A method for detecting floating targets at sea using local energy accumulation in Doppler channels

Abstract

The present invention relates to a method for detecting floating targets at sea using local energy accumulation of Doppler channels, and belongs to the technical field of radar signal processing and target detection. The present invention performs high-resolution power spectrum estimation on radar echo signals through MEM. Secondly, the Doppler frequency range of the buoy is calculated based on the roughly estimated radial velocity of the buoy, and the power spectrum density within the range is accumulated to form a characteristic value representing the overall energy of the target. Finally, the extracted energy features are detected through CA‑CFAR, and the detection results are post-processed in combination with regional connectivity analysis to further confirm the target area and eliminate false alarm signals. The present invention has higher accuracy in buoy target feature analysis and target detection under high sea conditions.

Description

Technical Field

[0001] The invention relates to a method for detecting floating targets at sea by utilizing local energy accumulation of Doppler channels, and belongs to the technical field of radar signal processing and target detection. Background Art

[0002] Marine target detection has important application value in fields such as national defense and ocean monitoring. However, in high sea conditions, strong clutter on the sea surface and complex environmental interference often make it difficult for radar systems to accurately identify target signals. These clutter and interference not only reduce the detection accuracy, but also increase the false alarm rate, posing a severe challenge to the stability and reliability of the radar system. Therefore, it is particularly important to develop radar signal processing methods that can effectively suppress sea clutter and improve target detection performance (Key. Review of radar marine target characteristics [J]. Journal of Radar, 2020, 9(4): 674-683.).

[0003] At present, in order to solve the problem of sea clutter interference, researchers have proposed a variety of radar signal processing methods. Among them, the Maximum Entropy Method (MEM) has become a commonly used spectrum analysis technology because it can provide high-resolution power spectrum estimation under limited data length (Zhang, Xian-Da. Modern signal processing. Walter de Gruyter GmbH&Co KG, 2022.). MEM estimates the power spectral density (PSD) based on the statistical characteristics of the signal by constructing an autoregressive (AR) model, which can effectively extract the frequency characteristics of the target and suppress sea clutter, so that the target frequency characteristics can be accurately captured. Gan Lingli et al. used MEM to process automotive millimeter-wave radar signals and obtained results with higher accuracy and stronger anti-interference ability than fast Fourier transform (Gan Lingli, Xu Jia, Zhang Fei. Processing of automotive millimeter-wave radar signals based on maximum entropy method. Journal of Detection and Control, 2006, (05): 65-68.). In addition, constant false alarm rate (CFAR) technology is widely used in target detection, among which CA-CFAR has good detection performance in uniform background clutter environment (Qin Tianci, Wang Zhongxun, Huang Yong, et al. Review of radar constant false alarm target detection processing technology [J]. Journal of Detection and Control, 2023, 45(03): 1-11.).

[0004] Although these methods have improved the effect of target detection to a certain extent, there are still many challenges in complex high sea conditions. When dealing with sea clutter with significant frequency changes, a single spectrum estimation or detection technology is difficult to take into account both high detection accuracy and anti-clutter capabilities. Most existing methods focus on the detection of a single feature and fail to fully utilize the frequency and energy characteristics of the target, resulting in insufficient detection stability and accuracy. In addition, with the improvement of radar data resolution and the increase in detection needs, the requirements for detection speed and efficiency are also getting higher and higher. Summary of the invention

[0005] In order to effectively solve the above problems, the present invention proposes a method for detecting floating targets at sea using local energy accumulation of Doppler channels. Aiming at the problem of channel buoy detection under high sea conditions, the signal characteristics are enhanced by combining the spectral characteristics of the buoy and the Doppler frequency range corresponding to the radial velocity, accumulating the power spectral density, and introducing regional connectivity analysis, aiming to improve the buoy target detection effect under complex sea conditions and strong sea clutter backgrounds. Experimental results show that this method significantly improves the target detection accuracy and system adaptability under complex sea conditions and strong sea clutter backgrounds, and provides more accurate regional guidance for sea surface target detection.

[0006] The method for detecting floating targets at sea by using local energy accumulation of Doppler channels of the present invention is special in that it comprises the following steps:

[0007] Step 1: Use MEM to perform high-resolution power spectrum estimation on the radar echo signal to extract the spectrum characteristics of the target and provide key feature information for subsequent steps;

[0008] Step 2: Calculate the Doppler frequency range based on the radial velocity of the buoy, and accumulate the power spectrum density within the range to form a characteristic value representing the overall energy of the target;

[0009] Step 3: Use CA-CFAR to detect the extracted energy features and dynamically adjust the detection threshold to cope with the complex clutter background;

[0010] Step 4: Perform regional connectivity analysis on the detection results to further confirm the target area and eliminate possible false alarm signals.

[0011] Preferably, the specific steps of step one are:

[0012] To construct the AR model required by MEM, a fixed order is selected , to determine the complexity of the model and calculate the autocorrelation function of the signal It is a key step in realizing MEM and describes the signal at different delays. The following correlations;

[0013] (1);

[0014] in, is the total length of the radar signal, is a discrete time series signal representing radar echo data;

[0015] After obtaining the autocorrelation function, the parameters of the AR model are estimated using the Yule-Walker equation in matrix form: :

[0016] (2);

[0017] This equation is based on the autocorrelation function of the signal With AR model parameters The relationship between , thereby constructing a model for signal prediction;

[0018] Residual sequence It reflects the difference between the actual signal and the signal predicted by the AR model, by calculating the residual Sum of squares to estimate the variance of the clutter :

[0019] (3);

[0020] (4);

[0021] Finally, using the estimated and , calculate the power spectral density of the signal , and convert to dB scale:

[0022] (5);

[0023] (6).

[0024] Preferably, the specific steps of step 2 are:

[0025] By knowing the radial velocity of the buoy, the corresponding Doppler frequency range is calculated and the overall energy characteristics of the target are extracted. The relationship between the Doppler frequency shift and the radial velocity is as follows:

[0026] (7);

[0027] (8);

[0028] in, is the Doppler shift, is the radial velocity of the buoy. A positive value indicates that the target is moving away from the radar, and a negative value indicates that the target is approaching the radar. is the wavelength of the signal, is the radar transmitting frequency, from which the Doppler frequency range is determined:

[0029] (9);

[0030] In this frequency range, the power spectral density is accumulated, and the accumulated power spectral density It is expressed as follows:

[0031] (10);

[0032] It should be smaller than the frequency ridge of clutter to avoid the influence of clutter interference on target characteristics.

[0033] Preferably, the specific steps of step three are:

[0034] The sliding window of CA-CFAR consists of three parts: the leading window, the lagging window, and the cell under test (CUT). The leading and lagging windows are used to estimate the clutter power, respectively, and the cell under test is the current target position to be detected.

[0035] The CA-CFAR algorithm introduces two parameters: the number of training units and number of protection units ; Number of training units Determines the number of units used to estimate clutter power, the number of protection units It is used to prevent the target signal near the detection unit from affecting the clutter estimation, and to estimate the clutter power level through the reference unit of the front and rear windows. :

[0036] (11);

[0037] in, Is the first The measured value of a unit, It is the first The measured value of each unit;

[0038] Threshold Factor is a parameter used to adjust the detection threshold, based on the set false alarm probability To decide:

[0039] (12);

[0040] Detection threshold By threshold factor and clutter estimate To calculate:

[0041] (13);

[0042] Signal strength of the detection unit With detection threshold For comparison, the detection decision is based on the following formula. If the formula is established, it is determined that the target is detected:

[0043] (14);

[0044] Preferably, the specific steps of step 4 are:

[0045] The detection results obtained by CA-CFAR are analyzed for regional connectivity. First, the detection results are converted into a logical matrix , that is, a binary matrix containing 0 and 1, where 1 indicates that a potential target is detected and 0 indicates that the target is not detected:

[0046] (15);

[0047] Each connected region It is composed of a group of interconnected "1"s, indicating possible target areas. For each detected connected area , calculate the size of the area , which is the number of distance cells contained in the region:

[0048] (16);

[0049] Size the area The preset minimum target area size and the maximum target area size Compare and retain the region as a valid target region if the following conditions are met:

[0050] (17);

[0051] Among the retained target areas, the center position of each target area is calculated and checked to see if it falls within the known buoy position range. If the above conditions are met, it is recorded as the final detected target area:

[0052] (18);

[0053] The present invention proposes a method for detecting floating targets at sea using local energy accumulation of Doppler channels. The radar echo signal is estimated by high-resolution power spectrum through MEM. Secondly, the Doppler frequency range is calculated based on the roughly estimated radial velocity of the buoy, and the power spectrum density within the range is accumulated to form a characteristic value representing the overall energy of the target. Finally, the extracted energy characteristics are detected by CA-CFAR, and the detection results are post-processed in combination with regional connectivity analysis to further confirm the target area and exclude false alarm signals. The present invention combines MEM and CA-CFAR technologies, uses the high-resolution spectrum estimation provided by MEM, and performs constant false alarm rate detection with dynamically adjusted detection thresholds on the power spectrum; the method calculates the Doppler frequency range based on the roughly estimated radial velocity of the target, and accumulates the power spectrum density within the range to form an overall energy characteristic value to enhance the significance of the target signal and reduce the influence of channel clutter, while simplifying parameter adjustment, which is suitable for practical application requirements. Compared with traditional detection methods, the method of the present invention has higher accuracy in buoy target feature analysis and target detection under high sea conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 This is an algorithm flow chart of the method for detecting floating targets at sea using the local energy accumulation of the Doppler channel of the present invention;

[0055] Figure 2 It is the frequency distribution and fitting ridgeline diagram of the present invention;

[0056] Figure 3 It is a three-dimensional diagram of power spectrum density of the present invention;

[0057] Figure 4 This is the final test result diagram of the present invention. DETAILED DESCRIPTION

[0058] For better understanding and implementation, a specific embodiment is given below to describe in detail a method for detecting floating targets at sea using local energy accumulation of Doppler channels of the present invention. The processing flow of the present invention is described in detail below:

[0059] Step 1: Use MEM to perform high-resolution power spectrum estimation on the radar echo signal to extract the spectrum characteristics of the target.

[0060] To construct the AR model required by MEM, a fixed order is selected , to determine the complexity of the model. Calculate the autocorrelation function of the signal It is a key step in realizing MEM and describes the signal at different delays. The following correlation.

[0061] (1);

[0062] in, is the total length of the radar signal, is a discrete time series signal representing radar echo data.

[0063] After obtaining the autocorrelation function, the Yule-Walker equation is used to estimate the parameters of the AR model .

[0064] (2);

[0065] This equation is based on the autocorrelation function of the signal With AR model parameters The relationship between , thereby constructing a model for signal prediction.

[0066] Residual sequence It reflects the difference between the actual signal and the signal predicted by the AR model. Sum of squares to estimate the variance of the clutter .

[0067] (3);

[0068] (4);

[0069] Finally, using the estimated and , calculate the power spectral density of the signal , and converted to dB scale.

[0070] (5);

[0071] (6);

[0072] The experimental data was collected by the SPPR50P-HH radar under sea conditions of level 4 with a wave height of 1.7 meters [5]. The radar that collected the data operated in staring mode with a pulse repetition frequency of 2000 Hz and a duration of 65.5 seconds. The data mainly contained echo signals from two offshore buoys, namely the red cylindrical light buoy 1 located at 2.97 nautical miles and the green cylindrical light buoy 2 located at 3.20 nautical miles.

[0073] As shown by the experimental results, Figure 2The result of MEM power spectrum estimation of radar data when the order of power spectrum estimation is set to 12. It can be observed intuitively from the figure that the buoy signal shows significant characteristics in the low-frequency range. In addition, the clutter frequency in the targetless data is analyzed, the frequency peak of each range unit is marked, and the fitted straight line is used as the clutter ridge line, as shown by the white dotted line in the figure.

[0074] To more intuitively display the energy distribution between frequency and distance units, Figure 3 A three-dimensional graph of the power spectrum density is given. The figure clearly shows two obvious peaks, while the amplitudes at other frequency and distance units are relatively small, which shows that the MEM method can effectively capture the spectral characteristics of the target signal and greatly suppress the interference of sea clutter. The two significant peak positions correspond to the actual positions of the two buoys. This efficient spectrum estimation process provides a solid data foundation for subsequent target detection.

[0075] Step 2: Calculate the Doppler frequency range based on the radial velocity of the buoy, and accumulate the power spectrum density within the range to form a characteristic value representing the overall energy of the target;

[0076] Under high sea conditions, the movement of the buoy is affected by the waves, and its radial velocity will cause the radar echo to produce a Doppler shift. By knowing the radial velocity of the buoy, the corresponding Doppler frequency range can be calculated to extract the overall energy characteristics of the target. The relationship between Doppler shift and radial velocity is as follows:

[0077] (7);

[0078] (8);

[0079] in, is the Doppler shift. is the radial velocity of the buoy, a positive value indicates that the target is moving away from the radar, and a negative value indicates that the target is approaching the radar. is the wavelength of the signal. is the radar transmission frequency. From this, the Doppler frequency range can be determined.

[0080] (9);

[0081] In this frequency range, the power spectrum density is accumulated to form an eigenvalue representing the overall energy of the distance unit, which is used for subsequent CA-CFAR detection. It is expressed as follows:

[0082] (10);

[0083] In order to reasonably set the Doppler frequency range of the buoy, it is necessary to consider the Doppler frequency characteristics of the clutter. It should be as small as possible from the frequency ridge of the clutter to avoid the influence of clutter interference on the target characteristics.

[0084] It is known that the rough radial velocity of the buoy is about 0.8 m / s, and the operating frequency range of the radar is 9.3 GHz to 9.5 GHz. To simplify the analysis, the center frequency of 9.4 GHz is used for calculation. The wavelength is calculated to be about 0.0319 m based on the radar operating frequency. Using the Doppler frequency shift formula, the Doppler frequency range of the buoy is about -50.16 Hz to 50.16 Hz. Figure 2 The main frequency of the clutter represented by the white dotted line is concentrated around 100 Hz, while the Doppler frequency range of the buoy is significantly lower than this ridge line. Such a frequency difference helps to distinguish the buoy signal from the sea clutter interference, making it more effective to distinguish the target signal from the clutter interference during detection.

[0085] Step 3: CA-CFAR is used to detect the extracted energy features and dynamically adjust the detection threshold to cope with complex clutter background;

[0086] The sliding window of CA-CFAR consists of three parts: the leading window, the lagging window, and the cell under test (CUT). The leading and lagging windows are used to estimate the clutter power, while the cell under test is the current target position to be detected.

[0087] In order to accurately estimate the clutter power, the CA-CFAR algorithm introduces two important parameters: the number of training units and and number of protection units . Number of training units determines the number of units used to estimate the clutter power, while the number of protection units Used to prevent the target signal near the detection unit from affecting the clutter estimation. Number of protection units They are not directly involved in the calculation of clutter power, but their location and amount are critical to ensuring the accuracy of clutter estimation. The clutter power level is estimated by using the reference cells in the front and rear windows. .

[0088] (11);

[0089] in, Is the first The measured value of a unit, It is the first The measured value of a unit.

[0090] Threshold Factor It is an important parameter for adjusting the detection threshold, which is determined by the false alarm probability to decide.

[0091] (12);

[0092] Detection Threshold By threshold factor and clutter estimate to calculate.

[0093] (13);

[0094] Finally, the signal strength of the detection unit Need and detection threshold The comparison is made and the detection decision is based on the following formula. If the formula is established, it is determined that the target is detected.

[0095] (14);

[0096] Step 4: Post-process the detection results through regional connectivity analysis to further confirm the target area and eliminate false alarm signals.

[0097] In order to further identify and confirm the detected target area, the present invention introduces a regional connectivity analysis module in post-processing. Specifically, the detection results obtained by CA-CFAR indicate whether there are potential targets in each distance unit. In order to extract the actual target area, it is necessary to perform regional connectivity analysis on these detection results.

[0098] First, convert the detection results into a logical matrix , which is a binary matrix containing 0 and 1, where 1 indicates that a potential target is detected and 0 indicates that the target is not detected.

[0099] (15);

[0100] Each connected region It is composed of a group of interconnected "1"s, indicating possible target areas. For each detected connected area , we calculate the size of this region , which is the number of distance cells contained in the region.

[0101] (16);

[0102] Size the area The preset minimum target area size and the maximum target area size A comparison is performed and if the following conditions are met, the region is retained as a valid target region.

[0103] (17);

[0104] Furthermore, in the retained target areas, the center position of each target area is calculated and checked to see if it falls within the known buoy position range. If the above conditions are met, it is recorded as the final detected target area.

[0105] (18);

[0106] Figure 4 This is the result of the target detection method based on the above four modules proposed by the present invention. The blue solid line in the figure represents the power spectrum density curve, which shows multiple peaks and troughs. These peaks correspond to the parts with stronger energy in the signal, representing the echo signals that may be the target. The green dotted line represents the adaptive threshold curve of the constant false alarm rate algorithm. It can be seen in the figure that the changing trend of the adaptive threshold is roughly consistent with the power spectrum density curve, indicating that the adaptive mechanism of the false alarm rate algorithm can effectively adjust the threshold to cope with changes in different clutter levels. The red solid line marks the target position finally detected. These target positions are basically consistent with the actual position interval of the buoy, which verifies the effectiveness of the algorithm.

[0107] For the same target data, the detection accuracy of the algorithm proposed in this application is 12.5% ​​higher than that of the traditional CA-CFAR. This result further verifies that the method proposed in this application has higher detection accuracy in complex sea conditions.

Claims

1. A method for detecting floating targets at sea using local energy accumulation of Doppler channels, characterized in that The following steps are involved: Step 1: Use MEM to perform high-resolution power spectrum estimation on the radar echo signal to extract the spectrum characteristics of the target and provide key feature information for subsequent steps; Step 2: Calculate the Doppler frequency range based on the radial velocity of the buoy, and accumulate the power spectrum density within the range to form a characteristic value representing the overall energy of the target; Step 3: Use CA-CFAR to detect the extracted energy features and dynamically adjust the detection threshold to cope with the complex clutter background; Step 4: Perform regional connectivity analysis on the detection results to further confirm the target area and eliminate possible false alarm signals.

2. The method for detecting floating targets at sea by using local energy accumulation of Doppler channels according to claim 1, characterized in that The specific steps of step one are: To construct the AR model required by MEM, a fixed order is selected , calculate the autocorrelation function of the signal : (1)； in, is the total length of the radar signal, is a discrete time series signal representing radar echo data; Estimating the parameters of the AR model using the Yule-Walker equation in matrix form : (2)； This equation is based on the autocorrelation function of the signal With AR model parameters The relationship between , thereby constructing a model for signal prediction; By calculating the residuals at all time points Sum of squares to estimate the variance of the clutter : (3)； (4)； Using the estimated and , calculate the power spectral density of the signal , and convert to dB scale: (5)； (6)。 3. The method for detecting floating targets at sea by using local energy accumulation of Doppler channels according to claim 1, characterized in that The specific steps of step 2 are: By knowing the radial velocity of the buoy, the corresponding Doppler frequency range is calculated and the overall energy characteristics of the target are extracted. The relationship between the Doppler frequency shift and the radial velocity is as follows: (7)； (8)； in, is the Doppler shift, is the radial velocity of the buoy. A positive value indicates that the target is moving away from the radar, and a negative value indicates that the target is approaching the radar. is the wavelength of the signal, is the radar transmitting frequency, from which the Doppler frequency range is determined: (9)； In this frequency range, the power spectral density is accumulated, and the accumulated power spectral density It is expressed as follows: (10)； It should be smaller than the frequency ridge of clutter to avoid the influence of clutter interference on target characteristics.

4. The method for detecting floating targets at sea by using local energy accumulation of Doppler channels according to claim 1, characterized in that The specific steps of step three are: The sliding window of CA-CFAR consists of three parts: the front window, the rear window and the detection unit. The front and rear windows are used to estimate the clutter power respectively, and the detection unit is the current target position to be detected. The CA-CFAR algorithm introduces two parameters: the number of training units and number of protection units ; Number of training units Determines the number of units used to estimate clutter power, the number of protection units It is used to prevent the target signal near the detection unit from affecting the clutter estimation, and to estimate the clutter power level through the reference unit of the front and rear windows. : (11)； in, Is the first The measured value of a unit, It is the first The measured value of each unit; Threshold Factor is a parameter used to adjust the detection threshold, based on the set false alarm probability To decide: (12)； Detection threshold By threshold factor and clutter estimate To calculate: (13)； Signal strength of the detection unit With detection threshold For comparison, the detection decision is based on the following formula. If the formula is established, it is determined that the target is detected: (14)。 5. The method for detecting floating targets at sea by using local energy accumulation of Doppler channels according to claim 4, characterized in that The specific steps of step 4 are: The detection results obtained by CA-CFAR are analyzed for regional connectivity. First, the detection results are converted into a logical matrix , that is, a binary matrix containing 0 and 1, where 1 indicates that a potential target is detected and 0 indicates that the target is not detected: (15)； Each connected region It is composed of a group of interconnected "1"s, representing possible target areas. For each detected connected area , calculate the size of the area , which is the number of distance cells contained in the region: (16)； Size the area The preset minimum target area size and the maximum target area size Compare and retain the region as a valid target region if the following conditions are met: (17)； Among the retained target areas, the center position of each target area is calculated and checked to see if it falls within the known buoy position range. If the above conditions are met, it is recorded as the final detected target area: (18)。

Images