Spatio-Temporal Inverse Filtering For Small Underwater Object Detection Based On Time Reversal Processing

Recently, active target detection and classification have attracted a great deal of concentration from all over the world, especially active detection, location and classification of underwater objects in shallow sea. Besides the noise of environment or watercraft, the key factor affecting the performance of small object detection in shallow water is the reverberation caused by rough seabed. So the main task of active sonar is distilling the object echo for target detection and location in the background of strong reverberation. It is an inverse filtering: the process of retrieving the unknown input of a known system by using its observed output. Moreover, the object echo can be resumed and the spatiotemporal inverse filter is achieved with deconvolution, if the acoustic field can be described as a right model. However, the right model which could describe the acoustic field can be gained in the changeful ocean, but the mismatch of forecasted acoustic field and actual acoustic field degrades the performance of the filter because of the uncertainty of ocean environment.The properties including time reversal invariant of wave equation and spatial reciprocity ensure that the wave can focus in spatial and temporal with time reversal. In static media, the signal transmitted from a source is received by the time reversal array. The received signal is time reversed and transmitted. Then the signal received again focuses on the source point. Time Reversal Processing (TRP) is a signal processing method based on the focusing property. It attached more attention in underwater acoustic communication, active detection and so on. TRP is used to restrain reverberation and enhance echo-to-reverberation in active detection. Compared with model-based matched filtering(MBMF), matched field processing(MFP) and optimum spatio-temporal matched filtering based on the model of plane wave or spherical wave, TRP educes ocean itself into signal processing. It can not only keep the optimum spatio-temporal matched filtering, but also avoid the loss of performance resulting from the mismatch in static complex wavefield.This paper is under the circumstance of underwater object detection and bases on the focusing property of time reversal. The conclusion that TRP could achieve spatio-temporal inverse filtering for small underwater object detection is educed based on theory analysis, compute simulation and experiment data. This thesis discusses time reversal focusing with suspended and sank objects and the focusing property at first. Then the relation between TRP and spatio-temporal inverse filter is devoted by analyzing the singular value distribution of the time reversal operator. Moreover, an adaptive time reversal algorithm which could achieve such an inverse filter is devoted. Finally, experiments made in the laboratory waveguide testifythat the echo-to-reverberation can be enhanced with 2-4 dB using TRP when the underwater object is located.The main innovations of the thesis are as follows:(1) Experiments in the laboratory waveguide present that TRP could achieve spatio-temporal inverse filter. The feasibility that TRP corresponds to an inverse filter of the propagation is studied. The conclusion that TRP is an easy and robust way to achieve spatio-temporal inverse filtering is educed based on theory analysis, compute simulation and experiment data.(2) Iterative time reversal can enhance echo-to-reverberation for detecting small underwater objects. The research that the TRP is used to detect small underwater object in active sonar is at a primary moment. This method corrects received signal adaptively to improve the result. The conclusion that iterative time reversal can enhance echo-to-reverberation for detecting small underwater objects is testified based on experiment data.