Global Modeling and Local Matching: A Dynamic Fusion Approach

In digital twin technology, precise point cloud alignment is crucial for effective LiDAR data modeling. This study presents a fusion algorithm that tackles issues of low accuracy and slow processing in large-scale point cloud management. The algorithm utilizes the SAC-IA for initial alignment and the enhanced ICP for further refinement, incorporating three-dimensional shape context (3DSC) to enhance its efficacy. The approach initiates by downsampling data to enhance processing efficiency, employs 3DSC to match local features, uses SAC-IA for initial coarse alignment, and applies ICP to achieve precise alignment within the overall global modeling framework. Experimental results demonstrate significant improvements in alignment speed and accuracy, offering a new strategy for efficiently handling complex point cloud data, enhancing both computational efficiency and alignment robustness for advanced digital twin applications.