Research On Technologies Of Real-time Three-dimensional Acoustic Imaging Based On Sparse Array

Underwater real-time three-dimensional(3-D)acoustic imaging technology has become increasingly prominent in the fields of seabed geomorphology mapping,submarine construction,underwater vehicle obstacle avoidance,underwater robots,oceanic engineering and military applications,and it has broad application prospects.Currently,the cross array and planar array are two common array formats in practical underwater acoustic imaging systems.The cross array has the advantage of a small number of array elements.In underwater real-time imaging applications,it is generally used for short-range detection.However,the latest real-time beamforming algorithm and sparse optimization design based on cross array only consider the far-field conditions,and could hardly guarantee the beam pattern performance at any distance in near field.On the other hand,the 3-D acoustic imaging system based on planar array has the advantage of high frame rate and the ability of real-time imaging for long-distance detection.The fast beamforming algorithm and sparse array are generally used to reduce the computational load and hardware complexity.However,the fast beamforming algorithms are inefficient for sparse arrays currently,and the sparsity rate is still insufficient.Therefore,based on the cross array and planar array,the real-time 3-D acoustic imaging technology and sparse array have been deeply researched in this thesis,and the research results are applied to real underwater system.1.To improve the near-field beam pattern performance of the cross array in real-time applications,a real-time beamforming algorithm based on cross array in both far-and near-field is proposed.A multiple-frequency transmitting(MFT)beamforming is adopted to ensure real-time imaging.At the same time,the whole 3-D scene is divided into multiple focal regions according to the depth of field,and the beam signals are transmitted to each focal region sequentially to improve the near-field beam pattern performance.To process the near-field wide-band signals efficiently,the receiving beamforming extended the traditional chirp zeta transform(CZT)to near-field wide-band beamforming with adoption of pruning fast Fourier transform(FFT).The algorithm improves the computational efficiency and guarantees the beam pattern performance of cross array in both far and near field.2.To suppress the side lobes of the beam pattern at unfocused distance and further reduce the number of array elements,a novel method of non-grid sparse array for whole field 3-D acoustic imaging is proposed.On the basis of simulated annealing algorithm,a new energy function is presented based on the analysis of far-and near-field beam pattern of the cross array.The new energy function contains the information of each distance,and contributes to suppressing the side lobes of beam pattern in whole field.A second optimization is then proposed to reduce the required array elements when working in far field.Afterwards,to further reduce the number of array elements,a small position perturbation is introduced in simulated annealing,which breaks the constraints of element position on grid and improves the degree of freedom in the sparse optimization process.Compared with other methods,the proposed algorithm obtains a sparse array that has the fewer elements and better beam pattern performance in whole field.3.To improve the computational efficiency of the fast beamforming algorithms for sparse arrays and the sparsity rate of the planar array,a pruning distributed parallel subarray(P-DPS)beamforming algorithm and a sparse array design method are proposed respectively.The algorithm transforms the traditional DPS beamforming into a series of one-dimensional FFT operations.Based on the analysis of the input and output characteristics of beamforming for fully array and sparse array,the pruning FFT technology is adopted to effectively remove the unnecessary operations.Then,to further reduce the complexity of the planar array system,a sparse planar array that is suitable for P-DPS algorithm is synthesized.Compared with other methods,the proposed algorithm obtains the similar beam pattern with the lower computational load and the fewer elements of the sparse array.The above research results have been carried out in practical 3-D acoustic imaging system and the underwater experiments verifies the effectiveness of the algorithm.