Research On Fast Target Detection Technology Of Navigation Radar Based On Feature Fusion

In order to build a maritime power,research in the field of ship navigation has attracted much attention.Among them,target detection is a key factor in this field.Moreover,target detection is widely used in various fields.In the military field,fast and accurate target detection can increase the commander’s time for analysis and judgment,providing effective technical support for effective military strikes.the efficiency of target detection is related to whether the ship can avoid the obstacle urgently during the navigation process.The traditional target detection is mainly based on the constant false alarm rate target detection,but in the X-band navigation radar target detection,point-by-point detection is required to distinguish the target,so it is inevitable that when the amount of data is huge,the detection is inefficiency.On the basis of the national defense scientific research project "Shipborne X-band Navigational Radar Inversion Wave Technology",this paper proposes a fast detection technology of navigation radar targets based on feature fusion to overcome the defects in detection time and efficiency of traditional technology.In this paper,according to the characteristics of X-band navigation radar echo,the radial radar echo data is selected in the first place,then preprocessed the data,selecting the 13 features used in the experiment.The detection performance of the Support Vector Machine(SVM,Support Vector Machine)algorithm and the Radial Basis Function(RBF,Radial Basis Function)neural network algorithm in the rapid target detection is compared,and the RBF neural network is finally selected as the basic algorithm for training the classifier in the experiment.Then,aiming at the problem of low detection accuracy of the feature RBF classifier,this paper discusses the number of features for training the classifier,and designs a fast target detection algorithm based on multi-features.By selecting a few features from the selected thirteen features and do different combination experiments.After a lot of experiments,the multi-feature target detection was achieved,the feature with the best detection performance in the multi-feature target fast detection is found,which is the optimal subset of multiple features.By discussing The performance of the classifier trained by the optimal subset,analyzing the feature combination of the feature subset of the multi-feature optimal subset,the applicability of the optimal subset RBF classifier to the target detection problem is confirmed.Finally,Finally,in the fast detection of navigation radar targets,although the optimal subset of multi-features can greatly improve the detection accuracy of the classifier,there still exists a problem of high-dimensional calculation,so a navigation radar target based on feature fusion is proposed,which combines high-dimensional features with low-dimensional features,and then trains a classifier.After analyzing the problem of retaining the number of features after fusion,a large number of experiments have proved that retaining three features after feature fusion is optimal.The fusion of the optimal subsets of multiple features is performed,after the multiple optimal subsets are fused into three features,part of the information of the original features will be lost due to feature fusion,resulting in a decrease in the detection accuracy of the classifier.Fortunately,this decrease is acceptable.The detection accuracy of the classifier is higher than the other three-feature optimal classifier.After experimenting the effect of fusion of all multi-feature combinations,the detection accuracy of some classifiers is partially reduced,the feature fusion can reduce the over-fitting problem of the classifier and increase the detection accuracy of the classifier when there are too many features.