depending on the structure of the computer used and its calculation speed, the filtering operation of the basic functions is performed in series (successive passages of the different basic signals sampled in a single filter) or in parallel (each signal of base is processed by a specific filter)

  or a combination of both serial and parallel methods.

  At the end of these operations, we now have two tables in memory of the calculator representing two of the three terms of equation (8). We can therefore calculate the third term which is for each of the points E, D; the table of A, coefficient of i projection of the signal S r on the basic functions of

each point Eg Dj.

  Function 3 is the projection operator which calculates the Ajs from equation (9), and tables in memory of the sampled SM signal and the basic functions of each EL point Dj For each of the values of i and j , we carry out the calculation described by the equation (') and thus obtain a table of the coefficients A> which like the tables

  previous is obtained every 0.2 seconds.

  The operator 5 performs the calculation, according to the invention, of

  the energy identified in the signal ie project-

  able on the basic functions. This operator calculates the numerical values of equation (10) for each of the bridges E; D, ie calculates and establishes a

table every 0.2 seconds.

  In parallel with the function 5, the function 4 calculates the total energy E of the p samples of the signal Sr coming from

  of the magnetometer by equation (12) every 0.2 s.

  The operator 6 calculates for each of the points E j, the ratio of the energies 9 and previous and 56 and thus establishes the table CZj detection criteria report & O j / c. As before, such a table is obtained

every 0.2 seconds.

  In a variant and depending on the nature of the noise, the detection criterion may advantageously be defined by C = t &, being the residual energy, that is, the energy of the noise obtained in the absence of detection. In this case the operator 4 is blocked by the presence of the detection signal C i and retains the value = & obtained previously at the detection and for the duration

  during which C remains greater than or equal to.

  The operator 7 compares each of the values of the criteria C; J preceding the threshold g and retains only the criteria

  higher than x now called table of Ct.

  The operator 8, compares two by two each of the previous criteria CtE and retains only the criterion CAt max, that is to say whose numerical value is the highest in the

previous table.

  This result is obtained every 0.2 seconds.

  The information extracted from these tables are: - information of presence of at least one coefficient CtZ in the table of Cl! which is the signal of detection, it is a logic signal used for example to alert an operator of the detection, the numerical value of (Ct) maximum, the values of E and and corresponding to (E_ DL) which are the level of confidence information of the detection and location of the object with respect to the magnetometer. In FIG. 5, the set of functions performed at

  from function 2 is denoted by 11.

  FIG. 6 represents the succession and the organization of the operations of an exemplary embodiment of the invention in the case of the use of two measurement sensors SM. and Sr, leading to a complete localization and measurement of the magnetic characteristics of the detected object. Like the previous one, this application of the invention applies to the two exemplary embodiments for airborne detection or underwater objects by

a fish towed by a boat.

  The numerical values chosen from the different parameters

are the same as before.

  The clock H which defines the synchronization of the operations

  calculating the pilot application now the sample

  signal Sm, and SM, by the two encoder samplers 1, the transfer of each of these signals into

  the buffers respectively MA and ML.

  Each of the signals M., and MI now discretized and digitized is filtered as before by two filters FA and FL and is then processed by an operator 11 which groups

  all the functions described above.

  As before, each of the operators 11 delivers, at the rate of 0.2 seconds, the information C E D and the array of (Ai), j for each of the signals Sm, and Se. respectively Cú E. DA (A & L) A and Ct E D; (A L), the function 12, groups the information CA and Ci confidence level of the detection provided independently by

  the detection chains Sm, and S ,.

  As needed, 12 is achieved by a "or" gate, in which case the probability of detection is maximized.

  by a "and" door, which minimizes false alarms.

  In the first case a single detection S, ^ or Smest sufficient for 12 delivers the detection signal C In the second case, the two detections must be

  present simultaneously so that 12 delivers the signal.

  In the same way, the numerical value of the criterion confidence level C of the detection transmitted by 12 can for the same reasons be: either the largest or the smallest of CA or Ca The function 13 receives the numerical estimation information EA and Eb detection distances of the object from the two detection chains. Normally EA-_ E. 13 compares E, and EX supplied with each cycle of

  calculates every 0.2 s and if E, And transmits the information

  E = EA E, otherwise, for the purpose of minimizing

  bet false alarms transmits an inhibition signal

  prohibiting the generation of the detection signal.

  Function 14 receives every 0.2 s both informa-

  oblique distance D. and DL from the two detection chains and calculates the localization parameters

  D and 5 in cylindrical coordinates by equations 13.

  the operator 14 performs this calculation every 0.2 s.

  The operator 15 performs the estimation of the detected magnetic object, that is to say the calculation of its dipole magnetic momentAt from the following information now available: - D (DA or D) - The sets of projection coefficients on the basic functions Ao, A ,, At, for the point E = Ek, D = D o is located the detected object (A &), or A (h,) The measurements hx hy hz of the components following the Measuring range provided for example by the 3-axis magnetometer (16) The operator 15 from the latter information and one of the sets of parameters DA (A AL) A or Di (A ht :) numerically computes the equation 14 whose all terms are known, and provides the value of the dipole moment of

  the object M by its three components / A x74y Az.

  The device described thus provides the user with a fully automated signal - a discrete detection alert signal, - the confidence level of the detection C, - the distance to the instant object plane E, the distance corresponding to the point of smaller distance D. When two magnetometers are used, the device additionally provides: the complete relative coordinates of the object E, D, fR - the measurement of the dipole magnetic moment vector

(or quadrupole) of the object.

  This information is provided in real time, ie with a maximum delay of 0.2 s in the case of our two

examples of application.

  This information is available in the computing device every 0.2 s and is transmitted visually to the operator in a conventional manner via a coupling member of the computer to a display device by a

  Dus digital data transmission.

  Depending on the characteristics of the computer used, the parameters not described so far can vary to a very large extent and affect the final performance obtained. For example, we

  Here are two examples of how to use calcula-

  different power levels for each of the two proposed applications: airborne and

  underwater detection by towed fish.

  In the first example we use a calculator

  very fast capable of 5 million arithmetic operations

  per second in floating point on words of

  32 bits (24 bits of mantissa, 8 bits of exponent).

  This computing capacity allows us to choose a large number of E. D points; a priori in the observation plan

W (or both W and W planes).

  The choice was made on 125 abscissae following E and 6 ordered according to D ie a total of 750 points

El Dj distributed evenly.

  This corresponds to a total detection horizon of 25 s (the computation cycle is 0.2 seconds) which we distribute 5 seconds before passing into the target plane and 20 seconds after this passage. For the example of the airborne detection, the ordinates (D, to De) of the points Es Di are chosen equal to 50, 100, 200, 400, 800,

1600 meters.

  In this example, the basic functions corresponding to

  each point EL Di are calculated as described above.

at each calculation cycle.

  The criteria C tabs also contain 750 numeric values. The power of this calculator allows to manage and visualize on cathode ray tube these tables

  criteria as shown in Figure 6.

  According to the method, the visualization is performed in the plane W defined by the directions E and O. The calculator from the criteria table C; j, conventionally performs the interpolations allowing

  visualize isocritical curves.

  Figure 7a shows an example of the isocritical curves obtained in the detection of an object located at Di = 250

  meters. Four isocritical curves are obtained corre-

  spawning at four energy levels identified equal to once, twice, 4 times and 8 times the threshold. The figure is plotted for the point EA located substantially 2.5 seconds before the passage of the aircraft in the object plane, ie at a distance of about 250 m before

the object plane.

  Figure 7b shows the detection of the same object about 12 seconds after the passage of the aircraft in the object plane. We find that the energy level identified has increased by 4 levels of criteria and that in particular the estimate of Dj is very precise and improves with

the time of acquisition of the signal.

  Under the above conditions, the accuracy of localisa-

  is approximately 20 meters in E and 3% C, ie less than

  m in D at the observation distance Dmax = 1600 meters.

  Moreover, the detection threshold X in this example is of the order of 0.2, that is to say that the device is capable of detecting objects with a signal energy of almost zero false alarms. 20% of the noise energy which results in a gain in range of about 3 compared to the methods of the prior art based on the comparison of the amplitude of the noise.

  signal and noise with respect to a threshold.

  In a second realization, we used a

  32-bit floating-point calculator with 600,000 operating capacity

arithmetic calculations per second.

  For the purposes of the invention, the necessary memory capacities are: RAM approximately 1000 words of 32 bits Program 1000 words of 32 bits Constants 3500 words of 32 bits In this case we have chosen a number limited to 8 points EL DJ Moreover, the basic functions corresponding to each of these 8 points are not calculated, but precalculated

and stored.

  the EL Dj points are divided into: Ed = -2 seconds, ie 200 m before object plane Et = +5 seconds, ie 500 m after passing through the object plane 'for the 4 distances DA to D4 equal to 100, 200, 400 and

800 meters.

  We thus obtain every 0.2 seconds a CAA CA coefficient table. C11 C14 corresponding to E. and C :. C1 CL Ct4 corresponding to Es, as indicated in figure Sa in the obstruction plan W.

  Due to the small number of points in D, it is necessary to

  area to perform, when there is detection, ie

  when a criterion at least Cj is greater than U

  kill an interpolar in O and C; as shown in figure ab.

  In this figure, the four values of CA, CL C, and C> are placed. corresponding to Da D coordinates; Di D4 for

the abscissa E4.

  The value D = Di corresponding to the maximum CA max of the curve is determined by quadratic interpolar interpolation.

  through the 4 points above.

  CA max is the confidence level of detection and DZX is the shortest distance between object and magnetometer sensor.

L ----

  FIG. Bc represents the detection of the same object obtained at the instant t = + 5s, ie 5s after the passage

in the object plane.

  The total energy of the identified signal having increased, the numerical value of the criteria corresponding to El, is C; 4 C; C; C, are much greater than their values obtained for E. In the same way, the maximum of the quadric passing through the 4 points of coordinates D. Ci, Dl Cu, Dl Cz, D, C4 gives the estimate D ,, of the following distance D of the detected object and the final confidence level Czmax of

detection.

  In this example, the precisions obtained are of the order of 0.4 seconds, ie 40 meters in E 15% or maximum 120 meters in D.

  for the estimated maximum distance of 800 m.

  Although, slightly degraded by comparison with the previous example, the method allows the use with almost zero false alarms of Y detection thresholds of the order of 0.6, which remains largely superior to the methods of the invention. prior art with a detection range

substantially doubled.

-39-

Claims (19)

  1. A method for detecting, locating and identifying magnetic objects comprising at least one magnetometric sensor delivering an electrical signal Sm whose temporal variation is generated by the magnetic moment (s) of the object sought in its relative movement with respect to the magnetometer and characterized by the fact that: - an observation plane W is defined such that this plane   contains M position of the magnetometric sensor, N posi-   the object sought, v velocity vector relative to   an instant t between said sensor and said object.   in this plane W, a set of points of current coordinates EL D. is defined - it is projected mathematically for each point   of coordinates E. D-, the signal S. received on bases.   - the identified energy + of the signal SM is calculated   projectable on the basis of each of the coordination points data E. Dj.   a detection criterion Cij is calculated corresponding to   at each point Ez D and defined as being the ratio of the energy of the considered point ú to the total energy E - the maximum value C k of the criterion C] j is sought,   which is such that k and 1 represent the coordinates Ek and.   D i of the point E D 'corresponding to the position of the sought object. The method of claim 1, wherein the exact analytic representation is a base game orthonormal and orthogonal.   3. The method of claim 1 wherein the object sought having a dipolar behavior, the bases are expressed by the expressions:   = 2 41g Q- with S ,: SO o - o '/ 1   in which the coefficients A.o A. Ab are functions   nonlinear effects of the terrestrial magnetic field vector Ht and the magnetic moment / of the searched object and in which the parameter u is defined as the ratio of the distance traveled along the relative trajectory and the shortest distance of passage between the object   searched and the corresponding magnetometric sensor.   4. The method according to claim 1, wherein, the sought object having a quadrupole behavior, the bases are expressed by the expressions: U = tW X / 2 '% = 8 ffi'? 5/2 =.: 5 (2) $ f. "XZ + u. CA /; + 1 av2c 42, f AQC. + Fl 3 P -41- in which the coefficients A0 A , At A-. A are non-linear functions of the magnetic field vector   earth and the magnetic moment.2r of the object sought.   where the parameter u is defined as the ratio of the distance traveled along the relative trajectory and the shortest distance between   the object sought and the magnetometric sensor corresponding to   laying. 5. A method according to claim 1 wherein the detection criterion C- - for each EL point Du is calculated by the expressions: ELe = P   o: A 4T is the transposed matrix of the matrix A Sj coefficients Aj.   eTest the matrix transposed from the matrix t from the values of the bases 9 p is the number of samples used in the measurement horizon T.   6. Method according to one of claims 1, 2 and 5   in which we use the unique parameter u = E: ratio of the distance traveled according to the relative trajectory and the shortest distance of passage between the sensor   magnetometric and the object sought. -42-   7. Process according to one of Claims 1 to 6   in which the probability of presence of   the object sought in the plan W containing the desired object   the magnetometric sensor and the relative velocity vector between said object and said sensor, for each   points of this plan chosen a priori.   8. Process according to one of Claims 1 to   in which the measurement signals and the basic signals from the analytical model for each of the chosen points   in the plane W are filtered identically.   9. Process according to one of claims 1 to 8   in which the comparison of the measured signal and the signals from the analytical model corresponding to each estimation point is performed in real time at each instant-t   defined by the H sampling and projection clock. analytic.   10. Process according to one of claims 1 to 9   wherein the criterion CLj is compared with a threshold and triggers a detection indication if the value of said criterion is greater than the value of said threshold; this being done simultaneously for each instant, sampling and for each of the points of the plane W.   11. Process according to one of claims 1 to 10   in which the confidence level of the   detection as the ratio of the maximum value C i. criteria C- and the threshold,   12. Process according to one of claims I to 11   in which a reduced number of points of the observation plane W are used.   13. Process according to one of Claims 1 to 12   in which two magnetometric sensors M, and M_ spaced   each other of a known distance from one another simul-   the measurements of the temporal signals S and St of the searched object N, the complete localization of this object being obtained by solving the triangle D of D and D.   being the shortest distance between   the M- and M sensors and the object sought in the   observation plane W1 and W, of each of the two sensors.   14. The method of claim 13 wherein the complete knowledge of the location of the object sought is used to calculate the magnetic moment of said object from the values of the projection coefficients A Qj following an expression of the type: ['8 3 h (y + bah) * h3 _kh2 _h.   expression in which: D = D or De depending on whether the coefficients A, AA A_ corresponding to C are used for the observation plane WA or W. 25. h, h, h3 are the components according to the measurement principle of the local terrestrial field Ht known a priori or measured for example by a directional fluxgate magnetometer triaxial along said axes of the measurement measure.   15. Process according to one of Claims 1 to 14   in which the detection criterion X is calculated from   energy and noise measured in the absence of detection.   -44- 16. A system characterized by the use of a digital computer comprising an arithmetic unit, a program memory and a data memory, input and output devices, the calculation program leading to the detection, the localization and the evaluation of the object being arranged to apply the method of one Claims 1 to 15.   17. System according to claim 16 wherein the calculator has a capacity of calculating 5,000,000 arithmetic operations per second floating point on words of 32 bits, the number of points chosen in each of the observation planes W, WA, W, is around from 750.   18. System according to claim 16 wherein the calculator has a calculation capacity of 800,000 arithmetic operations per second, the number of points chosen in each of the observation planes W, W4, Wt. being of the order of 8 to 20.   The system of claim 16 wherein the input members (1) are 12-bit encoder samplers operating at the rate of 5 acquisitions per about second.