Point Cloud Registration with Spatial and Color Data

Overview

This project advances point cloud registration by combining spatial and color information, enhancing alignment accuracy, especially in color-rich datasets. It includes RGB-D data acquisition from cameras, notebooks for same-source registration, and experiments with SIFT for point cloud registration. The project utilizes the RGB-D Object Dataset and data from Intel Realsense cameras, showcasing methodologies like SIFT and comparing results with the FPFH method for precise and robust point cloud alignment.

RGB-D Data acquisition from camera

• RS_data_acquisition.ipynb

Point cloud registration

• Same-source registration.ipynb
• Experiment on Point cloud registration using SIFT
  • Experiment on dataset: Arisa_PCRegist_SIFT-dataloop.ipynb
  • Compare transformation result with fpfh method: Arisa_PCRegist_SIFT-checkloopresults.ipynb

Dataset

• RGB-D Object Dataset : https://rgbd-dataset.cs.washington.edu/dataset/rgbd-dataset/
• RGB-D Data from Intel Realsense depth camera L515 and D435i