WSCG '2011: Communication Papers Proceedings

Permanent URI for this collection

http://hdl.handle.net/11025/10814

Browse

Showing 1 - 20 out of 28 results

A Method for Storing Clustering Information of Model Simplification in GPUs
(Václav Skala - UNION Agency, 2011) Kanaya, Takayuki; Taniguchi, Tomoaki; Teshima, Yuji; Nishio, Koji; Kobori, Kenichi; Baranoski, Gladimir; Skala, Václav
A vertex-clustering simplification is a kind of model simplification. It is difficult for the vertex-clustering simplification to simplify complex models in real-time, although it is known as a very fast method. In addition, it is also difficult for the vertex-clustering simplification to control the number of faces. It synthesizes vertices in each cluster. Therefore, models sometimes consist of the unexpected number of faces. In recent years, Graphics Processing Units (GPUs) have grown so significantly in performance that both the computational speed and the computational accuracy improve spectacularly. GPUs have programmable units such as vertex shaders and geometry shaders. With shaders, GPUs can be used not only for graphics rendering but also for general purposes. In this paper, we propose a real-time simplification algorithm for complex models of 3D objects by using a GPU whose performance gets better these days. First, vertex-clustering information is stored to video memory on a GPU. Next, the faces are reduced by the vertex-clustering information using a programmable shader, depending on the level of detail which a user defined. We also discuss a method to control the number of faces easily.
Three-dimensional shape descriptors and matching procedures
(Václav Skala - UNION Agency, 2011) Foulds, H.; Drevin, G. R.; Baranoski, Gladimir; Skala, Václav
Shape descriptors are used to identify objects in the same way that human fingerprints are used to identify people. Features of an object are extracted by applying functions to the digital representation of the object. These features are structured as a vector which is known as the shape descriptor (feature vector) of that object. The objective when constructing a shape descriptor is to find functions that will yield shape descriptors that can be used to uniquely identify or at least classify an object. A measure of similarity is required to identify or classify an object. The similarity between two objects is computed by applying a distance function to the shape descriptors of the two objects. The objective of this paper is to examine two of the possible techniques in three-dimensional shape descriptor construction based on Fourier analysis, and to find a descriptor that is able to not only classify, but also identify objects.
Off-line Handwritten Arabic Words Segmentation Based on Structural Features and Connected Components Analysis
(Václav Skala - UNION Agency, 2011) Elzobi, Moftah; Al-Hamadi, Ayoub; Al Aghbari, Zaher; Baranoski, Gladimir; Skala, Václav
A precise and efficient segmentation for handwritten Arabic text is a vital prerequisite for the accuracy of the subsequent recognition phase. In this paper, we present a dualphase segmentation approach. The proposed approach starts first by detecting and resolving sub-words overlapping, then a topological features based segmentation is applied by means of a set of heuristic rules. Because of its crucial importance, the segmentation phase is preceded by a handwritten specific preprocessing phase, that considers issues like word’s skew- and slant- correction. The proposed approach has been successfully tested on a database of handwritten Arabic words, that contains more than 3000 words images. The results were very promising and indicating the efficiency of our approach.
Visualization and Analysis of Inverse Kinematics Algorithms Using Performance Metric Maps
(Václav Skala - UNION Agency, 2011) Cardwell, Oliver; Mukundan, Ramakrishnan; Baranoski, Gladimir; Skala, Václav
Iterative inverse kinematics (IK) algorithms are commonly used in graphics animations involving goal-directed motion of joint chains and articulated character models. A well-known algorithm is the Cyclic Coordinate Descent. For certain joint chain configurations and target positions, iterative methods can generate undesirable joint rotations. Similarly, certain target positions may require large number of iterations, or may not even be reachable. This paper presents a novel concept called performance metric maps as a tool for visualizing and analysing the performance characteristics of an iterative IK algorithm under parametric variations. The proposed method is particularly useful in determining how well an algorithm converges within a given region of the workspace. The paper presents the visualization aspects of the metric maps, and the results of comparative performance analysis of two IK algorithms.
Detecting unstructured elements in 3D scanned scenes
(Václav Skala - UNION Agency, 2011) González, M. J.; Lucena, M.; Fuertes, J. M.; Segura, R.; Rueda, A. J.; Baranoski, Gladimir; Skala, Václav
This paper presents a technique for detecting unstructured areas in scanned 3D data. For many applications, outdoor 3D scanned data has to be filtered in order to eliminate undesirable artefacts, such as wires, vegetation, small objects, etc. Usually, this task is done manually, but it would be desirable to provide an automatic tool to reduce the preprocessing cost. The proposed technique, which consists in two stages, based on anisotropic diffusion and plane regression respectively, allows us to select most of the uninteresting data. It also has been shown good results with real data.
Rain Removal from Videos using the Temporal- Spatial Statistical Properties
(Václav Skala - UNION Agency, 2011) Kalia, Robin; Jaikar, Amol; Baranoski, Gladimir; Skala, Václav
Detection and removal of rain streaks from videos has recently become a great and challenging topic of research. This paper discusses a new technique for the removal of rain from videos using the temporal-spatial statistical properties. For this the temporal statistical properties of the pixels affected by rain are made use of, and then an efficient and easy algorithm is implemented which takes care of the effective removal of rain from videos. This technique works very well for videos with still and moving backgrounds involving moving objects with a fixed camera position. For the videos which involve the motion of the camera, the technique works well for a small rate of change of background in the camera frames. Our algorithm does not use variable and conditional parameters like the shape, size, velocity, and spatio-temporal physical model of raindrops, and camera’s parameters like the aperture, focal length, and exposure time. The test results quantitatively and qualitatively illustrate that the performance of our algorithm is quite efficient in comparison to the previously existing algorithms which are state of the art techniques used for the purpose of removing rain from videos.
Approximate importance sampling of functions reconstructed from spherical harmonics
(Václav Skala - UNION Agency, 2011) Berger, Martin; Baranoski, Gladimir; Skala, Václav
The ability to generate random samples that match a spherical PDF given in terms of spherical harmonic coefficients is very important in many fields of computer graphics. Recent work has shown that generating such samples can be done efficiently, but the published methods are not robust in the presence of reconstruction errors which manifest themselves as negative values of the PDF. In our paper, we extend the approach so that it can handle such errors, and generates uniform distribution of samples in the negative parts of the sampled function while preserving a distribution that matches the original function elsewhere. The overall distribution approximates the original function and guarantees that there are no parts of the spherical domain which remain unsampled. This property makes the scheme suitable for use in unbiased Monte Carlo rendering.
CUDA expression templates
(Václav Skala - UNION Agency, 2011) Wiemann, Paul; Wenger, Stephan; Magnor, Marcus; Baranoski, Gladimir; Skala, Václav
Many algorithms require vector algebra operations such as the dot product, vector norms or component-wise manipulations. Especially for large-scale vectors, the efficiency of algorithms depends on an efficient implementation of those calculations. The calculation of vector operations benefits from the continually increasing chip level parallelism on graphics hardware. Very efficient basic linear algebra libraries like CUBLAS make use of the parallelism provided by CUDA-enabled GPUs. However, existing libraries are often not intuitively to use and programmers may shy away from working with cumbersome and errorprone interfaces. In this paper we introduce an approach to simplify the usage of parallel graphics hardware for vector calculus. Our approach is based on expression templates that make it possible to obtain the performance of a hand-coded implementation while providing an intuitive and math-like syntax. We use this technique to automatically generate CUDA kernels for various vector calculations. In several performance tests our implementation shows a superior performance compared to CPU-based libraries and comparable results to a GPU-based library.
When It Makes Sense to Use Uniform Grids for Ray Tracing
(Václav Skala - UNION Agency, 2011) Hapala, Michal; Karlík, Ondřej; Havran, Vlastimil; Baranoski, Gladimir; Skala, Václav
Commonly used hierarchical data structures such as bounding volume hierarchies and kd-trees have rather high build times, which can be a bottleneck for applications rebuilding or updating the acceleration structure required by data changes. On the other hand uniform grids can be built almost instantly in linear time, however, they can suffer from severe performance penalty, in particular in scenes with non-uniformly populated geometry. We improve on performance using a two-step approach that combines both approaches: first we build a uniform grid and test its performance. Second, using an estimate on the number of rays to be queried we either continue using the grid or build a hierarchical data structure instead. This way we select a more efficient data structure given a particular implementation of the algorithms which yields with high probability an overall smaller computational time. We evaluate the properties of this method for a set of 28 scenes.
Biquadratic S-Patch in Bézier form
(Václav Skala - UNION Agency, 2011) Kolcun, Alexej; Baranoski, Gladimir; Skala, Václav
Mutual conversions between triangular and quadrilateral meshes need the same degree of both diagonal and boundary curves of quadrilateral meshes. New approach to quadrilateral patches, S-Patches, offers such possibility. The Bézier approach of Smart patches (S-Patch) in the biquadratic case is analyzed. Dependencies among the control points are derived. BS-Patches are presented. Close relation between Bézier triangles and BSPatches is found. Condition for smooth concatenation of biquadratic BS-Patches is formulated.
Automated 3D Visualization of Electron Microscope Tomograms
(Václav Skala - UNION Agency, 2011) Yamashita, Chikara; Nishio, Kijo; Kobori, Ken-ichi; Baranoski, Gladimir; Skala, Václav
In this paper, we propose a 3D visualization method for ultra-high voltage electron microscope tomography intended for use with biological samples. The most important process for constructing 3D images from UHVTEM tomograms is the extraction of contours from 2D sliced images. However, automatic extraction of contours is difficult because of typical noise and artifacts. The proposed method automatically extracts contours by the 3D level set method. In general, the result of the extraction with the level set method depends on the definition of initial contours and parameters. These parameters are usually set manually. Our method automatically generates the initial contours and decides the fittest parameters for the level set method. We verify the effectiveness of our method by applying the technique to two types of unicellular organisms, to compare the results of the proposed method with manual extraction. The automated method successfully identified most cell tissues, with the exception of a limitation in the imaging of tubular-shaped cell structures.
Mimicking POV-Ray Photorealistic Rendering with Accelerated OpenGL Pipeline
(Václav Skala - UNION Agency, 2011) Pečiva, J.; Zemčík, P.; Navrátil, J.; Baranoski, Gladimir; Skala, Václav
Traditional ray tracing algorithms tend to provide photorealistic results but at high computing costs. Rendering times of minutes or days are not exceptional. On the other side, hardware accelerated OpenGL rendering can provide real-time interaction with virtual environment with unnoticeable rendering times. This paper attempts to bring these two together and attempts to give an answer on the difficulty of implementing real-time photorealistic rendering. The paper presents case study on mimicking of POV-Ray photorealistic rendering with accelerated OpenGL pipeline. The study shows the opportunity to accelerate some photorealistic algorithms by real-time approaches while, on the other side, it locates the parts that are difficult to replace by traditional real-time rendering paradigms. Particularly, it is shown how to implement primary and shadow rays and POV-Ray-like material model using accelerated OpenGL pipeline using modern shader technology. On the other side, the difficulties of reflected and refracted rays implementation using real-time rendering approaches is discussed.
Chroma Reconstruction from Inaccurate Measurements
(Václav Skala - UNION Agency, 2011) Balinsky, Alexander; Mohammad, Nassir; Baranoski, Gladimir; Skala, Václav
Non-linear filter responses of natural colour images have been shown to display non-Gaussian heavy tailed distributions which we call sparse. These filters operate in the YUV colour space on the chroma channel U (and V) using weighting functions obtained from the gray image Y. In this paper we utilise this knowledge for denoising the chroma channels of a colour image from inaccurate measurements. In our model the U (and V) elements are affected by noise, with a good version of the gray image Y obtainable through existing methods. We show that accurate reconstruction of the chroma components can be accomplished by solving an L1 constrained optimisation problem, where the sparse filter response on natural images is used as a regularization term. This scheme gives comparable results to leading commercial and state of the art denoising algorithms, and exceeds for chroma noise that does not correlate with the luminance structure.
Speeding up probabilistic inference of camera orientation by function approximation and grid masking
(Václav Skala - UNION Agency, 2011) Werneck, Nicolau Leal; Reali Costa, Anna Helena; Baranoski, Gladimir; Skala, Václav
This article presents modifications to an existing technique for camera orientation estimation intending to make it faster for use in real time applications and also for analysis of large image sets. The technique is based on likelihood maximization of a probability function that has the image gradient as the observed data and the camera orientation as parameter values. The camera orientation is inferred from the vanishing points of the image, and the directions of the edges in the environment are assumed to be in three mutually orthogonal directions. The first proposed modification is to substitute the expression that is calculated at each pixel by a computationally lighter approximation. The second proposal is to take in consideration only a few of the pixel lines and columns of the image during the calculations, performing a grid windowing of the image. This article presents the derivation and reinterpretation of the likelihood function approximation and also a performance evaluation.
Investigating measures for transfer function generation for visualization of MET biomedical data
(Václav Skala - UNION Agency, 2011) Svensson, L.; Sintorn, I.-M.; Svensson, S.; Nyström, I.; Baranoski, Gladimir; Skala, Václav
In this paper, the question of automatically setting transfer functions for volume images is further explored. More specifically, the focus is automatic visualization of Molecular Electron Tomography (MET) volume images using one-dimensional transfer functions. We investigate how well a few general measures based on density, gradient, curvature and connected component information are suited for generating these transfer functions. To assess this, an expert has set suitable transfer function levels manually and we have studied how these levels relate to different characteristics of the selected measures for 29 data sets. We have found that the measures can be used to automatically generate a transfer function used to visualize MET data, to give the user an approximate view of the components in the image.
Improved Algorithm for Principal Curvature Estimation in Point Clouds due to Optimized Osculating Circle Fitting based on Geometric Algebra
(Václav Skala - UNION Agency, 2011) Getto, Roman; Hildenbrand, Dietmar; Baranoski, Gladimir; Skala, Václav
In this paper we introduce our improvements and innovations to an algorithm for the estimation of principal curvatures in point clouds. The major part of the improvement is achieved by the use of a new osculating circle fitting. In this paper, first of all, we explain the algorithm for Principal Curvature Estimation, as well as the previous osculating circle fitting. Then we present the new osculating circle fitting and an additional possibility of improving the results by a method, which adjusts some variables of the algorithm at runtime, to adapt the algorithm to the denseness of the particular point cloud. Furthermore, we present some results of the new methods. To complete this paper, we give a conclusion and an outlook.
Gait Recognition in the Presence of Occlusion: A New Dataset and Baseline Algorithms
(Václav Skala - UNION Agency, 2011) Hofmann, Martin; Sural, Shamik; Rigoll, Gerhard; Baranoski, Gladimir; Skala, Václav
Human gait is an important biometric feature for identification of people. In this paper we present a new dataset for gait recognition. The presented database overcomes a crucial limitation of other state-of-the-art gait recognition databases. More specifically this database addresses the problem of dynamic and static inter object occlusion. Furthermore this dataset offers three new kinds of gait variations, which allow for challenging evaluation of recognition algorithms. In addition to presenting the database we present two baseline algorithms (Color histograms, Gait Energy Image) to perform person identification using gait. These algorithms already show promising results on the presented database.
Integration of Reconstruction Error Obtained by Local and Global Kernel PCA with Different Role
(Václav Skala - UNION Agency, 2011) Hotta, Kazuhiro; Baranoski, Gladimir; Skala, Václav
This paper presents a scene classification method using the integration of the reconstruction errors by local Kernel Principal Component Analysis (KPCA) and global KPCA. There are some methods for integrating local and global features. However, it is important to give obvious different role to each feature. In the proposed method, global feature with topological information represents the rough composition of scenes and local feature without position information represents fine part of scenes. Experimental results show that accuracy is improved by using the reconstruction errors obtained from the different point of views. The proposed method is much better than only local KPCA, global KPCA and linear Support Vector Machine (SVM) of bag-of-visual words with the same basic feature. Our method is also comparable to conventional methods using the same database.
Example-based Deformation with Support Joints
(Václav Skala - UNION Agency, 2011) Yamanaka, Kentaro; Yano, Akane; Morishima, Shigeo; Baranoski, Gladimir; Skala, Václav
In character animation field, many deformation techniques have been proposed. Example-based deformation methods are widely used especially for interactive applications. Example-based methods are mainly divided into two types. One is Interpolation. Methods in this type are designed to interpolate examples in a pose space. The advantage is that the deformed meshes can precisely correspond to the example meshes. On the other hand, the disadvantage is that larger number of examples is needed to generate arbitrary plausible interpolated meshes between each example. The other is Example-based Skinning which optimizes particular parameters referencing examples to represent example meshes as accurately as possible. These methods provide plausible deformations with fewer examples. However they cannot perfectly depict example meshes. In this paper, we present an idea that combines techniques belonging to the two types, taking advantages of both types. We propose an examplebased skinning method to be combined with Pose Space Deformation (PSD). It optimizes transformation matrices in Skeleton Subspace deformation (SSD) introducing “support joints”. Our method itself generates plausible intermediate meshes with a small set of examples as well as other example-based skinning methods. Then we explain the benefit of combining our method with PSD. We show that provided examples are precisely represented and plausible deformations at arbitrary poses are obtained by our integrated method.
Novel Trilateral Approach for Depth Map Spatial Filtering
(Václav Skala - UNION Agency, 2011) Voronov, Alexander; Vatolin, Dmitriy; Smirnov, Maxim; Baranoski, Gladimir; Skala, Václav
In this paper, we present our approach for spatial filtering of depth map extracted from camera motion. An original depth map may have some artifacts owing to imperfect motion estimation. Our goal was to make the depth map uniform in smooth areas and to refine object boundaries without blurring edges. To solve this problem we propose the trilateral filter, whose convolution kernel is composed of a distance kernel, a color-based kernel and a depthbased kernel. Experiments demonstrate that this approach yields rather good results. Also, we compare our results with those of a typical bilateral filter.

Browse

Recent Submissions