Catálogo de publicaciones - libros

To perform 3D animation with carving on surfaces of 3D models is a new branch of 3D modeling. A set of new methods are carried out in this paper. One based on recursion can solve the image transformation problem with curve constrain. The edge corroding method can gain a relatively external depth image from a 2D image. Also another based on partial resampling is used to paste a 2.5D model to any other 3D model, and has made some good results. The work of this paper has greatly shortened the path from 2D editing to 3D surface animation.

A novel algorithm and approach to kinematic inbetweening for motion animation is proposed. To represent motion of a character in computer animation, a parametric curve is used to specify motion path of character while a velocity profile curve describing the varied speed of character along the path. Given the motion path and speed, kinematic inbetweening technique is executed to generate motion inbetweens from which skeleton sequences for character animation are obtained by inverse kinematical technique. In this paper, the principle and a novel algorithm for kinematic inbetweening is proposed and several demonstrations provided.

In off-road simulation, the terrain is being modified as a result of its interaction with the vehicles. Previous methods dealing with tire tracks are to animate tire tracks and have no interactivity. In this paper, we describe a method to render tire tracks in dynamic terrain. In the base of hierarchical structure for terrain LOD, hierarchical structure starts at a resolution sufficient to represent the initial terrain, and then increases the resolution wherever necessary at the modified terrain regions. To avoid the cracks at the boundary between terrain cells, which are modified and increased the resolution, an algorithm based on window-scanning avoids cracks between neighboring cells in the 2×2 window is given. This algorithm avoids the transfer of cracks effectively through guaranteeing the results after window scanning will not impact the edges processed previously. The results of experiments show that not only this algorithm avoids crack effectively, but also the cost of computing is very low.

Real-time deformation plays an important role in interactive applications such as video games. A new method of real-time dynamic deformation based on linear modal analysis is introduced in this paper. Compared to the existing techniques using the vertex program of the GPU to calculate vertex positions, our approach stores the modal displacement matrix in a series of 2D floating-point textures and the fragment program calculates vertex positions. We also encode the mesh neighbourhood information as an index form, and store it as textures. With a texture containing vertex positions, the fragment program can calculate vertex normals by local averaging. Experiments show that the proposed technique fully utilizes the parallel processing ability of the GPU, and it is suitable for the applications of real-time deformation.

This paper presents a new method for constructing parametric curve to interpolate a set of data points in plane. The constructed parametric curve has the precision of cubic polynomial function in the sense that if the given data points are taken from a cubic polynomial, then the constructed curve reproduces the cubic polynomial exactly. The comparisons of the precision of the new method with other ones are included.

This paper proposes a sketch-based modeling and animation system. Most traditional modeling and animation systems require users to have professional techniques and outstanding 3D perception in order to perform precise adjustments on the rigid skeleton positions and angles for each key frame, which usually costs several hours to complete an animation. We here propose a modeling and animation system for non-expert users with only sketch interactions. In our system, we design a new soft curve skeleton to represent the model animation and deformation. By using this new curve skeleton, users can establish skeleton by sketch and design animation conveniently, and it will cost only several minutes to finish the animation task. Our system allows non-expert users to easily design and create non-precise models and animations.

In this paper, we presented a point sampled surface reconstruction method based on local geometry. First, an adaptive Binary Space Partition (aBSP) tree was built based on the local shape complexity which was judged by three factors: local main curvature, the normal cone and the number of points. Different local complexities produce different aBSP-trees. Then, each leaf node in the aBSP-tree was approximated by a weighted quadric function and the center points set of aBSP tree’s leaf nodes could be treated as one approximation to the original point set at a given complexity of the local shape. Finally, a ray tracer based on point was used to render these aBSP-trees, which avoids triangulating of implicit surfaces before they are rendered. The experimental results show that the method can reconstruct sharp features very conveniently and effectively.

Progressive data transmission is critical for implementing interactive walkthrough of large scale distributed virtual environments on Internet. This paper proposes an incremental AOI (Area of Interests) algorithm to determine dynamically which VRML objects should be downloaded while the viewpoint is moving in a VRML environment at each step. Our method has three major steps: (1) to divide the entire VRML ground uniformly into a rectangular mesh (m*n grids) along x-axis and y-axis respectively and store all the VRML objects overlapped by each grid into an adjacent list; (2) statically to determine all the grids overlapped by an AOI according to spatial coherence; (3) dynamically to determine newly visible and invisible grids overlapped by a moving AOI at each step, according to the temporal coherence between two consecutive AOI circles; (4) at each step, only newly overlapped VRML objects should be downloaded and rendered incrementally. Thus, the required VRML data downloads can be decreased dramatically at each moving step. Further, the proposed AOI algorithm is an integer incremental algorithm without multiplication, divisions and floating operations, and also it is suitable for hardware implementation.

Content based image retrieval (CBIR) mostly uses color correlogram for image features because it extracts not only the color distribution of pixels in images like color histogram, but also extracts the spatial information of pixels in the images. The size of color correlogram depends on the number of quantized colors used for feature extractions. Usually, 64 quantized colors are used, and hence the size of the correlogram is 64x64 for unit distance. In this paper, we reduce the size of color correlogram to 9x9 by quantizing the color pixels using the median of pixels within a small 3x3 block. The proposed algorithm has smaller size of the correlogram and gives comparable image retrieval results.

In this paper we present a novel feature-based texture design scheme using deformation techniques. Firstly, we apply a compass operator to extract the feature map from the input small sample texture. Secondly, we use the feature-guided patch searching algorithm to find satisfactory candidate patches taking both color errors and feature errors into account. When the new feature map is created, the designed texture is obtained simultaneously. Thirdly, a completion-based texture design method is employed to design a variety of large deformed textures. A designer can repeat the above steps to design satisfactory textures. The proposed algorithm has the ability to design a variety of versatile textures from a single small sample texture by measuring the structural similarity. Our experimental results demonstrate that our proposed technique can be used for other texture synthesis applications, such as Wang Tiles based cyclic texture design.