An Improved ICP Algorithm for Point Cloud Registration

Speed and accuracy of point cloud registration are priorities in research. After finding points by close point criterion, we use the nearest criterion to remove point pairs which contain same points. This paper proposes an improved iteration close point algorithm that uses matching points´ center of gravity as reference points and point pair distance constraint to reject false point pairs for point cloud registration. We compared this improved algorithm with another method which uses point cloud center of gravity as reference point pair and iteration close point algorithm. Experimental results show that the registration rate, accuracy and convergence rate of the algorithm which we proposed are all improved.