Target detection method, device and equipment based on frequency domain Hurst exponent

Abstract

The present invention provides a target detection method, device and equipment based on frequency domain Hurst exponent. The method comprises: obtaining a frequency spectrum of a radar echo sequence; extracting frequency domain Hurst exponent features from the frequency spectrum of the radar echo sequence; determining a detection statistic based on the frequency domain Hurst exponent features; generating a decision threshold corresponding to the false alarm probability based on a preset false alarm probability of a constant false alarm detector; determining that a target unit is detected when the detection statistic is greater than or equal to the decision threshold; and determining that the target unit is not detected when the detection statistic is less than the decision threshold. Compared with the related art that uses a likelihood ratio detector to detect targets, the embodiments of the present invention use a constant false alarm detector based on frequency domain Hurst exponent features to detect target units, which can effectively control the false alarm rate and increase the probability of correct detection of target units under a certain false alarm rate.

Description

Technical Field

[0001] The present invention relates to the field of radar signal processing technology, and in particular to a target detection method, device and equipment based on frequency domain Hurst exponent. Background Art

[0002] Detecting small targets floating on the sea surface is one of the main tasks of sea detection radar. Such small targets mainly include buoys, small boats, mines and aircraft wreckage, which have the characteristics of weak radar echoes.

[0003] Currently, likelihood ratio detectors are usually used to detect small targets floating on the sea surface. The likelihood ratio detector designed based on statistical theory is actually an energy detector.

[0004] However, the performance of likelihood ratio detectors is easily affected by the signal-clutter ratio (SCR) and the distribution type of sea clutter amplitude. In situations such as high resolution, low ground-grazing angle, and high sea state, there are usually a large number of sea spikes. Sea clutter has obvious non-stationary, non-uniform, and non-Gaussian characteristics. Current likelihood ratio detectors cannot effectively detect small floating targets and have a high false alarm rate. Summary of the Invention

[0005] The present invention provides a target detection method, device and equipment based on frequency domain Hurst exponent, which are used to solve the problem in the prior art that likelihood ratio detectors cannot effectively detect small floating targets and have a high false alarm rate.

[0006] The present invention provides a target detection method based on frequency domain Hurst exponent, comprising:

[0007] Obtain the spectrum of the radar echo sequence;

[0008] Extracting frequency domain Hurst exponent features from the frequency spectrum of the radar echo sequence;

[0009] Determining a detection statistic based on the frequency domain Hurst exponent characteristics;

[0010] Based on a preset false alarm probability of a constant false alarm detector, generating a decision threshold corresponding to the false alarm probability;

[0011] When the detection statistic is greater than or equal to the decision threshold, it is determined that the target unit is detected; when the detection statistic is less than the decision threshold, it is determined that the target unit is not detected.

[0012] According to a target detection method based on frequency domain Hurst exponent provided by the present invention, obtaining the spectrum of the radar echo sequence includes:

[0013] Obtaining the frequency spectrum of the radar echo sequence through fast Fourier transform (FFT);

[0014] The number of frequency points of the FFT is greater than or equal to a first threshold.

[0015] According to a target detection method based on frequency domain Hurst exponent provided by the present invention, obtaining the spectrum of the radar echo sequence includes:

[0016] Determine the target polarization mode of the radar based on the priority order of the polarization modes;

[0017] The priority order includes: the vertical transmission and vertical reception VV polarization mode has the highest priority, the horizontal transmission and vertical reception HV polarization mode has the same priority as the vertical transmission and horizontal reception VH polarization mode, and the horizontal transmission and horizontal reception HH polarization mode has the lowest priority;

[0018] The target polarization mode is adopted to obtain the frequency spectrum of the radar echo sequence.

[0019] According to a target detection method based on frequency domain Hurst exponent provided by the present invention, the number of pulses of the radar echo sequence is greater than or equal to a second threshold.

[0020] According to a target detection method based on frequency domain Hurst exponent provided by the present invention, determining a detection statistic based on the frequency domain Hurst exponent feature includes:

[0021] Based on the logarithmic transformation value Y0 of the frequency domain Hurst exponent feature, the detection statistic t is determined using formula (1):

[0022]

[0023] Among them, R represents the number of reference units, i represents the i-th parameter unit, and Y i Characterizes the logarithmic transformation value of the frequency domain Hurst exponent characteristic of the i-th parameter unit, j represents the j-th parameter unit, Y j Characterizes the logarithmic transformation value of the frequency domain Hurst exponent characteristic of the j-th parameter unit.

[0024] According to a target detection method based on frequency domain Hurst exponent provided by the present invention, the false alarm probability of a constant false alarm detector is preset, and a decision threshold corresponding to the false alarm probability is generated, including:

[0025] The false alarm probability P based on the pre-set constant false alarm detector fa , use formula (2) to generate the decision threshold T corresponding to the false alarm probability:

[0026]

[0027] Where Γ() represents the gamma function, and x represents the detection statistic when the detection unit is not the target unit.

[0028] According to a frequency-domain Hurst exponent-based target detection method provided by the present invention, the constant false alarm detector is a Log-t CFAR detector.

[0029] The present invention also provides a target detection device based on frequency domain Hurst exponent, comprising:

[0030] An acquisition module, used to acquire the spectrum of the radar echo sequence;

[0031] An extraction module, configured to extract frequency domain Hurst exponent features from the frequency spectrum of the radar echo sequence;

[0032] A determination module, configured to determine a detection statistic based on the frequency domain Hurst exponent feature;

[0033] A generating module, configured to generate a decision threshold corresponding to a false alarm probability based on a preset false alarm probability of a constant false alarm detector;

[0034] The detection module is configured to determine that the target unit is detected when the detection statistic is greater than or equal to the decision threshold; and to determine that the target unit is not detected when the detection statistic is less than the decision threshold.

[0035] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the target detection method based on the frequency domain Hurst exponent as described above is implemented.

[0036] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the target detection method based on the frequency domain Hurst exponent as described above is implemented.

[0037] The present invention also provides a computer program product, comprising a computer program, wherein when the computer program is executed by a processor, the computer program implements any of the above-mentioned frequency-domain Hurst exponent-based target detection methods.

[0038] The target detection method, device and equipment based on the frequency domain Hurst exponent provided by the present invention, compared with the related art of using a likelihood ratio detector to detect targets, first use the frequency domain Hurst exponent characteristics extracted from the spectrum of the radar echo sequence to determine the detection statistic. Since the frequency domain Hurst exponent characteristics can describe the fractal characteristics of the spectrum of the radar echo sequence as a whole, taking into account the advantages of coherent accumulation and fractal characteristics, the fractal characteristics, as a type of non-energy characteristics, can overcome the influence of SCR to a certain extent and better distinguish target units from clutter units. Therefore, the detection statistics determined based on the frequency domain Hurst exponent characteristics are more accurate. In addition, the embodiment of the present invention improves the generation method of the decision threshold and uses a constant false alarm detector to determine the decision threshold corresponding to the false alarm probability, which can effectively control the false alarm rate and improve the correct detection probability of the target unit under a certain false alarm rate. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] In order to more clearly illustrate the technical solutions in the present invention or the prior art, a brief introduction is given below to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0040] Figure 1 This is one of the flow charts of the target detection method based on frequency domain Hurst exponent provided by the present invention;

[0041] Figure 2 Schematic diagram of typical spectrum partition functions corresponding to different FFT frequency points provided by the present invention;

[0042] Figure 3 It is a schematic diagram of the influence of different FFT frequency points provided by the present invention on the statistical distribution of Hurst exponents in the frequency domain of sea clutter units and target units;

[0043] Figure 4 is a schematic diagram of typical spectrum partition functions corresponding to different polarization modes provided by the present invention;

[0044] Figure 5 Schematic diagram of the effect of different polarization modes provided by the present invention on the frequency domain Hurst exponent statistical distribution of sea clutter units and target units;

[0045] Figure 6 Schematic diagram of typical spectrum partition functions corresponding to different pulse numbers provided by the present invention;

[0046] Figure 7 It is a schematic diagram of the influence of different pulse numbers on the frequency domain Hurst exponent statistical distribution of sea clutter units and target units provided by the present invention;

[0047] Figure 8 This is the second flow chart of the target detection method based on frequency domain Hurst exponent provided by the present invention;

[0048] Figure 9 Schematic diagram of the structure of the target detection device based on frequency domain Hurst exponent provided by the present invention;

[0049] Figure 10 It is a structural schematic diagram of the electronic device provided by the present invention. DETAILED DESCRIPTION

[0050] To make the objectives, technical solutions, and advantages of the present invention more clear, the technical solutions of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the embodiments described are only some of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts shall fall within the scope of protection of the present invention.

[0051] The target detection method, device and apparatus based on frequency domain Hurst exponent of the present invention are described below with reference to the accompanying drawings.

[0052] Figure 1 This is one of the flow charts of the target detection method based on frequency domain Hurst exponent provided by the present invention, such as Figure 1 As shown, the target detection method based on the frequency domain Hurst exponent includes steps 101 to 105; wherein:

[0053] Step 101: Acquire the spectrum of the radar echo sequence;

[0054] Step 102: extracting frequency domain Hurst exponent features from the spectrum of the radar echo sequence;

[0055] Step 103: Determine a detection statistic based on the frequency domain Hurst exponent characteristics;

[0056] Step 104: Based on the preset false alarm probability of the constant false alarm detector, generate a decision threshold corresponding to the false alarm probability;

[0057] Step 105: If the detection statistic is greater than or equal to the decision threshold, it is determined that the target unit is detected; if the detection statistic is less than the decision threshold, it is determined that the target unit is not detected.

[0058] Related technologies typically use likelihood ratio detectors to detect small floating targets on the sea surface. Designed based on statistical theory, likelihood ratio detectors are essentially energy detectors. Consequently, their performance is highly susceptible to the effects of SCR and the amplitude distribution of sea clutter. In situations such as high resolution, low ground-grazing angles, and high sea conditions, where numerous sea spikes are common, sea clutter exhibits significant non-stationary, non-uniform, and non-Gaussian characteristics. Current likelihood ratio detectors are unable to effectively detect small floating targets and suffer from a high false alarm rate.

[0059] Fractal features, as a type of non-energy feature, can overcome the effects of SCR to a certain extent and have garnered widespread attention since their introduction. When the SCR is low, time-domain fractal features are less effective. To further improve the SCR of time-domain signals and enhance the fractal differentiation between sea clutter and target echoes, numerous transform-domain fractal features have been proposed, leveraging the fractal properties of sea clutter spectra, autoregression (AR) spectra, and fractional Fourier transform spectra. The frequency-domain Hurst exponent feature is one of these.

[0060] The frequency-domain Hurst exponent can comprehensively describe the fractal characteristics of the spectrum, combining the advantages of coherent accumulation and fractal characteristics, and has strong practical application value. Existing target detection methods based on the frequency-domain Hurst exponent have two shortcomings. First, there is no in-depth study of the impact of parameter selection on the frequency-domain fractal characteristics of sea clutter, which makes the frequency-domain Hurst exponent characteristics difficult to apply in practice. Second, people often determine the decision threshold through Monte Carlo simulation or non-parametric Constant False Alarm Rate (CFAR) methods. Monte Carlo simulation methods require calculating a large amount of data and are time-consuming. Non-parametric CFAR methods suffer from significant detection performance losses in order to achieve constant false alarm. Both methods are unsuitable for practical applications of the frequency-domain Hurst exponent.

[0061] The purpose of the embodiments of the present invention is to provide a target detection method based on the frequency-domain Hurst exponent, which combines the frequency-domain Hurst exponent and a constant false alarm detector. In addition, the present invention also provides specific rules for selecting parameters when extracting the frequency-domain Hurst exponent from the spectrum of a radar echo sequence. After the frequency-domain Hurst exponent feature is extracted, a Log-t CFAR detector is introduced to determine the decision threshold to achieve the purpose of constant false alarm detection.

[0062] Specifically, in an embodiment of the present invention, on the one hand, the spectrum of the radar echo sequence is obtained, and the frequency domain Hurst exponent characteristics are extracted based on the spectrum of the radar echo sequence, and then the detection statistic is obtained based on the frequency domain Hurst exponent characteristics; on the other hand, based on the false alarm probability of a preset constant false alarm detector, a decision threshold corresponding to the false alarm probability is generated, and then by comparing the detection statistic with the decision threshold, it is determined whether the target unit is detected. Specifically, if the detection statistic is greater than or equal to the decision threshold, it is determined that the target unit is detected, and if the detection statistic is less than the decision threshold, it is determined that the target unit is not detected.

[0063] Optionally, the target unit is, for example, a small target floating on the sea surface.

[0064] In an embodiment of the present invention, compared with the use of a likelihood ratio detector to detect targets in the related art, a detection statistic is first determined using the frequency domain Hurst exponent feature extracted from the spectrum of the radar echo sequence. Since the frequency domain Hurst exponent feature can describe the fractal characteristics of the spectrum of the radar echo sequence as a whole, taking into account the advantages of coherent accumulation and fractal features, and the fractal feature as a type of non-energy feature can overcome the influence of SCR to a certain extent, and better distinguish target units from clutter units, the detection statistic determined based on the frequency domain Hurst exponent feature is more accurate. In addition, the embodiment of the present invention improves the generation method of the decision threshold and uses a constant false alarm detector to determine the decision threshold corresponding to the false alarm probability, which can effectively control the false alarm rate and improve the correct detection probability of the target unit under a certain false alarm rate.

[0065] Optionally, the method of acquiring the spectrum of the radar echo sequence may include:

[0066] Obtaining the frequency spectrum of the radar echo sequence through Fast Fourier Transform (FFT);

[0067] The number of frequency points of the FFT is greater than or equal to a first threshold.

[0068] Alternatively, the spectrum of the radar echo sequence can be obtained through FFT, and then the spectrum of the radar echo sequence can be modeled as a "random walk" model. In the double logarithmic coordinate system log2F(m) and log2(m) of the partition function F(m) and the frequency sampling interval m, the slope obtained by linearly fitting the partition function and the frequency sampling interval on the frequency-unscaled interval can be used as the frequency-domain Hurst exponent. Parameters such as the number (length) of pulses in the radar echo sequence and the number of frequency points in the FFT affect the range of the frequency-unscaled interval. The frequency-unscaled interval can be understood as the frequency within which the fractal characteristics are present. The correct selection of the frequency-unscaled interval profoundly affects target detection performance.

[0069] Specifically, assume that a high-resolution sea detection radar transmits a coherent pulse train with N pulses in a certain beam direction and receives a radar echo sequence x′(n) with N pulses in a certain range unit, where n = 1, 2, ···, N. The spectrum X(k) of the radar echo sequence is obtained by FFT, where k = 1, 2, ···, K. It is modeled as a “random walk” model to verify whether it satisfies the following formula (3):

[0070]

[0071] Among them, F(m) represents the partition function, m represents the frequency sampling interval, that is, the frequency scale, and H represents the Hurst exponent of the spectrum sequence, that is, the frequency domain Hurst exponent of the radar echo. The Hurst exponent can describe the fractal characteristics as a whole.

[0072] If formula (3) is satisfied, the spectrum X(k) of the radar echo sequence can be considered to be a fractal with a frequency unscaled interval.

[0073] In fact, the self-similarity of the sea clutter spectrum only exists in its frequency-free range, that is, in the frequency-free range, the sea clutter spectrum has fractal characteristics. In the double logarithmic coordinate system, the slope obtained by linear fitting log2F(m) and log2(m) in the frequency-free range is the frequency domain Hurst exponent, which can be described by formula (4):

[0074]

[0075] In practical applications, whether the spectrum of the radar echo sequence has fractal characteristics can also be determined by judging whether log2F(m) and log2(m) have a linear relationship.

[0076] There are many factors that affect the fractal characteristics of sea clutter spectrum. The embodiment of the present invention mainly introduces the influence of the pulse number (length) of the radar echo sequence, the number of frequency points of FFT and the polarization mode of the radar on the frequency domain fractal characteristics.

[0077] The following first explains the influence of the number of FFT frequency points on the frequency domain fractal characteristics.

[0078] The number of pulses in the radar echo sequence is 2 11 For example, Figure 2 Schematic diagram of typical spectrum partition functions corresponding to different FFT frequency points provided by the present invention, such as Figure 2 shown.

[0079] It should be noted that the range of the frequency unscaled interval can usually be obtained through the diagram of log2F(m) and log2(m). Figure 2In the direction from log2(m) 0 to 10, for example, take 2 4 As the lower limit of the frequency unscaled interval, take 2 8 As the upper limit of the range of the frequency unscaled interval, in the double logarithmic coordinate system, the frequency unscaled interval can be understood as an approximately linear interval, and the partition function of this interval is approximately a straight line.

[0080] It can be seen that as the number of FFT frequency points increases, the range of the frequency-free range gradually increases, and the linearity of the partition function does not change significantly. Since the Fourier transform is a linear transform, increasing the number of FFT frequency points does not significantly change the self-similar structure of the spectrum. Therefore, the range of the frequency-free range increases approximately linearly with the increase in the number of FFT frequency points, but the linearity of the partition function remains basically unchanged.

[0081] Figure 3 This is a schematic diagram of the effect of different FFT frequency points on the frequency domain Hurst exponent statistical distribution of sea clutter units and target units provided by the present invention, as shown in FIG. Figure 3 As shown in the figure, as the number of FFT frequency points increases, the convergence and tailing of the sea clutter frequency domain Hurst exponent distribution do not change significantly, but the overall separation of the frequency domain Hurst exponents extracted from sea clutter units and target units increases significantly. This indicates that when the false alarm probability is constant, the probability of correctly detecting the target unit tends to increase.

[0082] Optionally, the number of frequency points of the FFT is greater than or equal to a first threshold, which is set to 2, for example. 16 .

[0083] In practical applications, in order to improve the detection performance of the target unit and facilitate the selection of the frequency scale-free interval, the number of FFT points can be fixed to 2. 16 When the number of pulses in the radar echo sequence is large enough, the frequency unscaled interval is basically 2 5 ~2 10 .

[0084] The following first explains the influence of the radar polarization mode on the frequency domain fractal characteristics.

[0085] Optionally, the method of acquiring the spectrum of the radar echo sequence may include:

[0086] Determine the target polarization mode of the radar based on the priority order of the polarization modes;

[0087] The priority order includes: the vertical transmit vertical receive (Vertical Transmit Vertical Receive, VV) polarization mode has the highest priority, the horizontal transmit vertical receive (Horizontal Transmit Vertical Receive, HV) polarization mode has the same priority as the vertical transmit horizontal receive (Vertical Transmit Horizontal Receive, VH) polarization mode, and the horizontal transmit horizontal receive (Horizontal Transmit Horizontal Receive, HH) polarization mode has the lowest priority;

[0088] The target polarization mode is adopted to obtain the frequency spectrum of the radar echo sequence.

[0089] Specifically, for the polarization mode of the radar, the number of pulses in the radar echo sequence is 2 11 , and the number of FFT frequency points is 2 16 For example, Figure 4 is a schematic diagram of typical spectrum partition functions corresponding to different polarization modes provided by the present invention, such as Figure 4 As shown, the polarization mode has no significant effect on the range of the frequency unscaled interval, which is 2 5 ~2 10 .

[0090] Figure 5 Schematic diagram of the effect of different polarization modes on the frequency domain Hurst exponent statistical distribution of sea clutter units and target units provided by the present invention, as shown in FIG. Figure 5 As shown in the figure, VV polarization has the best clustering and the shortest tail for the sea clutter frequency domain Hurst exponent distribution. HV and VH polarizations have similar results, with HH polarization being the worst. Furthermore, the overall separation of the frequency domain Hurst exponents extracted from sea clutter units and target units is essentially the same for all four polarization modes. This indicates that when using frequency domain Hurst exponents for target detection, VV polarization data performs best, followed by HV and VH polarizations, with HH polarization performing the worst.

[0091] Based on the above analysis, the VV polarization mode with the best effect can be set to have the highest priority, followed by the HV polarization mode or the VH polarization mode, and the HH polarization mode with the worst effect has the lowest priority.

[0092] The following first explains the influence of the number of pulses in the radar echo sequence on the frequency domain fractal characteristics.

[0093] Optionally, the number of pulses in the radar echo sequence may be greater than or equal to a second threshold.

[0094] Specifically, for the pulse number of the radar echo sequence, the frequency points of FFT are set to 2 16 For example, Figure 6 is a schematic diagram of a typical spectrum partition function corresponding to different pulse numbers provided by the present invention, such as Figure 6 As shown, the number of pulses has an impact on the frequency unscaled range. The more pulses there are, the better the linearity of the frequency unscaled range. When the number of pulses is large enough, for example, greater than 2 10 When the frequency unscaled interval is always kept at 2 5 ~2 10 When the number of pulses is small, for example, far less than 2 10 The range of the frequency unscaled interval will be significantly reduced when the number of pulses is less. This is mainly because the frequency resolution is worse when the number of pulses is less, and the calculation error of the spectrum partition function is larger.

[0095] In the actual detection process, in order to reduce the calculation error of the frequency domain Hurst exponent, the number of pulses of the radar echo sequence can be set to be greater than or equal to the second threshold. For example, the second threshold is 2 10 and above.

[0096] Figure 7 This is a schematic diagram of the effect of different pulse numbers on the frequency domain Hurst exponent statistical distribution of sea clutter units and target units provided by the present invention, as shown in FIG. Figure 7 As shown in the figure, as the number of pulses used increases, the convergence of the sea clutter frequency domain Hurst exponent distribution becomes better, the tail becomes shorter, and the overall separation of the frequency domain Hurst exponents extracted from sea clutter units and target units increases. This indicates that when the false alarm probability is constant, the probability of correctly detecting the target unit tends to increase.

[0097] After the above analysis, the following parameter selection rules can be given: when using the frequency domain Hurst exponent to detect target units, in order to improve detection performance, the VV polarization mode should be preferred as the target polarization mode of the radar, followed by HV and VH polarizations, and the worst is HH polarization; the optimal FFT point number is 2 16 , when the number of pulses in the radar echo sequence is large enough, generally not less than 2 10 When the number of pulses is , the optimal frequency unscaled interval is 2 5 ~2 10 ; When the number of pulses in the radar echo sequence is much less than 2 10 Or when the number of FFT frequency points changes greatly, the range of the optimal frequency unscaled interval will be affected.

[0098] In addition, the above parameter selection rules can be used when the sea state level changes between 2 and 4, but the embodiment of the present invention does not limit the use of the parameter selection rules provided by the embodiment of the present invention to the time when the sea state level changes between 2 and 4.

[0099] Optionally, the constant false alarm detector may be a Log-t CFAR detector.

[0100] Specifically, the construction of a constant false alarm detector can be based on the statistical distribution of the Hurst exponent in the sea clutter frequency domain. Based on 47 sets of measured data from the IPIX (Intelligent PIxel Processing X-band) radar dataset and the Naval Aviation University's "Radar Sea Detection Data Sharing Program" dataset, the present invention uses six classic statistical distribution models (Gaussian distribution, Rayleigh distribution, lognormal distribution, Weibull distribution, K distribution, and KK distribution) to verify the optimal statistical distribution type of the Hurst exponent in the sea clutter frequency domain.

[0101] The frequency domain Hurst exponent of sea clutter units was extracted from 47 sets of measured data. Six distribution models were used to fit them respectively. The fitting error value of the chi-square test was used to judge the applicability of the model fitting. The experimental results are shown in Table 1.

[0102] Table 1 Statistics of minimum error of Hurst exponential distribution fitting in frequency domain of measured sea clutter

[0103]

[0104] As can be seen, among the 47 sets of measured data, the optimal distribution type for 45 sets is the lognormal distribution, and only 2 sets have the Weibull distribution. This indicates that the optimal distribution type of the Hurst exponent in the measured sea clutter frequency domain is basically the lognormal distribution. The 47 sets of data cover sea states 2 to 4 and four polarization modes. Clearly, different sea states and polarization modes do not change the optimal distribution type of the sea clutter frequency domain Hurst exponent. In fact, changing the pulse number does not affect the optimal distribution type either.

[0105] The Log-t CFAR detector provides a near-optimal single-pulse detection strategy for CFAR detection in Weibull or lognormal clutter with unknown shape and scale parameters. Therefore, embodiments of the present invention incorporate the Log-t CFAR detector as a constant false alarm detector, addressing the practical limitations of previous methods for determining decision thresholds using Monte Carlo simulation or non-parametric CFAR.

[0106] Optionally, the implementation of determining the detection statistic based on the frequency domain Hurst exponent feature may include:

[0107] Based on the logarithmic transformation value Y0 of the frequency domain Hurst exponent feature, the detection statistic t is determined using formula (1):

[0108]

[0109] Among them, R represents the number of reference units, i represents the i-th parameter unit, and Y i Characterizes the logarithmic transformation value of the frequency domain Hurst exponent characteristic of the i-th parameter unit, j represents the j-th parameter unit, Y j Characterizes the logarithmic transformation value of the frequency domain Hurst exponent characteristic of the j-th parameter unit.

[0110] Optionally, the implementation method of generating a decision threshold corresponding to the false alarm probability based on a preset false alarm probability of a constant false alarm detector may include:

[0111] The false alarm probability P based on the pre-set constant false alarm detector fa , use formula (2) to generate the decision threshold T corresponding to the false alarm probability:

[0112]

[0113] Where Γ() represents the gamma function, and x represents the detection statistic when the detection unit is not the target unit.

[0114] Optionally, by comparing the detection statistic t with the decision threshold T, the presence or absence of the target unit can be determined. The decision rule is, for example, formula (5):

[0115]

[0116] The following example illustrates a target detection method based on the frequency domain Hurst exponent provided by an embodiment of the present invention.

[0117] Figure 8 This is the second flow chart of the target detection method based on the frequency domain Hurst exponent provided by the present invention. Figure 8 As shown, the following steps are included:

[0118] Step S1: Extracting frequency domain Hurst exponent features.

[0119] Specifically, by performing an FFT on the radar echo, the spectrum of the radar echo sequence is obtained and modeled as a "random walk" model. The slope of the partition function and the frequency sampling interval, obtained by linearly fitting them on the frequency-unscaled interval in a logarithmic coordinate system, is the frequency-domain Hurst exponent. The number of pulses in the radar echo sequence and the number of frequency points in the FFT both affect the range of the frequency-unscaled interval. The correct selection of the frequency-unscaled interval profoundly impacts target detection performance. This paper provides a general method for parameter selection.

[0120] Step S2: constructing a constant false alarm detector.

[0121] Specifically, the present invention verifies that the optimal statistical distribution type of the Hurst exponent in the sea clutter frequency domain is the log-normal distribution. Changes in sea state level, polarization mode, and radar echo sequence length will not affect the optimal distribution type of the Hurst exponent in the sea clutter frequency domain. Therefore, the present invention introduces the Log-t CFAR detector as a constant false alarm detector. According to the given false alarm probability P fa Generate the required decision threshold T.

[0122] Step S3: Detection of target unit.

[0123] Specifically, the detection statistic t is calculated based on the frequency domain Hurst exponents extracted from the unit to be detected and the reference unit, and the presence or absence of the target unit is determined by comparing t with T.

[0124] Among them, the unit to be detected is, for example, a unit in the pulse number-distance correspondence table. For example, the unit to be detected is a unit corresponding to the pulse number M and the distance N. Then, the reference unit of the unit to be detected can be considered as the unit with the pulse number M and the distance N+1, or the unit with the pulse number M and the distance N-1, or both the unit with the pulse number M and the distance N+1 and the unit with the pulse number M and the distance N-1.

[0125] In the embodiments of the present invention, the following beneficial effects are achieved:

[0126] 1) Taking into account the advantages of coherent accumulation and fractal characteristics, it can effectively overcome the influence of low SCR and achieve effective detection of small targets floating on the sea surface;

[0127] 2) Aiming at practical applications, the rules for parameter selection when extracting the frequency domain Hurst exponent are given;

[0128] 3) A method for generating a decision threshold when using the frequency domain Hurst exponent for constant false alarm detection is proposed, which solves the problem that the previous method of determining the decision threshold through Monte Carlo simulation or non-parametric CFAR is not suitable for practical applications, and has better small target detection performance and stability.

[0129] The target detection device based on the frequency domain Hurst exponent provided by the present invention is described below. The target detection device based on the frequency domain Hurst exponent described below and the target detection method based on the frequency domain Hurst exponent described above can refer to each other.

[0130] Figure 9 FIG is a schematic diagram of the structure of the target detection device based on the frequency domain Hurst exponent provided by the present invention. Figure 9 As shown, the target detection device 900 based on the frequency domain Hurst exponent includes:

[0131] An acquisition module 901 is used to acquire the spectrum of the radar echo sequence;

[0132] An extraction module 902 is configured to extract a frequency domain Hurst exponent feature from the frequency spectrum of the radar echo sequence;

[0133] A determination module 903 is configured to determine a detection statistic based on the frequency domain Hurst exponent feature;

[0134] A generating module 904 is configured to generate a decision threshold corresponding to a false alarm probability based on a preset false alarm probability of a constant false alarm detector;

[0135] The detection module 905 is configured to determine that the target unit is detected when the detection statistic is greater than or equal to the decision threshold; and to determine that the target unit is not detected when the detection statistic is less than the decision threshold.

[0136] In an embodiment of the present invention, compared with the use of a likelihood ratio detector to detect targets in the related art, a detection statistic is first determined using the frequency domain Hurst exponent feature extracted from the spectrum of the radar echo sequence. Since the frequency domain Hurst exponent feature can describe the fractal characteristics of the spectrum of the radar echo sequence as a whole, taking into account the advantages of coherent accumulation and fractal features, and the fractal feature as a type of non-energy feature can overcome the influence of SCR to a certain extent, and better distinguish target units from clutter units, the detection statistic determined based on the frequency domain Hurst exponent feature is more accurate. In addition, the embodiment of the present invention improves the generation method of the decision threshold and uses a constant false alarm detector to determine the decision threshold corresponding to the false alarm probability, which can effectively control the false alarm rate and improve the correct detection probability of the target unit under a certain false alarm rate.

[0137] Optionally, the acquisition module 901 is specifically configured to: acquire the frequency spectrum of the radar echo sequence by fast Fourier transform (FFT);

[0138] The number of frequency points of the FFT is greater than or equal to a first threshold.

[0139] Optionally, the acquisition module 901 is further specifically configured to:

[0140] Determine the target polarization mode of the radar based on the priority order of the polarization modes;

[0141] The priority order includes: the vertical transmission and vertical reception VV polarization mode has the highest priority, the horizontal transmission and vertical reception HV polarization mode has the same priority as the vertical transmission and horizontal reception VH polarization mode, and the horizontal transmission and horizontal reception HH polarization mode has the lowest priority;

[0142] The target polarization mode is adopted to obtain the frequency spectrum of the radar echo sequence.

[0143] Optionally, the number of pulses of the radar echo sequence is greater than or equal to a second threshold.

[0144] Optionally, the determining module 903 is specifically configured to:

[0145] Based on the logarithmic transformation value Y0 of the frequency domain Hurst exponent feature, the detection statistic t is determined using formula (1):

[0146]

[0147] Among them, R represents the number of reference units, i represents the i-th parameter unit, and Y i Characterizes the logarithmic transformation value of the frequency domain Hurst exponent characteristic of the i-th parameter unit, j represents the j-th parameter unit, Y j Characterizes the logarithmic transformation value of the frequency domain Hurst exponent characteristic of the j-th parameter unit.

[0148] Optionally, the generating module 904 is specifically configured to:

[0149] The false alarm probability P based on the pre-set constant false alarm detector fa , use formula (2) to generate the decision threshold T corresponding to the false alarm probability:

[0150]

[0151] Where Γ() represents the gamma function, and x represents the detection statistic when the detection unit is not the target unit.

[0152] Optionally, the constant false alarm detector is a Log-t CFAR detector.

[0153] Figure 10 Schematic diagram of the structure of the electronic device provided by the present invention, such as Figure 10 As shown, the electronic device 1000 may include: a processor 1010, a communication interface 1020, a memory 1030, and a communication bus 1040, wherein the processor 1010, the communication interface 1020, and the memory 1030 communicate with each other via the communication bus 1040. The processor 1010 may call the logic instructions in the memory 1030 to execute a target detection method based on the frequency domain Hurst exponent, which includes:

[0154] Obtain the spectrum of the radar echo sequence;

[0155] Extracting frequency domain Hurst exponent features from the frequency spectrum of the radar echo sequence;

[0156] Determining a detection statistic based on the frequency domain Hurst exponent characteristics;

[0157] Based on a preset false alarm probability of a constant false alarm detector, generating a decision threshold corresponding to the false alarm probability;

[0158] When the detection statistic is greater than or equal to the decision threshold, it is determined that the target unit is detected; when the detection statistic is less than the decision threshold, it is determined that the target unit is not detected.

[0159] In addition, the logic instructions in the above-mentioned memory 1030 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when sold or used as an independent product. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0160] On the other hand, the present invention further provides a computer program product, which includes a computer program. The computer program can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can perform the target detection method based on the frequency domain Hurst exponent provided by the above methods, which includes:

[0161] Obtain the spectrum of the radar echo sequence;

[0162] Extracting frequency domain Hurst exponent features from the frequency spectrum of the radar echo sequence;

[0163] Determining a detection statistic based on the frequency domain Hurst exponent characteristics;

[0164] Based on a preset false alarm probability of a constant false alarm detector, generating a decision threshold corresponding to the false alarm probability;

[0165] When the detection statistic is greater than or equal to the decision threshold, it is determined that the target unit is detected; when the detection statistic is less than the decision threshold, it is determined that the target unit is not detected.

[0166] In another aspect, the present invention further provides a non-transitory computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the method for detecting a target based on the frequency domain Hurst exponent provided by the above methods is implemented. The method includes:

[0167] Obtain the spectrum of the radar echo sequence;

[0168] Extracting frequency domain Hurst exponent features from the frequency spectrum of the radar echo sequence;

[0169] Determining a detection statistic based on the frequency domain Hurst exponent characteristics;

[0170] Based on a preset false alarm probability of a constant false alarm detector, generating a decision threshold corresponding to the false alarm probability;

[0171] When the detection statistic is greater than or equal to the decision threshold, it is determined that the target unit is detected; when the detection statistic is less than the decision threshold, it is determined that the target unit is not detected.

[0172] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, i.e., they may be located in one location or distributed across multiple network units. Some or all of the modules may be selected based on actual needs to achieve the objectives of the present embodiment. Persons of ordinary skill in the art will be able to understand and implement the present invention without inventive effort.

[0173] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a necessary general hardware platform, or of course, by hardware. Based on this understanding, the essence of the above technical solution or the part that contributes to the existing technology can be embodied in the form of a software product. The computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a magnetic disk, an optical disk, etc., and includes a number of instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or certain parts of the embodiments.

[0174] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A target detection method based on frequency domain Hurst exponent, characterized in that: include: Obtain the spectrum of the radar echo sequence; Extracting frequency domain Hurst exponent features from the frequency spectrum of the radar echo sequence; Determining a detection statistic based on the frequency domain Hurst exponent characteristics; The determining of the detection statistic based on the frequency domain Hurst exponential feature includes: a logarithmic transformation value based on the frequency domain Hurst exponential feature , the test statistic is determined using formula (1) : （1） in, Characterizes the number of reference units, Characterization parameter units, Characterization The logarithmic transformation value of the frequency domain Hurst exponent characteristic of the parameter unit, Characterization parameter units, Characterization The logarithmic transformation value of the frequency domain Hurst exponent characteristic of the parameter unit; Based on the false alarm probability of the preset constant false alarm detector, a decision threshold corresponding to the false alarm probability is generated; the false alarm probability based on the preset constant false alarm detector, generating the decision threshold corresponding to the false alarm probability, including: based on the false alarm probability of the preset constant false alarm detector , use formula (2) to generate the decision threshold corresponding to the false alarm probability : （2） in, () represents the gamma function, The detection statistic that characterizes when the detection unit is not the target unit; When the detection statistic is greater than or equal to the decision threshold, it is determined that the target unit is detected; when the detection statistic is less than the decision threshold, it is determined that the target unit is not detected.

2. The target detection method based on frequency domain Hurst exponent according to claim 1, characterized in that: The acquiring of the spectrum of the radar echo sequence comprises: Obtaining the frequency spectrum of the radar echo sequence through fast Fourier transform (FFT); The number of frequency points of the FFT is greater than or equal to a first threshold.

3. The target detection method based on frequency domain Hurst exponent according to claim 1, characterized in that: The acquiring of the spectrum of the radar echo sequence comprises: Determine the target polarization mode of the radar based on the priority order of the polarization modes; The priority order includes: the vertical transmission and vertical reception VV polarization mode has the highest priority, the horizontal transmission and vertical reception HV polarization mode has the same priority as the vertical transmission and horizontal reception VH polarization mode, and the horizontal transmission and horizontal reception HH polarization mode has the lowest priority; The target polarization mode is adopted to obtain the frequency spectrum of the radar echo sequence.

4. The target detection method based on frequency domain Hurst exponent according to claim 1, characterized in that: The number of pulses of the radar echo sequence is greater than or equal to a second threshold.

5. The target detection method based on frequency domain Hurst exponent according to any one of claims 1 to 4, characterized in that: The constant false alarm detector is a Log-t CFAR detector.

6. A target detection device based on frequency domain Hurst exponent, characterized in that: include: An acquisition module, used to acquire the spectrum of the radar echo sequence; An extraction module, configured to extract frequency domain Hurst exponent features from the frequency spectrum of the radar echo sequence; A determination module, configured to determine a detection statistic based on the frequency domain Hurst exponent feature; The determining of the detection statistic based on the frequency domain Hurst exponential feature includes: a logarithmic transformation value based on the frequency domain Hurst exponential feature , the test statistic is determined using formula (1) : （1） in, Characterizes the number of reference units, Characterization parameter units, Characterization The logarithmic transformation value of the frequency domain Hurst exponent characteristic of the parameter unit, Characterization parameter units, Characterization The logarithmic transformation value of the frequency domain Hurst exponent characteristic of the parameter unit; A generating module is used to generate a decision threshold corresponding to the false alarm probability based on a preset false alarm probability of a constant false alarm detector; the generating module is used to generate a decision threshold corresponding to the false alarm probability based on a preset false alarm probability of a constant false alarm detector, including: , use formula (2) to generate the decision threshold corresponding to the false alarm probability : （2） in, () represents the gamma function, The detection statistic that characterizes when the detection unit is not the target unit; The detection module is configured to determine that the target unit is detected when the detection statistic is greater than or equal to the decision threshold; and to determine that the target unit is not detected when the detection statistic is less than the decision threshold.

7. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the target detection method based on the frequency domain Hurst exponent as described in any one of claims 1 to 5 is implemented.

8. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the target detection method based on frequency domain Hurst exponent is implemented as claimed in any one of claims 1 to 5.

Images