Font Size: a A A

Research On The Method Of Ship Target Detection Using Dual-Mode Fusion

Posted on:2018-06-27Degree:MasterType:Thesis
Country:ChinaCandidate:J J HuangFull Text:PDF
GTID:2428330566451608Subject:Pattern Recognition and Intelligent Systems
Abstract/Summary:PDF Full Text Request
In the background of increasing complexity of maritime situation,it is becoming more and more important to strengthen the ability of monitoring and surveillance of the oceans.Reconnaissance aircraft are often equipped with a variety of detection sensors,the research of multi-mode fusion target detection is of great significance to improve the ability of reconnaissance.Synthetic Aperture Radar(SAR),with all day,all-weather,sustained large-scale observation capability,is one of the important means of marine supervision.The imaging of visible light sensor is clear,informative,providing a strong guarantee for the purpose of accurate identification.The combination of SAR image and visible light image,learning from each other,can effectively improve the system target detection capability.On the basis of comprehensively summarizing the predecessors' work,this paper has carried on the thorough research to the SAR,the visible light dual-mode fusion ship target detection method.The main contents of the paper are as follows:(1)In the study of ship target detection method for SAR images,aiming at the problem that the traditional CFAR does not have the multi-scale detection capability,an improved CFAR ship target detection method based on objectness is proposed.Using The position and size of the potential target obtained by the method of objectness,to detect the target in the local window and dynamically adjust the CFAR background window size,makes the algorithm with multi-scale detection capabilities,and improves the efficiency of the algorithm.(2)In the study of ship target detection method for visible light images,a method of ship target detection based on CNN and SVM is proposed to solve the problem of poor classification ability of traditional CNN with single connection layer.This method uses the excellent classification ability of SVM to overcome the problem of the poor classification ability of CNN.Contrastive experiment with the traditional typical method shows that the method has higher accuracy.(3)In the study of SAR and visible light dual-mode fusion ship target detection method,methods of SAR-guided visible light sensor imaging and SAR-guided visible light detection is proposed.Aiming at the registration problem of SAR and visible light image,a method of scaling the target geographic coordinates and pixel coordinates of SAR,visible image is given.Through secondary detection by using visible light image,a large number of false alarms in SAR images are effectively removed,improving the robustness of the system.
Keywords/Search Tags:Ship target detection, Dual mode fusion, Synthetic aperture radar, Objectness, CNN
PDF Full Text Request
Related items