Research On Terahertz Imaging 3D Point Cloud Reconstruction Method Based On Edge Detection

The characteristics of terahertz wave determine its important value in the field of nondestructive testing.In order to achieve high efficiency terahertz 3D tomography with large field of view,fast adaptive and large depth of field,a high precision imaging motion control system needs to be researched.In order to realize3 D visualization of detection results,it is necessary to study 3D point cloud reconstruction method of terahertz imaging and develop multi-functional integrated software.In order to better extract the scattered contour point cloud from the ordered point cloud,the edge detection algorithm needs to be improved.In order to select the most suitable reconstruction method for terahertz point cloud,it is necessary to compare various point cloud reconstruction algorithms and analyze their advantages and disadvantages.The main research contents are as follows:(1)A reconstruction method of terahertz 3D point cloud based on edge detection is proposed.A feasible reconstruction process of terahertz imaging point cloud based on edge detection is summarized from terahertz point cloud acquisition and characteristics analysis of its data structure to efficient surface reconstruction.Firstly,the edge of the target contour is obtained by slice edge detection of the ordered point cloud,and then it is transformed into the spatial scattered point cloud for denoising and simplification pretreatment.Finally,the THREE-DIMENSIONAL reconstruction of the terahertz point cloud is completed by triangulation and Poisson reconstruction algorithm.(2)To meet the segmentation requirements of terahertz 3D tomographic images,an improved adaptive threshold Canny edge detection algorithm based on 3D block matching was proposed according to the disadvantages of the traditional Canny operator,which would blur the edges and require manual setting of high and low thresholds.The BM3 D algorithm is combined with guided filtering to reduce the loss of image edge information,and the maximum interclass variance method(OTSU)is used to determine the high and low thresholds of the algorithm adaptively for the gradient graph block matching results.(3)A terahertz 3D tomography software prototype system was designed and developed.In terms of hardware,the synchronization control of each module of the imaging system was studied.A high-speed motion control circuit was designed to achieve accurate synchronization control of terahertz devices in 3D space.The software realizes the precise control of the imaging system,real-time display of motion status,point cloud data reading,edge detection,point cloud simplified denoising,point cloud surface reconstruction and 3D model visualization.The reconstruction results of different algorithms are analyzed and compared to select the optimal reconstruction method.