Catálogo de publicaciones - libros

In this paper, a new panoramic image generation algorithm is proposed based on more realistic image formation processes. Perspective projection, lens distortion, vignetting and illumination effects are incorporated into the proposed panoramic modeling. Intrinsic and extrinsic camera parameters are estimated by the proposed stable camera parameter estimation algorithm derived from panning camera constraints. This paper shows that accurate panoramic images can be reconstructed based on the proposed camera modeling and parameters estimation. The effectiveness of the proposed algorithm is also shown with several image sequences in terms of reconstruction error from the generated panoramic image.

This article describes a method to insert virtual objects into a real video stream based on feature tracking and camera pose estimation from a set of single-camera video frames. To insert or modify 3D shapes to target video frames, the transformation from the 3D objects to the projection of the objects onto the video frames should be revealed. It is shown that, without a camera calibration process, the 3D reconstruction is possible using multiple images from a single camera under the fixed internal camera parameters. The proposed approach is based on the simplification of the camera matrix of intrinsic parameters and the use of projective geometry. The method is particularly useful for augmented reality applications to insert or modify models to a real video stream. Several experimental results are presented on real-world video streams, demonstrating the usefulness of our method for the augmented reality applications.

In this paper, we are proposing a new mechanism for controlling 3D (three dimensional) avatars to create user-designed peculiar motions of avatars in real-time using general interfaces, such as a mouse, a keyboard, or a joystick. The main idea is based on the new way of interactive control that is the combined usage of keyboard and mouse simultaneously. In order to generate natural human motions of avatars, we adopted the center line concept of art drawing and some influencing physics algorithms which developed intensively in the field of biped humanoid robot research. We demonstrate that user-designed motions of avatar can be created in real-time using proposed interaction method with keyboard and mouse. Also, we show that a rich set of peculiar behaviors can be generated from a ready-made motion with motion capture data or created and stored in our system. Note that the generated peculiar motions can be more natural if we appropriately apply our center line concept and physics algorithms.

The paper proposed a new environment matting algorithm to model the appearance of transparent object under different background. We used the frequency response to compute the relationship between the area of foreground object and background image. Moreover, the Kaczmarz method was applied to obtain more accurate weight matrices. In the experiments, we demonstrated that the algorithm can effectively improve the quality of compositing picture and has higher PSNR than the previous method. Besides, we also showed that the proposed algorithm can maintain the quality even the high threshold is set to reduce the computation time.

Our work focuses on the simplification of MPEG-4 avatar models. Similar to other general purposed 3D models, these avatars often claim complex, highly detailed presentation to maintain a convincing level of realism. However, the full complexity of such models is not always required, especially when a client terminal — for the reason of portability and cost reduction — cannot or does not necessarily support high complex presentation. First, we deploy the well-known 3D simplification based on quadric error metric to create a simplified version of the avatar in question, taking into account that the avatar is also a 3D model based on manifold mesh. Within this general scope, we introduce a new weight factor to overcome an uncertainty in choosing target for decimation. Next, exploiting the biomechanical characteristic of avatars — having the underlying skeleton structure — we propose an adaptation of the simplifying technique to avatars. The concept of bones is taken into account as either a boundary constraint or a cost-component for the quadric error. Encouraging results can be obtained with these modified procedures.

We perform an in-depth analysis of current state-of-the-art waveletbased 3D model coding techniques and then present a new one that outperforms them in terms of compression efficiency and, more importantly, provides full spatial and SNR scalability: PLTW (Progressive Lower Tree Wavelet) coding. As all SNR scalable bit-streams, ours can be used in heterogeneous networks with a wide range of terminals, both in terms of processing power and bandwidth. But because of being spatially scalable, the PLTW bit-stream does not impose on the less powerful terminals the need of building detail trees as deep as required by the maximum LOD, because the wavelet coefficients are sent on a per-LOD basis, thus achieving a “local” SNR scalability within a “global” spatial scalability. In particular, we show that our technique provides a substantial advantage over the only similar one in a current ISO standard (MPEG-4), and thus suggest that PLTW be considered for its future versions.

Graphic models represented by 3D polygonal mesh (with geometry, color, normal vector, and texture coordinate information) are now heavily used in interactive multimedia applications. As an international standard, MPEG-4 3D mesh coding (3DMC) is a compression tool for 3D mesh models. Texture mapping on a 3D mesh model is now getting popular, since it can produce high quality reconstruction even with small number of polygons used. Texture coordinates (TCs) play an important role in texture mapping. In compressing TCs using 3DMC, we found that the lossless compression is not guaranteed, which jeopardizes the texture mapping due to the precision error between the original and reconstructed TC values. In this paper, we proposed an adaptive quantization scheme for efficient TC compression in 3DMC. By adaptively choosing the step size for quantization, the proposed method can guarantee the lossless compression.

Metamorphosis or morphing is a technique to accomplish a gradual shape transformation between a source object and a target object. This technique is usually used for animation, i.e., to create some special effects, but it can be also employed as a modeling tool where some existing shapes are combined in order to obtain new shapes. Several interactive or automatic morphing techniques were designed in the past but always with a specific object representation in mind. In this paper we propose a generic framework for 3D morphing, that allows multi-resolution surface extraction and provides a hybrid (surface-volume) dynamic representation which is suitable for encoding the static models, for supporting different morphing methods, and for encoding the intermediate 3D frames (generated during the morphing process) in a MPEG-4 stream.

Surface Light Fields(SLF) are light fields parameterized to the surface of a geometric model. Past research in this area has explored the use of parameterization of aquired geometric model to the base mesh, in the context of rendering from compressed representation of the SLF. We further extend this aproach to enable hardware-accelerated SLF rendering. We show that our aproach enables hardware-accelerated, scalable rendering with commodity graphics hardware.

We present a new lossless geometry coding method for 3D triangle-quad meshes, Adaptive Vertex Chasing. Previous localized geometry coding methods have demonstrated better compression ratios than the global approach but they are considered hard to use in practice. It is because a proper linear quantization of the local range with three inputs is time-consuming, totally dependent on a user’s trials and errors. Our new localized scheme replaces this quantization with an adaptive subdivision of the range with only one input, a subdivision level. The deeper level a user choose, the closer to the original the mesh will be restored. We also present a connectivity coder improved upon the current leading Angle-Analyzer’s with a context-modeling. Without losing the current level of efficiency, our new coder provides simple and systematic way to control the balance between distortions and the bit-rates.