Papers on Pointclouds

Rendering Techniques

  • Point-Based Rendering of Forest LiDAR -- Lance Simons, Stewart He et al

    • Summary: Simons et al introduces techniques for rendering forest pointcloud data while preserving the complex, 3D structure of forest canopy. Using multi-pass rendering, shadows, approximate occlusion, and silhouettes are produced and rendering alongside pointcloud data, producing semi-realistic imagery that enhances the geometry of dense forests. The benefits of these rendering techniques are evaluated via two applications: tree segmentation and forest structure clustering.

    • Data: The data utilized in the paper is not public. Nevertheless, each pointcloud tested contains between 1.7 and 2.9 million points, representing a subsection of a larger forest.

    • Performance: 20 FPS on NVIDIA GTX 480

    • Conclusion: This paper introduces a collection of rendering techniques which improve the contrast and visual fidelity of forest pointcloud data. Hence, they are particularly useful when visualizing forest canopies or evaluating tree segmentation performance. While novel, these multi-pass rendering techniques are not computationally intensive, so they can be implemented within VR, YURTs, or CAVEs with minimal performance hits.

    • Citations: 7

    • Journal/Conference/Workshop: Workshop on Visualisation in Environmental Sciences (EnvirVis) 2014

  • Visualization of LIDAR datasets using point-based rendering technique -- Bostjan Kovac, Borut Zalik

    • Summary: Kovac et al introduces data structures, out-of-core data management techniques, and a point-based rendering algorithm to efficiently render very large pointclouds. While more of a review, this paper evaluated how out-of-core techniques can be used to visualize hundreds of millions of points with a set amount of GPU memory.

    • Data: The data utilized in the paper is not public and varies between 5 million and 700 million points.

    • Performance: 57.2 FPS on ATI Radeon HD4850 512 MB with 5.2 million points.

    • Conclusion: This paper introduces necessary data structures and rendering techniques for visualizing very large pointcloud datasets. It is, however, more of a survey than a technical paper and introduces only a few novel techniques.

    • Citations: 30

    • Journal/Conference/Workshop: Computers & Geosciences 2010

  • Scalable visualization of massive point clouds -- Gerwin de Haan

  • Multiresolution foliage for forest rendering -- Qingqiong Deng, Xiaopeng Zhang, et al

  • Immersive Visualization and Analysis of LiDAR Data -- Oliver Kreylos, Gerald W. Bawden, et al

  • Point-based rendering of trees -- Guillaume Gilet, Alexandre Meyer, et al

  • Rendering Large Point Clouds in Web Browsers -- Markus Schutz