Abstract
As a powerful global registration method for point clouds, the 4-points congruent sets (4PCS) algorithm has been wildly used in the 3D scene reconstruction field. In this paper, we propose an adaptive 4PCS (A4PCS), which aims to provide a robustness rigid transformation for two or more overlapping laser scans. The proposed method only incorporates the distance information of the stereoscopic base set and a fast mechanism for congruent base extraction into 4PCS. To ensure the registration accuracy when dealing with restrictive situations, such as point clouds with small overlaps or scenes with symmetrical structures, a non-coplanar 4-points base set is adopted without extra time consumption. Besides, the adaptive set fine-tuning is introduced to the point pair searching process to accelerate the convergence of the algorithm. In addition, we replace the binary cost function of the original 4PCS with a modified estimator to strengthen the robustness of the proposed method. Experiments on thirteen pairs of point clouds for 3D indoor scenes, including ten regular size models and three scenes of one large-scale model, can demonstrate the accuracy and efficiency of the proposed method.
Highlights
With the advancing of laser scanner based applications, 3D registration has attracted great interest in the computer vision [1]–[3] and remote sensing [18], [19] related research fields
We experimentally evaluate the adaptive 4PCS (A4PCS) with respect to registration accuracy and computational efficiency
In this paper, we propose a novel fast automated global registration method based on 4-points congruent sets (4PCS) algorithm
Summary
With the advancing of laser scanner based applications, 3D registration has attracted great interest in the computer vision [1]–[3] and remote sensing [18], [19] related research fields. As an indispensable component of 3D reconstruction, the main task of the 3D registration is to match multiple point clouds scanned from different orientations of the same scene to a uniform coordinate system [8]–[10]. In this way, 3D registration can provide a basic architecture for 3D reconstruction to restore the entire scene. To align two scans captured in arbitrary positions with partial overlaps, the first thing to do with 3D registration is to estimate the correspondences between the two scans [5], [6], [16].
Sign-in/Register to view highlights & summary Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
