点云
计算机科学
RGB颜色模型
人工智能
迭代最近点
分割
修剪
Orb(光学)
计算机视觉
点(几何)
算法
图像(数学)
数学
几何学
农学
生物
作者
Truong Thi Huong Giang,Young-Jae Ryoo,Dae-Young Im
标识
DOI:10.1109/scisisis55246.2022.10001991
摘要
One or two RGB-D images cannot provide enough information to detect cut-off points in pruning systems. 3D semantic point clouds constructed from many RGB-D images represent real tomato plants and help the system find the cut-off points correctly. We proposed a method to create 3D semantic point clouds based on ORB-SLAM3, ICP (iterative closet point) algorithm, and semantic segmentation neural network. RGB-D images are converted to semantic images by the semantic segmentation neural network. Each pair of camera poses which is estimated by ORB-SLAM3, and an RGB-D semantic image is used to create a 3D point cloud. The ICP method is applied to stick and refine these point clouds to construct a full 3D semantic point cloud.
科研通智能强力驱动
Strongly Powered by AbleSci AI