Colocated MIMO Imaging Sonar Detection And Tracking For Underwater Moving Small Targets

In recent years,the use of divers and other small weapons in attacking and destroying the coastal facilities is growing,and the small target detection becomes to be one of the main tasks of the coastal protection.In a complex shallow sea and nearshore environment,strong noise,strong reverberation and strong clutter make the small target detection sonar easily appear "myopia" or "crying wolf' phenomenon,in other words,the detection probability is not enough,the false alarm probability is high,that is the troublesome problem in small target detection.Small target detection sonar belongs to the imaging sonar filed,involving the high-resolution imaging,reverberation suppression,moving target detection and parameter estimation in low SNR conditions,small moving target tracking and small targets recognition.How to improve the imaging resolution to meet the requirements of alert distance,and how to ensure the detection and tracking performance under the low SNR conditions are the main content of the dissertation.The dissertation contents mainly has four parts:colocated MIMO(Multi-Input-Multi-Output)sonar high resolution imaging,acoustic image sequence moving target detection,split-beam-phase image sequence moving target detection,and moving target detection and tracking.The first part of the dissertation is about the improvement of azimuth resolution and SNR gain using MIMO imaging sonar.The dissertation brings the colocated MIMO technology into high-resolution imaging sonar to obtain virtual aperture,and improve the azimuth resolution and get SNR gain.Firstly,the MIMO sonar signal model was established,and the principle of the colocated MIMO imaging sonar was discussed,and the near-field focusing broadband beamforming imaging algorithms based on colocated MIMO sonar was derived.The simulation analysed the performance of colocated MIMO broadband near-field imaging method based on waveform diversity.The results of the pool experiment based on the system composed by 2 transmitting elements and 18 receiving elements validate the imaging performance and the anti-reverberation capability of the colocated MIMO imaging sonar.Traditional Detec-Before-Track(DBT)method is based on single image detection,and the detection probability is low,and much false clutter is detected.The Track-Before-Detect(TBD)method is suitable for dim targets detection and tracking,and therefore the second and third part the dissertation are about TBD algorithms for underwater moving small targets.In the second part a moving target detection algorithm based on image sequence is proposed.The 3-dimensional range-azimuth-frame time image sequences obtained by colocated MIMO imaging sonar were projected into a two-dimensional range-azimuth image,then an improved CFAR detector was used to detect the target tracks from the accumulated two-dimensional range-azimuth image,that was the first threshold of the detection.Then the Hough transform was used to project the detected result into the parameter space,then the non-coherent accumulation was completed,and the second threshold detection was completed in the parameter space.Through the numerical simulation,we can get:The detection performance of the maximum accumulation mode is better than the sum accumulation mode and the accumulation-after-detection mode.The performance of the improved 2-dimensional CFAR detector is better than the Cell-Average(CA)and Trimmed-Mean(TM)CFAR detector.The performance of the detection based on the maximum accumulation CFAR-Hough transform is better than detection based on single image.Finally,the pool experiments for underwater moving small target detection were executed to validate the performance of the proposed algorithm.In the third part of the dissertation,a moving target detection algorithm based on the split-beam-phase image sequence is proposed.On the basis of the research of the split-beam-phase imaging method,the Standard Deviation Reciprocal(denoted as SDR)of the split-beam-phase was applied to the target detection.The 2-dimensional Range-Frame information of the split-beam-phase SDR Image Sequence(denoted as SDRIS_R-F)was applied to the moving target detection and the interference elimination.The split-beam-phase SDRIS R-F image corresponding to some azimuths were accumulated,and the azimuth corresponding to Maximum per frame and per range cell(denoted as SDRIS_R-F_Max)was applied to finish the detection when the target changes azimuth during movement.The proposed algorithm does not require solving the problem of the phase ambiguity,and the SNR gain of the split-beam-phase SDR imaging method is higher than conventional beamforming method.Finally the pool experiments were executed to validate the feasibility and the performance of the proposed algorithm.In the fourth part of the dissertation,integrating the first,second with the third part content,a solution of underwater small moving target detection and tracking is given.The use of colocated MIMO imaging sonar of the first part can improve the azimuth resolution and SNR gain,in order to facilitate the subsequent detection and tracking.Two TBD algorithms proposed in the second and third part,respectively based on the acoustic image sequences and the split-beam-phase image sequences,can be used to extract the initial track.The detected initial track can guide for the subsequent focus detection area,so that the performance and accuracy of the detection and tracking can be ensured.On the other hand,passing through a limited period,the target short-term track can be extract using the extract method of the initial track.The target short-term track can be compared with the tracks which extracted by the real-time detection and tracking results.The comparison result can help to correct the yawed track and continue the termination track,and improve the reliability and low false alarm rate of the target track.Finally the pool experiments were executed to validate the feasibility of the given method.