Research Specifications

Point cloud registration has an application in a wide variety of industrial sectors including manufacturing, building, architecture, robotics, and even medical imaging and analysis. The iterative closest point (ICP) is a well-known algorithm that is used for registration purposes. However, ICP is slow and the result is highly dependent on the initial condition. In recent years, several hybrid ICP algorithms have been proposed and investigated to overcome the disadvantages. Using metaheuristic optimization algorithms such as genetic algorithm (GA) and simulated annealing (SA) to hybrid the registration procedure is among the best solutions to address the shortages of the conventional ICP algorithm. To increase the efficiency and precision of the point cloud registration, it is needed to compare the performances of these new hybrid algorithms in depth. In the present study, the performance of two popular classes of hybrid ICP algorithms including GA- ICP and SA-ICP are presented based on the recent research that is available in the literature. The reason for choosing these two optimization approaches is the popularity and relevant simplicity of these two optimization algorithms. The efficiency and precision of proposed algorithms are compared and benefits and shortages are discussed. Finally, the future works that cloud help to propose point cloud registration algorithms with higher efficiency and precision are suggested. The results suggest that the simulated annealing is more efficient compared to other metaheuristic optimization methods if the setting parameters are tuned well.