Algorithmic Investigation Of Underwater Small Targets Active Detection And Tracking

With the development of marine industry,the application of sonar system is widely extended.Sonar system is used in detection,positioning,tracking,recognition of underwater targets,communication and navigation in early years.But now submarine topographic mapping,maritime search and rescue,marine geological survey and marine fishery are increasingly depending on it.In the military,detecting and tracing the targets with strong threat,such as submarines,torpedoes and divers,is the focus issue to all countries all the time.It is of great significant meaning and practical value to improve the performance of detecting algorithms and provide more reliable detection technology for practical engineering practice.When detecting the signal using sonar,radar and communication equipment,the matched filter(MF)is a classical detector which has been favored by scholars at home and abroad.It is the best liner filter with the maximum output signal to noise ratio under the condition that the background interference is the band-limited white noise,which makes it a foundation for the application of matched filter in signal detection.This paper applies the detecting algorithm called the frequency domain adaptive matched filter(FDAMF)to detect the small targets underwater.The algorithm combines the adaptive line enhancer to express the noise in frequency domain and improve the output signal to noise ratio at the same time,which is based on the classical matched filter.Then in the consideration of the adaptive algorithms used in adaptive line enhancer,this paper compares and analysis the performance of the several adaptive filtering algorithms.Choosing a better adaptive filtering algorithm can make it works well in a lower signal to noise ratio.This paper theory derivation firstly,which is the base of detecting liner modulated frequency signal using FDAMF.And then simulating and analysis the detection performance between MF and FDAMF.Finally,it is verified that the FDAMF also perform well when processing the experimental data.Aiming to obtain a much clearer trajectory picture,meanwhile two post-processing algorithms were proposed.However,it cannot meet the requirements to detect singly whether there is a target or not.The more important thing is to trance and distinguish the target exactly in real time.Basing on the signal detecting results above,this paper carried out a further study on target tracking.Some classical tracking algorithms have been used because of their inherent advantages.Mean-shift algorithm and Kalman filter are the typical representatives.Firstly,the mean-shift algorithm is applied to track the target,and the target trajectory is obtained by processing the experimental data.Then analysis the performance of tracking the target.Meanwhile,Kalman filter is introduced to further improve the trace efficiency.Because it can adjust the predict deviation of mean-shift.Combining the mean-shift algorithm with the Kalman filter to solve the problems with the influence caused by strong interference and partial or complete occlusion.