坐标系
摄影测量学
点云
灰度
计算机视觉
直线(几何图形)
人工智能
立体视觉
计算机科学
三维重建
奇异值分解
图像(数学)
数学
几何学
作者
Yuan Chen,Wensong Jiang,Zai Luo,Li Yang
出处
期刊:Measurement
[Elsevier BV]
日期:2024-02-05
卷期号:227: 114238-114238
被引量:9
标识
DOI:10.1016/j.measurement.2024.114238
摘要
The accuracy of 3D reconstruction will be reduced due to sensor occlusion or tracking failure when it comes to complex structural measurement. To overcome this problem, a novel 3D reconstruction (NR) method is proposed based on a binocular-line laser and a stereo target. First, the coordinate system of a stereo target (TCS) is calibrated by the global photogrammetry method. Second, the marks on the stereo target are identified by using an image grayscale processor to locate its coordinates in the camera coordinate system (CCS). Third, TCS and CCS are matched and localized through distance iteration and singular value decomposition (SVD). Fourth, the point cloud is stitched by coordinate transformation between the CCS and global coordinate system (GCS). The experimental results show that the MAE of the suggested NR method is 0.051mm, and the maximum measurement SD is 0.172 mm. The NR method is superior to previous planar target location measurements.
科研通智能强力驱动
Strongly Powered by AbleSci AI