Visualisation and geometric modelling are today widely applied to the field of architecture. However, they remain often considered as means to communicate about the edifice rather than means to investigate the edifice. Our research focuses on the issues raised by the development of visualisation and modelling techniques on the Internet for documenting and representing edifices of the architectural heritage. Our position is that 3D representations can greatly favour the readability and the accessibility of data related to an edifice, on the condition that this representation shows architectural concepts before geometrical ones. Such interpretative visualisations of edifices can then be used to analyse simulations of buildings partly or totally destroyed. What is more, such visualisations can facilitate the construction of an information system about the architectural heritage, in which 3D models are used as interfaces to a database. The global disposal we propose to discuss includes: • An object-oriented model of architectural concepts, • A VRML (Virtual Reality Modelling Language) architectural modeller, • A resource database containing data on the edifices represented, • Interfaces to the database on Internet, among which VRML models. Our contribution details those various aspects and the experiments we have carried out on chosen buildings of the city of Cracow.
In this paper we discuss the mechanisms that have been used to deal with the dynamic nature of data in the CyberNet project. The purpose of this project is to study how three-dimensional (3D) metaphoric visualization may help the user in the process of monitoring large amounts of dynamic information. We present the dynamic data model and the visualization model and then focus on the problem of updating a 3D world without disorienting the user.
At the present time, one of the best methods for semiautomatic image segmentation seems to be the approach based on the fuzzy connectedness principle. First, we identify some deficiencies of this approach and propose a way to improve it, through the introduction of competitive learning. Second, we propose a different approach, based on watersheds. We show that the competitive fuzzy connectedness-based method outperforms the noncompetitive variant and generally (but not always) outperforms the watershed-based approach. The competitive variant of the fuzzy connectedness-based method can be a good alternative to the watersheds.
Closed streamlines are an integral part of vector field topology, since they behave like sources respectively sinks but are often neither considered nor detected. If a streamline computation makes too many steps or takes too long, the computation is usually terminated without any answer on the final behavior of the streamline. We developed an algo-rithm that detects closed streamlines during the integration process. Since the detection of all closed streamlines in a vector field requires the computation of many streamlines we extend this algorithm to a parallel version to enhance computational speed. To test our implementation we use a numerical simulation of a swirling jet with an inflow into a steady medium. We built two different Linux clusters as parallel test systems where we check the performance increase when adding more processors to the cluster. We show that we have a very low parallel overhead due to the neglectable communication expense of our implementation.
A method for automatic horizon tracking in reflection seismic data images across discontinuities is described. Horizon tracking is an important task of the structural interpretation of seismic images, however the automatic tracking across discontinuities has still not been solved satisfactorily. The reason for this is the difficulty involved in locating non-ambiguous local correlation features as a result of the small amount of local information contained in seismic reflection images. The method described here provides an enhancement over a solely local feature based analysis by including a high-level deterministic analysis based upon geological and geometrical constraints. Application of the method to typical seismic data images resulted in the successful matching of all major horizons across several normal faults.
We show that the results obtained through clusteringbased image segmentation of single or multi-component images can be improved by a fuzzy relaxation of the degrees of membership in the image space. We illustrate the point through two clustering techniques: the fuzzy Cmeans (FCM) technique and a clustering technique based on the estimation of the probability density function (pdf).
This work addresses the problem of automatic tracking of pedestrians observed by a fixed camera in outdoor scenes. Tracking isolated pedestrians is not a difficult task. The challenge arises when the tracking system has to deal with temporary occlusions and groups of pedestrians. In both cases it is not possible to track each pedestrian during the whole video sequence. However, the system should be able to recognize each pedestrian as soon as he/she becomes visible and isolated from the group. This paper presents methods to tackle these difficulties. The proposed system is based on a hierarchical approach which allows the application of the same methods for tracking isolated pedestrians and groups.
We propose a key frame extraction mechanism to aid the Structure from Motion (SfM) problem when dealing with image sequences from video cameras. Due to high frame rates (15 frames per second or more) the baseline between frames can be very small and the number of frames can become unpractical to deal with effectively. The mechanism described in this paper is a preprocessing step designed to make an ideal image sequence from larger sequence of video frame data. Based on a proven tracking mechanism, the algorithm remains quite simple yet effective for identifying and extracting salient frame data for the SfM problem in effect removing degeneracy cases.
This paper presents a method to obtain a single continuos polygonal mesh representing a branched structure taken from a tree and modelled by a L-systems-based method. A refinement process is made once the model of the tree is obtained. This process produces smooth transitions in the tree areas joining the branches. That is, the overlapping problems and geometry discontinuities are avoided. Because of this process, a continuous polygonal mesh that represents all the tree branches is generated. This mesh allows mesh simplification algorithms.
In this paper, we present an easy, efficient and practical algorithm, which extracts the feature silhouette from a photograph for virtual human modeling. Our segmentation algorithm is derived from the MumfordShah segmentation technology and the level set formulation. We accelerate the silhouette extraction process using the multi-pyramid level method. After that, a feature extraction algorithm is introduced to determine the feature points on the human contours. At the end of the paper, some results of the virtual human modeling are shown to demonstrate the functionality. Compared with other approaches, our silhouette extraction methods and feature extraction are automatic, and the speed of extraction is fast.
