Multiple Views Of Data Research Articles

Multiview segmentation and tracking of dynamic occluding layers

We present an algorithm for the layered segmentation of video data in multiple views. The approach is based on computing the parameters of a layered representation of the scene in which each layer is modelled by its motion, appearance and occupancy, where occupancy describes, probabilistically, the layer’s spatial extent and not simply its segmentation in a particular view. The problem is formulated as the MAP estimation of all layer parameters conditioned on those at the previous time step; i.e., a sequential estimation problem that is equivalent to tracking multiple objects in a given number views. Expectation–Maximisation is used to establish layer posterior probabilities for both occupancy and visibility, which are represented distinctly. Evidence from areas in each view which are described poorly under the model is used to propose new layers automatically. Since these potential new layers often occur at the fringes of images, the algorithm is able to segment and track these in a single view until such time as a suitable candidate match is discovered in the other views. The algorithm is shown to be very effective at segmenting and tracking non-rigid objects and can cope with extreme occlusion. We demonstrate an application of this representation to dynamic novel view synthesis.

Multiple view semi-supervised dimensionality reduction

Multiple view data, together with some domain knowledge in the form of pairwise constraints, arise in various data mining applications. How to learn a hidden consensus pattern in the low dimensional space is a challenging problem. In this paper, we propose a new method for multiple view semi-supervised dimensionality reduction. The pairwise constraints are used to derive embedding in each view and simultaneously, the linear transformation is introduced to make different embeddings from different pattern spaces comparable. Hence, the consensus pattern can be learned from multiple embeddings of multiple representations. We derive an iterating algorithm to solve the above problem. Some theoretical analyses and out-of-sample extensions are also provided. Promising experiments on various data sets, together with some important discussions, are also presented to demonstrate the effectiveness of the proposed algorithm.

Extending the trade‐off between spatial resolution and variance in quantitative ultrasonic backscattering imaging (QUS) using full angular spatial compounding.

Tissue microstructure properties can be estimated using backscattered power measurements and quantitative ultrasound (QUS) techniques. The objective of this study was to evaluate the use of full angular (i.e., 360°) spatial compounding as a means of extending the trade‐off between QUS variance and spatial resolution. Simulations and experimental results were conducted with a 10 MHz, f/4 transducer. In simulations, a synthetic phantom consisting of two eccentric cylindrical regions with Gaussian inclusions with 50 and 25 μm average scatterer diameter (ASD) values was analyzed. The use of multiple view data reduced the corresponding ASD standard deviations from 13.7 and 19.6 μm to 2.5 and 3.7 μm, respectively. In experimental measurements, a phantom that contained glass spheres with diameters between 45 and 53 μm was evaluated. The ASD standard deviation obtained using single view data was 10.4 μm. When using 32 angles of view and reducing both region‐of‐interest (ROI) dimensions by a factor of 2, the ASD standard deviation was reduced to 4.8 μm. The results presented here demonstrate that both QUS spatial resolution and precision can be improved simultaneously by using full angular spatial compounding. This method finds direct applicability for breast tissue characterization, for which full angular coverage is available.

Omni-Directional Shape Measurement of a Small Three-Dimensional (3-D) Metallic Object with Reflective Surface Based on Shape-from-Focus

An imaging system was build using a telecentric optics in image space and a CCD image sensor that was movable along the optical axis in order to keep a perfect perspective projection during focusing. It was successfully applied to establish omni-directional measurement of a drill bit as an example of shiny metallic objects based on the shape-from-focus method. Negative data from glare regions were replaced with overlapping positive data obtained from neighboring viewpoints. Usage of a table that precisely rotates the object simplified registration of the multiple-view data. Finally, a fine mesh connecting surface points (corresponding pixels in images) can be generated to produce a three-dimensional graphic model with texture.

MSDmotif: exploring protein sites and motifs

BackgroundProtein structures have conserved features – motifs, which have a sufficient influence on the protein function. These motifs can be found in sequence as well as in 3D space. Understanding of these fragments is essential for 3D structure prediction, modelling and drug-design. The Protein Data Bank (PDB) is the source of this information however present search tools have limited 3D options to integrate protein sequence with its 3D structure.ResultsWe describe here a web application for querying the PDB for ligands, binding sites, small 3D structural and sequence motifs and the underlying database. Novel algorithms for chemical fragments, 3D motifs, ϕ/ψ sequences, super-secondary structure motifs and for small 3D structural motif associations searches are incorporated. The interface provides functionality for visualization, search criteria creation, sequence and 3D multiple alignment options. MSDmotif is an integrated system where a results page is also a search form. A set of motif statistics is available for analysis. This set includes molecule and motif binding statistics, distribution of motif sequences, occurrence of an amino-acid within a motif, correlation of amino-acids side-chain charges within a motif and Ramachandran plots for each residue. The binding statistics are presented in association with properties that include a ligand fragment library. Access is also provided through the distributed Annotation System (DAS) protocol. An additional entry point facilitates XML requests with XML responses.ConclusionMSDmotif is unique by combining chemical, sequence and 3D data in a single search engine with a range of search and visualisation options. It provides multiple views of data found in the PDB archive for exploring protein structures.

Propagation of engineering changes to multiple product data views using history of product structure changes

The present paper proposes a comprehensive procedure for engineering change propagation in order to maintain consistency between various product data views. A product data model is also proposed for the propagation procedure, which integrates base product definitions for product design, and product data views for other manufacturing or customer support. The product data view in the proposed model enables manufacturing or customer support engineers to define their own product data views, without copying the existing product definition. Integrated with other components, the engineering changes in the data model provide structure-oriented change history, effectivity management for production, and integration with product configurations. Based on the integrated product data model, the proposed procedure propagates engineering changes to product data views using the history of product structure changes. The propagation procedure maintains consistency of product data during collaboration between various design, manufacturing and customer support departments, who have different product data views. Prototype database applications together with an example of illustrative change propagation are also presented.

レンジ画像による３次元都市空間データの自動計測方法に関するシミュレーション

In this paper, we described an approach for 3D object reconstruction using simulated laser range finder (LRF) data. For the purpose of automatic registration of multiple views of LRF data without apriori knowledge of the views, such as location and direction, planar faces are extracted from the LRF data, and by automatically identifying the corresponding pairs of planar faces, the spatial relationships (the rotation matrixes and the shift vectors) between the views are estimated. To extract planar faces reliably from erroneous LRF data, the MDL (minimum description length) principle is used. In registrating multiple views, two steps-local and global registration-are used to find transformations to avoid the accumulation of registration errors between all the views.

Extracting spectral albedo from NOAA-9 AVHRR multiple view data using an atmospheric correction procedure and an expert system

A 100 km by 200 km area of the Sahel was selected to demonstrate a new method of extracting hemispherical reflectance (albedo) from directional satellite data. Utilizing a stratification of the Gao region of Mali based on soil texture, satellite data for homogeneous areas were selected. These data were employed in a knowledge-based expert system called VEG, which is designed to handle a wide variety of types of input data. The calculated directional reflectance factors at ground level are very similar to ground measurements noted in the literature.

Logical composition of object-oriented interfaces

This paper describes an approach to object-oriented interface design that goes beyond mere object decomposition. In our user interface management system we use logic and filters to declaratively specify and control a space of ways that objects may be composed to create interfaces. A filter is a package of constraints and associated typed objects that express the relationship of data and representation objects. Conceptually our system is completely based on constraints. Filters provide the high bandwidth constraints to maintain the components of the direct-manipulation interface while the logic forms the low bandwidth constraints to combine and provide communication between these components. The use of Horn-clause logic to compose separate interface objects facilitates both the distribution of computation onto multiple processors and the generation of multiple views of data. Intelligent backtracking implemented in the logic allows for user- and system-initiated undo operations to correct errors and/or try alternative approaches to a problem. We illustrate the power and flexibility of this approach by describing a floor layout and design system.

Data base design methodology for structural analysis and design optimization

A methodology to design data bases for finite element analysis and structural design optimization is described. The methodology considers three views of data organization-conceptual, internal, and external. Tabular and matrix forms of data are included. The relational data model is used in the data base design. Entity, relation, and attributes are considered to form a conceptual view of data. First, second, and third normal forms of data are suggested to design an internal model. Several aspects such as processing, iterative needs, multiple views of data, efficiency of storage and access time, and transitive data are considered in the methodology.