【摘要】：目前基于點云面的三維重建方法中,重建的區(qū)域性選擇存在著兩個問題:重建區(qū)域過大會導致目標物體不明確,效果不佳,運行時間長;重建區(qū)域過小會導致目標物體不完整,信息丟失。針對重建窗口過大時,采用改進的Snake的區(qū)域性重建算法,即通過輪廓提取只對窗口內的目標物進行重建;針對重建窗口過小時,采用基于投影面的點云拼接算法,即通過重建后的點云進行拼接的方法使目標物體恢復完整。以上兩點改進彌補了點云三維重建及拼接時出現的應用局限性和不穩(wěn)定性,減少了重建時間,提高重建有效性,魯棒性。
[Abstract]:In the current 3D reconstruction method based on point cloud surface, there are two problems in the regional selection of reconstruction: the target object is not clear, the effect is not good, and the running time is long; If the reconstructed area is too small, the target object will be incomplete and the information will be lost. When the reconstruction window is too large, the improved regional reconstruction algorithm of Snake is adopted, that is, only the object in the window is reconstructed by contour extraction. The point cloud mosaic algorithm based on projection plane is used to reconstruct the window too small, that is to say, the target object is restored to complete by the point cloud splicing after reconstruction. The above two improvements make up for the limitation and instability of point cloud 3D reconstruction and stitching, reduce the reconstruction time, improve the effectiveness and robustness of the reconstruction.
【作者單位】： 廣西師范大學電子工程學院;廣西師范大學藥用資源化學與藥物分子工程教育部重點實驗室;桂林電子科技大學廣西自動檢測技術與儀器重點實驗室;
【基金】：國家自然科學基金資助項目(21327007) 廣西自然科學基金資助項目(2013GXNSFBA019278)
【分類號】：TP391.41
，

本文編號：2344938