Tilo Ochotta

Robust Smooth Feature Extraction from Point Clouds

Defining sharp features in a given 3D model facilitates a better understanding of the surface and aids visualizations, reverse engineering, filtering, simplification, non-photo realism, reconstruction and other geometric processing applications. We present a robust method that identifies sharp features in a point cloud by returning a set of smooth curves aligned along the edges. Our feature extraction is a multi-step refinement method that leverages the concept of robust moving least squares to locally fit surfaces to potential features. Using Newtons method, we project points to the intersections of multiple surfaces then grow polylines through the projected cloud. After resolving gaps, connecting corners, and relaxing the results, the algorithm returns a set of complete and smooth curves that define the features. We demonstrate the benefits of our method with two applications: surface meshing and point-based geometry compression.

Compression of point-based 3D models by shape-adaptive wavelet coding of multi-height fields

In order to efficiently archive and transmit large 3D models, lossy and lossless compression methods are needed. We propose a compression scheme for coordinate data of point-based 3D models of surfaces. A point-based model is processed for compression in a pipeline of three subsequent operations, partitioning, parameterization, and coding. First the point set is partitioned yielding a suitable number of point clusters. Each cluster corresponds to a surface patch, that can be parameterized as a height field and resampled on a regular grid. The domains of the height fields have irregular shapes that are encoded losslessly. The height fields themselves are encoded using a shape-adaptive wavelet coder, producing a progressive bitstream for each patch. A rate-distortion optimization provides for an optimal bit allocation for the individual patch codes. With this algorithm design compact codes are produced that are scalable with respect to rate, quality, and resolution. In our encodings of complex 3D models competitive rate-distortion performances were achieved with excellent reconstruction quality at under 3 bits per point (bpp).

Spline-based feature curves from point-sampled geometry

Defining sharp features in a 3D model facilitates a better understanding of the surface and aids geometric processing and graphics applications, such as reconstruction, filtering, simplification, reverse engineering, visualization, and non-photo realism. We present a robust method that identifies sharp features in a point-based model by returning a set of smooth spline curves aligned along the edges. Our feature extraction leverages the concepts of robust moving least squares to locally project points to potential features. The algorithm processes these points to construct arc-length parameterized spline curves fit using an iterative refinement method, aligning smooth and continuous curves through the feature points. We demonstrate the benefits of our method with three applications: surface segmentation, surface meshing and point-based compression.

Image-Based Surface Compression

We present a generic framework for compression of densely sampled three‐dimensional (3D) surfaces in order to satisfy the increasing demand for storing large amounts of 3D content. We decompose a given surface into patches that are parameterized as elevation maps over planar domains and resampled on regular grids. The resulting shaped images are encoded using a state‐of‐the‐art wavelet image coder. We show that our method is not only applicable to mesh‐ and point‐based geometry, but also outperforms current surface encoders for both primitives.

Hardware Rendering of 3D Geometry with Elevation Maps

We present a generic framework for realtime rendering of 3D surfaces. We use the common elevation map primitive, by which a given surface is decomposed into a set of patches. Each patch is parameterized as an elevation map over a planar domain and resampled on a regular grid. While current hardware accelerated rendering approaches require conversion of this representation back into a triangle mesh or point set, we propose to render the elevation maps directly in a hardware accelerated environment. We use one base data set to render each patch in the common vertex and fragment shader pipeline. We implement mesh- or point-based rendering by using a base mesh or a base point set respectively. This provides the basis for the underlying primitive for the final rendering. We show the benefits of this method for splat rendering by replacing attribute blending through a simplified and fast attribute interpolation. This results in rendering acceleration as well as an improvement in visual quality when compared to previous approaches