WSCG 2018: Full Papers Proceedings
Permanent URI for this collection
Browse
Recent Submissions
Item Markerless structure-based multi-sensor calibration for free viewpoint video capture(Václav Skala - UNION Agency, 2018) Papachristou, Alexandros; Zioulis, Nikolaos; Zarpalas, Dimitrios; Daras, Petros; Skala, VáclavFree-viewpoint capture technologies have recently started demonstrating impressive results. Being able to capture human performances in full 3D is a very promising technology for a variety of applications. However, the setup of the capturing infrastructure is usually expensive and requires trained personnel. In this work we focus on one practical aspect of setting up a free-viewpoint capturing system, the spatial alignment of the sensors. Our work aims at simplifying the external calibration process that typically requires significant human intervention and technical knowledge. Our method uses an easy to assemble structure and unlike similar works, does not rely on markers or features. Instead, we exploit the a-priori knowledge of the structure’s geometry to establish correspondences for the little-overlapping viewpoints typically found in free-viewpoint capture setups. These establish an initial sparse alignment that is then densely optimized. At the same time, our pipeline improves the robustness to assembly errors, allowing for non-technical users to calibrate multi-sensor setups. Our results showcase the feasibility of our approach that can make the tedious calibration process easier, and less error-prone.
Item Real-time visual off-road path detection(Václav Skala - UNION Agency, 2018) Krämer, Marc Steven; Kuhnert, Lars; Kuhnert, Klaus-Dieter; Skala, VáclavIn this paper, we propose a fast and real-time capable system for visual off-road path detection. We equipped our robot AMOR with a single monocular camera and explored unstructured environments like woods. In these areas, it is almost harder to identify and classify drivable and non-drivable parts in an image. In urban regions, roads can be detected by lane markers or delimitations whereas the boundaries of a forest path blend into the environment almost seamlessly. In our work, we developed a software system that is based on mostly simple and low computationally intensive algorithms. We developed and tested the functions with a large dataset of camera images and also generated a manually Ground Truth for the evaluation.Item Single image summary of time-varying Earth-features(Václav Skala - UNION Agency, 2018) Tripathi, Gaurav; Katayoon, Etemad; Samavati, Faramarz; Skala, VáclavThe Earth’s surface is live and dynamic due to natural and manmade events. Tracking and visualizing Earth-features (e.g. water, snow, and vegetation) is an important problem. Earth observation satellite imagery like Landsat 8 makes the tracking feasible by providing detailed multispectral imagery at regular intervals. In this paper, we explore a single image summary approach to detecting changes in Earth-features by using the Landsat 8 dataset. In our system, we use appropriate thresholds for spectral indices to identify features, reference datasets, and combine multiple images using predefined color palettes to generate a single image summary of features for a region. Furthermore, we illustrate the benefit of our method over traditional visualizations with case-studies for the Lake Urmia, the Amazon Rainforest, and the Bering Glacier.Item Fuzzy image inpainting aimed to medical images(Václav Skala - UNION Agency, 2018) Vlašánek, Pavel; Skala, VáclavThis paper focuses on a reconstruction of low color depth images using fuzzy mathematics. For demonstration purposes, we chose medical images taken from magnetic resonance imaging (MRI) and computed tomography (CT). The proposed technique is based on idea of diffusion where pixels surrounding damaged region are used to determine the corrupted ones. As it is illustrated, the classical diffusion techniques are not so effective. In the paper, we describe the reason why and propose the solution in a form of the new algorithm. The algorithm is demonstrated and visually compared with another ones.Item Barycentric combinations based subdivision shaders(Václav Skala - UNION Agency, 2018) Morlet, Lucas; Neveu, Marc; Lanquetin, Sandrine; Gentil, Christian; Skala, VáclavWe present a new representation of uniform subdivision surfaces based on Iterated Functions Systems formalism. Main advantages of this new representation are the formalization of topological subdivision, multiscale representation of limit surface, separation of iterative space where the attractor is computed once for all and modeling space where the attractor is projected many times. An important consequence of this approach is that all uniform subdivision schemes are handled in the same way whatever there are primal or dual, approximating or interpolating. Subdivision surfaces are no longer viewed as a set of rules but as a list of barycentric combinations to apply on neighborhoods of the coarse mesh. These combinations are representative subsets of the attractor which is deduced from a Controlled Iterated Functions System automaton. From this new point of view we present in this paper a straightforward implementation to directly compute a tessellation of the subdivision surface from a control mesh. This implementation takes full advantage of Graphics Processing Units high capability of computation and Tessellation Stage of OpenGL/GLSL rendering pipeline to generate on the fly a tessellation of the limit surface with a chosen Level of Details.Item Assessing objective image quality metrics for bidirectional texture functions(Václav Skala - UNION Agency, 2018) Azari, Banafsheh; Bertel, Sven; Wüthrich, Charles A.; Skala, VáclavBidirectional Texture Functions (BTFs) are view- and illumination-dependent textures used in rendering for accurate simulation of the complex reflectance behavior of fabrics. One major issue in BTF rendering is the large number and size of images which requires lots of storage. "Visually lossless" compression offers the potential to use higher compression levels without noticeable artifacts, but requires a rate-control strategy that adapts to image content and loss visibility. In this contribution, we investigate the applicability of objective image quality metrics to predict levels of perception degradation for compressed BTF textures. We apply traditional error-sensitivity and structural similarity based approaches to predict levels of perceptibility for compressed BTF textures to achieve visually lossless compression. To confirm the validity of the present study, the results of an experimental study on how decreasing the BTF texture resolution influences the perceived quality of the rendered images with the results of the applied image quality metrics are compared. In order to compare two representatives from each group were selected. The Visible Differences Predictor (VDP) and Visual Discrimination Model (VDM) are typical examples of an image quality metric based on error sensitivity, whereas the Structural SIMilarity index (SSIM) and Complex Wavelet Domain Structural Similarity Index (CWSSIM) are specific examples of a structural similarity quality measure.Item Calibrating low-cost structured-light 3D sensors(Václav Skala - UNION Agency, 2018) Chakib, Reda; Mérillou, Nicolas; Vincent, Pierre-Jean; Mérillou, Stephane; Skala, VáclavConsumer-grade RGB-D cameras are widely accessible, but they suffer from a lack of accuracy when compared to professional-grade 3D scanning solutions. In this paper, we propose a new method for calibrating an Intel RealSense SR300 camera, adaptable to other structured light sensors. The method uses classical checkerboard calibration and a coordinate-measuring machine (CMM) based setup with a calibrating plane. It delivers better results than the manufacturers settings.Item Generation of implicit flow representations for interactive visual exploration of flow fields(Václav Skala - UNION Agency, 2018) Molchanov, Vladimir; Linsen, Lars; Skala, VáclavA stream function is an implicit flow representation in form of a function, whose values are constant along streamlines of the underlying velocity field. To generate a stream function, a common approach is to use a streamline tracking technique after assigning scalar function values on the inflow/outflow domain boundary (pre-processing step). However, non-trivial flows generally have streamlines that do not start or end at the domain boundary. We propose an automatic approach that defines a stream function along such streamlines. To do so, we construct optimal termination surfaces inside the domain and assign scalar values to all streamlines crossing these surfaces. Furthermore, we propose a proper functional to characterize the quality of the approximated stream function. Using a variational approach, we derive a partial differential equation for the minimization of the derived functional. This minimization procedure is an effective tool to improve the stream function. It can also be used to significantly improve the pre-computation times by creating a high-quality high-resolution stream function from a low-resolution estimate. Once the implicit flow representation is established and improved, we can efficiently extract flow geometry such as stream ribbons, stream tubes, stream surfaces, etc. by applying fast marching algorithms. Tracking time recorded during the pre-processing step can be coupled with the stream function or used directly to extract time surfaces. Thus, the entire flow field can be explored interactively. There is no need for time-consuming particle tracking and mesh refinement during the visual exploration process.Item 3D object classification and parameter estimation based on parametric procedural models(Václav Skala - UNION Agency, 2018) Getto, Roman; Fina, Kenten; Jarms, Lennart; Kuijper, Arjan; Fellner, Dieter W.; Skala, VáclavClassifying and gathering additional information about an unknown 3D objects is dependent on having a large amount of learning data. We propose to use procedural models as data foundation for this task. In our method we (semi-)automatically define parameters for a procedural model constructed with a modeling tool. Then we use the procedural models to classify an object and also automatically estimate the best parameters. We use a standard convolutional neural network and three different object similarity measures to estimate the best parameters at each degree of detail. We evaluate all steps of our approach using several procedural models and show that we can achieve high classification accuracy and meaningful parameters for unknown objects.Item Procedural fracture of shell objects(Václav Skala - UNION Agency, 2018) Domaradzki, Jakub; Martyn, Tomasz; Skala, VáclavWe propose a novel algorithm to fracture brittle objects that are characterized by an empty interior and thick surface (which we denote as shell objects), such as: vases, pots, pitchers, antique ceramic, etc. Our method augments the previous ones based on fracture patterns and utilizes sparse voxel octrees (SVOs) as a highly efficient and detailed object representation. In our method, the fracture pattern relies on Voronoi diagrams and is calculated on-the-fly. The outcomes of applying the fracture pattern differ from the ones obtained with the previous methods in that it solves the problem of planar faces of the newly generated pieces of geometry, allowing them to have concave shapes. Without any precomputation, we are able to achieve various and interesting fractures that are unique to each destructed object. Finally, our approach is intuitive, adaptable and fast, which makes it a good candidate for applications in computer game industry.Item From geometric modeling to digital Earth(Václav Skala - UNION Agency, 2018) Samavati, Faramarz; Skala, Václav
