Catálogo de publicaciones - libros

In current 3D graphics architectures, the bus between the triangle server and the rendering engine GPU is clogged with triangle vertices and their many attributes (normal vectors, colors, texture coordinates). We develop a new 3D graphics architecture using data compression to unclog the bus between the triangle server and the rendering engine. The data compression is achieved by replacing the conventional idea of a GPU that renders triangles with a GPU that tessellates surface patches into triangles.

Nowadays computer graphics hardware presents a series of characteristics, such as AGP memory, vertex cache, etc., that can be used for real-time rendering. The aim of this paper is to conduct a comparative study of different techniques that are shown in the OpenGL graphics standard together with hardware features that enable the visualization of the geometry of complex objects to be accelerated. These techniques are applied to the multiresolution modeling, which requires techniques that can be implemented with dynamic geometry.

Graphics processing units (GPUs) in recent years have evolved to become powerful, programmable vector processing units. Furthermore, the maximum processing power of current generation GPUs is roughly four times that of current generation CPUs (central processing units), and that power is doubling approximately every nine months, about twice the rate of Moore’s law. This research examines the GPU’s advantage at performing convolutionbased image processing tasks compared to the CPU. Straight-forward 2D convolutions show up to a 130:1 speedup on the GPU over the CPU, with an average speedup in our tests of 59:1. Over convolutions performed with the highly optimized FFTW routines on the CPU, the GPU showed an average speedup of 18:1 for filter kernel sizes from 3x3 to 29x29.

In this paper, we present an adaptive bilateral filtering algorithm that can be used to remove unavoidable noise from 3D mesh data generated by initial stages. Selecting the parameters for bilateral filters automatically, this algorithm smoothes meshes in the normal field using anisotropic character of local neighborhood triangles. Experimental results demonstrate that the proposed method remove light noise from meshes and reserve fine features of meshes as good as best results of other methods, with the advantage of none user-assisted parameters setting. Visual comparisons display that the method proposed in this paper performs better than other smoothing method for heavy noisy mesh.

This paper presents a new Bayesian approach to the problem of finding correspondences of moving objects in a multiple calibrated camera environment. Moving objects are detected and segmented in multiple cameras using a background learning technique. A Point Based Feature (PBF) of each foreground region is extracted, in our case, the top. This features will be the support to establish the correspondences. A reliable, efficient and fast computable distance, the , is proposed to measure the closeness of sets of points belonging to different views. Finally, matching the features from different cameras originating from the same object is achieved by selecting the most likely PBF in each view under a Bayesian framework. Results are provided showing the effectiveness of the proposed algorithm even in case of severe occlusions or with incorrectly segmented foreground regions.

This paper presents DRS (Distributed Rendering System), a software library for rendering dynamic 3D scenes on low cost Virtual Reality (VR) computers. This library allows the distribution of object descriptions and geometries over several graphics nodes of a PC cluster for synchronized rendering. The objects can be dynamic: in particular DRS can manage deformable objects over time. Data compression is put to profit to optimize transfer of objects geometry over the cluster nodes. This library allows real-time VR rendering on PC clusters for applications where objects are highly deformable, such as solid modeling or simulation. DRS is available as a set of C++ classes to manage the 3D objects as well as their synchronized distribution and rendering. We present experimental results in the form of examples of applications successfully ported to inexpensive distributed Virtual Reality hardware thanks to DRS.

Surfaces subdivision is an important means for geometric computing of surfaces in CAD. This paper proposes a new quadric subdivision for canal surfaces in this paper, , that subdivides canal surfaces as a set of truncated revolute quadric with joint spheres. Experimental results show that the RQ-sphere decomposition is better than existing methods.

Recent developments in volume visualization using standard graphics hardware provide an effective and interactive way to understand and interpret the data. Mainly based on texture mapping, these hardware-accelerated visualization systems often use a cell-projection method based on a tetrahedral decomposition of volumes usually sampled as a regular lattice. On the contrary, the method we address in this paper considers the slicing problem as a restricted solution of the marching cubes algorithm [1,2]. Our solution is thus simple, elegant and fast. The nature of the intersection polygons provides us with the opportunity to retain only 4 of the 15 canonical configurations defined by Lorensen and Cline and to propose a special look-up table.

This paper describes a technique that speeds up both the modelling and the sampling processes for a ramified object. By introducing the notion of substructure, we divide the ramified object into a set of ordered substructures, among which only a part of basic substructures is selected for implicit modelling and point sampling. Other substructures or even the whole object can then be directly instantiated and sampled by simple transformation and replication without resorting to the repetitive modelling and sampling processes.

An environment model that is constructed using a single image has the problem of a blurring effect caused by the fixed resolution, and the stretching effect of the 3D model caused when information that does not exist on the image occurs due to the occlusion. This paper introduces integration method using multiple images to resolve the above problem. This method can express parallex effect and expand the environment model to represent wide range of environment.