Improving Transparent Visualization of Large-Scale Laser-Scanned Point Clouds by Using Poisson Disk Sampling

In recent years, due to the development of laser-measurement technology, 3D laser-scanned point clouds for cultural assets are often used as the recording format of digital archiving. The acquired point clouds are large-scale and precisely record complex 3D internal structures of the scanned objects. Visualization quality of such point clouds highly depends on density distributional uniformity, that is, the uniformity of the inter-point distances. The visualization quality can be improved by making point distances uniform. In addition, the visibility further improves by emphasizing edges. In this study, first the uniformity and flexible tuning of the point density based on Poisson disk sampling are examined. The resultant high-quality point clouds are then applied to transparent visualization. Moreover, edge emphasis visualization is realized by combining the principal component analysis calculating feature amount and Poisson disk sampling.