阅读说明：本技术 一种基于特征融合的海面小目标检测方法及系统 (Sea surface small target detection method and system based on feature fusion ) 是由 许聪 刘海成 王峥 于 2021-08-09 设计创作，主要内容包括：本发明公开了一种基于特征融合的海面小目标检测方法,用于对海监视场景,包括如下步骤：步骤1,提取雷达序列的统计复杂性特征,得到雷达序列特征；步骤2,提取雷达图像的空间特征,得到雷达图像特征；步骤3,对所述步骤1的雷达序列特征和所述步骤2的雷达图像特征进行融合处理；步骤4,对所述步骤3的融合结果进行分类,利用分类算法得到相应的分类结果。与现有检测方法相比,本申请通过特征融合提高了检测性能,且大大减少了计算量。(The invention discloses a sea surface small target detection method based on feature fusion, which is used for monitoring a view scene of a sea, and comprises the following steps: step 1, extracting the statistical complexity characteristics of a radar sequence to obtain radar sequence characteristics; step 2, extracting the spatial characteristics of the radar image to obtain radar image characteristics; step 3, fusing the radar sequence characteristics of the step 1 and the radar image characteristics of the step 2; and 4, classifying the fusion result obtained in the step 3, and obtaining a corresponding classification result by using a classification algorithm. Compared with the existing detection method, the detection performance is improved through feature fusion, and the calculated amount is greatly reduced.)

1. A sea surface small target detection method based on feature fusion is used for viewing scenes of a sea prison and is characterized by comprising the following steps:

step 1, extracting the statistical complexity characteristics of a radar sequence to obtain radar sequence characteristics;

step 2, extracting the spatial characteristics of the radar image to obtain radar image characteristics;

step 3, fusing the radar sequence characteristics of the step 1 and the radar image characteristics of the step 2;

and 4, classifying the fusion result obtained in the step 3, and obtaining a corresponding classification result by using a classification algorithm.

2. The sea surface small target detection method based on feature fusion as claimed in claim 1, wherein the step 1 further comprises:

step 101, performing sequential mode conversion on a radar sequence;

step 102, constructing a transfer graph;

step 103, calculating the statistical complexity.

3. The sea surface small target detection method based on feature fusion as claimed in claim 2, wherein the step 101 further comprises: giving a radar sequenceThe time delay embedding expression of the embedding dimension D and the time delay tau is as follows:

wherein T is 1, 2, …, N is T- (D-1) τ, andmapping to sequence number vectorsx_tIn ascending order.

4. A method for detecting small objects on the sea surface based on feature fusion as claimed in claim 3, characterized in that D and τ are set as D e [3, 10] and τ 1.

5. The sea surface small target detection method based on feature fusion as claimed in claim 1, wherein said step 102 further comprises: embedding the degree of separation of the radar sequence from the estimated value corresponding to the sequence into the edge weight, assumingThe method comprises the steps of obtaining a distribution function of sea clutter by using an estimation algorithm based on moments, calculating Kolmogorov-Smimov values between a radar sequence value and an estimation value of the radar sequence value, and calculating the Kolmogorov-Smimov values respectivelyAndand taking the difference between the two sums as the edge weight, the transition probabilities are as follows:

wherein the content of the first and second substances,is represented bySequence number vectorTo sequence number vectorThe transition probability of (a) is,is represented by a sequence number vectorTo sequence number vectorThe edge weight of (a) is calculated,representing the sum of all edge weights.

6. The sea surface small target detection method based on feature fusion as claimed in claim 1, wherein the step 103 further comprises:

statistical complexity characterizes the association and dependency between elements of a given sequence, defined as

Wherein the content of the first and second substances,is a statistical complexity feature, H (P) is normalized Shannon entropy, Q (P, U) is an imbalance factor, P is a set of all transition probabilities, U is a uniform distribution, P_lAnd u_lThe elements in the P and U sets respectively,and is an integer which is the number of the whole,and is an integer which is the number of the whole,

7. the sea surface small target detection method based on feature fusion as claimed in claim 1, wherein the step 2 further comprises:

step 201, constructing a radar image, calculating a radar distance and angle measurement map, reserving a measurement value if the measurement value exceeds a first preset threshold value, setting the measurement value to 0 if the measurement value does not reach the first preset threshold value, and overlapping a plurality of distance and angle maps to obtain a final radar image;

spatial features are computed, mathematical morphology operators are intended to provide useful image spatial features, where four different operators, erosion, dilation, opening and closing, are applied, and the circular structuring element radius range is set to {2, 4, 6}, such that each pixel contains a feature set of size 12, step 202.

8. The method of claim 7, wherein the first predetermined threshold is equal to the estimated average sea clutter value.

9. A sea surface small target detection system based on feature fusion for viewing scenes of a sea prison, comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the steps of the method according to any one of claims 1 to 8 are implemented when the computer program is executed by the processor.

10. A computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, carries out the steps of the method of any one of claims 1 to 8.

Technical Field

The invention relates to the technical field of radar detection, in particular to a sea surface small target detection method and system based on feature fusion.

Background

The detection of small targets on the sea surface is an urgent problem to be solved in sea surveillance. Existing detection methods can be broadly divided into two categories: one based on radar observation sequences and the other based on radar images. The most common method in the former is to divide the radar sequence into piecewise stationary sequences and then use conventional detection methods. Common algorithms include change point based detection algorithms and other statistical methods. Recently, new detection algorithms based on feature classification have emerged. The latter is mainly a detection algorithm based on SAR images. A deep learning method is generally adopted to detect the SAR image target. In addition, radar spectrogram-based detection algorithms also exist. And distinguishing the target echo and the clutter by finding out the characteristic difference between the two echoes. However, there are few detection algorithms where both are fused.

Recently, feature-based detection algorithms have attracted many scholars. The feature sources evolve from manual settings gradually towards deep learning. Common features include geometric features, texture features, moment features, scattering statistics features, scale invariant features, HOG features, deep learning features, and the like.

So far, detection algorithms based on single features or feature fusion do not achieve ideal detection results. Mainly because the features used do not adequately exploit the features of the radar sequence or radar image. While deep learning features perform better, it requires a large amount of labeling data and there is no uniform characterization criteria yet.

Disclosure of Invention

The present invention is directed to solving at least one of the problems of the prior art. Therefore, the invention discloses a sea surface small target detection method based on feature fusion, which is used for monitoring a view scene of a sea prison and comprises the following steps:

step 1, extracting the statistical complexity characteristics of a radar sequence to obtain radar sequence characteristics;

step 2, extracting the spatial characteristics of the radar image to obtain radar image characteristics;

step 3, fusing the radar sequence characteristics of the step 1 and the radar image characteristics of the step 2;

and 4, classifying the fusion result obtained in the step 3, and obtaining a corresponding classification result by using a classification algorithm.

Still further, the step 1 further comprises:

step 101, performing sequential mode conversion on a radar sequence;

step 102, constructing a transfer graph;

step 103, calculating the statistical complexity.

Still further, the step 101 further comprises: giving a radar sequenceThe time delay embedding expression of the embedding dimension D and the time delay tau is as follows:

wherein T is 1, 2, …, N is T- (D-1) τ, andmapping to sequence number vectorsx_tIn ascending order.

Further, D and τ are set to D ∈ [3, 10] and τ ═ 1.

Still further, the step 102 further comprises: embedding the degree of separation of the radar sequence from the estimated value corresponding to the sequence into the edge weight, assumingThe method comprises the steps of obtaining a distribution function of sea clutter by using an estimation algorithm based on moments, calculating Kolmogorov-Smimov values between a radar sequence value and an estimation value of the radar sequence value, and calculating the Kolmogorov-Smimov values respectivelyAndand taking the difference between the two sums as the side weight, the transition probabilities are as follows:

wherein the content of the first and second substances,is represented by a sequence number vectorTo sequence number vectorThe transition probability of (a) is,is represented by a sequence number vectorTo sequence number vectorThe edge weight of (a) is calculated,representing the sum of all edge weights.

Still further, the step 103 further comprises:

statistical complexity characterizes the association and dependency between elements of a given sequence, defined as

Wherein the content of the first and second substances,is a statistical complexity feature, H (P) is normalized Shannon entropy, Q (P, U) is an imbalance factor, P is a set of all transition probabilities, U is a uniform distribution, P_lAnd u_lThe elements in the P and U sets respectively,and is an integer which is the number of the whole,and is an integer which is the number of the whole,

still further, the step 2 further comprises:

step 201, constructing a radar image, calculating a radar distance and angle measurement map, reserving a measurement value if the measurement value exceeds a first preset threshold value, setting the measurement value to 0 if the measurement value does not reach the first preset threshold value, and overlapping a plurality of distance and angle maps to obtain a final radar image;

spatial features are computed, mathematical morphology operators are intended to provide useful image spatial features, where four different operators, erosion, dilation, opening and closing, are applied, and the circular structuring element radius range is set to {2, 4, 6}, such that each pixel contains a feature set of size 12, step 202. Further, the first preset threshold is equal to the estimated average sea clutter.

The invention also discloses a sea surface small target detection system based on feature fusion, which is used for monitoring the visual scene of the sea, and comprises a memory, a processor and a computer program which is stored on the memory and can be run on the processor, and is characterized in that the processor realizes the steps of any one of the methods when executing the computer program.

The invention further discloses a computer-readable storage medium, on which a computer program is stored which, when being executed by a processor, carries out the steps of the above-mentioned method.

Compared with the existing detection method, the detection performance is improved through feature fusion, and the calculated amount is greatly reduced.

Drawings

The invention will be further understood from the following description in conjunction with the accompanying drawings. The components in the figures are not necessarily to scale, emphasis instead being placed upon illustrating the principles of the embodiments. In the drawings, like reference numerals designate corresponding parts throughout the different views.

FIG. 1 is a schematic logic flow diagram of the present invention;

fig. 2 is a flowchart of a method for constructing a radar image according to an embodiment of the present invention.

Detailed Description

Example one

As shown in fig. 1, the present embodiment provides a method for detecting a small sea surface target based on feature fusion, which is used for monitoring a view scene of a sea prison, and includes the following steps:

step one, extracting the statistical complexity characteristic of a radar sequence;

first, sequential mode conversion is performed on a radar sequence. Giving a radar sequenceThe time delay embedding expression of the embedding dimension D and the time delay tau is as follows:

wherein t is 1, 2, …, N,will be provided withMapping to sequence number vectorsx_tIn ascending order. Typically D and τ are set to D ∈ [3, 10]]And τ ═ 1.

Next, a transition graph is constructed. Since the sea clutter has non-uniform non-stationary characteristics, the separation of the radar sequence from its estimate is embedded in the edge weights. Suppose thatDoes not contain eyesAnd obtaining a distribution function of the sea clutter by using a moment-based estimation algorithm. Kolmogorov-Smirnov values between the radar sequence values and their estimated values were calculated. Respectively calculateAndand the difference between the two sums is taken as the edge weight. The transition probabilities obtained were as follows:

wherein the content of the first and second substances,is represented by a sequence number vectorTo sequence number vectorThe transition probability of (a) is,is represented by a sequence number vectorTo sequence number vectorThe edge weight of (a) is calculated,representing the sum of all edge weights.

Finally, the statistical complexity is calculated. Statistical complexity characterizes the association and dependency between elements of a given sequence, defined as

Wherein the content of the first and second substances,is a statistical complexity feature, H (P) is normalized Shannon entropy, Q (P, U) is an imbalance factor, P is a set of all transition probabilities, U is a uniform distribution, P_lAnd u_lThe elements in the P and U sets respectively,and is an integer which is the number of the whole,and is an integer which is the number of the whole,

and step two, extracting the spatial characteristics of the radar image.

First, a radar image is constructed. As shown in fig. 2, a radar distance angle measurement map is calculated, and if the measurement value exceeds the threshold value, it is retained, whereas it is set to 0. The threshold value is equal to the average value of the sea clutter estimated in the step one. And superposing the plurality of distance angle graphs to obtain a final radar image.

Second, spatial features are computed, mathematical morphology operators are aimed at providing useful image spatial features, where four different operators, erosion, dilation, opening and closing, are applied, and the circular structuring element radius range is set to {2, 4, 6}, such that each pixel contains a set of features of size 12.

And step three, fusing radar sequence characteristics and radar image characteristics.

And step four, obtaining a classification result by using a classification algorithm.

Example two

The embodiment provides a sea surface small target detection method based on feature fusion, which is used for monitoring a view scene of a sea prison and comprises the following steps: step 1, extracting the statistical complexity characteristics of a radar sequence to obtain radar sequence characteristics; step 2, extracting the spatial characteristics of the radar image to obtain radar image characteristics; step 3, fusing the radar sequence characteristics of the step 1 and the radar image characteristics of the step 2; and 4, classifying the fusion result obtained in the step 3, and obtaining a corresponding classification result by using a classification algorithm. Still further, the step 1 further comprises: step 101, performing sequential mode conversion on a radar sequence; step 102, constructing a transfer graph; step 103, calculating the statistical complexity. Still further, the step 101 further comprises: giving a radar sequenceThe time delay embedding expression of the embedding dimension D and the time delay tau is as follows:

wherein T is 1, 2, …, N is T- (D-1) τ, andmapping to sequence number vectorsx_tIn ascending order.

Further, D and τ are set to D ∈ [3, 10] and τ ═ 1.

Still further, the step 102 further comprises: embedding the degree of separation of the radar sequence from the estimated value corresponding to the sequence into the edge weight, assumingNot containing targets, using moment-based estimationObtaining a distribution function of the sea clutter by an algorithm, calculating a Kolmogorov-Smimov value between a radar sequence value and an estimation value thereof, and respectively calculatingAndand taking the difference between the two sums as the edge weight, the transition probabilities are as follows:

wherein the content of the first and second substances,is represented by a sequence number vectorTo sequence number vectorThe transition probability of (a) is,is represented by a sequence number vectorTo sequence number vectorThe edge weight of (a) is calculated,representing the sum of all edge weights.

Still further, the step 103 further comprises:

statistical complexity characterizes the association and dependency between elements of a given sequence, defined as

Wherein the content of the first and second substances,is a statistical complexity feature, H (P) is normalized Shannon entropy, Q (P, U) is an imbalance factor, P is a set of all transition probabilities, U is a uniform distribution, P_lAnd u_lThe elements in the P and U sets respectively,and is an integer which is the number of the whole,and is an integer which is the number of the whole,

still further, the step 2 further comprises:

step 201, constructing a radar image, calculating a radar distance and angle measurement map, reserving a measurement value if the measurement value exceeds a first preset threshold value, setting the measurement value to 0 if the measurement value does not reach the first preset threshold value, and overlapping a plurality of distance and angle maps to obtain a final radar image;

spatial features are computed, mathematical morphology operators are intended to provide useful image spatial features, where four different operators, erosion, dilation, opening and closing, are applied, and the circular structuring element radius range is set to {2, 4, 6}, such that each pixel contains a feature set of size 12, step 202.

Further, the first preset threshold is equal to the estimated average sea clutter.

The invention also discloses a sea surface small target detection system based on feature fusion, which is used for monitoring the visual scene of the sea, and comprises a memory, a processor and a computer program which is stored on the memory and can be run on the processor, and is characterized in that the processor realizes the steps of any one of the methods when executing the computer program.

The invention further discloses a computer-readable storage medium, on which a computer program is stored which, when being executed by a processor, carries out the steps of the above-mentioned method.

It should also be noted that the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other like elements in a process, method, article, or apparatus that comprises the element.

As will be appreciated by one skilled in the art, embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.

Although the invention has been described above with reference to various embodiments, it should be understood that many changes and modifications may be made without departing from the scope of the invention. It is therefore intended that the foregoing detailed description be regarded as illustrative rather than limiting, and that it be understood that it is the following claims, including all equivalents, that are intended to define the spirit and scope of this invention. The above examples are to be construed as merely illustrative and not limitative of the remainder of the disclosure. After reading the description of the invention, the skilled person can make various changes or modifications to the invention, and these equivalent changes and modifications also fall into the scope of the invention defined by the claims.