Research on Adaptive Point Cloud Mapping and Filtering Method Based on Particle Swarm Algorithm

Aiming at the issues of traditional sparse point cloud map lacking feature details and the poor environmental adaptability of fixed filtering parameters, this paper proposes an adaptive dense point cloud mapping and filtering method based on the particle swarm optimization algorithm. Firstly, the keyframe selection strategy is optimized. By comprehensively considering factors such as time, distance and co-visible feature points, the quality of keyframes is ensured while reducing data redundancy. Then, the dense point cloud mapping thread is established to construct an accurate environmental data basis for subsequent applications. Subsequently, the particle swarm algorithm is introduced into the point cloud filtering process. Through adaptive and dynamic adjustment of filtering parameters, it breaks through the limitations of fixed parameter filtering and generates a point cloud map with a concise configuration and rich details. Finally, the performance of the proposed algorithm is verified using the TUM dataset and real environments. Experimental results show that our algorithm can construct lightweight and reliable dense point cloud map under limited computing resources, providing safe and efficient navigation support for unmanned platforms.