3D shape modeling using a self-developed hand-held 3D laser scanner and an efficient HT-ICP point cloud registration algorithm

Firstly, we develop a cost-efficient hand-held three-dimensional (3D) laser scanner for optical 3D laser scan data acquisition. Then, an automatic registration algorithm is used for 3D laser scanning based 3D shape modeling. Inspired by the use of twist to parameterize rigid motion in workpiece localization, we present the Hong–Tan based ICP (Iterative Closest Point) automatic registration algorithm (named HT-ICP) for partially overlapping point clouds. Using the point clouds from Stanford 3D Scanning Repository, we compare HT-ICP with the original ICP algorithm and its main variants, and experimental results show that the HT-ICP algorithm improves both the speed and accuracy of registration. Then we give the performance analysis with increasing amount of noise, and show the power of the “4PCS”+“HT-ICP” strategy for working directly on the raw noisy data. Furthermore, in the process of complete 3D shape modeling of Venus-head-statue, we demonstrate the effectiveness of the HT-ICP algorithm when aligning the actually acquired noisy point clouds from our self-developed low-precision hand-held scanner.

► HT-ICP automatic registration algorithm for partially overlapped range images is proposed.
► We give the rigorous and detailed mathematical derivation of the HT-ICP algorithm.
► Comparative experiments show HT-ICP algorithm improves both accuracy and speed.
► We present the strategy (4PCS+HT-ICP) for aligning the raw noisy point clouds.
► We develop a cost-efficient hand-held 3D laser scanner for range data acquisition.