Transformer Connection Terminal Registration Based on Objective Function

Abstract In recent years, the traditional ICP algorithm has high requirements for initial position of point clouds, and low registration ability clouds with overlap. The Microsoft Kinect depth sensor was used to obtain cloud data target object from real scene. Then, pre-processing such as segmentation, filtering down sampling. coarse registration, feature sampling consistency make a better position. Finally, point-to-surface optimized by linear least squares proposed in fine registration. experimental results show that root mean square error improved is 0.761mm time 52.32ms. Compared based on SIFT points consistency, accuracy increased 21.0% 43.3%, speed 18.9% 30.2%.