Recently, the terrestrial laser scanner (TLS) is actively used to acquire point cloud data of historical sites. Then, visualization of the acquired data is becoming an important research target. In our previous work, we reported the stochastic point-based rendering (SPBR), which realizes quick and high-quality visualization of laser-scanned point cloud data. It is a novel rendering solution with transparent rendering effect and quick rendering speed. As an extension of this previous research of ours, this paper proposes a novel method for segmentation of laser-scanned data to realize visualization of multi-level transparency. The proposed data segmentation method utilizes the general features of human-made structures, and is based on logical operations of point cloud sets.