Ground-based LiDAR shows the potentiality of highly automatic building detection and reconstruction. Due to the main structure of the building are the walls. When a surface feature is extracted from a point cloud, the discrepancies of points to the surface are resulted from the undulation of the scanned object surface. We focus on the analysis of the characteristics of surface features. We use some methods like least squares fitting (LSF) and principal component analysis (PCA) to discuss the different factors such as point density, distribution, noise. The goal of this paper is to investigate the different scanning characteristics how to influence the extraction of the walls. Experimental results of computations with simulated data are shown for analysis.