A 3D structural point cloud registration method for individual plants incorporating prior information
To address the issues of high computational cost and insufficient robustness in sampling and error calculation of the sample consensus initial alignment(SAC-IA)point cloud registration algorithm,this paper proposes a novel coarse registration method for 3D structural point cloud of individual plants by incorporating prior information.The proposed method adopts a"voxel downsampling+passthrough filtering+statistical outlier removal"point cloud preprocessing strategy to preserve the key structural features of plants while effectively reducing the point cloud data volume.By introducing initial pose constraints based on prior knowledge,the SAC-IA algorithm is improved to reduce the number of iterations for unreasonable transformation matrices,thereby improving the registration efficiency and lowering the risk of mismatches.The experimental results demonstrate that the proposed method achieves a registration root mean square error of 6.242 and a processing time of 0.902 seconds,which saves 1.407 seconds compared to the SAC-IA method,significantly enhancing the computational efficiency and robustness of point cloud registration for three-dimensional structures of individual plants.
NETL
NSTL
万方数据
