Perceptually guided level of detail techniques
Gary W. Meyer, Lijun Qu · 2008
Finding a compact polygonal representation of a geometric object is an important problem for computer graphics. Until recently, most of the effort has been focused on finding a simplified polygonal mesh that is geometrically similar to the original mesh. In this thesis, we develop algorithms with an emphasis on finding a compact representation that is visually similar to the original geometric object. First, we propose the concept of a faceting signal, which characterizes the faceted appearance of a polygonal model under different viewing distances, viewing directions, lighting conditions, etc. We further analyze its properties in the frequency domain. A perceptual metric is proposed that indicates the visibility of the faceting signal. The importance of this analysis is illustrated by a perceptually guided surface texturing algorithm and a geometric level of detail management system. Second, we propose a perceptually guided remeshing algorithm that automatically distributes samples over a polygon mesh by taking the visual perceptual properties of the appearance data into account during the remeshing process. Due to the properties of the human visual system, especially visual masking, the artifacts in the final rendered mesh are invisible to the human observer. The approach also improves the quality of the images in a budget based system since the distribution of polygons across all of the objects in a scene is guided by the principles of visual perception. Third, we develop a mesh simplification algorithm that is guided by the masking potential of its surface textures. A masking importance value is determined for each vertex of the original mesh, and this value is used to weight the geometric error in a geometry based simplification algorithm. Our mesh simplification algorithm takes into account both the geometry of the object and the masking potential of its surface textures. We also present a rendering algorithm for textured point based models that takes into account the masking properties of the human visual system. In our system, high quality textures are mapped onto point-based models. Given a texture, its importance map is first computed by an algorithm inspired by the visual masking tool included in the JPEG2000 standard. This importance map indicates the masking potential of the texture. During runtime, point-based models are simplified and rendered according to this computed importance. Finally, we propose to use a visual discrimination metric to determine the resolution requirement of a texture for rendering. A visual discrimination metric is first used to compute a curve that represents the visual distortion-vs-resolution of a texture. Given a visual distortion threshold, the resolution of the texture can be determined based on the distortion-vs-resolution curve.