Face tracking with automatic model construction

This paper describes an active model with a robust texture model built on-line. The model uses one camera and it is able to operate without active illumination. The texture model is defined by a series of clusters, which are built in a video sequence using previously encountered samples. This model is used to search for the corresponding element in the following frames. An on-line clustering method, named leaderP is described and evaluated on an application of face tracking. A 20-point shape model is used. This model is built offline, and a robust fitting function is used to restrict the position of the points. Our proposal is to serve as one of the stages in a driver monitoring system. To test it, a new set of sequences of drivers recorded outdoors and in a realistic simulator has been compiled. Experimental results for typical outdoor driving scenarios, with frequent head movement, turns and occlusions are presented. Our approach is tested and compared with the Simultaneous Modeling and Tracking (SMAT) [1], and the recently presented Stacked Trimmed Active Shape Model (STASM) [2], and shows better results than SMAT and similar fitting error levels to STASM, with much faster execution times and improved robustness.     

The Changing Face of International Cryptography Policy: Part 16 — Export Control Regulation in the United States

In my September and November contributions, I deviated slightly from my usual theme and examined several of the technical and procedural matters associated with Digital Signature technology. Of course, Digital Signature represents one of the less politically sensitive applications of cryptographic technology, although it is undoubtedly destined to become one of the most pervasive.     

Computer-aided development of a real-time program

The refinement calculus is a well-established theory for formal development of imperative program code and is supported by a number of automated tools. Via a detailed case study, this article shows how refinement theory and tool support can be extended for a program with real-time constraints. The approach adapts a timed variant of the refinement calculus and makes corresponding enhancements to a theorem-prover based refinement tool.     

A comparison of the perceptual benefits of linear perspective and physically-based illumination for display of dense 3D streamtubes

Large datasets typically contain coarse features comprised of finer sub-features. Even if the shapes of the small structures are evident in a 3D display, the aggregate shapes they suggest may not be easily inferred. From previous studies in shape perception, the evidence has not been clear whether physically-based illumination confers any advantage over local illumination for understanding scenes that arise in visualization of large data sets that contain features at two distinct scales. In this paper we show that physically-based illumination can improve the perception for some static scenes of complex 3D geometry from flow fields. We perform human-subjects experiments to quantify the effect of physically-based illumination on participant performance for two tasks: selecting the closer of two streamtubes from a field of tubes, and identifying the shape of the domain of a flow field over different densities of tubes. We find that physically-based illumination influences participant performance as strongly as perspective projection, suggesting that physically-based illumination is indeed a strong cue to the layout of complex scenes. We also find that increasing the density of tubes for the shape identification task improved participant performance under physically-based illumination but not under the traditional hardware-accelerated illumination model.     

Scientific sketching for collaborative VR visualization design

We present four studies investigating tools and methodologies for artist-scientist-technologist collaboration in designing multivariate, virtual reality (VR) visualizations. Design study 1 identifies the promise of 3D drawing-style interfaces for VR design and also establishes limitations of these tools with respect to precision and support for animation. Design study 2 explores animating artist-created visualization designs with scientific 3D fluid flow data. While results captured an accurate sense of flow that was advantageous as compared to the results of study 1, the potential for visual exploration using the design tools tested was limited. Design study 3 reveals the importance of a new 3D interface that overcomes the precision limitation found in study 1 while remaining accessible to artist collaborators. Drawing upon previous results, design study 4 engages collaborative teams in a design process that begins with traditional paper sketching and moves to animated, interactive, VR prototypes "sketched" by designers in VR using interactive 3D tools. Conclusions from these four studies identify important characteristics of effective artist-accessible VR visualization design tools and lead to a proposed formalized methodology for successful collaborative design that we expect to be useful in guiding future collaborations. We call this proposed methodology Scientific Sketching.     

High precision electrostatic potential calculations for cylindrically symmetric lenses

A method is developed for a potential calculation within cylindrically symmetric electrostatic lenses using mesh relaxation techniques, and it is capable of considerably higher accuracies than currently available. The method involves (i) creating very high order algorithms (orders of 6, 8, and 10) for determining the potentials at points in the net using surrounding point values, (ii) eliminating the effect of the large errors caused by singular points, and (iii) reducing gradients in the high gradient regions of the geometry, thereby allowing the algorithms used in these regions to achieve greater precisions-(ii) and (iii) achieved by the use of telescopic multiregions. In addition, an algorithm for points one unit from a metal surface is developed, allowing general mesh point algorithms to be used in these situations, thereby taking advantage of the enhanced precision of the latter. A maximum error function dependent on a sixth order gradient of the potential is defined. With this the single point algorithmic errors are able to be viewed over the entire net. Finally, it is demonstrated that by utilizing the above concepts and procedures, the potential of a point in a reasonably high gradient region of a test geometry can realize a precision of less than 10(-10).     

Polarized light imaging of white matter architecture

Polarized light imaging (PLI) is a method to image fiber orientation in gross histological brain sections based on the birefringent properties of the myelin sheaths. The method uses the transmission of polarized light to quantitatively estimate the fiber orientation and inclination angles at every point of the imaged section. Multiple sections can be assembled into a 3D volume, from which the 3D extent of fiber tracts can be extracted. This article describes the physical principles of PLI and describes two major applications of the method: the imaging of white matter orientation of the rat brain and the generation of fiber orientation maps of the human brain in white and gray matter. The strengths and weaknesses of the method are set out.     

Some Statistical-Estimation Methods for Stochastic Finite-State Transducers

Formal translations constitute a suitable framework for dealing with many problems in pattern recognition and computational linguistics. The application of formal transducers to these areas requires a stochastic extension for dealing with noisy, distorted patterns with high variability. In this paper, some estimation criteria are proposed and developed for the parameter estimation of regular syntax-directed translation schemata. These criteria are: maximum likelihood estimation, minimum conditional entropy estimation and conditional maximum likelihood estimation. The last two criteria were proposed in order to deal with situations when training data is sparse. These criteria take into account the possibility of ambiguity in the translations: i.e., there can be different output strings for a single input string. In this case, the final goal of the stochastic framework is to find the highest probability translation of a given input string. These criteria were tested on a translation task which has a high degree of ambiguity.     

Ernest Lepore and Zenon Pylyshyn,What is Cognitive Science?

Minds and Machines -