A method and device for identifying the type of non-cooperative target

Abstract

The present invention provides a type recognition method and device for non-cooperative targets, the method comprising: collecting multiple features of a designated space object; converting the multiple features into standard normal distributions respectively to form multiple standardized features correspondingly; constructing a multi-feature function corresponding to the standardized features, and calculating the probability density function of the multi-feature function; calculating the membership function of each standardized feature based on a bilateral Gaussian membership function consistent with the standard normal distribution characteristics; calculating the probability density and membership of each feature based on the probability density function and the membership function, and calculating the fuzzy probability of the designated space object based on the probability density and membership of each feature; determining whether the designated space object is a first-class true target based on the fuzzy probability. The method of the present invention can quickly and accurately identify space targets.

Description

Technical Field

[0001] The present invention belongs to the technical field of target recognition, and in particular relates to a method and a device for identifying the type of a non-cooperative target. Background Art

[0002] Non-cooperative target recognition refers to recognition performed without any communication between the radar and the target. In this type of recognition, the radar and the system perform recognition only by correlating the parameters of the detectable target with the above parameters of the known (stored) target type. Among various sensing methods, radar has many advantages that make it an ideal choice for many applications, including all-weather operation, long-range detection, high accuracy, high resolution and strong anti-interference capabilities. Its unique characteristics make radar sensors particularly suitable for detecting, tracking and monitoring targets. Therefore, radar sensors are widely used to support non-cooperative target recognition (NCTR) missions. During NCTR operations, radar sensors are prone to various interferences, including false targets and debris. Therefore, the ability to distinguish between real targets and other space objects is crucial to ensure space security. To this end, radar systems must be equipped with advanced algorithms and technologies to accurately and reliably identify targets in complex electromagnetic environments.

[0003] For target recognition in NCTR system, effective target feature classification technology, namely pattern recognition method, is crucial. At present, several mature pattern recognition technologies have been successfully applied to NCTR tasks, including statistical pattern recognition method, neural network pattern recognition method and fuzzy pattern recognition method.

[0004] Statistical pattern recognition methods are based on the statistical characteristics of sample features for classification and recognition. Statistical pattern recognition methods are one of the most commonly used feature classification methods in NCTR, and can be divided into single classification algorithms and fusion classification algorithms. Single classification algorithms include k-nearest neighbor algorithm, Bayesian classifier, support vector machine, random forest, etc. Fusion classification algorithms are composed of multiple single classification algorithms, which effectively utilize the advantages of various classification techniques and improve the effectiveness and stability of the entire classification process. Common fusion algorithms include Bagging algorithm, Boosting algorithm and AdaBoost algorithm. However, statistical pattern recognition methods often require high assumptions about data distribution, making them susceptible to outliers.

[0005] Neural network pattern recognition methods use artificial neural networks for pattern classification and recognition, and have shown significant adaptability in NCTR applications. These methods are particularly good at handling complex nonlinear relationships and large-scale data, including multi-layer feedforward networks using BP algorithms and perceptron algorithms, radial basis function networks (RBFNs), self-organizing neural networks represented by fuzzy ARTMAP networks and self-organizing feature maps (SOFM), convolutional neural networks, and fuzzy neural networks. However, neural network-based pattern recognition methods often require a large amount of training data, which is difficult to obtain in reality, bringing great challenges to the recognition task.

[0006] Fuzzy pattern recognition methods involve the use of fuzzy set theory for pattern classification and recognition. These methods introduce fuzzy sets and membership functions into pattern classification, effectively solving the problem of uncertainty and ambiguity of target feature information. In addition, since fuzzy recognition requires less prior information on membership functions, it is very suitable for NCTR scenarios with limited prior knowledge. Fuzzy pattern recognition methods have been widely used in NCTR, such as Dempster-Shafer evidence theory, fuzzy Bayesian method, and fuzzy set method. Although fuzzy pattern recognition methods provide a unique method for target classification in NCTR systems, they also have an obvious limitation: the subjective determination of feature indicator weights. This aspect introduces subjectivity and randomness into the recognition process, which may affect its accuracy and reliability. In order to alleviate these problems, it is necessary to adopt more rigorous and objective methods to determine the weights.

[0007] Compared with these three typical pattern recognition methods, the algorithm for identifying formation fluid targets using fuzzy probability theory has opened up a new path for NCTR theory. However, through theoretical and practical verification, it is found that using chi-square distribution as a multi-feature function in the target recognition process may lead to the loss of the real target. Summary of the invention

[0008] The technical problem to be solved by the present invention is to provide a method and device for identifying the type of non-cooperative targets that can quickly and accurately identify non-cooperative targets in space.

[0009] The present invention includes a method for identifying the type of a non-cooperative target, comprising:

[0010] Collecting multiple features of a designated space object, wherein the designated space object is an object to be detected to determine whether it is a true target;

[0011] Performing standard normal distribution transformation on the multiple features respectively to form multiple standardized features accordingly;

[0012] In the case where a plurality of the standardized features satisfy a target probability distribution, constructing a multi-feature function corresponding to the standardized features, the target probability distribution being associated with a chi-square distribution, which includes a probability distribution to which the mean and variance of each standardized feature obey;

[0013] In the case where the multi-feature function satisfies the target probability distribution under the specified degrees of freedom, constructing a probability density function and a membership function corresponding to the target probability distribution;

[0014] Input each of the standardized features into a probability density function and a membership function to obtain a probability density and a membership corresponding to each of the features;

[0015] A fuzzy probability is calculated based on the probability density and the degree of membership corresponding to each of the features, and the fuzzy probability is used to determine whether the designated space object belongs to a first fuzzy set, and the first fuzzy set corresponds to a first type of true target;

[0016] It is determined whether the designated space object is a first-category true target based on the fuzzy probability.

[0017] In some embodiments, the collecting of multiple features of a designated space object includes:

[0018] At least the radar cross-section characteristics, high-resolution range image characteristics, and polarization characteristics of the designated space object are collected.

[0019] In some embodiments, each of the features corresponds to a sample, and the plurality of features are respectively subjected to standard normal distribution transformation to form a plurality of standardized features, including:

[0020] The multiple features are respectively transformed into standard normal distribution based on the following formula to form multiple standardized features:

[0021]

[0022] The Y i is the i-th standardized feature, X i is the i-th feature, μ i is the average value of the i-th feature, σ i is the standard deviation of the ith feature, is the average value of the sample corresponding to the i-th feature, S i is the sample variance of the sample corresponding to the i-th feature.

[0023] In some embodiments, the method further comprises:

[0024] The multiple standardized features are independent of each other and all obey a univariate p-norm distribution with a mean of 0 and a variance of 1, which indicates that the multiple standardized features satisfy the target probability distribution, and the target probability distribution is |Y1| p +|Y2| p +…+|Y n | p It obeys Chi-p distribution with n degrees of freedom, where n is the degree of freedom, which depends on the number of types of features collected and the power of the p value setting.

[0025] In some embodiments, the multi-feature function is:

[0026]

[0027] The n is the degree of freedom, which depends on the number of types of features collected, the power value of the p value setting, i is the i-th feature, Y i is the i-th standardized feature.

[0028] In some embodiments, the probability density function is:

[0029]

[0030] in,

[0031] The p2(y) is the probability density function of the standardized feature y, n is the degree of freedom, which depends on the number of types of the feature collected, the power value of the p value setting, when p=2, the probability density function is the same as the probability density function of the chi-square distribution, Γ(c) represents the gamma function, e represents the base of the natural logarithm, x is the characteristic parameter of the feature, c=3σ, σ is the standard deviation.

[0032] In some embodiments, the membership function is:

[0033]

[0034] The Y i is the i-th standardized feature, the μ B (Y i ) is the membership function of the i-th standardized feature;

[0035] The membership function of each of the standardized features is calculated based on the following formula to obtain the membership function corresponding to the specified space object:

[0036]

[0037] The n is the degree of freedom, which depends on the number of types of the features collected.

[0038] In some embodiments, the fuzzy probability is calculated based on the probability density and the degree of membership corresponding to each of the features, including:

[0039] The fuzzy probability is calculated based on the product of the probability density and the degree of membership corresponding to each of the features.

[0040] In some embodiments, determining whether the designated space object belongs to the first category of true targets based on the fuzzy probability includes:

[0041] Based on the fuzzy probability, it is judged whether the following discriminant is satisfied. If so, it indicates that the specified space object belongs to the first category of true targets. If all the fuzzy probabilities are 0, it indicates that the specified space object does not belong to the first category of true targets:

[0042] P(B j )≥1.0×10 1-p

[0043] P(B j )=max[P(B1),P(B2),…,P(B m )]

[0044] The P(B j ) is the fuzzy probability, the B j is the first type of true target, and m is the number of true target types.

[0045] Another embodiment of the present invention also provides a target identification device, including:

[0046] A collection module, used to collect multiple features of a specified space object;

[0047] A conversion module, used to convert the multiple features into standard normal distributions respectively, so as to form multiple standardized features accordingly;

[0048] A first construction module is used to construct a multi-feature function corresponding to the standardized feature and calculate a probability density function of the multi-feature function;

[0049] A first calculation module, used for calculating the membership function of each of the standardized features according to a bilateral Gaussian membership function consistent with the standard normal distribution characteristics;

[0050] A second calculation module, used to calculate the probability density and the membership of each feature according to the probability density function and the membership function;

[0051] A third calculation module, used for calculating the fuzzy probability of the specified space object according to the probability density and the membership degree corresponding to each of the features, wherein the fuzzy probability is used to determine whether the specified space object belongs to a first fuzzy set, and the first fuzzy set corresponds to the first type of true target;

[0052] A determination module is used to determine whether the specified space object is a first-category true target according to the fuzzy probability.

[0053] The beneficial effects of the present invention include first performing standardized conversion on the collected multiple features of the designated space object, then performing multi-feature function conversion on the converted features, and then calculating the corresponding probability density function and membership function, and finally calculating the fuzzy probability based on the probability density function and the membership function, and based on the fuzzy probability, it can be determined whether the designated space object is a true target. The method can effectively improve recognition accuracy and recognition efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0055] Figure 1 Schematic diagram of a flow chart of a method for identifying a type of a non-cooperative target according to an embodiment of the present invention.

[0056] Figure 2 It is a schematic diagram of the application flow of the method for identifying the type of a non-cooperative target according to an embodiment of the present invention.

[0057] Figure 3 This is a structural block diagram of a target identification device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0058] The specific embodiments of the present invention are described in detail below with reference to the accompanying drawings, but are not intended to limit the present invention.

[0059] It should be understood that various modifications may be made to the embodiments disclosed herein. Therefore, the following description should not be considered as limiting, but merely as an example of an embodiment. Other modifications within the scope of the present disclosure will occur to those skilled in the art.

[0060] The accompanying drawings, which are incorporated in and constitute a part of the specification, illustrate embodiments of the present disclosure and, together with the general description of the present disclosure given above and the detailed description of the embodiments given below, serve to explain the principles of the present disclosure.

[0061] These and other characteristics of the invention will become apparent from the following description of a preferred form of embodiment given as a non-limiting example, with reference to the accompanying drawings.

[0062] It should also be understood that, although the invention has been described with reference to some specific examples, those skilled in the art will be able to realize many other equivalent forms of the invention that have the characteristics recited in the claims and are therefore within the scope of protection defined thereby.

[0063] The features and advantages of the present disclosure will become more apparent from the following detailed description when taken in conjunction with the accompanying drawings.

[0064] Specific embodiments of the present disclosure are described hereinafter with reference to the accompanying drawings; however, it should be understood that the disclosed embodiments are merely examples of the present disclosure, which may be implemented in a variety of ways. Well-known and / or repeated functions and structures are not described in detail to avoid obscuring the present disclosure with unnecessary or redundant details. Therefore, the specific structural and functional details disclosed herein are not intended to be limiting, but merely serve as a basis and representative basis for the claims to teach those skilled in the art to use the present disclosure in a variety of ways with substantially any suitable detailed structure.

[0065] This specification may use the phrases "in one embodiment," "in another embodiment," "in a further embodiment," or "in other embodiments," all of which may refer to one or more of the same or different embodiments according to the present disclosure.

[0066] The embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0067] like Figure 1 As shown, the present invention includes a method for identifying the type of a non-cooperative target, comprising:

[0068] S1: Collect multiple features of a specified spatial object;

[0069] S2: converting the plurality of features into standard normal distributions respectively to form a plurality of standardized features accordingly;

[0070] S3: constructing a multi-feature function corresponding to the standardized feature, and calculating a probability density function of the multi-feature function;

[0071] S4: calculating a membership function of each of the standardized features based on a bilateral Gaussian membership function consistent with standard normal distribution characteristics;

[0072] S5: Calculate the probability density and the membership of each feature based on the probability density function and the membership function;

[0073] S6: Calculate the fuzzy probability of the specified space object based on the probability density and membership degree corresponding to each of the features, and the fuzzy probability is used to determine whether the specified space object belongs to a first fuzzy set, and the first fuzzy set corresponds to the first type of true target;

[0074] S7: Determine whether the designated space object is a first-category true target based on the fuzzy probability.

[0075] First, the multiple features of the designated space object in this embodiment are collected by radar. The designated space object is a non-cooperative target, and non-cooperative target recognition refers to recognition performed without any communication between the radar and the target. In this type of recognition, the radar and the system only perform recognition by comparing the parameters of the detectable target with the above parameters of the known (stored) target type. Based on the scheme of this embodiment, the multiple features of the designated space object collected can be standardized first, and then the converted features can be converted into multi-feature functions, and then the corresponding probability density function and membership function can be calculated. Finally, the fuzzy probability is calculated based on the probability density function and the membership function, and based on the fuzzy probability, it can be determined whether the designated space object is a true target. This method can effectively improve recognition accuracy and recognition efficiency.

[0076] In one embodiment, the collecting of multiple features of a designated space object includes:

[0077] S8: Collect at least the radar cross section characteristics (RCS), high resolution range profile characteristics (HRRP), and polarization characteristics (POL) of the designated space object.

[0078] The above features are not unique, and other types of features can also be collected. In this embodiment, only this feature is used as an example for illustration. In this embodiment, by analyzing and synthesizing these features, the radar system can accurately identify and classify non-cooperative targets.

[0079] Furthermore, for the stored true target samples, assume that there are m fuzzy sets, denoted as B j , where j = 1, 2, ..., m, represents a set of m different types of true targets. The random variable used to identify space objects is X i Indicated by, where i = 1, 2, ..., n, represents the n characteristic parameters of the space object. According to the selection principle of the standard sample of the true target, the standard sample data composed of the jth category of true targets is expressed as:

[0080] Where k = 1, 2, ..., q, represents q standard samples of the jth class of real targets. The average value of the characteristic parameters of each sample is calculated based on the following formula:

[0081] Considering that the characteristic parameters of non-cooperative targets vary with spatial positions, it is necessary to select different spatial points for measurement. Therefore: Where l = 1, 2, …, p, where p represents multiple samples selected at different spatial points of the standard true target. The central limit theorem shows that the characteristic parameters of a specific type of true target obey mutually independent normal distributions. In other words, any feature X i All meet According to the unbiased estimation theory, for all standard samples of the true target, any feature X i The sample mean of and sample variance S i are the population means μ i and the population variance Therefore, any feature X of the standard true target sample i can be transformed into a form that obeys the standard normal distribution, as shown in the formula: where Y i ~N(0,1).

[0082] Based on this, the system or radar system has a sample set corresponding to each of the features, and the sample set contains features of the same type collected at different times. The multiple features are respectively subjected to standard normal distribution transformation to form multiple standardized features, including:

[0083] S9: Performing standard normal distribution transformation on the multiple features respectively based on the following formula to form multiple standardized features accordingly:

[0084]

[0085] The Y i is the i-th standardized feature, X i is the i-th feature, μ i is the overall mean of the i-th feature, σ i is the standard deviation of the ith feature, the population mean and population variance may be preset values, or may be estimated from subsequent sample means and sample variances, is the average value of the sample corresponding to the i-th feature, S i is the sample variance of the sample corresponding to the i-th feature.

[0086] Furthermore, the existing schemes usually use the chi-square distribution algorithm to realize the calculation of fuzzy probability. However, when constructing the multi-feature function f1(X), the original intention of the existing chi-square distribution algorithm is that as the probability density function p1(y) of f1(X) increases, the spatial object belongs to B. jThe probability of the target being detected increases accordingly. However, when the number of target feature types, that is, the degree of freedom n, exceeds 2, it is not monotonically decreasing on the positive semi-axis. For a real target, its f1(X) must tend to 0, which will cause p1(y) to tend to 0. Calculating the fuzzy probability based on this data will cause the system to mistakenly infer that the spatial object is not the real target. This is why the chi-square distribution as a multi-feature function will lead to the omission of real targets. In order to overcome this problem, such as Figure 2 As shown, the solution proposed in this embodiment is to first extract the average value of the input features of a given spatial object. Subsequently, these features are normalized by referring to the mean and standard deviation of the standard true target samples. This normalization step allows the calculation of the chi-p distribution value across multiple features. Then, the probability density function of the chi-p distribution is integrated with the membership function to obtain the fuzzy probability of the spatial target. This integration can determine whether the target is true or false. Finally, the simulation results are analyzed in depth to determine the conditions under which the p value of the chi-p distribution reaches the optimal state.

[0087] Specifically, the multi-feature function constructed in this embodiment is actually an enhanced multi-feature function, which is an enhanced multi-feature function based on chi-p distribution. This function expands the concept of the sum of squared deviations to form a new multi-feature function containing n feature parameters, which is expressed as follows: Where U i >0, V i >0, and is an undetermined constant. Assume Therefore, the above formula can be written as: Combining the previous formula, we get the multi-feature function f2(X):

[0088]

[0089] Therefore, the multi-feature function is:

[0090]

[0091] The n is the degree of freedom, which depends on the number of types of features collected, the p value is the power value set, i is the i-th feature, Y i is the i-th standardized feature.

[0092] It can be clearly seen from the above formula that the mathematical meaning of the multi-characteristic function f2(X) is similar to that of the multi-characteristic function f1(X). Both values ​​reflect the characteristic parameters X of the space object. i and the fuzzy set B of the true target j Therefore, the smaller the f2(X) value is, the more likely the target belongs to the fuzzy set B. jObviously, when f2(X)=0 for a space object, it completely belongs to the jth class of true targets B. j .

[0093] Furthermore, the method further comprises:

[0094] S10: The plurality of standardized features are independent of each other and all obey a univariate p-norm distribution with a mean of 0 and a variance of 1, which indicates that the plurality of standardized features satisfy a target probability distribution, and the target probability distribution is |Y1| p +|Y2| p +…+|Y n | p It obeys Chi-p distribution with n degrees of freedom, where n is the degree of freedom, which depends on the number of types of features collected, and p is the set power value.

[0095] Based on the above content, we can know that according to the mathematical principles of p-norm distribution and its related sampling distribution, if the random variables Y1, Y2, …, Y n are independent of each other and all obey μ=0, σ 2 =1, then the univariate p-norm distribution is called |Y1| p +|Y2| p +…+|Y n | p It obeys Chi-p distribution with n degrees of freedom, denoted as χ p Distribution. The random variables Y1, Y2, ..., Y n are independent of each other and all obey the standard normal distribution. The standard normal distribution is a special case of the univariate p-norm distribution when p = 2. Therefore, the multi-feature function f2(X) obeys the Chi-p distribution with n degrees of freedom, where the degree of freedom n depends on the number of target features collected by the radar sensor.

[0096] The probability density function of the Chi-p distribution in this embodiment can be expressed as:

[0097]

[0098] in,

[0099] The p2(y) is the probability density function of the standardized feature y, n is the degree of freedom, which depends on the number of types of the feature collected, the p value is a set power value, when p=2, the probability density function is the same as the probability density function of the chi-square distribution, Γ(c) represents the gamma function, e represents the base of the natural logarithm, x is the characteristic parameter of the feature, c=3σ, σ is the standard deviation.

[0100] Substituting p=2 into the above formula, λ can be calculated as:

[0101] Substitute p=2 and the λ calculated from the above formula into the probability density function and only calculate the part where y>0. The calculation process is as follows:

[0102]

[0103] According to the above derivation, it can be proved that when p = 2, the probability density function of the chi-p distribution is the same as the probability density function of the chi-square distribution. Only the part of y>0 is calculated, and the probability density function of the Chi-p distribution is derived as shown below:

[0104]

[0105] From the above formula, we can see that both λ and Γ(·) are non-negative. Therefore, the coefficient λ in the above formula is n / Γ(n / p) is non-negative. Therefore, the positive or negative value of the above equation depends only on:

[0106]

[0107] Considering the requirement that the probability density function p2(y) is monotonically decreasing, f(y) must be less than 0, that is:

[0108]

[0109] Rearranging the above formula, we get:

[0110] On the one hand, since we only need to consider the case of y>0, we can infer that y·λ p > 0. On the other hand, when p ≥ n, it can be inferred that In summary, when p≥n, the above formula holds, indicating that

[0111] Based on the above derivation, this embodiment has successfully proved that when the p value of the Chi-p distribution is greater than or equal to the degree of freedom n, the probability density function p2(y) is monotonically decreasing. Based on the probability density function data corresponding to different p values ​​and different degrees of freedom n, it can be seen that for the same p value, as the degree of freedom n increases, the trend of the probability density function on the positive semi-axis changes from monotonically decreasing to first increasing and then decreasing. Specifically, when n≤p, the probability density function is monotonically decreasing; when n>p, the probability density function first increases and then decreases. The reason why the probability density function of the chi-square distribution shows the phenomenon of first increasing and then decreasing is because its fixed p value is 2, which corresponds to the number of allowed features, that is, the degree of freedom n is exactly 2. Therefore, the chi-p distribution successfully overcomes the shortcomings of the chi-square distribution in terms of the lack of true targets.

[0112] The membership function plays a key role in describing fuzzy sets, which bridges the gap between fuzzy mathematics and exact mathematics. In order to accurately calculate the fuzzy probability of non-cooperative space targets, a suitable membership function is required. Consider all the identification features X of space objects. i can be transformed into the characteristic Y that conforms to the standard normal distribution i Therefore, a binary Gaussian membership function is selected as a basis to construct the membership function of this embodiment:

[0113]

[0114] Considering that σ=1 of the standard normal distribution, according to the 3σ rule, let c=3σ. Then, the membership function is:

[0115]

[0116] The Y i is the i-th standardized feature, the μ B (Y i ) is the membership function of the i-th standardized feature;

[0117] The membership function of each of the standardized features is calculated based on the following formula to obtain the membership function corresponding to the specified space object:

[0118]

[0119] The n is the degree of freedom, which depends on the number of types of the features collected.

[0120] After obtaining the probability density function and the membership function, the system can input each standardized feature into each function to obtain the probability density and membership corresponding to each feature, and calculate the fuzzy probability accordingly. The fuzzy probability calculated based on the probability density and membership corresponding to each of the features includes:

[0121] S11: Calculate the fuzzy probability based on the product of the probability density and the degree of membership corresponding to each of the features.

[0122] That is, the space object is calculated to belong to the jth target B j Fuzzy probability: P(B j )=μ B (y)·p2(y).

[0123] Among them, P(B j ) is the fuzzy probability, p2(y) is the probability density function, μ B (y) is the membership function.

[0124] The determining, based on the fuzzy probability, whether the designated space object belongs to the first category of true targets comprises:

[0125] S12: judging whether the following discriminant is satisfied based on the fuzzy probability, if satisfied, it indicates that the specified space object belongs to the first category of true targets, if all the fuzzy probabilities are 0, it indicates that the specified space object does not belong to the first category of true targets:

[0126] P(B j )≥1.0×10 1-p

[0127] P(B j )=max[P(B1),P(B2),...,P(B m )]

[0128] The P(B j ) is the fuzzy probability, the B j is the first type of true target, and m is the number of true target types.

[0129] For example, when P(B j ) is 0, indicating that the specified spatial object almost does not belong to the fuzzy set B j ; When P(B j ) is 0, indicating that the object basically belongs to the fuzzy set B j Therefore, fuzzy probability can be used to identify the type of space objects. Assume that there are m types of true targets, corresponding to m fuzzy sets. According to the fuzzy probability formula, we can get P(B1), P(B2),…, P(B m ). If P(B j ) are all 0, it means that the space object does not belong to any target category. When the following equations are satisfied at the same time, it can be determined that the space object belongs to the jth target category: P(B j )≥1.0×10 1-p , P(B j )=max[P(B1),P(B2),…,P(B m )].

[0130] like Figure 3 As shown, another embodiment of the present invention also provides a target identification device, including:

[0131] A collection module, used to collect multiple features of a specified space object;

[0132] A conversion module, used to convert the multiple features into standard normal distributions respectively, so as to form multiple standardized features accordingly;

[0133] A first construction module is used to construct a multi-feature function corresponding to the standardized feature and calculate a probability density function of the multi-feature function;

[0134] A first calculation module, used for calculating the membership function of each of the standardized features according to a bilateral Gaussian membership function consistent with the standard normal distribution characteristics;

[0135] A second calculation module, used to calculate the probability density and the membership of each feature according to the probability density function and the membership function;

[0136] A third calculation module, used for calculating the fuzzy probability of the specified space object according to the probability density and the membership degree corresponding to each of the features, wherein the fuzzy probability is used to determine whether the specified space object belongs to a first fuzzy set, and the first fuzzy set corresponds to the first type of true target;

[0137] A determination module is used to determine whether the specified space object is a first-category true target according to the fuzzy probability.

[0138] In some embodiments, the collecting of multiple features of a designated space object includes:

[0139] At least the radar cross-section characteristics, high-resolution range image characteristics, and polarization characteristics of the designated space object are collected.

[0140] In some embodiments, each of the features corresponds to a sample, and the plurality of features are respectively subjected to standard normal distribution transformation to form a plurality of standardized features, including:

[0141] The multiple features are respectively transformed into standard normal distribution based on the following formula to form multiple standardized features:

[0142]

[0143] The Y i is the i-th standardized feature, X i is the i-th feature, μ i is the average value of the i-th feature, σ i is the standard deviation of the ith feature, is the average value of the sample corresponding to the i-th feature, S i is the sample variance of the sample corresponding to the i-th feature.

[0144] In some embodiments, the apparatus further comprises:

[0145] The multiple standardized features are independent of each other and all obey a univariate p-norm distribution with a mean of 0 and a variance of 1, which indicates that the multiple standardized features satisfy the target probability distribution, and the target probability distribution is |Y1|p +|Y2| p +…+|Y n | p It obeys Chi-p distribution with n degrees of freedom, where n is the degree of freedom, which depends on the number of types of features collected, and p is the distribution probability value.

[0146] In some embodiments, the multi-feature function is:

[0147]

[0148] The n is the degree of freedom, which depends on the number of types of the features collected, the p value is the distribution probability value, i is the i-th feature, and Y i is the i-th standardized feature.

[0149] In some embodiments, the probability density function is:

[0150]

[0151] in,

[0152] The p2(y) is the probability density function of the standardized feature y, n is the degree of freedom, which depends on the number of types of the feature collected, the p value is the distribution probability value, when p=2, the probability density function is the same as the probability density function of the chi-square distribution, Γ(c) represents the gamma function, e represents the base of the natural logarithm, x is the characteristic parameter of the feature, c=3σ, and σ is the standard deviation.

[0153] In some embodiments, the membership function is:

[0154]

[0155] The Y i is the i-th standardized feature, the μ B (Y i ) is the membership function of the i-th standardized feature.

[0156] In some embodiments, the fuzzy probability is calculated based on the probability density and the degree of membership corresponding to each of the features, including:

[0157] The fuzzy probability is calculated based on the product of the probability density and the degree of membership corresponding to each of the features.

[0158] In some embodiments, determining whether the designated space object is the first category of true target based on the fuzzy probability includes:

[0159] Based on the fuzzy probability, it is judged whether the following discriminant is satisfied. If so, it indicates that the specified space object belongs to the first category of true targets. If all the fuzzy probabilities are 0, it indicates that the specified space object does not belong to the first category of true targets:

[0160] P(B j )≥1.0×10 1-p

[0161] P(B j )=max[P(B1),P(B2),…,P(B m )]

[0162] The P(B j ) is the fuzzy probability, the B j is the first type of true target, and m is the number of true target types.

[0163] Another embodiment of the present invention further provides an electronic device, comprising a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the steps in the method for identifying the type of a non-cooperative target as described in any of the above embodiments are implemented.

[0164] Another embodiment of the present invention further provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps in the method for identifying the type of a non-cooperative target as described in any of the above embodiments are implemented.

[0165] It should be noted that the computer storage medium of the present application may be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access storage medium (RAM), a read-only storage medium (ROM), an erasable programmable read-only storage medium (EPROM or flash memory), an optical fiber, a portable compact disk read-only storage medium (CD-ROM), an optical storage medium, a magnetic storage medium, or any suitable combination of the above. In the present application, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, device or device. In the present application, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, in which a computer-readable program code is carried. This propagated data signal may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium, which may send, propagate or transmit a program configured to be used by or in conjunction with an instruction execution system, apparatus or device. The program code contained on the computer-readable medium may be transmitted using any appropriate medium, including but not limited to: wireless, antenna, optical cable, RF, etc., or any suitable combination of the above.

[0166] A person skilled in the art should understand that the discussion of any of the above embodiments is merely illustrative and is not intended to imply that the scope of protection of the present application is limited to these examples. In line with the concept of the present application, the technical features in the above embodiments or different embodiments may be combined, the steps may be implemented in any order, and there are many other variations of different aspects of one or more embodiments of the present application as described above, which are not provided in detail for the sake of simplicity.

Claims

1. A method for identifying the type of a non-cooperative target, characterized in that: include: Collect multiple features of a specified space object; Converting the multiple features into standard normal distributions respectively to form multiple standardized features accordingly; Constructing a multi-feature function corresponding to the standardized feature, and calculating a probability density function of the multi-feature function; Calculating a membership function for each of the standardized features based on a bilateral Gaussian membership function consistent with standard normal distribution characteristics; Calculating the probability density and the membership of each feature based on the probability density function and the membership function; Calculating a fuzzy probability of a specified space object based on the probability density and the degree of membership corresponding to each of the features, wherein the fuzzy probability is used to determine whether the specified space object belongs to a first fuzzy set, wherein the first fuzzy set corresponds to a first type of true target; Determining whether the designated space object is a first-category true target based on the fuzzy probability; Among them, the multi-feature function f2(X) is The n is the degree of freedom, which depends on the number of types of features collected, i is the i-th feature, and Y i is the i-th standardized feature, and the p value is the set power value; The collecting of multiple features of the designated space object includes: At least the radar cross-section characteristics, high-resolution range image characteristics and polarization characteristics of the designated space object are collected.

2. The method for identifying the type of non-cooperative target according to claim 1, characterized in that: Each of the features corresponds to a sample set, and the sample set contains the features collected at different times. The multiple features are converted into standard normal distributions to form multiple standardized features, including: The multiple features are respectively transformed into standard normal distribution based on the following formula to form multiple standardized features: The Y i is the i-th standardized feature, X i is the i-th feature, μ i is the overall mean of feature i, σ i is the population standard deviation of feature i, is the average value of the sample corresponding to the i-th feature, S i is the sample variance of the sample corresponding to the i-th feature.

3. The method for identifying the type of non-cooperative target according to claim 1, characterized in that: The method further comprises: The multiple standardized features are independent of each other and all obey a univariate p-norm distribution with a mean of 0 and a variance of 1, which indicates that the multiple standardized features satisfy the target probability distribution, and the target probability distribution is |Y1| p +|Y2| p +…+|Y n | p It obeys Chi-p distribution with n degrees of freedom, where n is the degree of freedom, which depends on the number of types of features collected and the power of the p value setting.

4. The method for identifying the type of a non-cooperative target according to claim 1, characterized in that: The probability density function is in, The p2(y) is the probability density function of the standardized feature y, n is the degree of freedom, which depends on the number of types of the feature collected. When p=2, the probability density function is the same as the probability density function of the chi-square distribution, Γ(c) represents the gamma function, e represents the base of the natural logarithm, x is the characteristic parameter of the feature, c=3σ, and σ is the standard deviation.

5. The method for identifying the type of non-cooperative target according to claim 1, characterized in that: The membership function is The Yi is the i-th standardized feature, and the μ B (Y i ) is the membership function of the i-th standardized feature.

6. The method for identifying the type of non-cooperative target according to claim 1, characterized in that: The fuzzy probability is calculated based on the probability density and the degree of membership corresponding to each of the features, including: The fuzzy probability is calculated based on the product of the probability density and the degree of membership corresponding to each of the features.

7. The method for identifying the type of non-cooperative target according to claim 1, characterized in that: The determining, based on the fuzzy probability, whether the designated space object is the first category of true targets comprises: Based on the fuzzy probability, it is judged whether the following discriminant is satisfied. If so, it indicates that the specified space object belongs to the first category of true targets. If all the fuzzy probabilities are 0, it indicates that the specified space object does not belong to the first category of true targets: P(B j )≥1.0×10 1-p P(B j )=max[P(B1),P(B2),…,P(B m )] The P(B j ) is the fuzzy probability, the B j is the first type of true target, and m is the number of true target types.

8. A target recognition device, characterized in that: include: A collection module, used to collect multiple features of a specified space object; A conversion module, used for converting the plurality of features into standard normal distributions respectively, so as to form a plurality of standardized features accordingly; A first construction module is used to construct a multi-feature function corresponding to the standardized feature and calculate a probability density function of the multi-feature function; A first calculation module, used for calculating the membership function of each of the standardized features according to a bilateral Gaussian membership function consistent with the standard normal distribution characteristics; A second calculation module, used to calculate the probability density and the membership of each feature according to the probability density function and the membership function; A third calculation module, used for calculating the fuzzy probability of the specified space object according to the probability density and the membership degree corresponding to each of the features, wherein the fuzzy probability is used to determine whether the specified space object belongs to a first fuzzy set, and the first fuzzy set corresponds to the first type of true target; A determination module, used to determine whether the specified space object is a first-category true target according to the fuzzy probability; Among them, the multi-feature function f2(X) is The n is the degree of freedom, which depends on the number of types of features collected, i is the i-th feature, and Y i The power value of the p-value setting for the i-th standardized feature; The collecting of multiple features of the designated space object includes: At least the radar cross-section characteristics, high-resolution range image characteristics and polarization characteristics of the designated space object are collected.

Images