Research On Ultrasonic 3D Objects Reconstruction Method Based On Superquadrics

Aiming at the problem that photoelectric imaging equipment cannot play an effective role in the reconstruction of the target objects in special environment,a new method based on ultrasonic echo information to reconstruct 3D object is proposed and studied in detail from three aspects: the reconstruction of the echo signal,the extraction of the feature points on the surface of the target and the fitting of the surface model parameters.The target object of STL(Stereolithography)file format is created automatically by selecting an appropriate gridding standard,which is based on the Computational Geometry Algorithm Library(CGAL).The angle ranges that are transmitting angle of the transmitting array and deflecting angle of the receiving array are determined by analyzing the acoustic field characteristics of the ultrasonic array,respectively.Acoustic ray tracing method is used to determine the acoustic transmission path of ultrasonic echo,and then the sound pressure of the sound line is calculated with the calibration sound pressure of a single ultrasonic sensor.The sound line density of the ultrasonic receiving array is increased by the interpolating method,and the echo signal of the receiving array is calculated by the superposition of the sound ray.The echo data corresponding to the minimum peak time and the maximum peak time of all received elements is selected as the original data for signal processing.And the original data is converted from the real signal to the complex signal by using Hilbert transform,then spatial smoothing algorithm is used to decoherence the echo signal.HQ(Hannan-Quinnn)criterion is selected to estimate the number of the target surface feature points by analyzing the performance of different information criterion algorithms.After comparing the resolution,time consumption and error of the two algorithms,Unitary-ESPRIT algorithm is selected to estimate the direction angle of the feature points.The results for resolution of elevation and azimuth are about 0.5° and 2°,respectively.The selected echo signal is segmented into time slices,and the direction angle of the feature points on each signal segment is estimated by using the Unitary-ESPRIT algorithm,and the transmission distance corresponding to the center value of the current time segment is calculated by using transit time and sound velocity,and the coordinates of the feature points are calculated according to the triangular geometry relationship.Based on the quardtree method and octree method of the KNN(K-Nearest Neighbors)algorithm,the boundary points are extracted from the coordinates of the feature points,and LM(Levenberg-Marquardt)algorithm is used to fit the parameters of the superquadric model after extracting the boundary point set by using the quadtree method,the model parameter values are substituted into the superquadrics surface function equation to reconstruct the object.The experimental results show that the method is effective in the reconstruction of sphere,cube,cuboid and cylinder,but the reconstruction effects of the pyramid and ellipsoid has much to be improved.