The Research On Radar Target Detection And Tracking Method

Marine radar is a necessary means of maritime navigation,and plays an important role in the navigation collision avoidance and ship location.The detection and tracking of radar target are the two basic functions of the radar system.In recent years,due to the rapid development of the radar signal detection theory and practice,have create favorable conditions for targets tracking,and at the same time also can remove the false targets by tracking the target matching.With the rapid development of computer technology and digital image processing technology,according to the characteristics of the radar image,we can design and select the appropriate tracking and detection method to segment the target of the radar image by means of digital image processing,and extract the target movement information.It is a stable and efficient way to realize the radar target detection and tracking.This paper first analyzes the characteristics of the marine radar images,select the appropriate image de-noising and enhancement method for these features of radar images.In this paper,we use the method based on Bayesian threshold de-noising to remove the radar image noise,and enhance the radar image after de-noising,to prevent missing target in tracking link.The experimental results show that when the threshold adjustment factor a = 0.2,the de-noising effect two or three wavelet transform of the best systems can better suppress image noise and Gaussian noise.When the threshold value of 200,the individual weak signal has been enhanced to the degree of blurring of the image is weak,reinforcing effect is relatively good.In radar target detection,using the multi-resolution characteristic of wavelet transform and the localization characteristic of wavelet analysis,we use the edge detection method base on B-spline wavelet multi-scale transform to segment the radar image.Then according to the characteristics of marine radar images,research and extract the features of target's location,area and moment invariant to complete the target matching in the image sequence,ensuring the reliability and accuracy of target tracking.Experimental results show that the target level of the outline of the radar images decomposed relatively clear image edge detection,edge detection effect is obvious.And target marking to establish of the target sequence.In radar target tracking,in view of the eclipse phenomenon of radar target in the image this paper researches tracking algorithm based on feature matching and Kalman filter tracking algorithm combined.The method uses the target status information that have matched to the correction prediction value of the Kalman filter,and constantly predict target statusinformation in the next frame radar image,also can use the prediction value to compensate for this location information of the target is lost in a frame image.So that we can use the algorithm to establish a complete chain of the target in the image sequence,and achieve the accurate tracking of target.Experiments show that the method predicted target position and the actual target position is not more than three pixels,to achieve the desired effect.