The Design And Implementation Of Feature-level Information Fusion Algorithm For Multi-mode Compound Guidance Technique

With the development of multi-sensor information fusion technology,multi-mode homing guidance technique has become a trend of precision guiding weapons.In the multi-mode compound guidance system,the the merits of different detectors are combined in the detection and tracking process.Thus the target idtentification performance are improved while the anti-jamming performance.In this thesis,algorithm design and implementation of the feature fusion module in the whole scheme is based on the multi-mode compound guidance system,which is composed of radar,infrared and laser guidance techniques.And the algorithm implementation is base on the existing hardware plat form.First,the information fusion scheme is presented for the multi-mode compound guidance system in this thesis,which is composed of radar,infrared and laser guidance techniques.Then the algorithm design and implementation of the feature fusion module in the whole scheme is completed.The proposed feature fusion algorithm is applied to the fusion of information acquired by the radar and infrared detectors.And then the fused features are ultized for target identification to determine the target identity.The proposed feature-level information fusion algorithm mainly cosists of three procedures,i.e.feature extraction,feature fusion and target identification.In the feature extraction stage,the radar and infrared features are extracted from the front radar and infrared detector information.Then the separability analysis of the extracted features are conducted to select features with high separability.Feature fusion adopt the feature fusion algorithm of OWA to fuse the features extracted from both radar and infraded detectors.Then the identification result can be obtained by the fused features.The target identification adopt the algorithm of SVDD and the sort algorithm of KNN to determine the target identity.After the algorithm design is completed,the whole algorithm is implemented on the existing hardware platform.The algorithm is developed according to the characteristics of hardware platform with TMS320C6678 as the core.Finally,the real-time and recognition performance of the feature fusion algorithm running in the hardware platform are tested and verified.Based on the principle of information fusion,according to the characteristic of missile-borne platform,the feature-level information fusion algorithm is designed and implemented.With the combination of advantages of radar and infrared heterogeneous sensors,the target identification performance is improved.