Color-Guided Coarse Registration Method Based on RGB-D Data

This paper proposes a coarse registration method based on RGB-D data. The feature points are obtained from the mixed feature. The corresponding points of the feature points are searched in target point cloud according to the feature descriptor. The feature points are divided into several partitions and the rigid transformation is calculated between the corresponding point pairs in each partition. The optimal rigid transformation is chosen from the rigid transformation of each partition. The mixed feature is constructed by geometric information and color information of the neighborhood points. The feature descriptor is built with the mixed feature and normalized RGB value. The experimental results demonstrated that the method is effective for RGB-D data.