Radar target detection method, system and equipment based on two-feature fusion and medium

Abstract

The invention provides a radar target detection method, system and device based on two-feature fusion and a medium, and belongs to the technical field of radar signal processing. The method comprises the following steps: constructing training data by using existing radar target detection data, extracting phase linearity based on a training data extraction process, and extracting a relative peak height of a Doppler spectrum by using a relative peak height extraction process; based on the phase linearity and the relative peak height, a fusion matrix of the two features is obtained by using a distribution compactness linear dimensionality reduction method; acquiring reference distance unit data and to-be-detected distance unit data in real time, respectively executing an extraction process of phase linearity and an extraction process of relative peak height, and generating fusion features by using the fusion matrix; and on the basis of the fusion features, generalized extreme value distribution parameters are estimated, detection statistics are constructed, a self-adaptive threshold is calculated in combination with a preset false alarm probability, and target judgment is completed by using comparison between the detection statistics and the threshold, so that constant false alarm detection under clutter suppression is realized.

Description

Technical Field

[0001] The present invention belongs to the technical field of radar signal processing, and more specifically, relates to a radar target detection method, system, device and medium based on two-feature fusion. Background Art

[0002] The detection of weak targets in a strong sea clutter background has always been a widely studied topic in the radar field. Radar target detectors can generally be divided into energy detectors and feature detectors. Energy detectors utilize the differences in the positions and / or energy accumulation degrees of targets and clutter in various domains to enhance the signal-to-clutter ratio of target units, thereby improving target detection performance; while feature detectors utilize the differences in a certain feature between targets and clutter, and enhance the contrast between target units and clutter units by extracting this feature, thereby improving detection performance.

[0003] In recent years, the research on feature detectors has been increasing. Some features have high sensitivity to low signal-to-clutter ratios, but most features may not have such good properties, especially under short time series conditions. Therefore, using a multi-feature joint detector for radar target detection can significantly improve the detection performance of weak sea targets by fully utilizing the information volume in the high-dimensional space.

[0004] Currently, the decision space of multi-feature joint detectors is usually solved using convex hull or concave hull learning algorithms. The accuracy of the obtained decision space depends to a large extent on the number of multi-feature samples under target-free conditions, the geometric shape of the feature sample distribution in the multi-dimensional feature space, and the accuracy of the convex hull / concave hull algorithm itself. Analysis of measured data shows that the existence of decision space estimation errors seriously affects the detection performance of multi-feature joint detectors. An effective solution is feature dimensionality reduction, especially fusing multiple features into one feature for detection. At this time, the decision space degenerates into a detection threshold. Under the condition of the same data volume, compared with estimating the decision space, the estimation error of the detection threshold will be significantly reduced. Summary of the Invention

[0005] Aiming at the above problems, the purpose of the present invention is to provide a radar target detection method, system, device and medium based on two-feature fusion. By fusing two features of phase linearity and relative peak height, and using linear dimensionality reduction and generalized extreme value distribution modeling, constant false alarm radar target detection under clutter suppression is achieved, and the detection performance is improved.

[0006] To achieve the above object, the present invention is realized through the following technical solutions: In a first aspect, an embodiment of the present application provides a radar target detection method based on two-feature fusion, including: Construct training data using existing radar target detection data, extract the phase linearity based on the training data extraction process, and extract the relative peak height of the Doppler spectrum using the relative peak height extraction process; Based on the phase linearity and relative peak height, obtain the fusion matrix of the two features using the linear dimensionality reduction method of distribution compactness; Obtain the reference range cell data and the data of the range cell to be detected in real time, respectively execute the extraction process of the phase linearity and the extraction process of the relative peak height, and generate the fusion feature using the fusion matrix; Based on the fusion feature, estimate the generalized extreme value distribution parameters and construct the detection statistic, calculate the adaptive threshold in combination with the preset false alarm probability, and complete the target decision using the comparison between the detection statistic and the threshold to achieve constant false alarm detection under clutter suppression.

[0007] In an alternative embodiment, the extraction process of the phase linearity includes: Represent the wrapped phase of the th data in the training data as:

[0008] Let the difference between adjacent wrapped phases be , and adjust it to between and through the following formula to eliminate jumps:

[0009] Then the unwrapped true phase is represented as:

[0010] Let the phase difference between two adjacent phases in the true phase sequence be , then the phase linearity is represented as:

[0011] where , then the normalized phase linearity is: .

[0012] In an alternative embodiment, the extraction process of the relative peak height includes: Let the th data of each range cell be , then the Doppler amplitude spectrum can be expressed by the following formula:

[0013] where is the Doppler frequency, is the pulse repetition period; Let the peak value of the Doppler amplitude spectrum be , then the relative peak height can be expressed as:

[0014] where, is the range of the Doppler side lobe, is the number of points of the Doppler frequency within the range; The normalized relative peak height is: .

[0015] In an optional embodiment, the fusion matrix of the two features is obtained by using the linear dimensionality reduction method based on the distribution compactness according to the phase linearity and the relative peak height, including: Generating a feature data set based on the phase linearity and the relative peak height ; where, is the number of samples in the feature data set, and the th feature data sample contains the phase linearity feature and the relative peak height feature , that is:

[0016] After linear dimensionality reduction processing, a fused feature data set can be obtained:

[0017] Let , respectively represent the expectation and variance of the fused feature , then the kurtosis calculation formula is as follows:

[0018] where, represents taking the expectation; Let The probability distribution function of is , then the interquartile range is expressed as:

[0019] where, , are the lower quartile and upper quartile of respectively; Then the objective function is defined as follows:

[0020] Among them, ; In the interval, traverse , when the objective function reaches the minimum value, the corresponding fusion matrix is the optimal weighted vector of the two features of PL and RPH.

[0021] In an alternative embodiment, the real-time acquisition of reference distance unit data and data of the distance unit to be detected includes: Set the marine radar to work in the coherent processing mode, and the number of coherent pulse trains is . After quadrature demodulation and matched filtering, the I / Q data in the distance unit to be detected is denoted as , and the I / Q data in the reference distance unit is denoted as ; Among them, represents the distance unit to be detected, is the number of reference distance units, and the subscript represents the th reference distance unit; Denote the possible target I / Q data in the distance unit to be detected as , and denote the clutter I / Q data in the distance unit to be detected and the reference distance unit as and respectively.

[0022] In an alternative embodiment, the estimation of the generalized extreme value distribution parameters and the construction of the detection statistic based on the fusion features include: Model the probability density function of the fusion features using the generalized extreme value distribution, specifically as follows:

[0023] Among them is the fusion feature variable, is the location parameter, is the scale parameter, is the shape parameter; Use the probability weighted moment estimation method to obtain the expression of the estimator of the shape parameter as:

[0024] Among them, , , , ; From reference distance units, obtain fusion feature samples , and after arranging them in ascending order, the sequence ; Using the probability weighted moment estimation method, the estimator expression of the scale parameter is:

[0025] Using the probability weighted moment estimation method, the estimator expression of the location parameter is:

[0026] Wherein, is the Gamma function; According to the form of the probability density function of the GEV distribution of the fused feature under the condition of no target data, the following detection statistic is designed:

[0027] When the number of samples of the fused feature tends to infinity, such that , , , the asymptotic probability density function of the detection statistic can be deduced as: .

[0028] In an alternative embodiment, the calculating the adaptive threshold in combination with a preset false alarm probability includes: Calculating the detection threshold corresponding to the false alarm probability through the following formula: .

[0029] In a second aspect, an embodiment of the present application further provides a radar target detection system based on two-feature fusion, including: A data extraction module, configured to construct training data by using existing radar target detection data, extract the phase linearity based on the training data extraction process, and extract the relative peak height of the Doppler spectrum by using the relative peak height extraction process; A fusion rule generation module, configured to obtain a fusion matrix of the two features by using a linear dimensionality reduction method of distribution compactness based on the phase linearity and the relative peak height; An actual measurement data fusion module, configured to obtain reference range cell data and data of the range cell to be detected in real time, respectively execute the extraction process of the phase linearity and the extraction process of the relative peak height, and generate a fusion feature by using the fusion matrix; A target detection module, which is used to estimate the generalized extreme value distribution parameters and construct a detection statistic based on the fused features, calculate an adaptive threshold in combination with a preset false alarm probability, and complete target decision-making by comparing the detection statistic with the threshold, so as to achieve constant false alarm detection under clutter suppression.

[0030] In a third aspect, an embodiment of the present application further provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the program, the steps of the radar target detection method based on two-feature fusion as described in any one of the above are implemented.

[0031] In a fourth aspect, an embodiment of the present application further provides a storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the radar target detection method based on two-feature fusion as described in any one of the above are implemented.

[0032] From the above technical solutions, it can be seen that the present invention has the following advantages: In the radar target detection method based on two-feature fusion provided by the present application, the target detection performance is improved by fusing two key features in radar target detection, namely phase linearity and relative peak height. First, a linear dimensionality reduction technique for distribution compactness is used to perform dimensionality reduction processing on phase linearity and relative peak height. In this process, through linear transformation, the two features are fused into a more representative fused feature, effectively reducing the dimensionality of the features and improving the efficiency of subsequent processing. Second, the generalized extreme value distribution is used to model the fused feature. By modeling the probability density function, the statistical characteristics of the fused feature can be more accurately described, providing a basis for the construction of subsequent detection statistics. Finally, based on the generalized extreme value distribution model, a detection statistic is constructed, and an adaptive threshold is calculated in combination with a preset false alarm probability. In this process, by comparing the detection statistic with the threshold, accurate decision-making on the target is achieved, effectively suppressing clutter interference, and improving the accuracy and reliability of radar target detection.

[0033] The present application deeply explores two key features in radar target detection, namely phase linearity and relative peak height, and uses linear dimensionality reduction techniques such as principal component analysis (PCA) or linear discriminant analysis (LDA) to perform dimensionality reduction and fusion processing on these two features. Through this technical means, a more discriminative fused feature is formed, effectively reducing the redundant information of the features, and improving the characterization ability of the features and the accuracy of radar target detection.

[0034] In this application, the generalized extreme value distribution (GEV) is used to model the fused features. Through the estimation of the probability density function and parameter fitting, the statistical characteristics of the fused features are more accurately described. This modeling process takes into account the possible distribution patterns of the fused features, enhancing the robustness and self - adaptability of the detection. In a complex clutter environment, the generalized extreme value distribution modeling enables the detection statistic to have a stronger resistance to clutter interference, improving the reliability and stability of the detection.

[0035] This application combines a preset false - alarm probability and uses the constant false - alarm rate (CFAR) detection technology to automatically adjust the detection threshold according to environmental changes. By real - time estimating the statistical characteristics of the background clutter and calculating the adaptive threshold accordingly, the constant false - alarm rate radar target detection is achieved. This detection method effectively suppresses clutter interference and noise effects, ensuring the stability and consistency of the detection performance, and enabling the method to maintain excellent detection effects under different environments and conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] To more clearly illustrate the technical solutions of the present invention, the drawings required for description will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0037] Figure 1 It is a schematic flow chart of the radar target detection method based on two - feature fusion provided by this application.

[0038] Figure 2 It is a principle block diagram of the radar target detection method based on two - feature fusion provided by this application.

[0039] Figure 3 It is a two - dimensional feature data point diagram provided by this application.

[0040] Figure 4 It is a histogram of the fused features provided by this application.

[0041] Figure 5 It is a schematic diagram of the fitting result of the probability density function of the fused feature data provided by this application.

[0042] Figure 6 It is a schematic structural diagram of the radar target detection system based on two - feature fusion provided by this application.

[0043] Figure 7 It is a schematic structural diagram of the electronic device provided by this application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0044] In the following detailed description of the specific steps of the radar target detection method based on two - feature fusion, various embodiments of the present disclosure will be more comprehensively described. The present disclosure can have various embodiments, and adjustments and changes can be made therein. However, it should be understood that there is no intention to limit the various embodiments of the present disclosure to the specific embodiments disclosed herein, but the present disclosure should be understood to cover all adjustments, equivalents, and / or alternative solutions that fall within the spirit and scope of the various embodiments of the present disclosure.

[0045] Hereinafter, the term "comprising" or "may comprise" that can be used in various embodiments of the present disclosure indicates the presence of the disclosed functions, operations, or elements, and does not limit the addition of one or more functions, operations, or elements. In addition, as used in various embodiments of the present disclosure, the terms "comprising", "having" and their cognates are only intended to indicate a specific feature, number, step, operation, element, component, or combination of the foregoing items, and should not be construed as precluding the existence or addition of one or more other features, numbers, steps, operations, elements, components, or combinations of the foregoing items first.

[0046] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without making creative efforts belong to the scope of protection of the present invention.

[0047] Please refer to Figure 1 The following is a flowchart of a method for a radar target detection method based on two - feature fusion in a specific embodiment. The method includes: S1: Use the existing radar target detection data to construct training data, extract the phase linearity based on the training data using the extraction process of phase linearity, and extract the relative peak height of the Doppler spectrum using the extraction process of relative peak height.

[0048] S2: Based on the phase linearity and relative peak height, obtain the fusion matrix of the two features using the linear dimensionality reduction method of distribution compactness.

[0049] S3: Use the signal model to obtain the reference distance unit data and the data of the distance unit to be detected in real - time, respectively execute the extraction process of phase linearity and the extraction process of relative peak height, and generate the fusion feature using the fusion matrix.

[0050] S4: Based on the fused features, estimate the parameters of the generalized extreme value distribution and construct a detection statistic. Combine the preset false alarm probability to calculate the adaptive threshold, and use the comparison between the detection statistic and the threshold to complete the target decision, so as to achieve constant false alarm detection under clutter suppression.

[0051] In this embodiment, the radar target detection method based on two-feature fusion is divided into two stages: training and detection. As Figure 2 shown, the purpose of the training stage is to use the training data to obtain the fusion rules of the two features, PL and RPH, through a linear dimensionality reduction method based on distribution compactness, where the training data refers to a large number of I / Q data of range cells without targets obtained under the same observation conditions. The detection stage is based on the data of the range cell to be detected and the reference range cell data, uses the fusion rules to form fused features, and on this basis constructs a detection statistic to complete the threshold-crossing detection.

[0052] In this embodiment, by cleverly fusing the two complementary features of phase linearity and relative peak height, and using the linear dimensionality reduction technology of distribution compactness to achieve effective feature fusion, and at the same time combining the generalized extreme value distribution to model the fused features, an efficient detection statistic is constructed, so as to achieve constant false alarm radar target detection in a complex clutter environment, greatly improving the accuracy, robustness and adaptability of target detection.

[0053] In an embodiment of the present invention, based on the extraction processes of phase linearity and relative peak height disclosed in step S1 and step S3, a possible embodiment will be given below to non-restrictively elaborate on its specific implementation.

[0054] In the present invention, the extraction processes of the disclosed phase linearity and relative peak height are essentially feature extraction processes. The basic principle is to use the N I / Q data obtained in each range cell to extract two features, the phase linearity and the relative peak height of the Doppler spectrum, and normalize the two features respectively with respect to their respective maximum and minimum values.

[0055] The phase linearity describes the linear degree of the phase in the long I / Q data, briefly denoted as PL. The basic idea is that when there is a target in the I / Q data sequence, its phase linear degree will increase. Based on the above idea, the extraction process of the phase linearity of the present invention includes: 1. Represent the wrapped phase of the training data or the th data in each range cell as:

[0056] Let the difference between adjacent wrapped phases be , and adjust it to to to eliminate jumps:

[0057] Then the unwrapped true phase is expressed as:

[0058] 2. Let the phase difference between two adjacent phases of the true phase sequence be , then the phase linearity is expressed as:

[0059] where , then the normalized phase linearity is: .

[0060] The relative peak height of the Doppler spectrum describes number of data in the Doppler domain, referred to as the relative peak height, denoted as RPH. The basic idea is that the target echo energy is more concentrated in the Doppler domain than the sea clutter energy. Based on the above idea, the extraction process of the relative peak height of the present invention includes: 1. Let the th data of each range cell be , then the Doppler amplitude spectrum can be expressed by the following formula:

[0061] where is the Doppler frequency, is the pulse repetition period; 2. Let the peak value of the Doppler amplitude spectrum be , then the relative peak height can be expressed as:

[0062] where is the range of the Doppler side lobe, is the number of points of the Doppler frequency within the range; The normalized relative peak height is: .

[0063] In an embodiment of the present invention, based on step S2, a possible embodiment will be given below to non-limitingly elaborate on its specific implementation.

[0064] In step S2, a linear dimensionality reduction method with distribution compactness is disclosed. The linear dimensionality reduction problem of the present invention is to reduce the two-dimensional feature space composed of PL and RPH to a one-dimensional feature space, which is equivalent to searching for a straight line in the two-dimensional feature space so that when the two-dimensional feature data is projected onto this straight line, the objective function is optimal. Among them, the straight line search process is realized by traversing all possible direction vectors to achieve, where represents the angle with the horizontal axis; and the key lies in the design of the objective function.

[0065] In the radar target detection problem, a compact clutter data probability density function is usually more conducive to improving the detection performance. Therefore, the present invention introduces kurtosis and interquartile range to measure the compactness of the probability density function of the clutter fusion features after linear dimensionality reduction, where kurtosis reflects the sharpness of the probability density function at the mean, and the interquartile range reflects the data concentration degree.

[0066] Based on the above principle, the linear dimensionality reduction method with distribution compactness disclosed by the present invention specifically includes the following processes: Generate a feature data set based on the phase linearity and relative peak height ; among them, the phase linearity and relative peak height are generated by performing feature extraction on the training data composed of the observation data of the non-target range cells.

[0067] Referring to the above feature data set, is the number of samples in the feature data set, and the th feature data sample includes the phase linearity feature and the relative peak height feature , that is:

[0068] After linear dimensionality reduction processing, a fused feature data set can be obtained:

[0069] Let , respectively represent the expectation and variance of the fused feature , then the kurtosis calculation formula is as follows:

[0070] Among them, represents taking the expectation; Let The probability distribution function of is

[0071] Among them, , are respectively the lower quartile and the upper quartile of; Then the objective function is defined as follows:

[0072] wherein, ; In the interval of, traverse , when the objective function reaches the minimum value, the corresponding direction vector is the optimal weighted vector of the two features of PL and RPH.

[0073] When the measured data is processed by the above linear dimensionality reduction method for distribution compactness, referring to Figure 3 the two-dimensional feature data point diagram of the measured data given, and the straight line corresponding to the optimal weighted vector is marked in the figure. Referring to Figure 4 the fusion feature histogram of, where blue represents data without target and red represents data with target. It can be seen that the Pearson correlation coefficient between the two features of PL and RPH extracted based on data without target is -0.12, and its absolute value is less than 1 / e, so it can be considered that the correlation between the two is weak.

[0074] In an embodiment of the present invention, based on the signal model disclosed in step S3, the following will give a possible embodiment to non-restrictively elaborate on its specific implementation scheme.

[0075] In step S3, the reference range cell data and the data of the range cell to be detected are obtained in real time by using the signal model, including: Set the marine radar to work in the coherent processing mode, and the number of coherent pulse trains is . After orthogonal demodulation and matched filtering processing, the I / Q data in the range cell to be detected is denoted as , wherein, represents the range cell to be detected; the I / Q data in the reference range cell is denoted as , where is the number of reference range cells, and the subscript represents the th reference range cell, and there is no target echo in the reference range cell.

[0076] The possible target I / Q data in the range cell to be detected is denoted as , and the clutter I / Q data in the range cell to be detected and the reference range cell are respectively denoted as and , and the two are statistically independent and identically distributed.

[0077] Thus, the target detection problem is formulated as the following binary hypothesis testing problem:

[0078] where represents that there is no target in the distance cell to be detected, represents that there is a target in the distance cell to be detected.

[0079] It should be particularly noted that, using the feature extraction method and the linear dimensionality reduction method for distribution compactness disclosed in the above embodiments, the binary hypothesis testing problem can be expressed as

[0080] where represents the fused feature formed after linearly reducing the two-dimensional feature space composed of the PL and RPH features; represents the fused feature of the target plus clutter in the distance cell to be detected, represents the clutter fused feature in the reference distance cell. On the premise of the assumption that the clutter in each distance cell is statistically independent and identically distributed, the clutter fused features obtained separately from each distance cell are also statistically independent and identically distributed.

[0081] In an embodiment of the present invention, based on step S4, a possible embodiment will be given below to non-restrictively elaborate on its specific implementation scheme.

[0082] This embodiment discloses a generalized extreme value distribution modeling and detection statistic design process, which specifically includes the following two parts: 1. Generalized extreme value distribution modeling of the fused feature: In this embodiment, the generalized extreme value distribution (Generalized Extreme Value Distribution, GEV) is used to model the probability density function of the fused feature generated after dimensionality reduction of the features of the target-free cells. The GEV distribution is a three-parameter distribution. Compared with single-parameter or two-parameter distributions, the high degree of freedom of GEV enables it to flexibly fit the tail characteristics of different types of distributions. The distribution function and probability density function of the GEV distribution are as follows:

[0083] where is the fused feature variable; is the location parameter, which determines the central position of the distribution; is the scale parameter, which determines the extent of the distribution; is the shape parameter, which determines the shape and tail characteristics of the distribution; Using the probability weighted moment estimation method, the estimator expression of the shape parameter is as follows:

[0084] where , , , ; From reference distance units, fusion feature samples are obtained, and after arranging them in ascending order, the sequence is obtained; Using the probability weighted moment estimation method, the estimator expression of the scale parameter is as follows:

[0085] Using the probability weighted moment estimation method, the estimator expression of the location parameter is as follows:

[0086] where is the Gamma function.

[0087] Figure 5 For the histogram of the fused feature data generated after dimensionality reduction of the feature of the non-target unit obtained in Figure 4 , first, the probability density function is obtained through histogram normalization, and then the GEV distribution, exponential distribution, Rayleigh distribution, lognormal distribution, and Weibull distribution are used for fitting respectively, and the mean square error between the theoretical value and the empirical value at the tail of the probability density function is calculated.

[0088] The results show that the fitting effect of the GEV distribution is the best, as shown in Table 1 below.

[0089] Table 1: Comparison table of fitting effects

[0090] 2. Design of the detection statistic: According to the form of the probability density function of the GEV distribution of the fused feature under the condition of non-target data, the following detection statistic is designed:

[0091] When the number of samples of the fused feature tends to infinity, such that , , , the asymptotic probability density function of the detection statistic can be deduced as: .

[0092] The above formula shows that when there is no target data, as the number of samples of the fused feature tends to infinity, the asymptotic probability density function of the detection statistic is independent of any background parameter, that is, the detection statistic is asymptotically CFAR.

[0093] Therefore, the false alarm probability is calculated by the following formula and the corresponding detection threshold: .

[0094] As Figure 6 shown, the following is an embodiment of a radar target detection system based on two-feature fusion provided by the present disclosure. This system and the radar target detection method based on two-feature fusion in the above embodiments belong to the same inventive concept. For the details not described in detail in the embodiment of the radar target detection system based on two-feature fusion, reference can be made to the embodiments of the radar target detection method based on two-feature fusion above.

[0095] A radar target detection system based on two-feature fusion, comprising: A data extraction module, configured to construct training data by using existing radar target detection data, extract the phase linearity based on the training data by using the extraction process of phase linearity, and extract the relative peak height of the Doppler spectrum by using the extraction process of relative peak height.

[0096] A fusion rule generation module, configured to obtain a fusion matrix of two features by using a linear dimensionality reduction method of distribution compactness based on the phase linearity and the relative peak height.

[0097] An actual measurement data fusion module, configured to obtain reference range cell data and data of the range cell to be detected in real time by using a signal model, respectively execute the extraction process of phase linearity and the extraction process of relative peak height, and generate a fusion feature by using the fusion matrix;.

[0098] A target detection module, configured to estimate the generalized extreme value distribution parameter and construct a detection statistic based on the fusion feature, calculate an adaptive threshold in combination with a preset false alarm probability, and complete target decision by comparing the detection statistic with the threshold, so as to achieve constant false alarm detection under clutter suppression.

[0099] The radar target detection system based on two-feature fusion provided by this embodiment deeply explores and utilizes two key features in radar target detection, namely phase linearity and relative peak height, and uses linear dimensionality reduction technology for effective fusion to form more discriminative fusion features, significantly improving the detection performance. At the same time, the generalized extreme value distribution is used to model the fusion features, more accurately describing their statistical characteristics, and enhancing the robustness and self-adaptability of the detection. By combining the preset false alarm probability to calculate the adaptive threshold, constant false alarm radar target detection is achieved, effectively suppressing clutter interference and noise effects, and ensuring the stability and reliability of the detection performance.

[0100] Figure 7 Schematic diagram of the hardware structure of an electronic device for implementing each embodiment of the present invention.

[0101] The radar target detection method based on two-feature fusion provided by the embodiments of this application can be applied to electronic devices. Those skilled in the art can understand that the structure of the electronic device involved in the embodiments of the present invention does not constitute a limitation on the electronic device. The electronic device may include more or fewer components than shown in the figure, or combine certain components, or have different component arrangements. In the embodiments of the present invention, the electronic device includes, but is not limited to, laptop computers, desktop computers, workbenches, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processors, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the embodiments of this application described herein and / or claimed.

[0102] The electronic device may include a processor, an external memory interface, an internal memory, a universal serial bus (USB) interface, a charging management module, a power management module, a battery, a wireless communication module, an audio module, a speaker, a microphone, a sensor module, keys, a camera, a display screen, and a SIM card interface, etc.

[0103] The processor may include one or more processing units. For example, the processor may include a central processing unit (CPU), an application processor (AP), a modem processor, a graphics processing unit (GPU), an image signal processor (ISP), a controller, a memory, a video codec, a digital signal processor (DSP), a baseband processor, and / or a neural-network processing unit (NPU), etc. Among them, different processing units may be independent devices or integrated in one or more processors.

[0104] Among them, the processor may be the nerve center and command center of the electronic device. The controller may generate operation control signals according to the instruction operation code and timing signals to complete the control of fetching and executing instructions.

[0105] A memory may also be provided in the processor for storing instructions and data. In some embodiments, the memory in the processor is a cache memory. This memory can store the instructions or data that the processor has just used or reused. If the processor needs to use the instruction or data again, it can directly call it from this memory. This avoids repeated accesses, reduces the waiting time of the processor, and thus improves the system efficiency.

[0106] The external memory interface can be used to connect an external memory card, such as a MicroSD card, to expand the storage capacity of the electronic device. The external memory card communicates with the processor through the external memory interface to achieve the data storage function. For example, files such as music and videos are saved in the external memory card.

[0107] The internal memory can be used to store computer-executable program code, and the computer-executable program code includes instructions. The processor executes various functional applications and data processing of the electronic device by running the instructions stored in the internal memory. The internal memory may include a program storage area and a data storage area. The internal memory may include a high-speed random access memory and may also include non-volatile memory, such as at least one disk storage device, a flash memory device, a universal flash storage (UFS), etc.

[0108] The wireless communication function of the electronic device can be implemented through an antenna, a wireless communication module, a modem processor, a baseband processor, etc.

[0109] The wireless communication module can provide solutions for wireless communications applied to electronic devices, including wireless local area networks (WLANs) (such as wireless fidelity (Wi-Fi) networks), Bluetooth (BT), global navigation satellite systems (GNSSs), frequency modulation (FM), near field communication (NFC), infrared technology (IR), etc.

[0110] The electronic device can implement audio functions through an audio module, speaker, receiver, microphone, headphone jack, application processor, etc.

[0111] The electronic device can implement a shooting function through an ISP, camera, video codec, GPU, display screen, application processor, etc.

[0112] The electronic device can implement a display function through a GPU, display screen, application processor, etc.

[0113] The GPU is a microprocessor for image processing, connecting the display screen and the application processor. The GPU is used to perform mathematical and geometric calculations for graphics rendering. The processor may include one or more GPUs, which execute program instructions to generate or change display information.

[0114] The display screen is used to display images, videos, etc. The display screen includes a display panel.

[0115] The above-mentioned electronic device implements the radar target detection method based on two-feature fusion in this application. By adopting a linear dimensionality reduction technique to fuse the phase linearity and relative peak height, and combining the generalized extreme value distribution modeling and constant false alarm detection technique, it achieves the beneficial effects of improving the accuracy of radar target detection, enhancing robustness and adaptability, and ensuring stable detection performance.

[0116] In the storage medium provided in this application, there is a program product that can implement the radar target detection method based on two-feature fusion.

[0117] The radar target detection method based on two-feature fusion includes: Using existing radar target detection data to construct training data, extracting the phase linearity based on the training data using the extraction process of phase linearity, and extracting the relative peak height of the Doppler spectrum using the extraction process of relative peak height; Based on the phase linearity and relative peak height, obtaining a fusion matrix of the two features using a linear dimensionality reduction method for distribution compactness; The reference range cell data and the to-be-detected range cell data are obtained in real time by using a signal model, the extraction processes of phase linearity and relative peak height are respectively executed, and fusion features are generated by using a fusion matrix; Based on the fusion features, the generalized extreme value distribution parameters are estimated and a detection statistic is constructed, an adaptive threshold is calculated in combination with a preset false alarm probability, and target decision is completed by comparing the detection statistic with the threshold, so as to achieve constant false alarm detection under clutter suppression.

[0118] In some possible implementation manners, the radar target detection method based on two-feature fusion of the present disclosure may be implemented in the form of a program product, which includes program codes. When the program product runs on a terminal device, the program codes are used to enable the terminal device to execute the steps according to various exemplary embodiments of the present disclosure described in the above "Exemplary Method" section of this specification.

[0119] The storage medium of the present disclosure may adopt any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. The readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the above. More specific examples (non-exhaustive list) of the readable storage medium include: an electrical connection with one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above.

[0120] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A radar target detection method based on the fusion of two features, characterized in that, Including: Construct training data using existing radar target detection data, extract phase linearity based on the training data extraction process, and extract the relative peak height of the Doppler spectrum using the relative peak height extraction process; Based on the phase linearity and relative peak height, obtain the fusion matrix of the two features using the linear dimensionality reduction method of distribution compactness; Obtain reference range cell data and data of the range cell to be detected in real time, respectively execute the extraction process of phase linearity and the extraction process of relative peak height, and generate fusion features using the fusion matrix; Based on the fusion features, estimate the generalized extreme value distribution parameters and construct a detection statistic, calculate the adaptive threshold in combination with the preset false alarm probability, and complete target decision-making by comparing the detection statistic with the threshold to achieve constant false alarm detection under clutter suppression.

2. The radar target detection method based on two-feature fusion according to claim 1, wherein The extraction process of the phase linearity includes: Represent the wrapped phase of the th data in the training data as follows: Let the difference between adjacent winding phases be , and adjust it to between and through the following formula to eliminate jumps: Then the unwrapped true phase is expressed as: Let the phase difference between two adjacent phases of the true phase sequence be , then the phase linearity is expressed as: Among them, , the normalized phase linearity is: 。 3. The radar target detection method based on two-feature fusion according to claim 2, wherein The extraction process of the relative peak height includes: Let the th data of each range cell be , then the Doppler amplitude spectrum is expressed by the following formula: Among them, is the Doppler frequency, is the pulse repetition period; Let the peak value of the Doppler amplitude spectrum be , then the relative peak height can be expressed as: Among them, is the range of Doppler side lobes, is the number of points of Doppler frequency within the range; The normalized relative peak height is: 。 4. The radar target detection method based on two-feature fusion according to claim 3, characterized in that The obtaining of the fusion matrix of the two features using the linear dimensionality reduction method of distribution compactness based on the phase linearity and relative peak height includes: Generate a feature data set based on the phase linearity and relative peak height ; Among them, is the number of samples in the feature dataset, the th feature data sample includes the phase linearity feature and the relative peak height feature , that is: After linear dimensionality reduction processing, construct a fusion feature data set: Let and represent the mean and variance of the fusion feature respectively. Then the kurtosis calculation formula is as follows: Among them, denotes taking the expectation; Let have a probability distribution function of , then the interquartile range is expressed as: wherein, and are respectively the lower quartile and the upper quartile of; Then define the objective function as follows: Among them, ; Traverse within the range of and when the objective function reaches the minimum value, the corresponding fusion matrix is the optimal weighted vector of the two features of PL and RPH. ​ 5. The radar target detection method based on two - feature fusion according to claim 4, wherein The obtaining of the reference range cell data and data of the range cell to be detected in real time includes: The marine radar is set to operate in the coherent processing mode, and the number of coherent pulse trains is . After quadrature demodulation and matched filtering, the I / Q data in the range cell to be detected is denoted as , and the I / Q data in the reference range cell is denoted as ; Among them, represents the distance unit to be detected, is the number of reference distance units, and the subscript represents the th reference distance unit; Denote the possible target I / Q data in the distance cell to be detected as , and denote the clutter I / Q data in the distance cell to be detected and the reference distance cell as and respectively.

6. The radar target detection method based on two-feature fusion according to claim 5, wherein The estimating of the generalized extreme value distribution parameters and constructing of the detection statistic based on the fusion features includes: Model the probability density function of the fusion features using the generalized extreme value distribution, specifically as follows: Among them is the fusion feature variable, is the position parameter, is the scale parameter, is the shape parameter; Using the probability weighted moment estimation method, the estimator expression of the shape parameter is as follows: Among them, , , , ; Obtained from reference distance units fused feature samples , and after arranging them in ascending order, a sequence is obtained; Using the probability weighted moment estimation method, the estimator expression of the scale parameter is as follows: Using the probability weighted moment estimation method, the estimator expression of the location parameter is as follows: wherein, is the Gamma function; According to the form of the GEV distribution probability density function of the fusion features under the condition of no target data, design the following detection statistic: When the number of samples of the fusion feature tends to infinity, such that , , , the asymptotic probability density function of the detection statistic is derived as: 。 7. The radar target detection method based on two - feature fusion according to claim 6, wherein, The calculating of the adaptive threshold in combination with the preset false alarm probability includes: The false alarm probability is calculated by the following formula The corresponding detection threshold: 。 8. A radar target detection system based on the fusion of two features, characterized in that, The system adopts the radar target detection method based on the fusion of two features according to any one of claims 1 to 7; The system includes: A data extraction module for constructing training data using existing radar target detection data, extracting phase linearity based on the training data extraction process, and extracting the relative peak height of the Doppler spectrum using the relative peak height extraction process; A fusion rule generation module for obtaining the fusion matrix of the two features using the linear dimensionality reduction method of distribution compactness based on the phase linearity and relative peak height; A measured data fusion module for obtaining reference range cell data and data of the range cell to be detected in real time, respectively executing the extraction process of phase linearity and the extraction process of relative peak height, and generating fusion features using the fusion matrix; A target detection module for estimating the generalized extreme value distribution parameters and constructing a detection statistic based on the fusion features, calculating the adaptive threshold in combination with the preset false alarm probability, and completing target decision-making by comparing the detection statistic with the threshold to achieve constant false alarm detection under clutter suppression.

9. An electronic device, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, wherein, When the processor executes the program, it implements the steps of the radar target detection method based on the fusion of two features according to any one of claims 1 to 7.

10. A storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the radar target detection method based on the fusion of two features according to any one of claims 1 to 7.

Images