A sea surface target recognition method using time domain approximate entropy features

Abstract

The application relates to a sea surface target recognition method using a time domain approximate entropy feature, and belongs to the technical field of radar signal processing and feature recognition. The method comprises the following steps: 1, target echo data buffering; 2, target time domain approximate entropy feature extraction echo data time domain analysis; 3, time domain approximate entropy feature optimization; and 4, target recognition using the time domain approximate entropy feature. The application aims to recognize sea surface targets under the condition that traditional target features cannot effectively recognize targets.

Description

Technical Field

[0001] This invention relates to a method for identifying sea surface targets using time-domain approximate entropy features, belonging to the field of radar signal processing and feature recognition technology. Background Technology

[0002] There are many types of sea surface targets, the most common being floating targets (small fishing boats, buoys, ice floes, etc.) and ship targets (large cruise ships, cargo ships, roll-on / roll-off cargo ships, etc.). The detection and identification of these targets are crucial in fields such as marine monitoring, environmental protection, and shipping safety. Especially for floating targets in complex sea surface scenarios, when using high-resolution range imaging, the angle, distance, and time of observation can affect the target's identifiability. The lack of stable identification features makes floating target identification difficult and impacts recognition performance.

[0003] Therefore, there is an urgent need to explore a more effective and stable feature information to break through the performance bottleneck of target recognition. Summary of the Invention

[0004] Based on radar target detection, this invention proposes a sea surface target identification method that applies temporal approximate entropy features, aiming to identify sea surface targets when traditional target features are difficult to use effectively.

[0005] A method for identifying sea surface targets using temporal approximate entropy features, characterized by the following steps:

[0006] Step 1: Target echo data buffering

[0007] Target detection is performed using the pulse compression processed echo data. The target time-domain echo sequence is extracted and buffered according to the distance cell information of the target.

[0008] Step 2: Extraction of target temporal approximate entropy features and temporal analysis of echo data

[0009] The target time-domain echo sequence is segmented by sliding window, and the data in each segment is reconstructed to calculate the corresponding approximate entropy of the time series. The obtained entropy values ​​are then integrated to form the target feature matrix.

[0010] Step 3: Time-domain approximate entropy feature optimization

[0011] For the two parameters in the feature extraction process, namely the dimension m of data reconstruction and the tolerance coefficient r in the threshold F, the optimal parameter range is determined according to the degree of influence of different parameters on the separability of the two types of targets, so as to achieve feature optimization and improve recognition performance.

[0012] Step 4: Target recognition using temporal approximate entropy features

[0013] Support Vector Machine (SVM) classification algorithm was used to identify two classes of targets, and the recognition performance was tested. The feature matrix composed of multiple targets was divided into training and test sets, and the hyperparameters of the model were optimized, including the penalty factor c and the RBF kernel function parameter gamma, to obtain better recognition results. Finally, the feature vectors of the test set were input into the model for testing to obtain the target recognition results.

[0014] Preferably, the specific steps of step 1 are as follows:

[0015] In tracking mode, the radar receives echo data. After pulse compression, clutter suppression, target detection, and tracking processing, k targets are detected in the current scene. The range cell containing the target's strongest scattering point is selected, and its pulse echo data sequence z is cached. k (m), for the k-th target, it is represented as {z k (1),z k (2),...,z k (L)}, where L represents the length of the coherent pulse train.

[0016] Preferably, the specific steps of step 2 are as follows:

[0017] For a target time-domain echo sequence z detected in step 1), k (m) is segmented, with each segment containing N pulses. After segmentation, I subsequences are obtained, denoted as {x1(n), x2(n), ..., x...}. I (n)|1≤n≤N}, where

[0018] For the truncated i-th time-domain sequence x i (n) Perform temporal approximate entropy feature extraction. The specific steps are as follows:

[0019] ① For the truncated i-th time-domain sequence x i (n), reconstruct its data with m as the dimension, dividing it into k = n - m + 1 sequences, i.e., X i =[x(i),x(i+1),…,x(i+m-1)];

[0020] ② Calculate the distance d between each sequence and all k sequences. ij =max|x i+k (t)-x j+k (t)|, where k = 0, 1, ..., m-1, forming a k×k d ij matrix;

[0021] ③ Define the threshold F and statistically analyze d. ij The ratio of the number of elements less than F in each row of the matrix to the total number of elements k = n - m + 1 is denoted as F. Then based on k Calculate its logarithmic mean Where F = r·SD, r is the tolerance coefficient, and SD is the standard deviation of the sequence;

[0022] ④ Increase the dimension from m to m+1, and repeat steps ① to ③;

[0023] ⑤ Calculate the approximate entropy ApEn(t)=Φ m (t)-Φ m+1 (t).

[0024] For all target time-domain echo sequences detected in step 1), the above feature extraction method is used to extract I corresponding feature vectors for each target. These feature vectors are then integrated to obtain the target feature matrix.

[0025] Preferably, the specific steps of step 3 are as follows:

[0026] Using the Bhattacharyya distance, which measures the similarity between two types of samples, as an indicator, the separability between the ship target sample space S1 and the floating target sample space S2 in the feature space is quantitatively analyzed. By analyzing and optimizing to select a suitable parameter range, the Bhattacharyya distance estimation method is as follows:

[0027] Calculate the mean vector μ and covariance matrix C of the two sample spaces.

[0028] μ1=mean(S1),C1=cov(S1) (1)

[0029] μ2=mean(S2),C2=cov(S2) (2)

[0030] In the formula, mean and cov represent the mean and covariance operators, respectively, and the Bach distance between the two sample spaces is d. B The estimated value of (S1,S2) is obtained by the following formula.

[0031]

[0032] In the formula, C = (C1 + C2) / 2, d μ = (μ1-μ2) / chol(C), where det represents the determinant operator and chole represents the Choleski decomposition operator.

[0033] Compared with existing technologies, the sea surface target identification method using temporal approximate entropy features described in this technical solution has the following advantages:

[0034] (1) The method proposed in this invention uses the approximate entropy of the target as a feature, measures the similarity between sequences by reconstructing time series data, and then studies the similarity difference between two adjacent reconstructed data to obtain more target feature information, thus solving the problem of relying on traditional target features but making it difficult to achieve recognition.

[0035] (2) Under the existing data conditions, the method proposed in this invention achieves an average accuracy of 95.05% in identifying ship targets and floating targets in an observation time of 0.5s, which is 40.63% higher than existing methods. Attached Figure Description

[0036] Figure 1 : A comparison of the Bach distance between two types of target samples under different dimensions m;

[0037] Figure 2 : A comparison of the Bhattacharyya distances between two types of target samples under different tolerance coefficients r;

[0038] Figure 3 : A scatter plot of approximate entropy features for two types of targets before and after feature optimization. Detailed Implementation

[0039] To better understand and implement this invention, specific embodiments are provided below to illustrate a sea surface target identification method applying temporal approximate entropy features. The processing flow of this invention is described in detail below:

[0040] 1) Target echo data buffer

[0041] In tracking mode, the radar receives echo data. After pulse compression, clutter suppression, target detection, and tracking processing, k targets are detected in the current scene. The range cell containing the target's strongest scattering point is selected, and its pulse echo data sequence z is cached. k (m), for the k-th target, it is represented as {z k (1),z k (2),…,z k (L)}, where L represents the length of the coherent pulse train.

[0042] 2) Extraction of approximate temporal entropy features from target echo data

[0043] For a target time-domain echo sequence z detected in step 1), k (m) is segmented, with each segment containing N pulses. After segmentation, I subsequences are obtained, denoted as {x1(n), x2(n), ..., x...}. I (n)|1≤n≤N}, where

[0044] For the truncated i-th time-domain sequence xi (n) Perform temporal approximate entropy feature extraction. The specific steps are as follows:

[0045] ① For the truncated i-th time-domain sequence x i (n), reconstruct its data with m as the dimension, dividing it into k = n - m + 1 sequences, i.e., X i =[x(i),x(i+1),…,x(i+m-1)];

[0046] ② Calculate the distance d between each sequence and all k sequences. ij =max|x i+k (t)-x j+k (t)|, where k = 0, 1, ..., m-1, forming a k×k d ij matrix;

[0047] ③ Define the threshold F and statistically analyze d. ij The ratio of the number of elements less than F in each row of the matrix to the total number of elements k = n - m + 1 is denoted as F. Then based on k Calculate its logarithmic mean Where F = r·SD, r is the tolerance coefficient, and SD is the standard deviation of the sequence;

[0048] ④ Increase the dimension from m to m+1, and repeat steps ① to ③;

[0049] ⑤ Calculate the approximate entropy ApEn(t)=Φ m (t)-Φ m+1 (t).

[0050] For all target echo sequences detected in step 1), the above feature extraction method is used to extract I corresponding feature vectors for each target. These feature vectors are then integrated to obtain the target feature matrix.

[0051] 3) Optimization of temporal approximate entropy features

[0052] In step 2), the two parameters in the feature extraction process, namely the dimension m of data reconstruction and the tolerance coefficient r in the threshold F, need to be optimized in combination with the impact on the separability of different targets, so as to achieve feature optimization.

[0053] The Bhattacharyya distance (B-distance), which measures the similarity between two classes of samples, is used as an indicator to quantitatively analyze the separability between the ship target sample space S1 and the floating target sample space S2 in the feature space. An appropriate parameter range is then selected through analysis and optimization. The B-distance estimation method is as follows:

[0054] Calculate the mean vector μ and covariance matrix C of the two sample spaces.

[0055] μ1=mean(S1),C1=cov(S1) (1)

[0056] μ2=mean(S2),C2=cov(S2) (2)

[0057] In the formula, mean and cov represent the mean and covariance operators, respectively, and the Bach distance between the two sample spaces is d. B The estimated value of (S1,S2) is obtained by the following formula.

[0058]

[0059] In the formula, C = (C1 + C2) / 2, d μ = (μ1-μ2) / chol(C). Here, det represents the determinant operator, and cholesky represents the Cholesky decomposition operator.

[0060] The following study uses measured data to investigate the effects of these two parameters on the separability of the two types of targets, in order to determine the appropriate parameter range.

[0061] (1) Reconstruct the dimension m of the data space

[0062] The Bach distance between the two types of target samples under different dimensions m was obtained using measured data. Figure 1 As shown, the larger the Bach distance, the better the feature separability. Obviously, when the dimension m is set to 2, the overall feature separability is good, but as m increases, the feature separability deteriorates.

[0063] (2) The tolerance coefficient r in the threshold F

[0064] The Bach distance between the two types of target samples under different tolerance coefficients r was obtained using measured data. Figure 2 As shown in the figure. Obviously, when the tolerance coefficient r is selected as 0.2, the overall feature separability is good, and the smaller the tolerance coefficient r is, the worse the overall feature separability is.

[0065] After feature optimization, by selecting an appropriate parameter range, the Bach distance between the two classes of samples is increased. The comparison of the extracted two-dimensional features before and after optimization is as follows: Figure 3 As shown, where Figure 3 (a) is the two-dimensional approximate entropy feature scatter envelope diagram of the two types of targets before feature optimization. Figure 3 (b) is the two-dimensional approximate entropy feature scatter envelope diagram after feature optimization. After optimization, the overlap between the two types of target features is significantly reduced, and the feature separability is improved.

[0066] 4) Target recognition using temporal approximate entropy features

[0067] In step 3), feature vectors of all targets were obtained. Classification algorithms from machine learning and deep learning were used to identify the two types of targets, and the recognition performance was tested. Classification algorithms included decision trees, support vector machines, Naive Bayes, convex hull algorithms, and convolutional neural networks.

[0068] Three sets of measured data were used for feature extraction, and their basic information is shown in Table 1. The echo data was segmented, with each segment consisting of 1024 pulses and a duration of approximately 0.5 seconds. A total of 128 feature vectors could be extracted for each target, and a total of 256 feature values ​​related to the two types of targets could be obtained from each set of data, forming a feature matrix.

[0069] Table 1. Description of Measured Data

[0070]

[0071] This invention employs a Support Vector Machine (SVM) recognition method. First, each feature vector is assigned a label value during recognition; for example, a ship target is labeled 1, and a floating target is labeled 2. The feature matrix is ​​then divided into a training set and a test set. Next, an SVM classifier model is constructed using the training set. The hyperparameters in the SVM, including the penalty factor c and the RBF kernel function parameter gamma, are optimized using an iterative method, with both parameters ranging from a minimum value of 2. -8 Initially, the maximum value is 2. 8 The process ends with the loop step size increasing exponentially by one. Five-fold cross-validation is performed on the model for each set of parameters, and the parameters with the best recognition performance are selected to obtain the classifier model. Finally, the test set data is input into the model for recognition. The test results will show the following four scenarios:

[0072] ①True Positive (TP): The model correctly predicts samples that are actually ship targets as ship targets;

[0073] ② False Negative (FN): The model incorrectly predicts samples that are actually ship targets as floating targets;

[0074] ③ False Positive (FP): The model incorrectly predicts samples that are actually floating targets as ship targets;

[0075] ④True Negative (TN): The model correctly predicts samples that are actually floating targets as floating targets.

[0076] The recognition results were statistically analyzed and are shown in Table 2. For ease of comparison, the recognition results using time-domain multi-waveform entropy as a feature are presented in Table 3. It can be seen that the average recognition accuracy achieved by existing methods is 54.42%, while the average accuracy of the method proposed in this invention reaches 95.05%. Under existing data conditions, the recognition accuracy is improved by 40.63% compared to existing methods.

[0077] Table 2. Statistics of Time-Domain Approximate Entropy Feature Target Recognition Results

[0078]

[0079] The above examples of the present invention are merely illustrative of the computational model and process of the present invention, and are not intended to limit the implementation of the present invention. Those skilled in the art will recognize that other variations or modifications can be made based on the above description. It is impossible to exhaustively list all possible implementations here. Any obvious variations or modifications derived from the technical solutions of the present invention are still within the scope of protection of the present invention.

Claims

1. A method for identifying sea surface targets using temporal approximate entropy features, characterized by the following steps: Step 1: Target echo data buffering Target detection is performed using the pulse compression processed echo data. The target time-domain echo sequence is extracted and buffered according to the distance cell information of the target. Step 2: Extraction of target temporal approximate entropy features and temporal analysis of echo data The target time-domain echo sequence is segmented by sliding window, and the data in each segment is reconstructed to calculate the corresponding approximate entropy of the time series. The obtained entropy values ​​are then integrated to form the target feature matrix. Step 3: Time-domain approximate entropy feature optimization For the two parameters in the feature extraction process, namely the dimension m of data reconstruction and the tolerance coefficient r in the threshold F, the optimal parameter range is determined according to the degree of influence of different parameters on the separability of the two types of targets, so as to achieve feature optimization and improve recognition performance. Step 4: Target recognition using temporal approximate entropy features The Support Vector Machine (SVM) classification algorithm was used to identify two types of targets, and the recognition performance was tested. The feature matrix composed of multiple targets was divided into training and test sets, and the hyperparameters of the model were optimized, including the penalty factor c and the RBF kernel function parameter gamma, to obtain better recognition results. Finally, the feature vectors of the test set were input into the model for testing to obtain the target recognition results.

2. The sea surface target identification method using temporal approximate entropy features according to claim 1, characterized in that... The specific steps of step 1 are as follows: In tracking mode, the radar receives echo data. After pulse compression, clutter suppression, target detection, and tracking processing, k targets are detected in the current scene. The range cell containing the target's strongest scattering point is selected, and its pulse echo data sequence z is cached. k (m), for the k-th target, it is represented as {z k (1),z k (2),...,z k (L)}, where L represents the length of the coherent pulse train.

3. A sea surface target identification method applying temporal approximate entropy features according to claim 1, characterized in that, The specific steps of step 2 are as follows: For a target time-domain echo sequence z detected in step 1 k (m) is segmented, with each segment containing N pulses. After segmentation, I subsequences are obtained, denoted as {x1(n), x2(n), ..., x...}. I (n)|1≤n≤N}, where For the truncated i-th time-domain sequence x i (n) Perform temporal approximate entropy feature extraction. The specific steps are as follows: ① For the truncated i-th time-domain sequence x i (n), reconstruct its data with m as the dimension, dividing it into k = n - m + 1 sequences, i.e., X i =[x(i),x(i+1),...,x(i+m-1)]; ② Calculate the distance d between each sequence and all k sequences. ij =max|x i+k (t)-x j+k (t)|, where k = 0, 1, ..., m-1, forming a k×k d ij matrix; ③ Define the threshold F and statistically analyze d. ij The ratio of the number of elements less than F in each row of the matrix to the total number of elements k = n - m + 1 is denoted as F. Then based on k Calculate its logarithmic mean Where F = r·SD, r is the tolerance coefficient, and SD is the standard deviation of the sequence; ④ Increase the dimension from m to m+1, and repeat steps ① to ③; ⑤ Calculate the approximate entropy ApEn(t)=Φ m (t)-Φ m+1 (t); For all target time-domain echo sequences detected in step 1, the above feature extraction method is used to extract I corresponding feature vectors for each target. These feature vectors are then integrated to obtain the target feature matrix.

4. A method for identifying sea surface targets using temporal approximate entropy features according to claim 1, characterized in that... The specific steps of step 3 are as follows: Using the Bhattacharyya distance, which measures the similarity between two types of samples, as an indicator, the separability between the ship target sample space S1 and the floating target sample space S2 in the feature space is quantitatively analyzed. By analyzing and optimizing to select a suitable parameter range, the Bhattacharyya distance estimation method is as follows: Calculate the mean vector μ and covariance matrix C of the two sample spaces. μ1=mean(S1),C1=cov(S1)(1) μ2=mean(S2),C2=cov(S2)(2) In the formula, mean and cov represent the mean and covariance operators, respectively, and the Bach distance between the two sample spaces is d. B The estimated value of (S1,S2) is obtained by the following formula: In the formula, C = (C1 + C2) / 2, d μ = (μ1-μ2) / chol(C), where det represents the determinant operator and chole represents the Choleski decomposition operator.

Images