Browsing by Author "Loke, R. E."
Now showing 1 - 10 of 17
Results Per Page
Sort Options
- A background-priority discrete boundary triangulation methodPublication . Loke, R. E.; Jansen, F. W.; du Buf, J. M. H.Discrete boundary triangulation methods generate triangular meshes through the centers of the boundary voxels of a volumetric object. At some voxel configurations it may be arbitrary whether a part of the volume should be included in the object or could be classified as background. Consequently, important details such as concave and convex edges and corners are not consistently preserved in the describing geometry. We present a "background priority" version of an existing "object priority" algorithm [6]. We show that the ad hoc configurations of the well-known Discretized Marching Cubes algorithm [13] can be derived from our method and that a combined triangulation with "object priority" and "background priority" better would preserve object details.
- Analysis of underwater acoustic data via 3-D segmentationPublication . Reed, Todd R.; Loke, R. E.; du Buf, J. M. H.The analysis of seabed structure is important in a wide variety of scientific and industrial applications. In this paper, underwater acoustic data produced by bottom-penetrating sonar (Topas) are analyzed using unsupervised volumetric segmentation, based on a three dimensional Gibbs-Markov model. The result is a concise and accurate description of the seabed, in which key structures are emphasized. This description is also very well suited to further operations, such as the enhancement and automatic recognition of important structures. Experimental results demonstrating the effectiveness of this approach are shown, using Topas data gathered in the North Sea off Horten, Norway.
- Diatom classification in ecological applicationsPublication . Loke, R. E.; du Buf, J. M. H.; Bayer, M. M.; Mann, D. G.We apply new contour features: (1) Point features by computing the convexity and curvature in small Contour neighborhoods. (2) Segment features by segmenting the contour into convex, concave and straight segments, and computing length and curvature measures for each segment. (3) Global features by computing the mean, maximum and minimum of all point and segment features. Features can be extracted from noisy contours with convex, concave and straight parts, but also from completely convex ones, for the purpose of shape analysis or identification (ID) tasks. Using only four global features, a nearest-mean classifier yielded a perfect ID rate of 100% on diatoms with minute differences in shape, which are difficult to identify, even for diatomists. (C) 2003 Pattern Recognition Society. Published by Elsevier Ltd. All rights reserved.
- Diatom recognition by convex and concave contour curvaturePublication . Loke, R. E.; Bayer, M. M.; Mann, D. G.; du Buf, J. M. H.In this paper we describe a new contour feature set. A contour is segmented into convex, concave and straight segments, after which length and curvature features are computed. A symmetry analysis allows the detection of the number of elementary segments. Results on two contour test sets were obtained: Using only four features, a simple nearest-mean classifier yielded a perfect identification (ID) rate of 100% on a small set consisting of shapes with minute differences, which are difficult to identify even for human experts. Using 10 features, it yielded 83.5% on a large set with very diverse shapes.
- Fast interpolation, segmentation and visualization of 3D sonar seabottom data by using tree structuresPublication . Loke, R. E.; du Buf, J. M. H.We describe an unsupervised processing pipeline for the analysis and visualization of sonar data. Recently developed 3D interpolation and segmentation methods are introduced, as well as a completely new surface construction method All methods are based on employing quad- and octrees, which results in a fast processing. Accurate results are obtained for complex boundary shapes. Processed seabed data are accurately fused with bathymetric data. The methods enable a fast, highly accurate and interactive seabed analysis.
- Fast segmentation of 3D data using an octreePublication . Rodrigues, J. M. F.; Loke, R. E.; du Buf, J. M. H.The algorithm developed uses an octree pyramid in which noise is reduced at the expense of the spatial resolution. At a certain level an unsupervised clustering without spatial connectivity constraints is applied. After the classification, isolated voxels and insignificant regions are removed by assigning them to their neighbours. The spatial resolution is then increased by the downprojection of the regions, level by level. At each level the uncertainty of the boundary voxels is minimised by a dynamic selection and classification of these, using an adaptive 3D filtering. The algorithm is tested using different data sets, including NMR data.
- Highlighting pipelines offshore Norway: visualization and quantitative analysisPublication . Loke, R. E.; du Buf, J. M. H.Sonar profiler data of the seabottom are 3D data sets that can cover huge survey areas with many gaps. This paper describes a multiresolution visualization framework that is being optimized for dealing with such data taking into account both the CPU time and the user interactivity. The paper describes the techniques employed: (a) the construction of a quadtree that allows to eliminate gaps by interpolating available 3D data, (b) a first but coarse visualization at a high tree level in order to rapidly change or adjust the ROI, and (c) a very efficient triangulation (mesh reduction) that allows for a fast interactivity even at the highest detail level. By using one single octree all this processing can be combined because (1) gaps can be filled by interpolation because they are smaller at higher tree levels, (2) connected components can be projected down the tree and refined using the data available there and (3) triangulations at higher tree levels can be used to steer those at lower levels to fill efficiently large and smooth surface areas. As a result, huge data sets can be visualized in near realtime with OpenGL/VRML using shading instead of wire-frames, and this enables a fast searching for objects in the seabottom. Real CPU times are presented for the visualization of a real sonar data set which was obtained in the North Sea, offshore Norway. These data are visualized at a low resolution, showing the overall shape of the seabottom, and at a high resolution, showing a (semi-)buried pipeline. In order to detect an object at such a high resolution additional techniques are applied to the data: (a) an interslice interpolation in order to cope with the increased data sparseness and (b) a maximum-homogeneity filtering in order to cope with the decreased SNR. After the extraction of the pipeline a thinning technique is applied in order to be able to quantify its length.
- Interactively visualizing 18-connected object boundaries in huge data volumesPublication . Loke, R. E.; du Buf, J. M. H.We present a multiresolution framework for the visualization of structures in very large volumes. Emphasis is given to an in the framework embedded, new algorithm for triangulating 18-connected object boundaries which preserves 6-connectivity details. Such boundaries cannot be triangulated by standard 6-connectivity algorithms such as Marching Cubes. Real sonar imaging results show that the framework allows to visualize global subbottom structure, but also high-resolution objects, with a reduced CPU time and an improved user interactivity.
- Polygon optimisation for the modelling of planar range dataPublication . Nunes, S.; Almeida, D.; Loke, R. E.; du Buf, J. M. H.In this paper we present efficient and fast algorithms for the reconstruction of scenes or objects using range image data. Assuming that a good segmentation is available, we concentrate on the polygonisation, triangulation and optimisation, i.e. both triangle reduction and adaptive edge filtering to improve edge linearity. In the processing, special attention is given to complex edge junctions. In a last step, vertex neighbourhoods are analysed in order to robustly attribute depth to the triangle list from the noisy range data.
- Quadtree-guided 3-D interpolation of irregular sonar data setsPublication . Loke, R. E.; du Buf, J. M. H.Bottom-penetrating sonar can be used to visualize large areas, for example by normal logging and printing of collected pings. In many applications, it is necessary to obtain an impression of three-dimensional (3-D) structures, but this is not easy because of the irregular spatial sampling due to coarse ship trajectories. Normally, the ping map and the ping data, cover only a very small part of a region of interest. In this paper, we describe a new method for interpolating irregularly spaced sonar data. The basic idea is to use a two-dimensional quadtree of the ping map in order to guide the 3-D interpolation process: since gaps between pings become smaller at higher tree levels, the volume can be filled by marking neighborhood relations in the quadtree and interpolating available pings when they become neighbors. Different marking schemes and their central processing unit times are compared. In the interpolation process, we apply cross correlations of ping data in order to construct continuity of sloping reflections. Our results show that excellent results can be obtained on real sonar data sets, even for volumes filled for less than 7%, for which processing times are reasonable even for large areas, and that the interpolated data can be used for volumetric interactive visualization.