Registration of LiDAR Point Cloud and Panoramic Image From Mobile Mapping System Considering Spatial Expression Differences

Fusing information from LiDAR point clouds and panoramic images in mobile mapping technology maximizes their advantages. However, the registration of heterogeneous data is a challenge that relies on constructing a loss function for mapping between corresponding points for iterative optimization to solve the mapping parameters between the point cloud and the image. However, current methods ignore the spatial expression differences of heterogeneous data, which reduce the accuracy of the registration and weaken the reasonableness of the evaluation results. To address this, this study innovatively constructs a spatial resolution expression model for the respective characteristics of LiDAR point cloud and panoramic image, and proposes a new strategy for registration and evaluation using minimum weighted pixel distance and differential evolution, which improves the registration accuracy and convergence speed. The introduction of a weighting factor in the registration evaluation reduces the influence of spatial expression differences on the evaluation results. In addition, multiple groups of registration experiments with different corresponding points distributed in different regions were conducted, and all of them achieve good results, indicating that the differences in corresponding point distributions have no influence on the proposed method, which proves the robustness of the proposed method. The achievements of this study not only enhance the efficiency and accuracy of multimodal data registration but also provide strong technical support and theoretical reference for the development of multimodal data fusion technology in the field of mobile measurement.