Sea surface target identification method in radar staring mode

Abstract

The invention relates to a sea surface target identification method in a radar staring mode, and belongs to the technical field of radar signal processing. In order to solve the defect of low accuracy of target-based one-dimensional range profile features of a radar target identification method in the prior art, the method comprises the following steps: step 1, echo data acquisition to obtain a range unit where a target is located; 2, echo data time domain processing and time domain feature extraction are carried out according to the distance unit where the target is located obtained in the step 1; step 3, extracting frequency domain features of the echo data; step 4, constructing a feature sequence; step 5, optimizing and training a classifier: optimizing parameters of the SVM classifier and training to obtain an optimal classifier model; and step 6, target identification: inputting to-be-identified data into the classifier model trained in the step 5 to realize identification of a ship target and a floating target. The method is high in applicability and high in accuracy.

Description

Technical Field

[0001] The invention relates to a sea surface target recognition method, in particular to a sea surface target recognition method in a radar staring mode, and belongs to the technical field of radar signal processing. Background Art

[0002] When radar detects the sea, it is of great significance in both the civil and military fields to accurately identify and classify sea surface targets and determine their specific categories, uses and threat levels. Common floating targets on the sea surface include small fishing boats, buoys, and ice floes. Traditional radar target recognition methods are based on the one-dimensional range image characteristics of the target. When the radar beam is irradiated along the side of the ship, such targets have similar one-dimensional range image characteristics to large cruise ships, cargo ships and other ship targets, which seriously affects the recognition performance. In addition, when the radar resolution decreases or the signal-to-clutter ratio (SCR) of the target is low, the accuracy of the recognition method that relies on the amplitude characteristics of the target is seriously reduced. Summary of the invention

[0003] The purpose of the present invention is to solve the problem of low accuracy of radar target recognition methods based on one-dimensional range image characteristics of targets in the prior art, and to provide a sea surface target recognition method in radar staring mode, which utilizes the periodic consistency of target transient power level and bandwidth change to distinguish floating targets from ship targets.

[0004] Starting from the coupling mechanism between the target and the sea surface, the differences between the ship target and the floating target on the sea surface under the staring mode are explored, and the statistics that can describe the differences between the two are constructed, which can effectively solve the above problems. Obviously, the floating target has a uniform mass distribution and a small density. Under the staring condition, the coupling degree with the waves is high, and the periodicity of the ups and downs with the waves is obvious; the ship target is large in size and complex in structure. Its own movement or uneven structure causes the distribution of scattering points to be unstable, the coupling degree with the waves is low, and the periodicity of the parameters is not obvious. Therefore, the corresponding relationship between the transient power of sea clutter and the periodic consistency of the bandwidth at the center of mass is more obvious on the floating target on the sea surface.

[0005] In order to solve the above problems, the present application is implemented through the following technical solutions: A method for identifying sea surface targets in a radar staring mode is characterized in that it comprises the following steps: Step 1: Collect echo data to obtain the distance unit where the target is located; Step 2: Perform time domain processing of echo data and time domain feature extraction according to the distance unit where the target is located obtained in step 1; Step 3, frequency domain feature extraction of echo data: in each data segment, calculate the Fourier transform FFT of each data segment to obtain the Doppler spectrum of the data segment, and use a non-parametric method to obtain the centroid sequence of the Doppler spectrum corresponding to each data segment and the root mean square bandwidth sequence BW at the centroid; Step 4, constructing a feature sequence: assigning a label value to each feature vector, normalizing the transient echo power level sequence obtained in step 2 and the RMS bandwidth sequence at the centroid obtained in step 3 to obtain a feature sequence, and dividing the feature sequence into a training set and a test set; Step 5: Optimize and train the classifier: Optimize the parameters of the SVM classifier and train it to obtain an optimal classifier model;

[0006] Step 6: Target recognition: Input the data to be recognized into the classifier model trained in step 5 to realize the recognition of ship targets and floating targets.

[0007] Furthermore, the step 1 is specifically as follows: the sea detection radar scans and detects the sea surface target. After detecting the target, the target is stared at at a fixed angle, and the radar receiver receives the target echo signal. The echo data set in the staring mode is recorded as , Where: is the number of frames of radar data collection, , Indicates The dimensions of the frame data are The time domain echo data, Indicates the number of pulses contained in each frame, Represents the number of distance units in the gaze scene; For moving targets with moving distance units, non-coherent accumulation is used to align the distance units frame by frame: , , That is the The distance unit where the target is located in the frame data.

[0008] Furthermore, the step 2 is as follows: set the number of pulses contained in each segment of data And the overlap rate of adjacent data segments, select the distance unit where the target is located , intercept the radar echo along the time dimension ; In each data segment In the segment, the data in the segment is processed by sliding window to remove the speckle component and calculate the transient echo power level sequence of the segment data: , Where: For the The complex value of the sampling points, express The transient power level sequence of the echo in the data segment, Indicates the length of a single data segment.

[0009] Furthermore, the specific calculation method of step 3 is as follows: Step 3.1, calculate the spectrum function of each segment of data: select the appropriate number of FFT points for each segment of data, calculate the short-time power spectrum, and obtain the spectrum function of each segment of data: , Where: Indicates In order to eliminate the influence of range side lobes on frequency energy, it is necessary to perform windowing in the time domain. represents the time domain window function, Indicates The radar echo data of the data segment, exp represents the exponential function with the natural constant e as the base, n represents the sampling point, T represents the pulse repetition frequency of the radar, Indicates the frequency point number. Indicates Frequency points; Step 3.2, calculate the Doppler centroid sequence and the bandwidth sequence at the centroid: use the non-parametric method in modern signal spectrum analysis to obtain the Doppler centroid sequence of the target and the RMS bandwidth sequence at the centroid : , , Where: Indicates the length of a single data segment. Indicates frequency, represents the short-time Doppler spectrum, Indicates The power level of the short-time Doppler spectrum of a data segment, that is: ; Where: Indicates The short-time power spectrum of a data segment.

[0010] Furthermore, the step 4 specifically includes the following steps: Step 4.1: Normalize the transient echo power level: , Where: Indicates the maximum value operation. A column vector representing the transient power level sequence of the normalized time domain echo of the target; Step 4.2: For the RMS bandwidth sequence at the centroid Perform normalization: , Where: Indicates the maximum value operation. Represents the column vector of the bandwidth sequence at the centroid after target normalization; Step 4.3: Normalize the column vector of the transient power level sequence of the time domain echo in step 4.1 and the column vector consisting of the bandwidth sequence at the centroid after normalization in step 4.2 The ratio of the length of the training set and the test set data to form the feature sequence is 3:1. The training set is expressed as: .

[0011] Furthermore, the step 5 is specifically as follows: Using the data in the training set, the hyperparameters in the SVM classifier are optimized using a cyclic iteration method, including the penalty factor c and the RBF kernel function parameter gamma. The range of both parameters starts from a minimum value of 2-8 and ends at a maximum value of 28. The cycle step is exponentially increased by one. For each set of parameters, the model is cross-validated with a five-fold cross validation using the test set in step 4. The parameters with the best recognition effect are selected to obtain the classifier model.

[0012] The sea surface target recognition method of the present application can utilize radar historical data to improve the utilization rate of target echo information; it is independent of the radar detection angle of view and has strong applicability; and it has high recognition accuracy: at the observation time of seconds, the recognition accuracy of this method for targets with large mass differences such as ship targets and floating targets can reach more than 90%. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Attached Figure 1 It is an implementation flow chart of the present invention; Attached Figure 2 is the sequence of transient power levels of ship targets and the sequence of bandwidth at the centroid proposed by the present invention; Attached Figure 3 It is the transient power level sequence of the floating target on the sea surface and the bandwidth sequence at the center of mass mentioned in the present invention. DETAILED DESCRIPTION

[0014] The following is a specific implementation of the present invention with reference to the accompanying drawings to further illustrate the composition of the present invention.

[0015] Embodiment 1. A method for identifying sea surface targets in radar staring mode, the specific process is as follows Figure 1 As shown, the following steps are included: Step 1: Collect echo data to obtain the distance unit where the target is located; Step 2: Perform time domain processing of echo data and time domain feature extraction according to the distance unit where the target is located obtained in step 1; Step 3, frequency domain feature extraction of echo data: in each data segment, calculate the Fourier transform FFT of each data segment to obtain the Doppler spectrum of the data segment, and use a non-parametric method to obtain the centroid sequence of the Doppler spectrum corresponding to each data segment and the root mean square bandwidth sequence BW at the centroid; Step 4, constructing a feature sequence: assigning a label value to each feature vector, with the ship target being 1 and the floating target being 2. Normalizing the transient echo power level sequence obtained in step 2 and the RMS bandwidth sequence at the centroid obtained in step 3 to obtain a feature sequence, and dividing the feature sequence into a training set and a test set. Step 5: Optimize and train the classifier: Optimize the parameters of the SVM classifier and train it to obtain an optimal classifier model; Step 6: Target recognition: Input the data to be recognized into the classifier model trained in step 5 to realize the recognition of ship targets and floating targets.

[0016] Furthermore, the step 1 is specifically as follows: the sea detection radar scans and detects the sea surface target. After detecting the target, the target is stared at at a fixed angle, and the radar receiver receives the target echo signal. The echo data set in the staring mode is recorded as , Where: is the number of frames of radar data collection, , Indicates The dimensions of the frame data are The time domain echo data, Indicates the number of pulses contained in each frame, Represents the number of distance units in the gaze scene; For moving targets with moving distance units, non-coherent accumulation is used to align the distance units frame by frame: , , That is the The distance unit where the target is located in the frame data.

[0017] Furthermore, the step 2 is as follows: set the number of pulses contained in each segment of data And the overlap rate of adjacent data segments, select the distance unit where the target is located , intercept the radar echo along the time dimension ; In each data segment In the segment, the data in the segment is processed by sliding window to remove the speckle component and calculate the transient echo power level sequence of the segment data: , Where: For the The complex value of the sampling points, express The transient power level sequence of the echo in the data segment, Indicates the length of a single data segment.

[0018] Furthermore, the specific calculation method of step 3 is as follows: Step 3.1, calculate the spectrum function of each segment of data: Select the appropriate number of FFT points for each segment of data, calculate the short-time power spectrum, and obtain the spectrum function of each segment of data: , Where: Indicates In order to eliminate the influence of range side lobes on frequency energy, it is necessary to perform windowing in the time domain. represents the time domain window function, Indicates The radar echo data of the data segment, exp represents the exponential function with the natural constant e as the base, n represents the sampling point, T represents the pulse repetition frequency of the radar, Indicates the frequency point number. Indicates Frequency points; Step 3.2, calculate the Doppler centroid sequence and the bandwidth sequence at the centroid: use the non-parametric method in modern signal spectrum analysis to obtain the Doppler centroid sequence of the target and the RMS bandwidth sequence at the centroid : , , Where: Indicates the length of a single data segment. Indicates frequency, represents the short-time Doppler spectrum, Indicates The power level of the short-time Doppler spectrum of a data segment, that is: ; Where: Indicates The short-time power spectrum of a data segment.

[0019] Furthermore, the step 4 specifically includes the following steps: Step 4.1: Normalize the transient echo power level: , Where: Indicates the maximum value operation. A column vector representing the transient power level sequence of the normalized time domain echo of the target; Step 4.2: For the RMS bandwidth sequence at the centroid Perform normalization: , Where: Indicates the maximum value operation. Represents the column vector of the bandwidth sequence at the centroid after target normalization; The ship target transient power level sequence and the bandwidth sequence at the center of mass are as follows: Figure 2 As shown; The transient power level sequence of floating targets on the sea surface and the bandwidth sequence at the center of mass are as follows: Figure 3 As shown; Step 4.3: Normalize the column vector of the transient power level sequence of the time domain echo in step 4.1 and the column vector consisting of the bandwidth sequence at the centroid after normalization in step 4.2 The ratio of the length of the training set and the test set data to form the feature sequence is 3:1. The training set is expressed as: .

[0020] Furthermore, the step 5 is specifically as follows: Using the data in the training set, the hyperparameters in the SVM classifier are optimized using a cyclic iteration method, including the penalty factor c and the RBF kernel function parameter gamma. The range of both parameters starts from a minimum value of 2-8 and ends at a maximum value of 28. The cycle step is exponentially increased by one. For each set of parameters, the model is cross-validated with a five-fold cross validation using the test set in step 4. The parameters with the best recognition effect are selected to obtain the classifier model.

[0021] The specific embodiments described herein are merely examples of the spirit of the present invention. Those skilled in the art may make various modifications or additions to the specific embodiments described or replace them in similar ways, but they will not deviate from the spirit of the present invention or exceed the defined scope.

Claims

1. A method for identifying sea surface targets in radar staring mode, characterized in that: The following steps are involved: Step 1: Collect echo data to obtain the distance unit where the target is located; Step 2: Perform time domain processing of echo data and time domain feature extraction according to the distance unit where the target is located obtained in step 1; Step 3, frequency domain feature extraction of echo data: in each data segment, calculate the Fourier transform FFT of each data segment to obtain the Doppler spectrum of the data segment, and use a non-parametric method to obtain the centroid sequence of the Doppler spectrum corresponding to each data segment and the root mean square bandwidth sequence BW at the centroid; Step 4, constructing a feature sequence: assigning a label value to each feature vector, normalizing the transient echo power level sequence obtained in step 2 and the RMS bandwidth sequence at the centroid obtained in step 3 to obtain a feature sequence, and dividing the feature sequence into a training set and a test set; Step 5: Optimize and train the classifier: Optimize the parameters of the SVM classifier and train it to obtain an optimal classifier model; Step 6: Target recognition: Input the data to be recognized into the classifier model trained in step 5 to realize the recognition of ship targets and floating targets.

2. The method for identifying sea surface targets in radar staring mode according to claim 1, characterized in that: The step 1 is specifically as follows: the sea detection radar scans and detects the sea surface target. After detecting the target, the target is stared at at a fixed angle. The radar receiver receives the target echo signal. The echo data set in the staring mode is recorded as , Where: is the number of frames of radar data collection, , Indicates The dimensions of the frame data are The time domain echo data, Indicates the number of pulses contained in each frame, Represents the number of distance units in the gaze scene; For moving targets with moving distance units, non-coherent accumulation is used to align the distance units frame by frame: , , That is the The distance unit where the target is located in the frame data.

3. The method for identifying sea surface targets in radar staring mode according to claim 1 or 2, characterized in that: The step 2 is specifically as follows: Set the number of pulses contained in each segment of data And the overlap rate of adjacent data segments, select the distance unit where the target is located , intercept the radar echo along the time dimension ; In each data segment In the segment, the data in the segment is processed by sliding window to remove the speckle component and calculate the transient echo power level sequence of the segment data: , Where: For the The complex value of the sampling points, express The transient power level sequence of the echo in the data segment, Indicates the length of a single data segment.

4. The method for identifying sea surface targets in radar staring mode according to claim 3, characterized in that: The specific calculation method of step 3 is as follows: Step 3.1, calculate the spectrum function of each segment of data: select the appropriate number of FFT points for each segment of data, calculate the short-time power spectrum, and obtain the spectrum function of each segment of data: ， Where: Indicates In order to eliminate the influence of range side lobes on frequency energy, it is necessary to perform windowing in the time domain. represents the time domain window function, Indicates The radar echo data of the data segment, exp represents the exponential function with the natural constant e as the base, n represents the sampling point, T represents the pulse repetition frequency of the radar, Indicates the frequency point number. Indicates Frequency points; Step 3.2, calculate the Doppler centroid sequence and the bandwidth sequence at the centroid: use the non-parametric method in modern signal spectrum analysis to obtain the Doppler centroid sequence of the target and the RMS bandwidth sequence at the centroid : ， ， Where: Indicates the length of a single data segment. Indicates frequency, represents the short-time Doppler spectrum, Indicates The power level of the short-time Doppler spectrum of a data segment, that is: ; Where: Indicates The short-time power spectrum of a data segment.

5. The method for identifying sea surface targets in radar staring mode according to claim 3, characterized in that: The step 4 specifically comprises the following steps: Step 4.1: Normalize the transient echo power level: , Where: Indicates the maximum value operation. A column vector representing the transient power level sequence of the normalized time domain echo of the target; Step 4.2: For the RMS bandwidth sequence at the centroid Perform normalization: ， Where: Indicates the maximum value operation. Represents the column vector of the bandwidth sequence at the centroid after target normalization; Step 4.3: Normalize the column vector of the transient power level sequence of the time domain echo in step 4.1 and the column vector consisting of the bandwidth sequence at the centroid after normalization in step 4.2 The ratio of the length of the training set and the test set data to form the feature sequence is 3:

1. The training set is expressed as: .

6. The method for identifying sea surface targets in radar staring mode according to claim 3, characterized in that: The step 5 is specifically as follows: Using the data in the training set, the hyperparameters in the SVM classifier are optimized using a cyclic iteration method, including the penalty factor c and the RBF kernel function parameter gamma. The range of both parameters starts from a minimum value of 2-8 and ends at a maximum value of 28. The cycle step is exponentially increased by one. For each set of parameters, the model is cross-validated with a five-fold cross validation using the test set in step 4. The parameters with the best recognition effect are selected to obtain the classifier model.

Images