Point Cloud Splicing Based on 3D-Harris Operator

The paper proposes an algorithm of point cloud splicing based on 3D-Harris operator. First, it extracts 3D interest points efficiently according to the combination of 3D-Harris operator and normal vector. Then it calculates the fast point feature histogram (FPFH) of all 3D interest points, as a result of which the initial transformation matrix is calculated through FPFH. Last, ICP algorithm is used to realize the accurate registration. At the last part, experiment shows that the method has high efficiency in point cloud splicing and performs well in scenes with more corners.