点击次数：

论文类型：期刊论文

发表时间：2015-05-01

发表刊物：IEEJ TRANSACTIONS ON ELECTRICAL AND ELECTRONIC ENGINEERING

收录刊物：SCIE、EI、Scopus

卷号：10

期号：3

页面范围：350-352

ISSN号：1931-4973

关键字：pairwise view registration; MLS surface; multidimensional optimization; hill-climbing method

摘要：A self-adaptive 3D pairwise view registration method is presented. Coarse registration of neighboring views is achieved according to their common continuous nonflat areas. These areas are calculated by local least-square (LS) surface fitting. Then, the principle of point-to-surface projection is used to search for the corresponding points, where the local moving least-square (MLS) surface is employed. Finally, multidimensional optimization based on the technique of hill-climbing is performed to refine the coarse registration. Experimental results show that this method gives good noise immunity and requires simple computation process and very few manual interventions. Thus it is suitable for large-scale free-form surfaces lacking obvious geometric characteristics. (c) 2014 Institute of Electrical Engineers of Japan. Published by John Wiley & Sons, Inc.