A method, device and terminal device for classifying and identifying a target

Abstract

The present application provides a method, apparatus and terminal device for classifying and recognizing a target, which are applicable to the technical field of data processing. The method includes: extracting characteristic values from target data collected by a radar, where the characteristic values at least include a first characteristic value and a second characteristic value; performing a first classification on the target according to the first characteristic value; performing a second classification on the target according to the second characteristic value; and combining the results of at least two classifications to recognize the category of the target. By extracting more than two characteristic values and jointly using more than two characteristic values for target classification and recognition, the present invention can effectively improve the accuracy of target classification and recognition.

Description

Technical Field

[0001] This application belongs to the technical field of data processing, and particularly relates to a method, device and terminal device for classifying and identifying a target. Background Technique

[0002] Radar target recognition, as an important indicator to measure modern radars, has always been concerned by people and certain progress has been made. Radar automatic target recognition has become the development direction of future radars. Since narrowband radars have a long detection range, they have been widely used in the recognition of ground targets. However, the accuracy of existing ground target classification and recognition is relatively low. Summary of the Invention

[0003] In view of this, embodiments of this application provide a method, device and terminal device for classifying and identifying a target, which are used to improve the accuracy of target classification and recognition.

[0004] The first aspect of the embodiments of this application provides a method for classifying and identifying a target, including:

[0005] Extract feature values from the target data collected by the radar, where the feature values at least include a first feature value and a second feature value;

[0006] Classify the target for the first time according to the first feature value;

[0007] Classify the target for the second time according to the second feature value;

[0008] Combine the results of at least two classifications to identify the category of the target.

[0009] In a possible implementation manner of the first aspect, the first feature value is a speed feature value;

[0010] The classifying the target for the first time according to the first feature value includes:

[0011] When the speed feature value is greater than a first preset value, the target is classified as a vehicle for the first time; when the speed feature value is not greater than the first preset value, the target is classified as uncertain for the first time;

[0012] The combining the results of at least two classifications to identify the category of the target includes:

[0013] When the first classification is a vehicle and the second classification is a vehicle, then identify the category of the target as a vehicle;

[0014] When the first classification is uncertain and the second classification is a person, then identify the category of the target as a person.

[0015] It should be understood that in the case where the first classification is uncertain, uncertain means that both a vehicle or a person is possible.

[0016] In a possible implementation of the first aspect, when the first classification is uncertain and the second classification is not a person, the method further includes:

[0017] When both the speed eigenvalue and the second eigenvalue are within further defined conditions, the category of the target is identified as a vehicle; otherwise, the category of the target is identified as unknown.

[0018] In a possible implementation of the first aspect, the eigenvalue further includes a third eigenvalue, and the method further includes:

[0019] When the first classification is a vehicle and the second classification is not a vehicle, or when the first classification is uncertain, the second classification is not a person, and the speed eigenvalue and the second eigenvalue are not both within further defined conditions, the target is classified for the third time according to the third eigenvalue;

[0020] Combining the results of at least two classifications to identify the category of the target includes:

[0021] When the first classification is a vehicle and the second classification is not a vehicle, if the third classification is a vehicle, the category of the target is identified as a vehicle; otherwise, the category of the target is identified as unknown. When the first classification is uncertain, the second classification is not a person, and the speed eigenvalue and the second eigenvalue are not both within further defined conditions, if the third classification is a person, the target is identified as a person; otherwise, the category of the target is identified as unknown.

[0022] In a possible implementation of the first aspect, one of the RCS eigenvalue and the Doppler spectrum entropy eigenvalue is used as the second eigenvalue, and the other is used as the third eigenvalue.

[0023] In a possible implementation of the first aspect,

[0024] The second classification of the target according to the second eigenvalue includes:

[0025] The second eigenvalue of the target is second-classified by an SVM classifier;

[0026] The third classification of the target according to the third eigenvalue includes:

[0027] The third eigenvalue of the target is third-classified by the SVM classifier.

[0028] In a possible implementation of the first aspect, the eigenvalue further includes a third eigenvalue, and the method further includes:

[0029] When the target is classified as a vehicle for the first time and not a vehicle for the second time, or when it is classified as uncertain for the first time and not a person for the second time, the target is classified for the third time according to the third eigenvalue;

[0030] Combining the results of at least two classifications to identify the category of the target, including:

[0031] When the target is classified as a vehicle for the first time and not a vehicle for the second time, if it is classified as a vehicle for the third time, the category of the target is identified as a vehicle; otherwise, the category of the target is identified as unknown. When it is classified as uncertain for the first time and not a person for the second time, if it is classified as a person for the third time, the target is identified as a person; otherwise, the category of the target is identified as unknown.

[0032] It should be understood that the embodiments of the present application are not only applicable to the classification and recognition of people and vehicles, but can also be extended to the classification and recognition of animals and vehicles.

[0033] The second aspect of the embodiments of the present application provides a target classification and recognition device, including:

[0034] An extraction module, configured to extract eigenvalues from the target data collected by the radar, where the eigenvalues at least include a first eigenvalue and a second eigenvalue;

[0035] A classification and recognition module, configured to perform a first classification on the target according to the first eigenvalue; perform a second classification on the target according to the second eigenvalue; combine the results of at least two classifications to identify the category of the target.

[0036] The third aspect of the embodiments of the present application provides a terminal device, where the terminal device includes a memory and a processor, and a computer program that can run on the processor is stored on the memory. When the processor executes the computer program, the steps of the target classification and recognition method described in any item of the first aspect above are implemented.

[0037] The fourth aspect of the embodiments of the present application provides a computer-readable storage medium, including: storing a computer program, and when the computer program is executed by a processor, the steps of the target classification and recognition method described in any item of the first aspect above are implemented.

[0038] The fifth aspect of the embodiments of the present application provides a computer program product, and when the computer program product runs on a terminal device, the terminal device is enabled to execute the steps of the target classification and recognition method described in any item of the first aspect above.

[0039] The beneficial effects of the embodiments of the present application compared with the prior art are as follows: By extracting more than two eigenvalues and jointly performing target classification and recognition using more than two eigenvalues, the accuracy of target classification and recognition can be effectively improved. Description of the Drawings

[0040] To more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the accompanying drawings required for use in the embodiments or the description of the prior art. Obviously, the accompanying drawings in the following description are only some embodiments of the present application. For those of ordinary skill in the art, without creative efforts, other accompanying drawings can also be obtained based on these drawings.

[0041] Figure 1 It is a schematic diagram of the implementation process of the target classification and recognition method provided by the embodiments of the present application;

[0042] Figure 2 It is a schematic diagram of the relative RCS characteristics of the vehicle and the person provided by the embodiments of the present application;

[0043] Figure 3.1 It is a Doppler spectrum diagram of the vehicle provided by the embodiments of the present application;

[0044] Figure 3.2 It is a Doppler spectrum diagram of the person provided by the embodiments of the present application;

[0045] Figure 3.3 It is a schematic diagram of the Doppler spectrum entropy value of the vehicle and the person provided by the embodiments of the present application;

[0046] Figure 3.4 It is an instantaneous frequency curve diagram of the vehicle provided by the embodiments of the present application;

[0047] Figure 3.5 It is an instantaneous frequency curve diagram of the person provided by the embodiments of the present application;

[0048] Figure 3.6 It is a schematic diagram of the principle of the SVM classifier provided by the embodiments of the present application;

[0049] Figure 4 It is a schematic diagram of the implementation process of the target classification and recognition method provided by the embodiments of the present application in an application scenario;

[0050] Figure 5.1 It is a schematic diagram of the distribution of the RCS characteristics of the vehicle and the person in an experimental result of the target classification and recognition method provided by the embodiments of the present application;

[0051] Figure 5.2 It is a relative RCS distribution histogram of the vehicle and the person in an experimental result of the target classification and recognition method provided by the embodiments of the present application;

[0052] Figure 5.3 It is a schematic diagram of the implementation process of the target classification and recognition method provided by the embodiments of the present application in an experiment;

[0053] Figure 6.1It is a schematic diagram showing the distribution of the Doppler spectrum entropy features of vehicles and humans, which is an experimental result of the target classification and recognition method provided by an embodiment of the present application;

[0054] Figure 6.2 It is a schematic diagram of the implementation process of the target classification and recognition method provided by an embodiment of the present application in an experiment;

[0055] Figure 7 It is a schematic diagram of the terminal device provided by an embodiment of the present application;

[0056] Figure 8 It is a schematic diagram of the target classification and recognition device provided by an embodiment of the present application. Detailed implementation manners

[0057] In the following description, specific details such as specific system architectures and technologies are presented for the purpose of illustration rather than limitation, so as to thoroughly understand the embodiments of the present application. However, those skilled in the art should clearly understand that the present application can also be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to avoid unnecessary details from interfering with the description of the present application.

[0058] In order to illustrate the technical solutions described in the present application, the following will be described through specific embodiments.

[0059] Figure 1 It shows a flowchart of the implementation of the target classification and recognition method provided in Embodiment 1 of the present application, Figure 1 The target classification and recognition method in includes steps 11 to 14.

[0060] Step 11: Extract feature values from the target data collected by the radar, where the feature values at least include a first feature value and a second feature value;

[0061] Step 12: Perform a first classification of the target according to the first feature value;

[0062] Step 13: Perform a second classification of the target according to the second feature value;

[0063] Step 14: Combine the results of at least two classifications to identify the category of the target.

[0064] In the embodiments of the present application, the extracted feature values may include, but are not limited to, speed feature values, RCS feature values, and Doppler spectrum entropy feature values.

[0065] Preferably, the first feature value is a speed feature value. Because when the speed feature value is used as the first determination factor, the accuracy of identifying the target category is higher when the ground conditions are complex.

[0066] The ground conditions are relatively complex, including: there are uniformly moving and variably moving objects in the ground environment.

[0067] Taking the classification targets of people and vehicles as an example, according to careful observation in daily life, it can be known that the movement speed of a person's limbs is not uniform when walking, but changes from fast to slow in one-step cycles. When a vehicle moves as the target type, its speed basically remains constant, equivalent to uniform motion.

[0068] When using only SVM (relative to RCS) and only SVM (spectrum entropy feature) to distinguish uniformly moving and variably moving objects, the misjudgment rate is relatively large.

[0069] For the above reasons, using the speed eigenvalue as the first determination factor results in a higher accuracy rate for identifying the target category.

[0070] Preferably, the second eigenvalue is the RCS eigenvalue. Because when the first eigenvalue is the speed eigenvalue, the accuracy rate of identifying the target category with the second eigenvalue being the RCS eigenvalue is higher than that with the second eigenvalue being the Doppler spectrum entropy eigenvalue.

[0071] As will be mentioned later regarding the third eigenvalue and the third classification, for the same reasons as above, when there are three classifications, preferably, the first eigenvalue is the speed eigenvalue, the second eigenvalue is the RCS eigenvalue, and the third eigenvalue is the Doppler spectrum entropy eigenvalue.

[0072] In the embodiments of the present application, the RCS eigenvalue refers to the eigenvalue based on RCS, which can be the RCS itself or the eigenvalue obtained after processing based on RCS; the speed eigenvalue refers to the eigenvalue based on the speed eigenvalue, which can be the speed itself or the eigenvalue obtained after processing based on the speed; the Doppler spectrum entropy eigenvalue refers to the eigenvalue based on the Doppler spectrum entropy, which can be the Doppler spectrum entropy itself or the eigenvalue obtained after processing based on the Doppler spectrum entropy.

[0073] In the embodiments of the present application, the first eigenvalue is the speed eigenvalue;

[0074] The first classification of the target according to the first eigenvalue includes:

[0075] When the speed eigenvalue is greater than the first preset value, the target is first classified as a vehicle; when the speed eigenvalue is not greater than the first preset value, the target is first classified as uncertain;

[0076] Combining the results of at least two classifications to identify the category of the target includes:

[0077] When the first classification is a vehicle and the second classification is a vehicle, then the category of the target is identified as a vehicle;

[0078] When it is classified as uncertain for the first time and as a person for the second time, the category of the target is identified as a person.

[0079] Among them, when it is classified as uncertain for the first time, being uncertain means that it may be either a vehicle or a person.

[0080] By using the speed eigenvalue as the first determination factor, the accuracy of identifying the category of the target can be improved. By comparing the speed eigenvalue with the first preset value and combining the result of the second classification, the final category of the target can be identified.

[0081] When the category of the target cannot be identified by combining the results of the first and second classifications after the first and second classifications, the category of the target can be identified through the following scheme:

[0082] In the embodiment of the present application, when it is classified as uncertain for the first time and not as a person for the second time, the method further includes:

[0083] When both the speed eigenvalue and the second eigenvalue are within the further limited conditions, the category of the target is identified as a vehicle; otherwise, the category of the target is identified as unknown.

[0084] For example, Figure 5.3 and Figure 6.2 both give the further limited conditions. Figure 5.3 It can be seen from that when the speed eigenvalue is not less than 2 m / s and the RCS is not less than 224, it can be used as the further limited condition. When both the speed eigenvalue and the second eigenvalue are within this limited condition, the category of the target is identified as a vehicle; otherwise, the category of the target is identified as unknown. Similarly, Figure 6.2 It can be seen from that when the entropy is not greater than 0.9 and the speed is not less than 2 m / s, it can be used as the further limited condition. When both the speed eigenvalue and the second eigenvalue are within this limited condition, the category of the target is identified as a vehicle; otherwise, the category of the target is identified as unknown.

[0085] In practical applications, the further limited conditions can be set with corresponding parameters according to the specific target to be identified.

[0086] When the category of the target cannot be identified by combining the results of the first and second classifications after the first and second classifications, it can also be identified through the following scheme:

[0087] In the embodiment of the present application, the eigenvalue further includes a third eigenvalue, and the method further includes:

[0088] When the target is classified as a vehicle for the first time and not a vehicle for the second time, or when the first classification is uncertain, the second classification is not a person, and the speed eigenvalue and the second eigenvalue are not both within the further defined conditions, the target is classified for the third time according to the third eigenvalue;

[0089] Combining the results of at least two classifications to identify the category of the target, including:

[0090] When the first classification is a vehicle and the second classification is not a vehicle, if the third classification is a vehicle, the category of the target is identified as a vehicle, otherwise the category of the target is identified as unknown; when the first classification is uncertain, the second classification is not a person, and the speed eigenvalue and the second eigenvalue are not both within the further defined conditions, if the third classification is a person, the target is identified as a person, otherwise the category of the target is identified as unknown.

[0091] In the embodiments of the present application, one of the RCS eigenvalue and the Doppler spectrum entropy eigenvalue is used as the second eigenvalue, and the other is used as the third eigenvalue. Preferably, the second eigenvalue is the RCS eigenvalue and the third eigenvalue is the Doppler spectrum entropy eigenvalue, and the recognition accuracy is higher.

[0092] In the embodiments of the present application, the second classification of the target according to the second eigenvalue includes:

[0093] The second eigenvalue of the target is classified for the second time by an SVM classifier;

[0094] The third classification of the target according to the third eigenvalue includes: [[ID=T20]]

[0095] The third eigenvalue of the target is classified for the third time by the SVM classifier.

[0096] When it is still impossible to identify the category of the target by combining the results of the first and second classifications after the first and second classifications, the following solution can also be used to identify:

[0097] In the embodiments of the present application, the eigenvalue further includes a third eigenvalue, and the method further includes:

[0098] When the first classification is a vehicle and the second classification is not a vehicle, or when the first classification is uncertain and the second classification is not a person, the target is classified for the third time according to the third eigenvalue;

[0099] Combining the results of at least two classifications to identify the category of the target, including:

[0100] When it is classified as a vehicle for the first time and not a vehicle for the second time, if it is classified as a vehicle for the third time, the category of the target is identified as a vehicle; otherwise, the category of the target is identified as unknown. When it is classified as uncertain for the first time and not a person for the second time, if it is classified as a person for the third time, the target is identified as a person; otherwise, the category of the target is identified as unknown.

[0101] This solution omits the judgment step of whether both the speed eigenvalue and the second eigenvalue are within further limiting conditions, and directly identifies through the third classification, relatively simplifying one step.

[0102] In the embodiments of the present application, the so-called classification can be to divide the category of the target according to a certain standard, and this standard can be diverse. For example, judging the magnitude and range of certain parameters, or classifying through an SVM classifier. Specifically, in the embodiments of the present application, the first classification can be performed by judging whether the speed eigenvalue is greater than a first preset value; the second classification can be performed on the second eigenvalue through an SVM classifier; the third classification can be performed on the third eigenvalue through an SVM classifier; and further classification can be performed by judging whether both the speed eigenvalue and the second eigenvalue are within further limiting conditions.

[0103] It can be seen that in the embodiments of the present application, at least two eigenvalues are extracted, classified at least twice respectively, and the category of the target is identified by combining the results of at least two classifications, which can effectively improve the accuracy of classification and recognition. A third eigenvalue can also be added to perform a third classification on the target, and the category of the target is identified through another comparison and confirmation, thereby more effectively improving the accuracy of classification and recognition.

[0104] In the embodiments of the present application, for the classification and recognition of targets of people and vehicles, preferably, the first eigenvalue is the speed eigenvalue, the second eigenvalue is the RCS eigenvalue, and the third eigenvalue is the Doppler spectrum entropy eigenvalue.

[0105] By using the speed eigenvalue as the classification element for the first classification, both the RCS eigenvalue or the Doppler spectrum entropy eigenvalue can effectively improve the accuracy of classification and recognition of the target.

[0106] By using the speed eigenvalue as the classification element for the first classification and combining the RCS eigenvalue and the Doppler spectrum entropy eigenvalue, the accuracy of classification and recognition of the target can be more effectively improved.

[0107] To better understand the solution of the embodiments of the present application, the eigenvalues extracted in step 11 are illustrated as follows:

[0108] 1. RCS eigenvalue:

[0109] Radar Cross Section (RCS) is a physical quantity that characterizes the backscattering ability of a target to incident electromagnetic waves. Different types of targets can be identified by applying the radar cross section characteristics.

[0110] Assume the power of the radar transmitter is P t , the antenna gain is G, the distance from the target to the radar is R, and the effective receiving area of the antenna is A e , the radar cross-sectional area of the target is σ, then the echo power P received at the radar is r for:

[0111]

[0112] Because the working environment of the radar does not change when collecting data, that is, P t , G and Ae can be regarded as constants, so we have:

[0113]

[0114] Then the radar cross-sectional area σ of the target can be expressed as:

[0115] σ=kP r R 4 (2.3)

[0116] From formula (2.3), we can see that the radar cross section area is only related to k and P in numerical calculation. r It is related to R, but because k is a constant, its influence is generally not considered in calculations. Furthermore, RCS is proportional to the fourth power of distance R and the echo power, so the relative value of RCS is very large and the fluctuation range is also relatively large. To facilitate comparison of the relative RCS values of different targets, we use the logarithm of RCS as the characteristic value, expressed as follows:

[0117] g r =10log 10 (σ)=10log 10 (P r R 4 ) (2.4)

[0118] Among them, the echo power P r The calculation method is as follows:

[0119] Since the echo energy of the target may spread to adjacent range cells, there are generally two methods for calculating the echo power: First, the data on the range cell where the maximum value of the target spectrum is located is used to calculate the echo power; Second, according to the spread of the target energy, the data means of several adjacent range cells are taken to calculate the echo power. The advantage of the first method is that the calculation is simple, but the disadvantage is that the calculation error is large. The advantage of the second method is that it can calculate the power of the target more accurately, but the disadvantage is that the calculation is complex. Preferably, the first method is used to obtain the echo power.

[0120] Among them, the calculation method of the target range is as follows:

[0121] After detecting the target through constant false alarm detection, the range gate information where the target is located can be obtained, and then the actual range can be calculated according to the range resolution cell size of the narrowband radar. Assuming that the range resolution cell of the narrowband radar is ΔR and the range gate where the target is located is R d , then the actual range R of the target is:

[0122] R = R d ·ΔR (2.5)

[0123] After obtaining the echo power and range of the target, the relative RCS of the target can be calculated according to Equation (2.4).

[0124] The relative RCS extraction results of the target based on the measured data can be obtained.

[0125] Taking the classified targets of people and vehicles as an example, based on the two sets of field measured data of a certain type of radar, 126 relative RCS features of vehicles and people are extracted respectively, and the results are as Figure 2 shown. It can be seen from the figure that the relative RCS features of vehicles and people overlap less, and the classification and recognition of vehicles and people can be realized based on this feature.

[0126] II. Doppler spectrum entropy eigenvalue

[0127] 1. Analysis of the target Doppler spectrum

[0128] The spectrum analysis of the signal is to transform the signal from the time domain to the frequency domain, and some characteristic information closely related to the frequency in the signal can be found.

[0129] The methods for signal spectrum analysis include two types: the parametric method based on models and the non-parametric method based on Fourier transform. The parametric methods mainly include the maximum likelihood method, the linear prediction frequency estimation technique, the AR model, the Prony method, and so on. The advantages of the parametric method are high frequency resolution, which can distinguish frequency components that are very close to each other in the signal frequency, no signal periodicity assumption, and suitability for short data processing. The disadvantages are the need to establish a process model, difficulties in model order determination, large computational amount, and so on. The non-parametric method based on Fourier transform has the advantages of high computational efficiency and easy implementation. The disadvantages are insufficient frequency resolution and the problem of spectral leakage.

[0130] The computational amount of the parametric method based on models is large. Preferably, the embodiments of the present application use the non-parametric method based on Fourier transform to analyze the Doppler spectrum of the signal. The non-parametric method based on Fourier transform has the problem of insufficient frequency resolution. Preferably, when analyzing the Doppler structure of the target, it is mainly analyzed from the overall shape of the target Doppler spectrum.

[0131] Analyze the Doppler spectra of vehicle and pedestrian targets. The Doppler spectrum diagram of the vehicle is as Figure 3.1 and the Doppler spectrum of the pedestrian Figure 3.2 as shown.

[0132] It can be seen from the figure that the Doppler spectrum of the vehicle basically has no side lobes and the waveform is very sharp, while the main lobe of the Doppler spectrum of the pedestrian is wider than that of the previous two types of targets and the side lobe fluctuations are larger. Therefore, to extract feature information from the target Doppler spectrum, a mathematical method that can reflect the overall shape change of the spectrum waveform needs to be found, and the computational amount of this mathematical method cannot be too large, and it can reduce the dimension of the spectrum waveform. Considering the above factors and the physical meaning of entropy, preferably, the embodiments of the present application use the entropy value to describe the change of the spectrum waveform.

[0133] 2. Extraction of the entropy value feature of the target spectrum

[0134] Entropy is a very important concept in information theory and is used in statistics to describe the irregularity or uncertainty degree of a system. If a random signal has N possible values and their occurrence probabilities are p1, p2,..., p N , then the entropy of the random signal is:

[0135]

[0136] It can be clearly seen from Equation (2.6) that entropy is non-negative, and only when the occurrence probability of a certain signal is 1, the entropy reaches the minimum value of zero.

[0137] Let x i (i = 1, 2,..., M) represent the modulus values of each point on the Doppler spectrum of the range gate where the target is located. First, for xi (i = 1, 2, ..., M) is normalized to obtain:

[0138]

[0139] Substituting Equation (2.7) into Equation (2.6), we get:

[0140]

[0141] The spectral entropy value of the target Doppler spectrum can be obtained from Equation (2.8).

[0142] 3. Extraction Results of Spectral Entropy Value

[0143] Taking the classification targets of people and vehicles as an example, based on the two sets of field measurement data of a certain type of radar, 109 Doppler spectral entropy values of vehicles and people are extracted respectively. The results are as Figure 3.3 shown. It can be seen from the figure that the Doppler spectral entropy features of vehicles and people have less overlap, and theoretically, the classification and recognition of vehicles and people can be achieved.

[0144] III. Velocity Feature Value

[0145] Here, the feature extraction based on the instantaneous frequency is taken as an example for illustration.

[0146] For the radar echo signal, its instantaneous frequency is:

[0147] f i (k) = 2vf0 / c (2.10)

[0148] where f0 is the carrier frequency of the antenna transmitted signal, v is the radial velocity of the target, and c is the speed of light.

[0149] It can be seen from Equation (2.10) that the instantaneous frequency is proportional to the target velocity. When the magnitude of the target velocity changes, the magnitude of the instantaneous frequency also changes accordingly.

[0150] Taking the classification targets of people and vehicles as an example, according to careful observation in daily life, it can be known that the movement speed of a person's limbs is not uniform when walking, but changes rapidly and slowly in a cycle of one step. The speed of the vehicle type target is basically constant during movement, equivalent to uniform motion. The instantaneous frequency curve of the vehicle is as Figure 3.4 and the instantaneous frequency curve of the person is as Figure 3.5 shown.

[0151] Preferably, taking the standard deviation of the extracted waveform as the feature quantity has small error and high classification and recognition rate. Let X = (x1, x2, ..., x N ) be the values of each point on the target instantaneous frequency waveform, and the calculation formula of the standard deviation is:

[0152]

[0153] Among them, is the mean value of X.

[0154] For a better understanding of the solution of the embodiments of the present application, the classification operation in the embodiments of the present application is further described as follows:

[0155] In the embodiments of the present application, the first classification, the second classification, and the third classification can all be performed by using the method based on Support Vector Machines (SVM).

[0156] Support Vector Machines (SVM) is a pattern recognition technology based on statistical learning theory and the principle of structural risk minimization. Its main idea is to map the samples in the input space to a high-dimensional feature space through a non-linear transformation, and to find the optimal classification surface for linearly separating the samples in this feature space.

[0157] The SVM classifier is derived from the optimal classification surface in the linearly separable case, and the principle is as Figure 3.6 shown. In the figure, the circles (‘○’) and squares (‘□’) represent two types of target samples, H is the classification line, H1 and H2 are the lines passing through the samples closest to H in each class and parallel to H respectively, where d is the Euclidean distance between H1 and H2, also known as the classification margin. In a three-dimensional or higher-dimensional classification space, H is also called the classification surface.

[0158] The optimal classification line means that the classification margin reaches the maximum on the basis of correctly separating the targets. Although the dotted line in the figure can correctly classify the targets, the classification margin is obviously not the smallest, so it is not the optimal classification line. HI and H2 can correctly distinguish the targets and have the largest classification margin, so H is the optimal classification line.

[0159] Let the equation of the classification surface be wx + w0 = 0, and the linearly separable sample set (x i , y i ), i = 1, 2,..., n, x ∈ R d , y ∈ {-1, +1}, satisfies

[0160] y i [wx i + w0] - 1 ≥ 0, i = 1, 2,..., n (3.4)

[0161] According to the definition of the distance from a point to a plane, the classification margin Making the margin maximum is equivalent to making the smallest. The classification surface that satisfies condition (3.4) and makes the smallest is the optimal classification surface, and the sample points on the two lines Hl and H2 are called support vectors.

[0162] There are different algorithms for different inner product functions in the SVM classifier. There are three main kernel functions, namely polynomial kernel function, radial basis function (RBF), and sigmoid kernel function.

[0163] For multi-class problems, two methods can be used: "one-against-one" and "one-against-all".

[0164] The specific process of the "one-against-one" method is to construct all possible two-class SVM classifiers. The training sample set of each classifier is only taken from the corresponding two classes. In this way, a total of N = k(k - 1) / 2 two-class SVM classifiers need to be constructed. When constructing the two-class SVM classifier between the i-th class and the j-th class, the data in the training set only comes from the corresponding two classes, and the points of the i-th class and the j-th class are respectively marked as +1 and -1. During testing, the test data is respectively substituted into the N = k(k - l) / 2 classifiers constructed before for testing, the scores of each category are accumulated, and the category corresponding to the highest scorer is selected as the category to which the test data belongs.

[0165] The "one-against-all" method is to construct k two-class SVM classifiers according to the number of classes k to be classified. Each class corresponds to one of them. The i-th two-class SVM classifier marks the samples in the i-th class as +1, while all other samples are marked as -1. During testing, the decision function values corresponding to each two-class classifier are respectively calculated for the test data, and the category corresponding to the largest function value is selected as the category to which the test data belongs.

[0166] In the embodiment of the present application, target data can be collected by a dual-radar microwave traffic detector (MTD), denoted as MTD data.

[0167] Preferably, before performing the step of extracting feature values, the following steps are further included:

[0168] Perform pre-false alarm processing on the MTD data. By performing two-level FFT on the echo data and briefly processing the data of some range cells and frequency channels, the influence of clutter can be minimized. Then, perform constant false alarm processing on the echo data preprocessed by two-level FFT to determine whether there is a target and obtain relevant information such as the distance and spectrum of the target. The radar echo data not only contains useful target information but also various interference information. The detection of radar targets is carried out in an environment of various interferences. Through constant false alarm processing, a higher target detection probability can be obtained from the echo information, which is of great significance for target feature extraction and recognition.

[0169] After the MTD data undergoes pre - false alarm processing and constant false alarm processing, the influence of clutter can be effectively reduced, and the detection probability of the target can be improved, which is more conducive to the extraction and recognition of eigenvalues.

[0170] The following combines Application Scenario Example 1. For example, Figure 4 taking the classification targets as people and vehicles, the application of the target classification and recognition method provided by the embodiments of the present application is described as follows:

[0171] 1) Collect target data, denoted as MTD data.

[0172] 2) Perform pre - false alarm processing and constant false alarm processing on the MTD data.

[0173] 3) Use the relative RCS feature, speed, and Doppler spectrum entropy algorithm for feature extraction, relative RCS feature - relative RCS value, speed - instantaneous speed feature quantity, Doppler spectrum entropy - Doppler spectrum entropy value.

[0174] 4) Process through the SVM classifier, set different conditions for different feature classifications;

[0175] The first judgment element speed is derived from the default speed demarcation point between people and vehicles. V = 10m / s is used as the demarcation point. If it is greater than 10m / s, it is initially judged as a vehicle; if it is less than 10m / s, it is judged as a person or a vehicle.

[0176] 5) When V > 10m / s, use the SVM classifier (RCS feature) for further verification. If it is a vehicle, it is judged as a vehicle; if it is not a vehicle, then use the SVM classifier (spectrum entropy feature) for further verification. If it meets the criteria, it is a vehicle; if it does not meet the criteria, it is an unknown item.

[0177] 6) When V < 10m / s, use the SVM classifier (RCS feature) for further verification. If it is a person, it is judged as a person; if it is not a person, and if V < 2m / s and RCS < 224 are satisfied, then it is a vehicle. If V < 2m / s and RCS < 224 are not satisfied, then use the SVM classifier (spectrum entropy feature) for further verification. If it meets the criteria, it is a person; if it does not meet the criteria, it is an unknown item.

[0178] The comprehensive recognition rate of the target based on the two features of relative RCS and speed is 98.96%. The comprehensive recognition rate of the target based on the two features of Doppler spectrum entropy and speed is 97.82%. The comprehensive recognition rate of the target based on the three combined features of relative RCS, Doppler spectrum entropy, and speed is 99.62%. Obviously, the accuracy of target classification and recognition has been improved.

[0179] It can be seen that both the relative RCS feature and the Doppler spectrum entropy feature can better distinguish cars from people, but the recognition rate is not very high. In the embodiments of the present application, algorithms such as relative RCS feature, speed, and Doppler spectrum entropy are used for feature extraction. The first judgment factor selected is speed, mainly because when using separate SVM (relative RCS) and separate SVM (spectrum entropy feature) for discriminating uniform and variable-speed moving objects in a complex ground environment, the misjudgment rate is relatively high. Therefore, speed is used as the first discrimination factor. The second discrimination factor is SVM (relative RCS), mainly based on the fact that the accuracy rate of the relative RCS + speed + SVM experiment > the accuracy rate of the spectrum entropy feature + speed + SVM experiment. Therefore, SVM (relative RCS) is used as the second discrimination factor.

[0180] Special case: When a person carries a metal object similar to a corner reflector and is detected by a certain type of radar. When the electromagnetic wave of a certain type of radar scans this object, the electromagnetic wave will refract and amplify at the metal corner, generating a very strong echo signal, and the radar will receive a very strong echo signal. The result of this situation is that the RCS value returned by target objects such as humans is relatively large when received and extracted, which is likely to be confused with large target objects such as vehicles, thus increasing the misjudgment rate of the target.

[0181] In this case, selecting speed as the first judgment factor will be more accurate and improve the accuracy rate of target recognition.

[0182] Especially when facing uniform and variable-speed moving objects in a complex ground environment, a higher recognition accuracy can be achieved based on multiple combined features of relative RCS, Doppler spectrum entropy, and speed; even for a low-resolution radar, using the method of multiple combined features can accurately detect ground moving objects, reducing the equipment replacement cost and project cost; compared with the method of using a single relative RCS or Doppler spectrum entropy or speed feature, using the method of multiple combined features will improve the recognition accuracy. Combining the three algorithms with SVM multi-level hierarchical verification can reduce the misjudgment rate for uniform and variable-speed moving objects in a complex ground environment and increase the anti-interference ability.

[0183] In the embodiments of the present application, the first eigenvalue, the second eigenvalue, and the third eigenvalue are all from the speed eigenvalue, the Doppler spectrum entropy eigenvalue, and the RCS eigenvalue, and the first eigenvalue, the second eigenvalue, and the third eigenvalue are different types of eigenvalues. Therefore, in practice, it can be set correspondingly according to needs. When the targets are cars and people, preferably, the first eigenvalue is the speed eigenvalue, the second eigenvalue is the RCS eigenvalue, and the third eigenvalue is the Doppler spectrum entropy eigenvalue.

[0184] To prove the effects of the embodiments of the present application, relevant experiments were carried out, which are described as follows:

[0185] I. Target Classification and Recognition Results Based on RCS Features and Speed

[0186] Based on the data collected from two field experiments of a certain type of radar, 126 relative RCS features of vehicles and 126 relative RCS features of humans were extracted in the experiment. As Figure 5.1 shown in the distribution of RCS features of vehicles and humans. Figure 5.2 It is the relative RCS distribution histogram of vehicles and humans.

[0187] Figure 5.1 There is a large difference in RCS between the first half and the second half. It is preliminarily judged that it may be the influence of the pulse repetition frequency. The pulse repetition frequency of the first half is 109 μs, and the pulse repetition frequency of the second half is 108 μs.

[0188] Its classification and recognition process is as Figure 5.3 shown.

[0189] Randomly select 100 RCS samples of vehicles and 100 RCS samples of humans from 126 RCS feature samples of vehicles and humans as training samples, and the remaining 52 samples as test samples. A total of 100 random experiments are carried out. The results are shown in Table 4.1. It can be seen from the table that based on speed and relative RCS features, vehicles and humans can be classified and recognized well, and the recognition rate is higher than 90%.

[0190] Table 4.1 Classification Results of Vehicles and Humans Based on Relative RCS Features

[0191]

[0192] It should be noted that a total of 100 random experiments were carried out. 57 represents the number of times of 100% recognition rate, 32 represents the number of times of 98.08% recognition rate, 11 represents the number of times of 96.15% recognition rate, and the comprehensive recognition rate is: (100% * 57 + 98.08 * 32 + 96.15 * 11) / 100 = 98.96%.

[0193] II. Target Classification and Recognition Results Based on Doppler Spectrum Entropy Feature + Speed

[0194] Based on the data collected from two field experiments of a certain type of radar, the Doppler spectrum entropy of the target is calculated by using 20 Doppler cells before and after the Doppler cell where the target is located. 109 Doppler spectrum entropies of vehicles and 109 Doppler spectrum entropies of humans were extracted in the experiment. As Figure 6.1 shown in the distribution of Doppler spectrum entropy features of vehicles and humans. It can be seen from the figure that the spectral entropy features of humans and vehicles overlap less, and using this feature can better distinguish humans and vehicles. Its classification and recognition process is as Figure 6.2 shown.

[0195] Randomly select 90 samples from the 109 Doppler spectrum entropy features of humans and vehicles as training samples, and the remaining 38 as test samples. A total of 100 random experiments are conducted, and the experimental results are shown in Table 4.2:

[0196] Table 4.2 Classification results of vehicles and humans based on speed and Doppler spectrum entropy features

[0197]

[0198] Among them, the meanings of the parameters in the table and the calculation method of the comprehensive recognition rate refer to the remarks under Table 4.1.

[0199] It can be seen from the experimental results that by extracting the speed and Doppler spectrum entropy features of the target, vehicles and humans can also be well distinguished.

[0200] III. Target classification and recognition results based on the joint features of speed, relative RCS, and Doppler spectrum entropy

[0201] Based on the data collected from two field experiments of a certain type of radar, 109 samples of each of the relative RCS features, speed, and Doppler spectrum entropy of humans are extracted, and 109 samples of each of the relative RCS features, speed, and Doppler spectrum entropy of vehicles are also extracted. When using the SVM classifier for classification, 90 samples are randomly selected from the 109 features of humans and vehicles as training samples, and the remaining 38 as test samples. A total of 200 random experiments are conducted, and the experimental results are shown in Table 4.3.

[0202] Table 4.3 Recognition results of humans and vehicles based on three features of speed, spectrum entropy, and RCS

[0203]

[0204] Among them, a total of 200 random experiments are conducted. 171 represents the number of times of 100% recognition rate, 29 represents the number of times of 97.37% recognition rate, and the others represent the number of times of 0% recognition rate. The comprehensive recognition rate is: (100% * 57 + 98.08 * 32 + 96.15 * 11) / 100 = 98.96%.

[0205] Comparing Table 4.3 with Table 4.1 and Table 4.2, it can be seen that the comprehensive recognition rate of target recognition by combining three features is higher than that of speed plus a single feature, and the probability of 100% recognition rate has also been greatly improved.

[0206] The comprehensive recognition rate of the target based on two features of relative RCS and speed is 98.96%. The comprehensive recognition rate of the target based on two features of Doppler spectrum entropy and speed is 97.82%. The comprehensive recognition rate of the target based on three combined features of relative RCS, Doppler spectrum entropy and speed is 99.62%. That is, whether using a single feature of acceleration or combining two features of acceleration, the classification and recognition of vehicles and humans can be better realized. The computational complexity of using combined features for recognition is higher than that of using a single feature, but the recognition accuracy is also higher.

[0207] Corresponding to the method in the above embodiment, Figure 8 The structural block diagram of the classification and recognition device of the target provided by the embodiment of the present application is shown. For the sake of convenience of description, only the part related to the embodiment of the present application is shown. Figure 8 The exemplary classification and recognition device of the target may be the execution subject of the classification and recognition method of the target provided in the foregoing Embodiment 1.

[0208] Referring to Figure 8 , the classification and recognition method device of the target includes:

[0209] An extraction module 81, configured to extract feature values from the target data collected by the radar, where the feature values at least include a first feature value and a second feature value;

[0210] A classification and recognition module 82, configured to perform a first classification on the target according to the first feature value; perform a second classification on the target according to the second feature value; and identify the category of the target by combining the results of at least two classifications.

[0211] For the process of each module in the classification and recognition device of the ground target provided by the embodiment of the present application to realize its respective functions, reference may be specifically made to the description in the foregoing related embodiments of the method, which will not be elaborated here.

[0212] It should be understood that the magnitudes of the sequence numbers of the steps in the above embodiments do not mean the order of execution. The order of execution of each process should be determined according to its function and internal logic, and should not constitute any limitation to the implementation process of the embodiments of the present application.

[0213] It should be understood that when used in the specification and the appended claims of the present application, the term "comprising" indicates the presence of the described features, wholes, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components and / or their combinations.

[0214] It should also be understood that the term " / and" as used in the specification and the appended claims of the present application refers to any combination and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0215] As used in the specification of this application and the appended claims, the term "if" can be construed, depending on the context, as "when", "once", "in response to determining", or "in response to detecting". Similarly, the phrase "if determined" or "if [the described condition or event] is detected" can be construed, depending on the context, as meaning "once determined", "in response to determining", "once [the described condition or event] is detected", or "in response to detecting [the described condition or event]".

[0216] In addition, in the description of the specification of this application and the appended claims, the terms "first", "second", "third", etc. are only used for distinguishing descriptions and should not be construed as indicating or implying relative importance. It should also be understood that although the terms "first", "second", etc. are used in the text in some embodiments of this application to describe various elements, these elements should not be limited by these terms. These terms are only used to distinguish one element from another. For example, the first table can be named the second table, and similarly, the second table can be named the first table, without departing from the scope of the various described embodiments. The first table and the second table are both tables, but they are not the same table.

[0217] Reference to "one embodiment" or "some embodiments" or the like described in the specification of this application means that a specific feature, structure, or characteristic described in connection with that embodiment is included in one or more embodiments of this application. Thus, statements such as "in one embodiment", "in some embodiments", "in other some embodiments", "in still other embodiments", etc. that appear in different places in this specification are not necessarily all referring to the same embodiment, but mean "one or more but not all embodiments", unless otherwise specifically emphasized in another way. The terms "comprising", "including", "having", and their variants all mean "including but not limited to", unless otherwise specifically emphasized in another way.

[0218] The method for classifying and identifying ground targets provided by the embodiments of this application can be applied to terminal devices such as mobile phones, tablet computers, wearable devices, vehicle-mounted devices, augmented reality (AR) / virtual reality (VR) devices, laptop computers, ultra-mobile personal computers (UMPCs), netbooks, personal digital assistants (PDAs), etc. The embodiments of this application do not impose any restrictions on the specific types of terminal devices.

[0219] For example, the terminal device may be a station (STA) in a WLAN, a cellular phone, a cordless phone, a Session Initiation Protocol (SIP) phone, a Wireless Local Loop (WLL) station, a Personal Digital Assistant (PDA) device, a handheld device with wireless communication capabilities, a computing device, or other processing devices connected to a wireless modem, a vehicle-mounted device, a vehicle-to-everything (V2X) terminal, a computer, a laptop computer, a handheld communication device, a handheld computing device, a satellite wireless device, a wireless modem card, a set-top box (STB), a customer premise equipment (CPE), and / or other devices for communicating on a wireless system, as well as next-generation communication systems, such as a mobile terminal in a 5G network or a mobile terminal in a future evolved Public Land Mobile Network (PLMN).

[0220] By way of example and not limitation, when the terminal device is a wearable device, the wearable device may also be a general term for devices that are intelligently designed for daily wear using wearable technology, such as glasses, gloves, watches, clothing, and shoes. A wearable device is a portable device that is either worn directly on the body or integrated into the user's clothing or accessories. A wearable device is not just a hardware device, but also achieves powerful functions through software support, data interaction, and cloud interaction. Broadly speaking, wearable intelligent devices include those with comprehensive functions and large sizes that can achieve complete or partial functions without relying on a smartphone, such as smart watches or smart glasses, as well as those that focus only on a certain type of application function and need to cooperate with other devices such as smartphones, such as various smart bracelets and smart jewelry for monitoring physical signs.

[0221] Figure 7 It is a schematic structural diagram of a terminal device provided by an embodiment of the present application. As Figure 7 shown, the terminal device 7 of this embodiment includes at least one processor 70 ( Figure 7 only one is shown in the figure), and a memory 71. A computer program 72 that can run on the processor 70 is stored in the memory 71. When the processor 70 executes the computer program 72, it implements the steps in the classification and recognition embodiments of the above various objectives, such as Figure 1 the steps 11 to 14 shown in the figure. Alternatively, when the processor 70 executes the computer program 72, it implements the functions of each module / unit in the above device embodiments, such as Figure 8Functions of the modules 81 to 82 shown

[0222] The terminal device 7 may be a computing device such as a desktop computer, a notebook, a palm computer, and a cloud server. The terminal device may include, but is not limited to, a processor 70 and a memory 71. Those skilled in the art can understand that Figure 7 These are merely examples of the terminal device 7 and do not constitute a limitation on the terminal device 7. It may include more or fewer components than shown in the figure, or combine certain components, or have different components. For example, the terminal device may further include an input and sending device, a network access device, a bus, etc.

[0223] The so-called processor 70 may be a central processing unit (CPU), or may also be other general-purpose processors, digital signal processors (DSPs), application specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0224] In some embodiments, the memory 71 may be an internal storage unit of the terminal device 7, such as the hard disk or memory of the terminal device 7. The memory 71 may also be an external storage device of the terminal device 7, such as a plug-in hard disk equipped on the terminal device 7, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. Further, the memory 71 may also include both the internal storage unit and the external storage device of the terminal device 7. The memory 71 is used to store an operating system, application programs, a boot loader, data, and other programs, such as the program code of the computer program, etc. The memory 71 may also be used to temporarily store data that has been sent or will be sent.

[0225] In addition, in each embodiment of the present application, the functional units may be integrated in one processing unit, or each unit may exist physically alone, or two or more units may be integrated in one unit. The above integrated units may be implemented in the form of hardware or in the form of software functional units.

[0226] An embodiment of the present application further provides a terminal device, which includes at least one memory, at least one processor, and a computer program stored in the at least one memory and executable on the at least one processor. When the processor executes the computer program, the terminal device implements the steps in any of the above method embodiments.

[0227] An embodiment of the present application further provides a computer-readable storage medium storing a computer program, which when executed by a processor can implement the steps in the above method embodiments.

[0228] An embodiment of the present application provides a computer program product, which when running on a terminal device enables the terminal device to implement the steps in the above method embodiments when executed.

[0229] If the integrated module / unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, to implement all or part of the processes in the above method embodiments of the present application, it can also be completed by a computer program instructing relevant hardware. The computer program can be stored in a computer-readable storage medium. When the computer program is executed by a processor, the steps in the above method embodiments can be implemented. Among them, the computer program includes computer program code, which can be in the form of source code, object code, executable file, or some intermediate form, etc. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disc, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal, and software distribution medium, etc.

[0230] In the above embodiments, the descriptions of the respective embodiments have their own emphases. For parts not detailed or recorded in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0231] Those of ordinary skill in the art can realize that the units and algorithm steps of the examples described in combination with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. A professional technician can use different methods for each specific application to implement the described functions, but such implementation should not be considered to exceed the scope of the present application.

[0232] The unit described as a separating component may or may not be physically separated. The component shown as a unit may or may not be a physical unit, that is, it may be located in one place or may be distributed over multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0233] The above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present application, and should all be included in the protection scope of the present application.

Claims

1. A method for classifying and recognizing a target, characterized in that, Including: Extracting eigenvalue from target data collected by radar, where the eigenvalue at least includes a first eigenvalue and a second eigenvalue; The first eigenvalue is a speed eigenvalue; Performing a first classification on the target according to the first eigenvalue, including: when the speed eigenvalue is greater than a first preset value, the target is first classified as a vehicle; when the speed eigenvalue is not greater than the first preset value, the target is first classified as uncertain; where uncertain means that it could be either a vehicle or a person; Performing a second classification on the target according to the second eigenvalue; Combining the results of at least two classifications to identify the category of the target, including: when the speed eigenvalue is greater than the first preset value, the target is first classified as a vehicle; when the speed eigenvalue is not greater than the first preset value, the target is first classified as uncertain; When the first classification is a vehicle and the second classification is not a vehicle, or when the first classification is uncertain, the second classification is not a person, and the speed eigenvalue and the second eigenvalue are not both within further limiting conditions, then a third classification is performed on the target according to a third eigenvalue; one of the RCS eigenvalue and the Doppler spectrum entropy eigenvalue is the second eigenvalue, and the other is the third eigenvalue; Combining the results of at least two classifications to identify the category of the target, including: When the first classification is a vehicle and the second classification is not a vehicle, if the third classification is a vehicle, then the category of the target is identified as a vehicle; when the first classification is uncertain, the second classification is not a person, and the speed eigenvalue and the second eigenvalue are not both within further limiting conditions, if the third classification is a person, then the target is identified as a person.

2. The method according to claim 1, characterized in that, When the first classification is uncertain and the second classification is not a person, the method further includes: When the speed eigenvalue and the second eigenvalue are both within further limiting conditions, then the category of the target is identified as a vehicle.

3. The method according to claim 1, wherein Performing the second classification on the target according to the second eigenvalue includes: Performing a second classification on the second eigenvalue of the target through an SVM classifier; Performing a third classification on the target according to the third eigenvalue includes: Performing a third classification on the third eigenvalue of the target through the SVM classifier.

4. The method according to claim 1, wherein The eigenvalue further includes a third eigenvalue, and the method further includes: When the first classification is a vehicle and the second classification is not a vehicle, or when the first classification is uncertain and the second classification is not a person, then a third classification is performed on the target according to the third eigenvalue; Combining the results of at least two classifications to identify the category of the target, including: When the first classification is a vehicle and the second classification is not a vehicle, if the third classification is a vehicle, then the category of the target is identified as a vehicle; when the first classification is uncertain and the second classification is not a person, if the third classification is a person, then the target is identified as a person.

5. A classification and recognition device for an object, characterized in that, Including: An extraction module for extracting eigenvalue from target data collected by radar, where the eigenvalue at least includes a first eigenvalue and a second eigenvalue; The first eigenvalue is a speed eigenvalue; The classification and recognition module is used to perform the first classification on the target according to the first eigenvalue, and includes: when the speed eigenvalue is greater than the first preset value, the target is classified as a vehicle for the first time; when the speed eigenvalue is not greater than the first preset value, the target is classified as uncertain for the first time; where uncertain means that it could be either a vehicle or a person; perform the second classification on the target according to the second eigenvalue; combine the results of at least two classifications to identify the category of the target, including: when the first classification is a vehicle and the second classification is a vehicle, then identify the category of the target as a vehicle; when the first classification is uncertain and the second classification is a person, then identify the category of the target as a person; when the first classification is a vehicle and the second classification is not a vehicle, or when the first classification is uncertain, the second classification is not a person, and the speed eigenvalue and the second eigenvalue are not both within further defined conditions, then perform the third classification on the target according to the third eigenvalue; use one of the RCS eigenvalue and the Doppler spectrum entropy eigenvalue as the second eigenvalue, and the other as the third eigenvalue; when the first classification is a vehicle and the second classification is not a vehicle, if the third classification is a vehicle, then identify the category of the target as a vehicle; when the first classification is uncertain, the second classification is not a person, and the speed eigenvalue and the second eigenvalue are not both within further defined conditions, if the third classification is a person, then identify the target as a person.

6. A terminal device, characterized in that, The terminal device includes a memory and a processor. A computer program that can run on the processor is stored on the memory. When the processor executes the computer program, the steps of the method according to any one of claims 1 to 4 are implemented.

7. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, the steps of the method according to any one of claims 1 to 4 are implemented.

Images