Research on Point Cloud Registration Method Based on Improved Harmony Search Algorithm

The ship hull contour surface is primarily measured and inspected by a three-dimensional laser scanner during the process of ship hull subsection loading. However, due to the characteristics of large volume, high precision requirements, and complex appendage surfaces, achieving multi-station data splicing calculation using traditional point cloud registration methods becomes challenging. To address this issue, a point cloud registration method based on an improved harmony search algorithm is proposed for ship hull contour scanning. By representing the point cloud transformation matrix as a group of harmony instruments, an iterative optimization harmony library is established to optimize the best point cloud registration parameters through combination and iterative optimization. Furthermore, in order to overcome the low efficiency and difficulty in calling out local optima associated with harmony algorithms, genetic algorithm concepts such as crossover and mutation are introduced to enhance the search ability and diversity of the harmony search algorithm's library and combination mode. Finally, through example calculations, we compare and analyze the improved harmony search algorithm with both traditional harmony search algorithms and classic ICP (Iterative Closest Point) point cloud registration algorithms. Experimental results demonstrate that the improved harmony search algorithm significantly enhances accuracy in point cloud registration with an average error reduction rate down to 4% compared to traditional harmony algorithms. Additionally, it exhibits superior efficiency with convergence speed shortened to one-sixth that of traditional harmony algorithms.