Satellite-borne SAR and electronic reconnaissance data fusion detection method and system

Abstract

The application provides a kind of spaceborne SAR and electronic reconnaissance signal data fusion method, comprising: obtaining SAR image using spaceborne SAR system, detecting SAR image, obtaining target rough detection result;Obtain spaceborne electronic reconnaissance radiation source detection result;Complete SAR data and electronic reconnaissance data time and space alignment;Using load multi-attribute to construct multi-attribute feature extraction network, the feature of SAR detection data and electronic reconnaissance detection data is modeled to obtain feature fuzzy vector, and the feature that can be used is extracted;According to the space relationship of load, determine the projection space of SAR data and electronic reconnaissance data feature matrix, carry out feature mapping, extract effective information in feature, remove redundant information, and strengthen target multi-attribute feature;The strengthened feature is clustered, false alarm is eliminated, detection is completed, and target feature information after target fusion is obtained, and spaceborne SAR and electronic reconnaissance data fusion detection are completed.

Description

Technical Field

[0001] This invention belongs to the field of space microwave remote sensing technology, and relates to a method and system for fusion detection of spaceborne SAR and electronic reconnaissance data. Background Technology

[0002] Synthetic Aperture Radar (SAR) is a microwave remote sensing radar that uses range-direction pulse compression and azimuth-direction Doppler effects for imaging. It transmits pulse signals and then processes the received echoes to obtain images. Electronic reconnaissance payloads are used to detect and identify external radiation sources, acquiring information about their location and type. Unlike optical and hyperspectral remote sensing equipment, microwave payloads are unaffected by adverse weather conditions such as clouds, rain, and fog, enabling all-weather, 24 / 7 surveillance. They are now widely used in marine applications, military reconnaissance, and scientific research.

[0003] However, using a single payload for target detection is limited by the dimensions of information acquisition, resulting in poor robustness in complex sea conditions. Furthermore, due to the variability of the environment, it is difficult to develop universal target detection algorithms. As the types of spaceborne payloads become increasingly diverse, the use of multi-source data for target detection is receiving more and more attention from researchers and is one of the future development directions. However, most of the multi-fusion models and detection schemes proposed so far only use the position or velocity information of multi-source data, resulting in low information utilization. Moreover, the fusion criteria are merely simple superposition, and the multi-dimensional information of multi-source data is not fully utilized. Summary of the Invention

[0004] The technical problem solved by this invention is: This invention proposes a method and system for fusing spaceborne SAR and electronic reconnaissance signal data based on target multi-attribute enhanced feature clustering. Through theoretical derivation and simulation analysis, a suitable target multi-attribute enhanced feature method is obtained. Then, according to the multi-attribute target fusion decision criterion, the enhanced features are clustered to realize multi-source data fusion detection and improve the robustness of target detection.

[0005] The technical solution of this invention is: a method for fusing spaceborne SAR and electronic reconnaissance signal data, comprising:

[0006] (1) Use a spaceborne SAR system to acquire SAR images, perform detection on the SAR images, and obtain coarse target detection results;

[0007] (2) Obtain the detection results of radiation sources by spaceborne electronic reconnaissance;

[0008] (3) Complete the temporal and spatial alignment of SAR data and electronic reconnaissance data;

[0009] (4) A multi-attribute feature extraction network is constructed using the multi-attribute payload. Relationship modeling is performed on the features of SAR detection data and electronic reconnaissance detection data to obtain feature fuzzy vectors and extract usable features.

[0010] (5) Based on the payload space relationship, determine the projection space of the feature matrix of SAR data and electronic reconnaissance data, perform feature mapping, extract effective information from the features, remove redundant information, and enhance the multi-attribute features of the target.

[0011] (6) Cluster the enhanced features, eliminate false alarms, complete the detection, and obtain the target feature information after target fusion, thus completing the fusion detection of spaceborne SAR and electronic reconnaissance data.

[0012] The feature fuzzy vector is a vector composed of the feature values ​​of the same target point in various feature dimensions, and the number of features corresponding to target p is n. k (p), k=1,...,K, then the feature fuzzy vector is

[0013] N(p) = [n1(p), n2(p), ..., n K (p)],

[0014] Where K is the number of fused attributes; multiple data sources form multiple feature fuzzy vectors, which are jointly composed of feature matrix A. K×H :

[0015]

[0016] Wherein, the characteristic matrix A K×H The element in the h-th row is n1(p) h ),n2(p h ),...,n k (p h ),...,n K (p h ), h=1,2,3,...,H,N(p h )=[n1(p h ),n2(p h ),...,n k (p h ),...,n K (p h )] represents the target p h The corresponding feature fuzzy vector, where H is the number of data sources.

[0017] The enhanced target multi-attribute features include:

[0018] Determine the mapping space R using the load space relationship. M R N , the characteristic matrix AK×H After mapping, mapping vectors M and N are obtained. Based on the obtained mapping features, enhanced features are obtained through multi-attribute interaction.

[0019] Calculate the feature interaction matrix C:

[0020] C = XY T ,

[0021] Where C is the interactive projection metric of the X features of data source 1 and the Y features of data source 2;

[0022] Based on the obtained interaction relationships, the mapped features are enhanced to obtain features M or N:

[0023]

[0024]

[0025] The clustering process for the enhanced features includes:

[0026] Clustering is performed on the enhanced feature vectors M and N, that is, calculating the enhanced feature blur vectors of all targets in the image corresponding to different data sources;

[0027] Targets with the same feature fuzzy vector are grouped into one class.

[0028] A spaceborne SAR and electronic reconnaissance signal data fusion system includes:

[0029] The first module is used to acquire SAR images using the spaceborne SAR system, detect SAR images, and obtain coarse target detection results; acquire the detection results of radiation sources by spaceborne electronic reconnaissance; and complete the temporal and spatial alignment of SAR data and electronic reconnaissance data.

[0030] The second module is used to construct a multi-attribute feature extraction network using the multi-attribute payload, to model the relationship between the features of SAR detection data and electronic reconnaissance detection data to obtain feature fuzzy vectors, and to extract usable features.

[0031] The third module is used to determine the projection space of the feature matrix of SAR data and electronic reconnaissance data according to the payload space relationship, perform feature mapping, extract effective information from the features, remove redundant information, and enhance the multi-attribute features of the target.

[0032] The fourth module is used to cluster the enhanced features, eliminate false alarms, complete the detection, and obtain the target feature information after target fusion, thus completing the fusion detection of spaceborne SAR and electronic reconnaissance data.

[0033] Furthermore, the feature fuzzy vector is a vector composed of the feature values ​​of the same target point in each feature dimension;

[0034] The number of features corresponding to target p is n k (p), k=1,...,K, then the feature fuzzy vector is

[0035] N(p) = [n1(p), n2(p), ..., n K (p)],

[0036] Where K is the number of fusion attributes;

[0037] Several types of data form several feature fuzzy vectors, which are jointly composed of feature matrix A. K×H :

[0038]

[0039] Wherein, the characteristic matrix A K×H The element in the h-th row is n1(p) h ),n2(p h ),...,n k (p h ),...,n K (p h ), h=1,2,3,...,H,N(p h )=[n1(p h ),n2(p h ),...,n k (p h ),...,n K (p h )] represents the target p h The corresponding feature fuzzy vector, where H is the number of data sources.

[0040] Furthermore, the enhanced target multi-attribute features include:

[0041] Determine the mapping space R using the load space relationship. M R N , the characteristic matrix A K×H After mapping, mapping vectors M and N are obtained. Based on the obtained mapping features, enhanced features are obtained through multi-attribute interaction.

[0042] Calculate the feature interaction matrix C:

[0043] C = XY T ,

[0044] Where C is the interactive projection metric of the X features of data source 1 and the Y features of data source 2;

[0045] Based on the obtained interaction relationships, feature enhancement is performed on the mapped features to obtain features M and features N:

[0046]

[0047]

[0048] Furthermore, the clustering process for the enhanced features includes:

[0049] Clustering is performed on the enhanced feature vectors M and N, that is, calculating the enhanced feature blur vectors of all targets in the image corresponding to different data sources;

[0050] Targets with the same feature fuzzy vector are grouped into one class.

[0051] The advantages of this invention compared to the prior art are:

[0052] (1) This invention provides a new method for multi-source data fusion detection of SAR and electronic reconnaissance. By using feature mapping, effective information is extracted from the features, redundant information is removed, and effective features of the payload are enhanced, thereby improving the robustness of multi-source data fusion detection. It is applicable to target detection in complex environments.

[0053] (2) The spaceborne SAR and electronic reconnaissance data fusion detection method proposed in this invention can be directly applied to light and small networked reconnaissance satellite systems, rapid response satellite systems, etc., and can provide strong support in both military and civilian applications. It has broad and important application prospects and value. Attached Figure Description

[0054] Figure 1 This is a flowchart of a multi-source data fusion detection method based on target multi-attribute enhanced feature clustering. Detailed Implementation

[0055] The present invention will be described in conjunction with the accompanying drawings.

[0056] like Figure 1 As shown, a method for fusing spaceborne SAR and electronic reconnaissance signal data based on target multi-attribute enhanced feature clustering is presented. The specific method is as follows:

[0057] (1) Use a spaceborne SAR system to acquire high-resolution SAR images, perform detection on the SAR images, and obtain coarse target detection results;

[0058] (2) Obtain the detection results of radiation sources by spaceborne electronic reconnaissance;

[0059] (3) Complete the temporal and spatial alignment of SAR data and electronic reconnaissance data;

[0060] (4) A multi-attribute feature extraction network is constructed using the multi-attribute payload. Relationship modeling is performed on the features of SAR detection data and electronic reconnaissance detection data to obtain feature fuzzy vectors and extract usable features.

[0061] (5) Based on the payload space relationship, determine the projection space of the feature matrix of SAR data and electronic reconnaissance data, perform feature mapping, extract effective information from the features, remove redundant information, and enhance the multi-attribute features of the target.

[0062] (6) Cluster the enhanced features, eliminate false alarms, complete the detection, and obtain the target feature information after target fusion, thus completing the fusion detection of spaceborne SAR and electronic reconnaissance data.

[0063] In step (4), the feature fuzzy vector is established, including:

[0064] Multiple features of SAR and electronic reconnaissance payloads are extracted and subjected to dimensional transformation to form a feature fuzzy vector. This feature fuzzy vector is defined as a vector composed of the eigenvalues ​​of the same target point across various feature dimensions, where the number of features corresponding to target p is n. k (p), k=1,...,K, then the feature fuzzy vector is

[0065] N(p) = [n1(p), n2(p), ..., n k (p),...,n K (p)],

[0066] Where K is the number of fusion attributes;

[0067] Multiple data sources form multiple feature fuzzy vectors, which are then jointly combined to form feature matrix A. K×H ,

[0068]

[0069] Wherein, the characteristic matrix A K×H Let A be a matrix with H rows and K columns, and its characteristic matrix A be... K×H The element in the h-th row is n1(p) h ),n2(p h ),...,n k (p h ),...,n K (p h ), h=1,2,3,...,H,N(p h )=[n1(p h ),n2(p h ),...,n k (p h ),...,n K (p h)] represents the target p h The corresponding feature fuzzy vector, where H is the number of data sources;

[0070] In step (5), feature information enhancement includes:

[0071] Determine the mapping space R using the load space relationship. M R N , the characteristic matrix A K×H After mapping, mapping vectors M and N are obtained. Based on the obtained mapping features, enhanced features are obtained through multi-attribute interactions. First, the feature interaction matrix C is calculated.

[0072] C = XY T ,

[0073] Where C is the interactive projection metric of the X feature of data source 1 and the Y feature of data source 2, and then feature enhancement is performed on the mapped features based on the obtained interactive relationship to obtain features M and N:

[0074]

[0075]

[0076] By following the steps above, we can robustly capture the potential interactions between features from multiple sources and achieve information fusion between multi-source attribute features.

[0077] In step (6), the enhanced feature fuzzy clustering includes:

[0078] Clustering is performed on the enhanced feature vectors M and N, which involves calculating the enhanced feature fuzzy vectors of all targets in the image corresponding to different data sources. Then, targets with the same fuzzy vectors are clustered together. In actual processing, due to the influence of noise clutter and other factors, the target fuzzy vectors will not completely overlap. It is necessary to distinguish the clustering error, obtain its standard fuzzy vector, and combine it with the feature accuracy to complete the clustering. Here, the feature accuracy depends on the accuracy of the payload's feature information. Based on it, the clustering threshold of the detected target can be determined, thereby generating the target's clustered fuzzy region. The feature accuracy of radar payload and electronic reconnaissance payload changes under different conditions, so the clustering threshold also needs to be specifically determined based on information such as the downward viewing angle, azimuth angle, and reconnaissance frequency at the detection time.

[0079] Based on the clustering results, targets from multiple data sources can be reliably and effectively correlated and detected, and the feature information of the fused targets can be obtained, thus completing the fusion detection of spaceborne SAR and electronic reconnaissance data.

[0080] A spaceborne SAR and electronic reconnaissance signal data fusion system includes:

[0081] The first module is used to acquire SAR images using the spaceborne SAR system, detect SAR images, and obtain coarse target detection results; acquire the detection results of radiation sources by spaceborne electronic reconnaissance; and complete the temporal and spatial alignment of SAR data and electronic reconnaissance data.

[0082] The second module is used to construct a multi-attribute feature extraction network using the multi-attribute payload, to model the relationship between the features of SAR detection data and electronic reconnaissance detection data to obtain feature fuzzy vectors, and to extract usable features.

[0083] The third module is used to determine the projection space of the feature matrix of SAR data and electronic reconnaissance data according to the payload space relationship, perform feature mapping, extract effective information from the features, remove redundant information, and enhance the multi-attribute features of the target.

[0084] The fourth module is used to cluster the enhanced features, eliminate false alarms, complete the detection, and obtain the target feature information after target fusion, thus completing the fusion detection of spaceborne SAR and electronic reconnaissance data.

[0085] Furthermore, the feature fuzzy vector is a vector composed of the feature values ​​of the same target point in each feature dimension;

[0086] The number of features corresponding to target p is n k (p), k=1,...,K, then the feature fuzzy vector is

[0087] N(p) = [n1(p), n2(p), ..., n K (p)],

[0088] Where K is the number of fusion attributes;

[0089] Several types of data form several feature fuzzy vectors, which are jointly composed of feature matrix A. K×H :

[0090]

[0091] Wherein, the characteristic matrix A K×H The element in the h-th row is n1(p) h ),n2(p h ),...,n k (p h ),...,n K (p h ), h=1,2,3,...,H,N(p h )=[n1(p h ),n2(p h ),...,n k (p h ),...,nK (p h )] represents the target p h The corresponding feature fuzzy vector, where H is the number of data sources.

[0092] Furthermore, the enhanced target multi-attribute features include:

[0093] Determine the mapping space R using the load space relationship. M R N , the characteristic matrix A K×H After mapping, mapping vectors M and N are obtained. Based on the obtained mapping features, enhanced features are obtained through multi-attribute interaction.

[0094] Calculate the feature interaction matrix C:

[0095] C = XY T ,

[0096] Where C is the interactive projection metric of the X features of data source 1 and the Y features of data source 2;

[0097] Based on the obtained interaction relationships, feature enhancement is performed on the mapped features to obtain features M and features N:

[0098]

[0099]

[0100] Furthermore, the clustering process for the enhanced features includes:

[0101] Clustering is performed on the enhanced feature vectors M and N, that is, calculating the enhanced feature blur vectors of all targets in the image corresponding to different data sources;

[0102] Targets with the same feature fuzzy vector are grouped into one class.

[0103] The parts of this invention not described in detail are well-known to those skilled in the art.

Claims

1. A method for fusing spaceborne SAR and electronic reconnaissance signal data, characterized in that, include: SAR images are acquired using a spaceborne SAR system, and detection is performed on the SAR images to obtain coarse target detection results; Acquire the results of spaceborne electronic reconnaissance detection of radiation sources; Complete the temporal and spatial alignment of SAR data and electronic reconnaissance data; A multi-attribute feature extraction network is constructed using multiple payload attributes. Relationship modeling is performed on the features of SAR detection data and electronic reconnaissance detection data to obtain feature fuzzy vectors, and usable features are extracted. Based on the payload space relationship, the projection space of the feature matrix of SAR data and electronic reconnaissance data is determined, feature mapping is performed, effective information in the features is extracted, redundant information is removed, and the multi-attribute features of the target are enhanced. The enhanced target multi-attribute features include: Determine the mapping space using the load space relationship. , , the feature matrix The mapped vector is obtained after mapping. , Enhanced features are obtained by interacting with multiple attributes based on the obtained mapping features; The number of fusion attributes; Number of data sources; Calculate the feature interaction matrix : ， in, For data source 1 Features and data source 2 Interactive projection metric of features; Based on the obtained interaction relationships, feature enhancement is performed on the mapped features to obtain the features. and characteristics : ， ； The enhanced features are clustered to eliminate false alarms, and the detection is completed. The target feature information after target fusion is obtained, and the fusion detection of spaceborne SAR and electronic reconnaissance data is completed.

2. The method for fusing spaceborne SAR and electronic reconnaissance signal data according to claim 1, characterized in that, The feature fuzzy vector is a vector composed of the feature values ​​of the same target point in various feature dimensions; Target The corresponding feature number is Then the feature fuzzy vector is ， Several types of data form several feature fuzzy vectors, which are then combined to form a feature matrix. : ， Among them, the feature matrix The element in the h-th row is , , Indicate target The corresponding feature fuzzy vector, This represents the number of data sources.

3. The method for fusing spaceborne SAR and electronic reconnaissance signal data according to claim 2, characterized in that, The clustering process for the enhanced features includes: Strengthening feature vectors , Clustering is performed, which involves calculating the enhanced feature blur vectors of all targets in the image corresponding to different data sources. Targets with the same feature fuzzy vector are grouped into one class.

4. A spaceborne SAR and electronic reconnaissance signal data fusion system, characterized in that, include: The first module is used to acquire SAR images using the spaceborne SAR system, detect SAR images, and obtain coarse target detection results; Acquire the results of spaceborne electronic reconnaissance detection of radiation sources; Complete the temporal and spatial alignment of SAR data and electronic reconnaissance data; The second module is used to construct a multi-attribute feature extraction network using the multi-attribute payload, to model the relationship between the features of SAR detection data and electronic reconnaissance detection data to obtain feature fuzzy vectors, and to extract usable features. The third module is used to determine the projection space of the feature matrix of SAR data and electronic reconnaissance data according to the payload space relationship, perform feature mapping, extract effective information from the features, remove redundant information, and enhance the multi-attribute features of the target. The enhanced target multi-attribute features include: Determine the mapping space using the load space relationship. , , the feature matrix The mapped vector is obtained after mapping. , Enhanced features are obtained by interacting with multiple attributes based on the obtained mapping features; The number of fusion attributes; Number of data sources; Calculate the feature interaction matrix : ， in, For data source 1 Features and data source 2 Interactive projection metric of features; Based on the obtained interaction relationships, feature enhancement is performed on the mapped features to obtain the features. and characteristics : ， ； The fourth module is used to cluster the enhanced features, eliminate false alarms, complete the detection, and obtain the target feature information after target fusion, thus completing the fusion detection of spaceborne SAR and electronic reconnaissance data.

5. A spaceborne SAR and electronic reconnaissance signal data fusion system according to claim 4, characterized in that, The feature fuzzy vector is a vector composed of the feature values ​​of the same target point in various feature dimensions; Target The corresponding feature number is Then the feature fuzzy vector is ， Several types of data form several feature fuzzy vectors, which are then combined to form a feature matrix. : ， Among them, the feature matrix The element in the h-th row is , , Indicate target The corresponding feature fuzzy vector, This represents the number of data sources.

6. The spaceborne SAR and electronic reconnaissance signal data fusion system according to claim 5, characterized in that, The clustering process for the enhanced features includes: Strengthening feature vectors , Clustering is performed, which involves calculating the enhanced feature blur vectors of all targets in the image corresponding to different data sources. Targets with the same feature fuzzy vector are grouped into one class.

Images