DE102010019936B4 - Method for qualifying an object and device for carrying out the method - Google Patents

Abstract

Method for qualifying an object in terms of a classification with a number N of classes (B1...Bi...BN), where N > 1, based on a number M of properties (E1...Ej...EM) of the object , where M > 0, with:for each property (Ej),detecting at least one property value (ej),for each class (Bi),determining a qualification value (q1...qi...qN) of the object taking into account each detected property value (ej),wherein qualification values (qi, qi') for at least two classes (Bi, Bi') of the classification are non-zero.

Description

In one respect, the present invention relates to a method for qualifying an object, in particular a flying object, a projectile, a vehicle and/or a person, with regard to a classification with at least a first class and a second class, based on at least one first property of the object. In a further aspect, the present invention relates to a computer program for performing a qualification of the object. In a further aspect, the present invention relates to a device for qualifying the object.

DE 197 31 111 A1 discloses a method for identifying and classifying a target. The target determined by means of physical detectors is recorded in a vector space. In the vector space, a target identification vector associated with the target is decomposed into a main tendency identity vector and a contradiction vector. The magnitude of the main tendency identity vector is weighted and then compared to a threshold. The target is identified or classified depending on the result of the comparison.

The publication "Comprehensive Approach to Improve Identification Capabilities" (RTO IST Symposium on "New Information Processing Techniques for Military Systems", Istanbul, Turkey, 9-11 October 2000; RTO MP-049) discloses an identification method in which data are combined, to determine probability vectors and risks.

A qualification of an observed object or an appropriate assignment of an observed flying object to an alliance based on incomplete and/or uncertain and/or falsified information can be based on processing of existing information, in particular a merging of individual pieces of information.

DE 10 2009 035 377 A1 (published February 3, 2011) discloses a facility for transforming object classification results. The device has a converter module that transfers an object classification result into a target probability vector. The object classification result of a classifier is in the form of a probability vector based on a first set of object classes. The device also has a configuration module into which a set of object classes and a linking specification for linking the set of object classes with a further set of object classes can be read.

DE 102 06 253 A1 relates to a unified evaluation of different feature dimensions within sensor data. For class A, measured values x are used as a sample for "object present", for class B measured values x are used as a sample for "object not present". The samples provide information about conditional probabilities P(x|A) and P(x|B). An equation provides the information on the probability that the object is present at the measurement position if a measured value x was determined there.

DE 10 2007 021 578 A1 relates to a method for classifying at least one traffic sign and discloses processing of image data from a plurality of images in an image sequence, with an object being determined. The picture elements that can be assigned to this object are determined for each picture. An individual probability is determined for a specific traffic sign. Based on the ascertained individual probabilities, an overall probability is ascertained that the object is the specific traffic sign.

In the DE 10 2009 009 571 A1 (published on August 26, 2010) describes a method for identifying and classifying an object. A number of different physical features are derived from an output signal for the object, and the object is assigned to at least one of N predetermined base classes on the basis of the derived physical features.

the DE 10 2008 005 099 A1 describes a classification module for classifying combat objects and an assessment module which is designed to assess the situation of a combat facility and/or a protected object by linking the data of the classified combat objects. Merging the evaluation of a defense situation and an attack situation will be described.

So far, no method for qualifying an object based on property values has been known that allows a decision to be made about a class to which the object is to be assigned, knowing that there is uncertainty about the correctness of this decision on the basis of the qualification.

The invention is defined in independent claim 1. The dependent method claims define embodiments of the method according to the invention. The other dependent claims define further embodiments of the invention, in particular as a computer program product and/or as an apparatus.

The method according to the invention is intended for qualifying an object. In particular, the method is suitable for qualifying a flying object, a projectile, a vehicle and/or a person. According to the invention, the qualification is based on a number M of properties E _j , j=1 the method captures property values e _j . For example, only one property is defined, eg: E ₁ = color blue.

In particular, in the case of application of an embodiment of the method according to the invention to a qualification or identification of an observed aircraft or other flying object with regard to the membership of possible alliances according to applied classification, for example, the behavior in the airspace and the type of aircraft observed can provide information regarding the alliance membership of the aircraft and insofar be used for the qualification according to the procedure. Examples of identification-relevant properties in the aforementioned application can be found, for example, in STANAG 4162, Annex A.

The object is qualified with regard to a classification with a number N of classes B _i , i=1 . . . N, ie the classification has at least a first class B ₁ and a second class B ₂ . In one embodiment, the object actually belongs to at least one class B, the classes B ₁ ... B _N , and in one embodiment only to class B, the classification. For example, the classification has three classes, defined in one embodiment as follows: B ₁ = friend, B ₂ = neutral, B ₃ = enemy.

According to the invention, a property value e _j is recorded for the property E _j on the object. In one embodiment, the object is examined for the expression or presence of the property E _j to obtain the property value e _j representing the presence or absence of the property E _j on the object. For example, in one embodiment, the property value e _j is binary and can only assume the values 1 (one) or 0 (zero). The value 1 (one) means, for example, that the property was recorded or found on the object; For example, the statement can be derived from a sensor signal that the object has the property and the property was thus observed on the object. Furthermore, the value 0 (zero) in this example means that the property on the object was not recorded and therefore not observed.

According to the invention, an assigned qualification value q ₁ . . . q _N of the object is determined for each class B ₁ . . . B _N .

As used herein, the phrase "for each class" or "for each property" denotes an application or use of the particular step or thing for more than just one class or property, as the context requires. Depending on the embodiment, it may be necessary in exceptional cases for not all classes or properties to be recorded. If, due to the circumstances of the individual case, not all classes or properties can be recorded, a neutral value of 0 (zero) or 1 (one) can be assumed for the classes or properties that are not to be recorded, depending on which value the result as a whole does not have influenced. For example, if the method forms a sum of values that are each assigned to a class, then the values that are not to be recorded as a summand are each 0; if the method forms a product of values that are each assigned to a class, the values that are not to be recorded as a factor are each 1.

In one embodiment, the method operates according to the principles of probability theory. The qualification value q assigned to a class B _i indicates a degree of conviction or certainty that the observed object belongs to class B _i .

In one embodiment, the respective qualification value q is determined taking into account the detected property value e _j of the at least one property E _j .

According to the invention, at least two qualification values q _i , q _i′ , i<>i′, which are each assigned to a class B _i or B _i′ , are not equal to zero. In other words, according to the invention, the qualification is carried out in such a way that the object is not qualified as being able to be assigned to only one class B _i with absolute certainty.

The method assigns a qualification value q i of the object to each of the classes B _i of the classification. In one embodiment with regard to the same object, the method does not necessarily always have to come to the same result; ie the method can, for example, assign different qualification values q i to the same object for this object of the same class B _i in successive applications. For example, this can be due to the fact that at least one of the recorded property values e' _j in the second application differs from the property value e _j recorded in the first application. Thus, the qualification value q _i can depend on time and/or other circumstances.

Each qualification value q _i can take a value from 0 (zero) to 1 (one), but not exactly 1 (one). The qualification value q represents the trustworthiness of a statement according to which the object is to be assigned or belongs to class B _i and no other class B _i' with i<>i'. Similar to a probability, the qualification value q _i thus indicates the extent to which the assignment of the object to class B _i can be relied on to match an actual assignment of the object.

If the method is used, the actual assignment of the object will generally be unknown to the user. Thus, several qualification values q offer the user of the method an aid, which the user can use to make a decision about the assignment of the object to one of the classes. In particular, the totality of qualification values q ₁ . . . q _N offers this help.

In one embodiment, the method comprises a decision step, after which the method decides a presumptive assignment of the object to only one of the classes B ₁ ... B _N on the basis of the plurality of qualification values q ₁ ... q _N .

In one embodiment, the method according to the invention works in a fundamentally positive manner, ie a specific class B _i , for example the association of the observed flying object with a specific alliance, is inferred only from the presence of a specific property E _j . An absence of the specific property E _j can be expressed, for example, in increased uncertainty when a decision is made about membership in the specific class B _i .

In one embodiment, the method comprises the steps, in particular for each property E _j , in particular with regard to each class B _i , determining a closing force value s _ij with regard to a conclusion from property E _j to class B _i , and determining a closing uncertainty value u _ij with regard to a Conclusion from the property E _j to the class B _i . In one embodiment, the method further includes, in particular for each class B _i , determining a class closing force value s, taking into account each closing force value s _ij with regard to the conclusion of each property E _j to the class B _i and each closing uncertainty value u _ij with regard to the conclusion of each property E _j on the class B _i .

The closing force value s _ij represents, for example, a measure such as a probability that the object can be assigned to class B _i if the property E _j is found on the object. In one embodiment, the value s _ij of the closing force represents approximately the extent or degree of inference security S(B _i |E _j ) with regard to each class B _i , with which the value e _j of the recorded property E _{j relates} to the respective class B _i can be closed. The closing force can vary with the value e _j of the detected property E _j , ie the value s _ij of the closing force does not necessarily have to be independent of the detected value e _j of the property E _j to which the closing force is assigned. The closing force value s _ij can be variably specified in such a way that the closing force value is changed depending on the time of day or other external circumstances. The method can determine the value s _ij of the respective closing force by reading out a value s _ij specified for this closing force for the application of the method to an object, for example from a memory, or by determining the value s _ij of the closing force based on other circumstances, for example of the use case determined. The closing force value s _ij can be determined empirically, for example, it can be an empirical value, a theoretical value or it can be predetermined in some other way.

The class closure force value s _i represents a degree of certainty or a probability that class B _i is correctly inferred based on at least one correctly recorded property value e _j .

In one embodiment of the method according to the invention, each property E _j is assigned an error value f _j which specifies a measure, such as a probability, with which the detected value e _j of the respective property E _j is incorrectly detected. In one embodiment, a reliability value z _j is determined which is complementary to the error value f _j , eg the following applies to the reliability value: z _j =1−f _j . The class closure force value s _i is determined for each class B _i taking into account each error value f _j or the complementary reliability value z _j .

The error value f _j assigned to the recorded property value e _j represents an uncertainty about the correctness of the recorded property value e _j , ie an uncertainty about the correctness of the observation of the property value e _j . In one embodiment, the error f _j can assume values from 0 to 1. The error value f _j =0 (zero) represents the greatest possible uncertainty, ie complete uncertainty as to whether the property value e _j correctly reflects the represented property E _j of the object; the error value f _j =1 (one) represents the greatest possible certainty, ie complete certainty that the property value e _j correctly reflects the represented property E _j of the object.

In one embodiment, not all properties relevant to an identification or qualification can be observed without errors and/or without contradictions. In one application of the method for qualifying flying objects, this can be due to the fact that sensors of a ground-based air defense system, for example, are only able to perceive a specific property E _j to a limited extent or not at all. Another reason may be that the observed object has a property E _{j '} that assigns the method itself, for example, to a different class B _{i '} or connects it to the other class than the class B _i to which the observed object itself belongs .

wobei u_ij den Schlussunsicherheitswert für den Eigenschaftswert e_j hinsichtlich der Klasse B_i bezeichnet, s_ij den Schlusskraftwert des Eigenschaftswertes e_j auf die Klasse B_i bezeichnet und f_j den Fehlerwert hinsichtlich des Erfassens des Eigenschaftswerts e_j bezeichnet.In one embodiment of the method according to the invention, the closing force value s _ij indicates a probability with which the actual assignability of the object to the class B _i can be inferred from the value e _j of the property E _j . Each final uncertainty value u _ij is determined, for example, according to the rule: $${\text{and}}_{\text{ij}}=1-\left({\text{s}}_{\text{ij}}*\left(1-{\text{f}}_{\text{j}}\right)\right),$$

where u _ij denotes the final uncertainty value for the property value e _j with regard to the class B _i , s _ij denotes the final force value of the property value e _j for the class B _i and f _j denotes the error value with regard to the detection of the property value e _j .

The foregoing embodiment expresses a principle on which the method of the present invention is based. The method then applies laws of probability theory, for example. To a certain extent, however, the method does not compare “probability” and “contra-probability”, but “certainty”, in particular in the sense of a degree of conviction, and a remaining “residual uncertainty”, in particular in the sense of ignorance or ignorance. If, for example, a closing force certainty S(A) describes or indicates the certainty that statement A is true, an associated, complementary concluding uncertainty U(A) = 1 - S(A) indicates the remaining uncertainty that statement A is true is, but not the probability that statement A is false.

In one embodiment, the method includes steps, according to which a class uncertainty value u _i is determined for each class B _i , in particular as a product of the final uncertainty values u _ij , with regard to class B _i . The class uncertainty value u _i represents a residual uncertainty that an assignment of the object to the class B _i does not apply. The class closure force value s _i can in turn be determined on the basis of the class uncertainty value u _i , in particular if the class closure force value s _i is complementary to the respective class uncertainty value u _i . For example, the equation s _i =1−u _{i applies to the class closing force value s i .}

In one embodiment, the qualification takes into account a conflict level K. In one embodiment, the conflict level K is derived from each property E _j . The method includes the determination of a conflict degree value k. The conflict degree value k represents, for example, the degree of conflict between information that suggests membership in different classes, for example in the case of an observation of a flying object and classification according to possible alliance memberships of the flying object, in different alliances.

In one embodiment, determining the degree of conflict value k includes the steps of forming, in particular for each class B _i , a product of the respective class uncertainty value u _i with the conclusion force values s _i of each remaining class B _{i '} with i'<> i, and forming a sum of these Products for each class B _i and a product of all class closure force values s _i .

In this embodiment, the degree of conflict k is equal to a sum of N+1 products, where N corresponds to the number of classes B _i , i=1...N. Factors of a first of these products are the values of the class closing forces S _i , factors of the other products are for each class B _i the values u _i of the respective class uncertainty and the values s _i' , where i'<> i, of the closing forces for each remaining one Class. In a formula representation, k = $$k={\displaystyle \prod _{i=1N}{s}_i}+{\displaystyle \sum _{i=1N}{\mathrm{and}}_i}+{\displaystyle \prod _{i=1i-\mathrm{1,}i+1N}{s}_i}$$

For example, in a case of three classes (i = ₁ ... ₃ ), the value _k = s1 *s2 *s3 + s1 *s2 * _{u3 + s1} * _u2 * _{s3 + u1} *s ₂ * s ₃ . The value k of the degree of conflict represents the extent of a contradiction with regard to the assignment of the object to either the first class or the second class or, if appropriate, to another class. The contradiction arises if the assignment of the object to a class, taking into account only the first properties, is incompatible with the assignment of the object to another class, taking into account only a second property that differs from the first property.

In one embodiment, the method includes for each class B _i forming an attributable value s' _i as a product of the closing force value s _i with the class uncertainty values u _i' of each remaining class B _i', where i'<>i. According to the so-called Dempster-Shafer rule, the respective conclusion force value s _i , which represents an assignment of the object to class B _i implied by the recorded properties E ₁ ... E _M , by multiplication with the class uncertainty values u _{i '} of the other classes B _{i '} to a "contradictory" value by the attributable value s' _i represented class assignment charged.

In one embodiment of the method, determining the qualification value q _i for each class B _i includes dividing the imputable value s' _i by the conflict degree value k. With the knowledge of the degree of contradiction of the entire available information, the certainties for the affiliation or assignability of the object to a class B _i , expressed by the attribution values s' _i , in particular according to the Dempster-Shafer rule, can to a certain extent be conflict-adjusted.

In one embodiment, the method includes forming a total uncertainty value u as a product of the class uncertainty values u _i for each class B _i . The total uncertainty value u represents a residual uncertainty or the extent of a lack of information or a degree of incompleteness of the information, on the basis of which the system may make a decision. In the special case of non-conflict, the total uncertainty value represents a probability or the extent of a possibility of an erroneous decision about assignment of the object to a single class.

In one embodiment, the method includes assigning the object to only one of the classes B _i considering the qualification value q _i for each class. In a further embodiment, the method makes an allocation decision taking into account the overall uncertainty value u.

According to another aspect, the invention includes a computer program product that has steps for performing the method according to the invention. According to a further aspect, the invention includes a storage medium with the computer program product according to the invention.

According to a further aspect, the invention includes a device for qualifying flying objects based on at least one property E _j with regard to at least two classes B _i , B _{i '} of a classification.

An embodiment of the device has an input suitable for receiving property data e' _j representing a property value e _j . The input is embodied, for example, as a connection port on a device housing, or the input is embodied, for example, as a conductor track section within an integrated circuit. The property value e _j is associated with the property E _j and indicates, for example, whether the property E _j was detected on the object or not, or the property value e _j indicates an extent or an intensity with which the property E _j on the object has been recorded or identified.

The device has a first memory circuit section for storing program instructions of a computer program. In particular, the computer program includes steps for carrying out the method according to the invention.

The device has a processor configured to execute the program instructions. The processor is therefore suitable for determining qualification data q' _i using the property data e' _j . The qualification data q' _i each represent a qualification value q _j related to the classification.

One embodiment of the device includes an output that is suitable for providing the determined qualification data q' _i . The output is embodied, for example, as a connection port on a device housing, or the output is embodied, for example, as a conductor track section within an integrated circuit. In one embodiment of the device according to the invention, the device has a second memory circuit section which is particularly suitable for storing error values f _j .

In a further embodiment, the device is set up to process the qualification data (q _i ) based on a specifiable rule. The device is therefore suitable for assigning the object, in particular the flying object, to only a single class. The device is therefore suitable in particular for an automatic decision regarding the association of the flying object with an alliance.

In the following, the invention is explained using a preferred exemplary embodiment in two applications, which are constructed as simple cases for the purpose of illustration and are specified with imaginary numbers.

In the exemplary embodiment, the method is applied to a qualification of an observed flying object, for example to qualify a combat aircraft, a drone, a cruise missile, a carrier rocket or the like. The qualification takes place in the exemplary use cases with regard to an alliance obedience of an airplane. Accordingly, a classification with three classes is defined: "Friend", "Neutral", "Enemy". The qualification is based on observing the color of the aircraft and hearing an answer to the question for a password. Accordingly, the following properties to be recorded are defined "color_BLUE", "color_RED" and "password_BLUE".

Only the "Friend" alliance has blue aircraft. Alliance B ₁ pilots also use the password "blue". However, not all pilots of the "Friend" alliance know the password "blue". There is also the possibility that the "enemy" alliance has eavesdropped on the "friend"alliance's password "blue" and is using it. Pilots from the "neutral" alliance also use the password "blue" if they are asked to provide the password by the "friend" alliance.

Only the "Enemy" alliance has red-colored aircraft. Furthermore, pilots of the alliance "enemy" use the password "red".

In the exemplary embodiment, an integrated circuit (hereinafter IC) is provided for carrying out the qualification method, which has a data input which is set up to receive data signals from a light sensor, for example a so-called CCD camera (charge coupled device camera). The IC also has a data memory in which a program is stored. The program has steps for performing an embodiment of the qualification method according to the invention. Also stored in the data memory are data representing specific values, which will be discussed in more detail below. The IC includes a circuit section designed as a processor. The processor is set up to execute the stored program instructions in such a way that the processor can determine qualification data which represent qualification values q ₁ , q ₂ and q ₃ related to the classification into “friend”, “neutral” and “enemy”. In accordance with the program instructions, the processor carries out arithmetic steps, among other things, the results of which the processor writes to and reads from the data memory as required. In the exemplary embodiment, the processor is able to make a decision on whether the observed aircraft belongs to an alliance on the basis of the qualification values. The IC is intended, for example, as a replaceable module or as an integral part of an air defense device.

• Die Klassifikation besteht aus den Klassen B₁ = „Freund“, B₂ = „Neutral“ und B₃ = „Feind“. Zu erfassen sind Werte der Eigenschaften E₁ = „Farbe_BLAU“, E₂ = „Farbe_ROT“ und E₃ = „Passwort_BLAU“. Die Eigenschaft E₁ kann etwa die Ausprägung „vorhanden“ und „nicht vorhanden“ annehmen, repräsentiert durch die Werte e₁ = 1 bzw. e₁ = 0, um ein Erfassen der Farbe „blau“ bzw. ein Nicht-Erfassen der Farbe „blau“ zu signalisieren. Dementsprechend kann der Wert e₂ der Eigenschaft E₂ ebenfalls 1 und 0 betragen (e₂ = 1, e₂ = 0). Und weiter kann der Wert e₃ der Eigenschaft E₃ des Passworts 1 oder 0 betragen (e₁ = 1, e₃ = 0).

The circumstances of the application result in the following values, which are specified, determined or read from a memory into the IC and/or are stored there as data representing the respective value, for example in a method step:

• The classification consists of the classes B ₁ = "Friend", B ₂ = "Neutral" and B ₃ = "Enemy". Values of the properties E ₁ = "color_BLUE", E ₂ = "color_RED" and E ₃ = "password_BLUE" are to be recorded. The property E ₁ can take on the characteristics “present” and “not present”, represented by the values e ₁ = 1 and e ₁ = 0, respectively, in order to detect the color “blue” or non-detection of the color “ blue”. Accordingly, the value e ₂ of the property E ₂ can also be 1 and 0 (e ₂ = 1, e ₂ = 0). And further, the value e ₃ of the property E ₃ of the password can be 1 or 0 (e ₁ = 1, e ₃ = 0).

If the aircraft is “blue” in color, it can be concluded with certainty that it belongs to the “Friend” alliance. Thus, the closing force value s ₁₁ with regard to a conclusion from the property E ₁ ("color_BLUE") to the alliance B ₁ ("friend") is: $${\text{s}}_{11}=\text{S}\left({\text{B}}_1|{\text{E}}_1\right)=1$$

At the same time, it cannot be concluded from property E ₁ ("color_BLUE") that the aircraft belongs to alliance B ₂ (neutral) or B ₃ (enemy): $$\begin{array}{l}{\text{s}}_{21}=\text{S}\left({\text{B}}_2|{\text{E}}_1\right)=0\\ {\text{s}}_{31}=\text{S}\left({\text{B}}_3|{\text{E}}_1\right)=0\end{array}$$

If the aircraft is "red" in color, it can be concluded with certainty that it belongs to Alliance B ₃ (enemy). Accordingly, the closing force value s ₃₂ with regard to a conclusion from the property E ₂ ("color_RED") to the alliance B ₃ ("enemy") is: $${\text{s}}_{32}=\text{S}\left({\text{B}}_3|{\text{E}}_2\right)=1$$

At the same time, it cannot be concluded from property E ₂ ("color_RED") that the aircraft belongs to the "friend" or "neutral" alliance: $$\begin{array}{l}{\text{s}}_{12}=\text{S}\left({\text{B}}_1|{\text{E}}_2\right)=0\\ {\text{s}}_{22}=\text{S}\left({\text{B}}_2|{\text{E}}_2\right)=0\end{array}$$

However, a recorded "BLUE" response from the pilot to the question about the password cannot be used to determine membership of the "Friend" alliance with certainty, since the pilot may not belong to the "Friend" alliance, but knows the password "BLUE". and calls for deception. Accordingly, the closing force value s ₁₃ with regard to a conclusion from the property E ₃ ("password_BLUE") to the alliance B ₁ ("friend") is: $${\text{s}}_{13}=\text{S}\left({\text{B}}_1|{\text{E}}_3\right)=0.6$$

Furthermore, with regard to a conclusion from the property E ₃ (“Password_BLUE”) to the alliance B ₂ (“neutral”) or B ₃ (“enemy”): $$\begin{array}{l}{\text{s}}_{23}=\text{S}\left({\text{B}}_2|{\text{E}}_3\right)=0.1\\ {\text{s}}_{33}=\text{S}\left({\text{B}}_3|{\text{E}}_3\right)=0.3\end{array}$$

Accordingly, a neutral answer, such as silence on the part of the pilot, does not necessarily mean that the pilot does not belong to the "Friend" alliance. The pilot could be part of the "Friend" alliance but have forgotten the password.

Experience has shown that the sensor of the CCD camera signals the color "blue" in one out of five cases, although the object captured by the camera is not blue; accordingly, an error value f ₁ =0.2 assigned to the property E ₁ (color_BLUE) or an assigned reliability value z ₁ =0.8.

Furthermore, the sensor of the CCD camera signals the color "red" in one out of five cases, although the object captured by the camera is not red; accordingly, an error value f ₂ =0.2 assigned to the property E ₂ (color_RED) or an assigned reliability value z ₂ =0.8.

With regard to the understanding of the answer to the question about the password, there are no doubts in the example, which is why an error value f ₃ =0 or an associated reliability value z ₃ =1 assigned to the property E ₃ (password).

The list above is only an example and is therefore neither final nor without alternatives. If required, other and/or further values can be represented by data and stored in the IC for processing.

A sequence of the exemplary method in the first exemplary application is described below.

The aircraft is recognized as such. A sensor element of the CCD camera captures the color of the detected aircraft and signals the value e ₁ = 1. The pilot is asked for the password, answers "blue" and thus signals the value e ₃ = 1.

In accordance with the above rule, the processor calculates a final uncertainty value u ₁₁ for the observed color blue (e ₁ = 1) with regard to the affiliation of the aircraft to alliance B ₁ (friend): $${\text{and}}_{\text{11}}=1-\left({\text{s}}_{\text{11}}*\left(1-{\text{f}}_1\right)\right)=1-\left(1*\left(1-0.2\right)\right)=0.2$$

There is no need to calculate a final uncertainty value u ₁₂ because the aircraft was not examined with regard to property E ₂ (color_RED); the associated final uncertainty value u ₁₂ is: $${\text{and}}_{\text{12}}=1-\left({\text{s}}_{\text{12}}*\left(1-{\text{f}}_2\right)\right)=1-\left(0*\left(1-0.2\right)\right)=1$$

The processor also calculates a final uncertainty value u ₁₃ for the recorded response "blue" (e ₃ = 1) to the question about the password regarding the aircraft's membership in alliance B ₁ (friend): $${\text{and}}_{\text{13}}=1-\left({\text{s}}_{\text{13}}*\left(1-{\text{f}}_3\right)\right)=1-\left(0.6*\left(1-0\right)\right)=0.4$$

The processor further calculates a class uncertainty value u ₁ for alliance B ₁ (friend) as a product of the final uncertainty values u ₁₁ , u ₁₂ , u ₁₃ : $${\text{and}}_1={\text{and}}_{11}*{\text{and}}_{12}*{\text{and}}_{13}=0.2*1*0.4=0.08$$

It becomes clear that the final uncertainty value of the property E ₂ (color_RED), with regard to which the aircraft was not examined, as a factor with the value 1 does not influence the class uncertainty value u ₁ .

Furthermore, the processor calculates a class closure force value s ₁ as a complementary value to the class uncertainty value u ₁ according to the rule: $${\text{s}}_1=1-{\text{and}}_1=1-0.08=0.92$$

Thus, considering the circumstances of the present case (possibility of error in detecting the color of the aircraft, detected password based on ignorance or deception of the pilot) is based on observation of only "friendly" characteristics (value of the detected color e ₁ = 1 and value of the detected The "security" based on the password e ₃ = 1) is not absolute, ie not one hundred percent, but - here - only ninety-two percent.

In the present case, no characteristics of the alliances B ₂ (neutral) or B ₃ (enemy) were observed. Thus the aircraft is neither B ₂ (neutral) nor B ₃ (enemy) and there is no conflict (k=0). This results in q ₁ = 0.92, q ₂ = 0, and q ₃ = 0 for the qualification values. The total uncertainty u is 0.08, since the class uncertainty values u ₂ and u ₃ , which are factors in the overall uncertainty, are 1 for the alliances B ₂ and B ₃ respectively. Thus, all observations suggest that the plane belongs to the "Friend" alliance. No observation indicates the opposite. The residual uncertainty, represented by the total uncertainty value u, that the aircraft could belong to another alliance as a "friend", is 0.08, which can mean good information, depending on the situation. If significant properties with a sufficiently high degree of certainty have been observed, the residual uncertainty can be sufficiently low to make a decision.

A sequence of the exemplary method in the second exemplary application is described below, in which the results of observations conflict with one another.

The aircraft is recognized as such. A sensor element of the CCD camera captures the color of the recognized aircraft and signals the value e ₂ =1 with regard to property E ₂ (color red). This means the value e ₁ =0 with regard to property E ₁ (color blue). The pilot is asked for the password, answers "blue" and thus signals the value e ₃ = 1.

The processor calculates a final uncertainty value u ₃₂ for the observed color red (e ₂ = 1) with regard to the affiliation of the aircraft to alliance B ₃ (enemy): $${\text{and}}_{\text{32}}=1-\left({\text{s}}_{\text{32}}*\left(1-{\text{f}}_2\right)\right)=1-\left(1*\left(1-0.2\right)\right)=0.2$$

Thus, despite the observation of the color red, which implies that the observed flying object belongs to alliance B ₃ (enemy), there remains a residual uncertainty, expressed by the final uncertainty value u ₃₂ , according to which there is still the possibility that the observed flying object is not in alliance B ₃ (enemy) attributable.

A calculation of a final uncertainty value u ₃₁ for the color blue (e ₁ =0) with regard to the affiliation of the aircraft to alliance B ₃ (enemy) is superfluous because the aircraft was not examined with regard to property E ₁ (color_BLUE); the associated final uncertainty value u ₃₁ is: $${\text{and}}_{\text{31}}=1-\left({\text{s}}_{\text{31}}*\left(1-{\text{f}}_1\right)\right)=1-\left(0*\left(1-0.2\right)\right)=1$$

The processor further calculates a concluding uncertainty value u ₃₃ for the response e ₃ =BLUE recorded to the query for the password with regard to the affiliation of the aircraft to alliance B ₃ (enemy): $${\text{and}}_{\text{33}}=1-\left({\text{s}}_{\text{33}}*\left(1-{\text{f}}_3\right)\right)=1-\left(0.3*\left(1-0\right)\right)=0.7$$

The processor further calculates a class uncertainty value u ₃ for alliance B ₃ (enemy) as a product of the final uncertainty values u ₃₁ , u ₃₂ , u ₃₃ : $${\text{and}}_3={\text{and}}_{31}*{\text{and}}_{32}*{\text{and}}_{33}=1*0.2*0.7=0.14$$

It becomes clear that the final uncertainty value of the property E ₁ (color_BLUE), with regard to which the aircraft was not examined, as a factor with the value 1 does not influence the class uncertainty value u ₃ .

In principle, the method can refrain from considering final uncertainty values of properties with regard to which the method cannot access any determination if, for example, no observation was made and no measurement result is available. For the purposes of a class uncertainty calculation, the method assumes a maximum final uncertainty value of 1 (one), which is included in the calculation as a factor or which is ignored in the calculation.

Thus, despite the observation of the color red, which implies that the observed flying object belongs to the "enemy" alliance, there remains a residual uncertainty, expressed by the final uncertainty value u ₃ , according to which there is still the possibility that the observed flying object is not in alliance B ₃ (enemy ) to be attributed.

Further, the processor calculates a class closure force value s ₃ with respect to alliance B ₃ (enemy) as a complement of the class uncertainty value u ₃ : $${\text{s}}_3=1-{\text{and}}_3=1-0.14=0.86$$

Considering the circumstances of the present case (possible error in detecting the color of the aircraft, detected password based on ignorance or deceit on the part of the pilot) is based on observation of both a "hostile" property (value of the detected color e ₂ = 1) and one The "security" based on the "friendly" property (value of the recorded password e ₃ = 1) is not absolute, ie not one hundred percent, but is - here - only about six sevenths.

If the processor also calculates a final uncertainty value u ₁₁ for the unobserved color blue (e ₁ = 0) with regard to the affiliation of the aircraft to alliance B ₁ (friend), the result is: $${\text{and}}_{\text{11}}=1-\left({\text{s}}_{\text{11}}*\left(1-{\text{f}}_1\right)\right)=1-\left(0*\left(1-0.2\right)\right)=1$$

The processor calculates a final uncertainty value u ₁₂ for the observed color red e ₂ = 1 with regard to the affiliation of the aircraft to alliance B ₁ (friend): $${\text{and}}_{\text{12}}=1-\left({\text{s}}_{\text{12}}*\left(1-{\text{f}}_2\right)\right)=1-\left(0*\left(1-0.2\right)\right)=1$$

The processor also calculates a final uncertainty value u ₁₃ for the response “blue” (e ₃ = 1) recorded to the question about the password with regard to the affiliation of the aircraft to alliance B ₁ (friend): $${\text{and}}_{\text{13}}=1-\left({\text{s}}_{\text{13}}*\left(1-{\text{f}}_3\right)\right)=1-\left(0.6*\left(1-0\right)\right)=0.4$$

The processor can now calculate a class uncertainty value u ₁ for alliance B ₁ (friend) as a product of the final uncertainty values u ₁₁ , u ₁₂ , u ₁₃ : $${\text{and}}_1={\text{and}}_{11}*{\text{and}}_{12}*{\text{and}}_{13}=1*1*0.4=0.4$$

Thus, despite the observation of the color red, which implies that the observed flying object belongs to the alliance "enemy", there remains a residual uncertainty, expressed by the class uncertainty value u ₁ , according to which there is still the possibility that the observed flying object is the alliance B ₁ (friend) attributable.

Furthermore, the processor calculates a class closure value s ₁ with regard to the alliance B ₁ (friend) as a complementary value to the class uncertainty value u ₁ : $${\text{s}}_1=1-{\text{and}}_1=1-0.4=0.6$$

Accordingly, the processor calculates final uncertainty values u ₂₁ , u ₂₂ , u ₂₃ for the unobserved color blue, the observed color red and the answer “blue” to the question about the password regarding the aircraft’s membership in alliance B ₂ (neutral): $$\begin{array}{l}{\text{and}}_{\text{21}}=1-\left({\text{s}}_{\text{21}}*\left(1-{\text{f}}_1\right)\right)=1-\left(0*\left(1-0.2\right)\right)=1\\ {\text{and}}_{\text{22}}=1-\left({\text{s}}_{\text{22}}*\left(1-{\text{f}}_2\right)\right)=1-\left(0*\left(1-0.2\right)\right)=1\\ {\text{and}}_{\text{23}}=1-\left({\text{s}}_{\text{23}}*\left(1-{\text{f}}_3\right)\right)=1-\left(0.1*\left(1-0\right)\right)=0.9\end{array}$$

Furthermore, the processor calculates a class uncertainty value u ₂ with regard to alliance B ₂ (neutral) as a product of the final uncertainty values u ₂₁ , u ₂₂ , u ₂₃ : $${\text{and}}_2={\text{and}}_{21}*{\text{and}}_{22}*{\text{and}}_{23}=1*1*0.9=0.9$$

Furthermore, the processor calculates a class closure value s ₂ with regard to alliance B ₂ (neutral) as a complementary value to the class uncertainty value u ₂ : $${\text{s}}_2=1-{\text{and}}_2=1-0.9=0.1$$

In the present case, characteristics of alliances B ₁ (friend) and B ₃ (enemy) were observed. Thus, the aircraft may be assigned to _B1 (friend) or _B3 (enemy) and there is a conflict. The processor calculates the total uncertainty value u and the value k of the degree of conflict, using previously determined values. If necessary, these values are read from the data memory:

The processor further calculates an imputable value s' ₁ , s' ₂ , s' ₃ for each alliance B ₁ , B ₂ , B ₃ according to the following rule: $$\begin{array}{l}{\text{it's}}_1={\text{s}}_1*{\text{and}}_2*{\text{and}}_3=0.6*0.9*0.14=0.0756\\ {\text{it's}}_2={\text{and}}_1*{\text{s}}_2*{\text{and}}_3=0.4*0.1*0.14=0.0056\\ {\text{it's}}_3={\text{and}}_1*{\text{and}}_2*{\text{s}}_3=0.4*0.9*0.972=0.3096\end{array}$$

$$\text{s}{\text{'}}_2<{\text{s}}_2,$$

und $$\text{s}{\text{'}}_3<{\text{s}}_3.$$

While each class closure force value s ₁ , s ₂ and s ₃ represents a statement about whether the observed aircraft belongs to the respective alliance B ₁ , B ₂ or B ₃ , the respective attributable value s' ₁ , s' ₂ or s' ₃ puts it into perspective this statement taking into account the existing indications of belonging to a different alliance. The relativized statement is weakened compared to the statement represented by the class closure force value. i.e. the respective attributable value is lower than the assigned class completion value: $$\text{s}{\text{'}}_1<{\text{s}}_1,$$

$$\text{s}{\text{'}}_2<{\text{s}}_2,$$

and $$\text{s}{\text{'}}_3<{\text{s}}_3.$$

The processor also determines a total uncertainty value u, which represents a lack of information: $$\text{and}={\text{and}}_1*{\text{and}}_2*{\text{and}}_3=0.4*0.9*0.14=0.0504$$

The processor then determines a degree of conflict k using the class conclusion forces s ₁ , s ₂ , s ₃ and the class uncertainty values u ₁ , u ₂ , u ₃ according to the rule: $$\begin{array}{l}\text{k}={\text{s}}_1*{\text{s}}_2*{\text{s}}_3\\ +{\text{s}}_1*{\text{s}}_2*{\text{and}}_3\\ +{\text{s}}_1*{\text{and}}_2*{\text{s}}_3\\ +{\text{and}}_1*{\text{s}}_2*{\text{s}}_3\\ \\ =0.6*0.1*0.86\\ +0.6*0.1*0.14\\ +0.6*0.9*0.86\\ +0.4*0.1*0.86=0.5588\end{array}$$

The degree of conflict k > 0 indicates that properties of more than one alliance were observed. Finally, the processor calculates a qualification value q ₁ , q ₂ , q ₃ for each alliance B ₁ , B ₂ , B ₃ : $$\begin{array}{l}{\text{q}}_1=\text{s}{\text{'}}_1/\text{k}=0.0756/0.5588=0.1353\\ {\text{q}}_2=\text{s}{\text{'}}_2/\text{k}=0.0056/0.5588=0.0100\\ {\text{q}}_3=\text{s}{\text{'}}_3/\text{k}=0.3096/0.5588=0.5540\end{array}$$

This results in a qualification of the observed flying object, according to which the observed aircraft is predominantly assigned to alliance B ₃ (enemy). In an alternative embodiment of the method, the processor determines the qualification values by the processor dividing the respectively assigned imputed value by the expression (1-k).

The certainty of belonging to alliance B ₃ (enemy) in the present application example is significantly higher than the certainty of belonging to alliance B ₁ (friend) or the certainty of belonging to alliance B ₂ (neutral). It seems most likely that the plane belongs to Alliance B ₃ (enemy). Accordingly, the IC emits a signal at an output, according to which action to intercept the observed aircraft is required; in other words, the IC decides on the basis of the ascertained qualification values whether the flying object probably belongs to an alliance and signals the decision.

Claims (16)

Method for qualifying an object in terms of a classification with a number N of classes (B ₁ ... B _i ... B _N ), where N > 1, based on a number M of properties (E ₁ ... E _j .. . E _M ) of the object, where M > 0, with: for each property (E _j ), detecting at least one property value (e _j ), for each class (B _i ), determining a qualification value (q ₁ ... q _i ... q _N ) of the object considering each detected property value (e _j ), where qualification values (q _i , q _i' ) for at least two classes (B _i , B _i' ) of the classification are non-zero. Method according to the preceding claim, comprising: for each property (E _j ) with respect to each class (B _i ), determining a conclusion force value (s _ij ) with regard to a conclusion from a j-th property (E _j ) to an i-th class ( B _i ), and determining a conclusion uncertainty value (u _ij ) with regard to a conclusion from the j-th property (E _j ) to the i-th class (B _i ), and for each class (B _i ), determining a class conclusion force value (s _i ) taking into account each final force value (s _ij ) with regard to the inference from each property (E ₁ ... E _M ) to the i-th class (B _i ) and each final uncertainty value (u _i1 , ... u _iM ) with regard to the Conclusion from each property (E ₁ ... E _M ) to the i-th class (B _i ). Method according to the preceding claim, in which each property (E _j ) is assigned an error value (f _j ), which represents a measure of the fact that the detected value (e _j ) of the respective property (E _j ) is incorrectly detected, and at which for each class (B _i ) determines the class closing force value (s _i ) taking into account each error value (f ₁ ... f _j ... f _M ). Method according to the preceding claim, in which the closing force value (s _ij ) represents a measure for the fact that from the value (e _j ) of the j-th property (E _j ) an actual assignment of the object to the i-th class (B _i ) can be inferred, and where each final uncertainty value (u _ij ) is determined according to the rule: $${\text{and}}_{\text{ij}}=1-\left({\text{s}}_{\text{ij}}*\left(1-{\text{f}}_{\text{j}}\right)\right),$$

where u _ij denotes the conclusion uncertainty value for the j th property value with respect to the ith class, s _ij denotes the conclusion force value of the j th property value with respect to the ith class and f _j denotes the error value with regard to detecting the j th property value. Method according to the preceding claim, comprising: for each class (B _i ), forming a final class uncertainty value (u _i ) as the ith product of the final uncertainty values (u _ij ) for each property (E _j ) with regard to the ith class (B _i ), and determining the class closure force value (s _i ) complementary to the class closure uncertainty value (u _i ). Method according to one of the preceding claims, in which the qualification takes into account a degree of conflict (K) derived from that of each property (E _j ), comprising: determining a value (k) of the degree of conflict (K). Method according to the preceding claim, in which the determination of the conflict degree value (k) comprises: for each class (B _i ), forming an i-th product of the i-th class conclusion uncertainty value (u _i ) with the conclusion force values (s _i' ) of each remaining one class (B _i' ; i'<>i), and forming a sum of the products for each class (B _i ) and a product of all class closure force values (s _i ). Method according to the preceding claim, with: for each class (B _i ), forming an attributable value (s' _i ) as the i-th product of the i-th final force value (s _i ) with the final class uncertainty values (u _i' ) of each remaining class ( B _i' ; i'<>i). Method according to the preceding claim, in which the determination of the qualification value (q _i ) for each class (B _i ) comprises: dividing the i-th imputation value (s' _i ) by the conflict degree value (k). Method according to one of the preceding claims, comprising: forming a total uncertainty value (u) as a product of the class uncertainty values (u _i ) for each class (B _i ). Method according to one of the preceding claims, comprising: assigning the object to only one of the classes (B _i ) taking into account the qualification value (q ₁ ... q _N ) for each class (B ₁ ... B _N ). Computer program product which has steps for performing the method according to one of the preceding method claims. Storage medium with the computer program product according to the preceding computer program product claim. Device for qualifying flying objects based on at least one property (E _j ) with regard to at least two classes (B _i , B _{i '} ) of a classification, with: an input for receiving property data (e' _j ), which has a property value (e _j ) for represent the property (E _j ), a first memory circuit section for storing program instructions of a computer program which, when executed, the device for qualifying an object with respect to a classification with a number N of classes (B ₁ ... B _i ... B _N ) , where N > 1, based on a number M of properties (E ₁ ... E _j ... E _M ) of the object, where M > 0, where for each property (Ej), at least one property value (e _j ) is recorded, and/or for each class (Bi) a qualification value (q ₁ ... q _i ... q _N ) of the object is determined, taking into account each recorded property value (e _j ), with qualification values (q _i , q _i' ) for at least two classes (B _i , B _i' ) of the classification are not equal to zero, a processor for executing the program instructions in order to use the property data (e' _j ) to determine qualification data (q' _i ) which contain at least one classification-related qualification value (q _j ) represent, and an output for providing the determined qualification data (q ' _j ). Device according to the preceding claim, wherein the device comprises a second memory circuit section for storing the error values (f _j ). Device according to one of Claims 14 or 15 , which is set up to process the qualification data (q _i ) based on a specifiable rule in order to assign the flying object to only a single class (B _i ).