LiDAR point cloud quality optimization method based on BIM and affine transformation

Laser Detection and Ranging (LiDAR) systems possess the capability to generate high-resolution three-dimensional (3D) data of indoor environments. The inherent uncertainties pertaining to relative spatial positioning and the centimeter-level precision of LiDAR ranging, however, contribute to discernible constraints within contexts requiring elevated degrees of precision, particularly in the domain of high-precision sensing applications. In response to this concern, this paper introduces an approach designed to mitigate and appraise the uncertainty associated with plane positioning through the utilization of point cloud fitting methodologies, concurrently integrating principles of building information modeling (BIM) and anisotropic affine transformations (AAT). Primarily, the methodology involves the extraction of precise plane characteristics employing the tenets of robustly weighted total least squares theory within the context of point cloud fitting. Subsequently, the method synergistically incorporates geometric information emanating from the Building Information Model alongside the accurately determined plane positioning data derived from LiDAR point clouds via AAT. This integration markedly enhances the precision of the ranging system's datasets. Ultimately, the assessment of ranging uncertainty is conducted by quantifying the deviations of individual points from the conforming plane and employing a probability approximative scheme grounded in higher-order moments. Experimental results demonstrate the method's precision and efficacy, offering a solution to the challenge of achieving higher perception precision in LiDAR-based ranging systems.