考虑图像采集的退役零件SfM三维重建精度研究

Precision of SfM 3D reconstruction of end-of-life parts considering image acquisition

运动恢复结构(SfM)是一种单目多视图三维重建技术。针对SfM多视图三维重建的速度及质量问题,提出了一种考虑图像采集的退役零件SfM三维重建精度优化方法。首先,构建了图像采集试验平台,以导靴和凸轮轴瓦盖为试验对象,选取图像质量、图像数量、拍摄距离及视角数量4个对重建质量有影响的参数为影响因素,每种因素取3个水平,设计了正交试验;然后,根据试验设计方案,对两个试验对象分别进行了9组图像采集,并用Reality Capture软件对其进行了三维建模;最后,根据重建点云和模型的准确性、重建的完整性与重建效率3个指标,对重建结果进行了评价,并进行了正交试验,通过极差分析获得了最优参数组合。研究结果表明:图像数量与视角数量对三维重建质量影响显著,拍摄距离与图像质量对三维重建质量影响不显著。该研究结果可为提高图像三维重建速度及质量提供参考。

Structure from motion(SfM)is a monocular multi-view technique for 3D reconstruction.Aiming at the problem of address speed and quality issues in SfM multi-view 3D reconstruction,an optimized method for improving the accuracy of SfM 3D reconstruction was proposed,taking the image capture of end-of-life components into consideration.Firstly,an experimental platform for image acquisition was established,guide shoes and camshaft cover were selected as the objects of study.Four key parameters were identified for their influence on the quality of reconstruction:image quality,number of images,shooting distance,and the number of viewpoints.Each of these factors was evaluated at three different levels,and an orthogonal experiment was designed.Then,in accordance with the experimental design,nine sets of images were collected for each of the two test subjects.These images were then processed using Reality Capture software to perform 3D modeling.Finally,the reconstruction results were evaluated based on three key metrics:the accuracy of the reconstructed point cloud and the model,the completeness of the reconstruction,and the efficiency of the reconstruction process.Orthogonal experiments were conducted,and the optimal parameter combination was obtained through range analysis.The research findings indicate that the number of images and the number of viewpoints significantly influence the quality of the 3D reconstruction.In contrast,the shooting distance and the quality of the images were found to have a less significant impact on the quality of the 3D reconstruction.These findings offer a reference for enhancing the speed and quality of image-based 3D reconstruction.