Mitsubishi Electric (France)

We present a new algorithm to detect pedestrian in still images utilizing covariance matrices as object descriptors. Since the descriptors do not form a vector space, well known machine learning techniques are not well suited to learn the classifiers. The space of d-dimensional nonsingular covariance matrices can be represented as a connected Riemannian manifold. The main contribution of the paper is a novel approach for classifying points lying on a connected Riemannian manifold using the geometry of the space. The algorithm is tested on INRIA and DaimlerChrysler pedestrian datasets where superior detection rates are observed over the previous approaches.

Video, like images, should support content aware resizing. We present video retargeting using an improved seam carving operator. Instead of removing 1D seams from 2D images we remove 2D seam manifolds from 3D space-time volumes. To achieve this we replace the dynamic programming method of seam carving with graph cuts that are suitable for 3D volumes. In the new formulation, a seam is given by a minimal cut in the graph and we show how to construct a graph such that the resulting cut is a valid seam. That is, the cut is monotonic and connected. In addition, we present a novel energy criterion that improves the visual quality of the retargeted images and videos. The original seam carving operator is focused on removing seams with the least amount of energy, ignoring energy that is introduced into the images and video by applying the operator. To counter this, the new criterion is looking forward in time - removing seams that introduce the least amount of energy into the retargeted result. We show how to encode the improved criterion into graph cuts (for images and video) as well as dynamic programming (for images). We apply our technique to images and videos and present results of various applications.

Adaptively Sampled Distance Fields (ADFs) are a unifying representation of shape that integrate numerous concepts in computer graphics including the representation of geometry and volume data and a broad range of processing operations such as rendering, sculpting, level-of-detail management, surface offsetting, collision detection, and color gamut correction. Its structure is uncomplicated and direct, but is especially effective for quality reconstruction of complex shapes, e.g., artistic and organic forms, precision parts, volumes, high order functions, and fractals. We characterize one implementation of ADFs, illustrating its utility on two diverse applications: 1) artistic carving of fine detail, and 2) representing and rendering volume data and volumetric effects. Other applications are briefly presented.

We describe a new paradigm for three-dimensional computer graphics, using projectors to graphically animate physical objects in the real world. The idea is to replace a physical object— with its inherent color, texture, and material properties—with a neutral object and projected imagery, reproducing the original (or alternative) appearance directly on the object. Because the approach is to effectively “lift” the visual properties of the object into the projector, we call the projectors shader lamps. We address the central issue of complete and continuous illumination of non-trivial physical objects using multiple projectors and present a set of new techniques that makes the process of illumination practical. We demonstrate the viability of these techniques through a variety of table-top applications, and describe preliminary results to reproduce life-sized virtual spaces.

Construction toys are a superb medium for geometric models. We argue that such toys, suitably instrumented or sensed, could be the inspiration for a new generation of easy-to-use, tangible modeling systems—especially if the tangible modeling is combined with graphical-interpretation techniques for enhancing nascent models automatically. The three key technologies needed to realize this idea are embedded computation, vision-based acquisition, and graphical interpretation. We sample these technologies in the context of two novel modeling systems: physical building blocks that self-describe, interpret, and decorate the structures into which they are assembled; and a system for scanning, interpreting, and animating clay figures.

Soft output detection for signals transmitted on linear channels is investigated. A particular emphasis is made for signal detection on multiple antenna channels. The a posteriori information at the detector output is evaluated from a shifted spherical list of point candidates. The spherical list is centered on the maximum likelihood point, which has the great advantage of stabilizing the list size. Thus, the sphere radius is selected in order to control the list size and to cope with the boundaries of the finite multiple antenna constellation. Our new soft output sphere decoder is then applied to the computation of constrained channel capacity and to the iterative detection of a coded transmission. For example, we achieved a signal-to-noise ratio at 1.25 dB from capacity limit on a 4/spl times/4 MIMO channel with 16-QAM modulation and a 4-state rate 1/2 parallel turbo code.

We present a class of image fusion techniques to automatically combine images of a scene captured under different illumination. Beyond providing digital tools for artists for creating surrealist images and videos, the methods can also be used for practical applications. For example, the non-realistic appearance can be used to enhance the context of nighttime traffic videos so that they are easier to understand. The context is automatically captured from a fixed camera and inserted from a day-time image (of the same scene). Our approach is based on a gradient domain technique that preserves important local perceptual cues while avoiding traditional problems such as aliasing, ghosting and haloing. We presents several results in generating surrealistic videos and in increasing the information density of low quality nighttime videos.

Video, like images, should support content aware resizing. We present video retargeting using an improved seam carving operator. Instead of removing 1D seams from 2D images we remove 2D seam manifolds from 3D space-time volumes. To achieve this we replace the dynamic programming method of seam carving with graph cuts that are suitable for 3D volumes. In the new formulation, a seam is given by a minimal cut in the graph and we show how to construct a graph such that the resulting cut is a valid seam. That is, the cut is monotonic and connected. In addition, we present a novel energy criterion that improves the visual quality of the retargeted images and videos. The original seam carving operator is focused on removing seams with the least amount of energy, ignoring energy that is introduced into the images and video by applying the operator. To counter this, the new criterion is looking forward in time - removing seams that introduce the least amount of energy into the retargeted result. We show how to encode the improved criterion into graph cuts (for images and video) as well as dynamic programming (for images). We apply our technique to images and videos and present results of various applications.

The purpose of this paper is to provide an operative way of achieving high data rates for impulse radio (IR) transmission based systems. Since applications targeted for ultra wide band (UWB) are liable to be low-cost, we especially focus on simple transceiver design. To that effect, we present an original demodulation scheme adapted to a multi-band ON-OFF keying modulation. From the receiver point of view, we impose relaxed channel estimation constraints and derive suitable signal processing schemes and simple hardware architectures. Indeed, we only consider the benefit from a limited a priori channel knowledge: approximative delay spread and energy level. The associated optimum demodulation turns out to be a non-trivial energetic threshold comparison whose precise theoretical computation admits an analytical solution proving its feasibility. Numerical results are eventually performed for IEEE 802.15.3a channel models and FCC requirements; they obviously demonstrate the potential of these techniques.

Capacitors usually determine the overall lifetime of power converters since they are mainly responsible for breakdowns. Their failure results from the deterioration of their dielectric, the production of gases, and, eventually, their explosion. This process leads to an increase in the capacitor equivalent series resistance (ESR) and a decrease in its capacitance value for both electrolytic and metalized polypropylene film (MPPF) capacitors. In this paper, a novel noninvasive technique for capacitor diagnostic in Boost converters is presented. It can easily be applied online and even in real time, and it is suitable for the operation in both continuous current mode (CCM) and discontinuous current mode (DCM). The technique is based on the double estimations of the ESR and the capacitance, improving the diagnostic reliability. This way, predictive maintenance is provided, and it is possible to alarm for capacitor replacement, avoiding downtime. As the method is intended for railway high-power applications, it has been conceived neither to add any additional hardware in the power stage nor to even slightly modify it. To demonstrate and validate the effectiveness and accuracy of the proposed technique, several simulations and experimental results are discussed in a small prototype. In this prototype, the software for real-time estimation is programmed in a low-cost digital signal processor (DSP).

Abstract Dogma holds that plasma cells, as opposed to B cells, cannot bind antigen because they have switched from expression of membrane-bound immunoglobulins (Ig) that constitute the B-cell receptor (BCR) to production of the secreted form of immunoglobulins. Here we compare the phenotypical and functional attributes of plasma cells generated by the T-cell-dependent and T-cell-independent forms of the hapten NP. We show that the nature of the secreted Ig isotype, rather than the chemical structure of the immunizing antigen, defines two functionally distinct populations of plasma cells. Fully mature IgM-expressing plasma cells resident in the bone marrow retain expression of a functional BCR, whereas their IgG + counterparts do not. Antigen boost modifies the gene expression profile of IgM + plasma cells and initiates a cytokine production program, characterized by upregulation of CCL5 and IL-10. Our results demonstrate that IgM-expressing plasma cells can sense antigen and acquire competence for cytokine production upon antigenic challenge.

The traditional high-level algorithms for rigid body simulation work well for moderate numbers of bodies but scale poorly to systems of hundreds or more moving, interacting bodies. The problem is unnecessary synchronization implicit in these methods. Jefferson's timewarp algorithm [22] is a technique for alleviating this problem in parallel discrete event simulation. Rigid body dynamics, though a continuous process, exhibits many aspects of a discrete one. With modification, the timewarp algorithm can be used in a uniprocessor rigid body simulator to give substantial performance improvements for simulations with large numbers of bodies. This paper describes the limitations of the traditional high-level simulation algorithms, introduces Jefferson's algorithm, and extends and optimizes it for the rigid body case. It addresses issues particular to rigid body simulation, such as collision detection and contact group management, and describes how to incorporate these into the timewarp framework. Quantitative experimental results indicate that the timewarp algorithm offers significant performance improvements over traditional high-level rigid body simulation algorithms, when applied to systems with hundreds of bodies. It also helps pave the way to parallel implementations, as the paper discusses.

Some apparently powerful algorithms for automatic label placement on maps use heuristics that capture considerable cartographic expertise but are hampered by provably inefficient methods of search and optimization. On the other hand, no approach to label placement that is based on an efficient optimization technique has been applied to the production of general cartographic maps—those with labelled point, line, and area features—and shown to generate labellings of acceptable quality. We present an algorithm for label placement that achieves the twin goals of practical efficiency and high labelling quality by combining simple cartographic heuristics with effective stochastic optimization techniques.

Special features such as ridges, valleys and silhouettes, of a polygonal scene are usually displayed by explicitly identifying and then rendering `edges' for the corresponding geometry. The candidate edges are identified using the connectivity information, which requires preprocessing of the data. We present a non-obvious but surprisingly simple to implement technique to render such features without connectivity information or preprocessing. At the hardware level, based only on the vertices of a given flat polygon, we introduce new polygons, with appropriate color, shape and orientation, so that they eventually appear as special features.

In this paper, we present an original work on sub-wavelength optical switching performed over a coherent multi-band orthogonal frequency-division multiplexing (MB-OFDM) super-channel operating at 100 Gbps. After having demonstrated that dual-polarization MB-OFDM (DP-MB-OFDM) is as efficient as single-carrier dual-polarization quaternary phase shift keying (DP-QPSK) technology to transport 100 Gbps data-rate over a 10 × 100-km G.652 fiber-based transmission line, we show that optical add-drop of OFDM sub-bands as narrow as 8 GHz inside a 100 Gbps DP-MB-OFDM signal constituted of four sub-bands is feasible in the middle of this 1000-km transmission line. The flexible optical add-drop multiplexer (FOADM) implemented here is constituted by the association of an ultra-narrow pass-band and stop-band optical filter. The design and realization of such ultra-selective optical filters is presented, while the impact of their physical features over the quality of transmission is discussed. To prove that several add-drop multiplexers can be cascaded, our FOADM is introduced into a G.652 fiber-based recirculating loop and the impact of the cumulated filtering transfer function as well as the crosstalk inside the OADM are investigated. A typical use case for the introduction of such FOADM into long-haul transport networks is given, and the capital expenditure (CAPEX) cost advantage for the multi-layer transport networks is highlighted. By the proof of concept delivered here, combination of super-channel and sub-wavelength optical switching pushes network flexibility far away of what is today proposed by system vendors, opening new horizons for an optimized use of multi-layer transport networks.

The debate on OFDM vs. single-carrier (SC) transmission started back in the 1980s at the time of the European Digital Audio Broadcasting (DAB) and Digital Video Broadcasting (DVB) projects. The same questions arose later in wireless communications, and OFDM transmission with TDMA was adopted in the IEEE 802.11a specifications for wireless local area networks and by the WiMAX Forum for fixed WiMAX systems. Later, orthogonal frequency-division multiple access (OFDMA) was adopted by the WiMAX Forum for mobile WiMAX systems and more recently by the 3GPP for the downlink of Long Term Evolution (LTE) systems. In contrast, single-carrier FDMA was adopted for the uplink of LTE. In this overview paper, we will review these historic developments and give some recent results on OFDMA and Single-Carrier FDMA.

In this paper, a lifetime model for bond wire contacts of insulated gate bipolar transistors (IGBT) power modules is reported. This model is based on power cycling tests obtained under accelerated conditions, and a finite-element model taking into account the electrical, thermal, and mechanical coupling. It allows us to estimate the bond wire lifetime for a large scale of junction temperature swing amplitudes (ΔT <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">j</sub> ) and stress durations (t <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">ON</sub> ). To build it, a numerical design of experiment was performed in both high and low stress values (ΔT <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">j</sub> ). Then, a strain-life curve has been constructed where the average strain values on a defined volume around the contact areas between top-metallization and the most exposed bond wires to fatigue and liftoff have been used. As a result, it has been shown that the total strain is linearly dependent with ΔT <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">j</sub> and power law dependent with t <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">ON</sub> . The combination of the strain-life relation and the strain dependency with stress parameters leads to the lifetime relationship. The obtained lifetime model has been satisfactorily validated with some additional experimental points obtained from literature and with a large range of values for t <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">ON</sub> . This methodology can be easily replicated to other structures and is quite generic.

As hardware designs get increasingly complex and time-to-market constraints get tighter there is strong motivation for high-level synthesis (HLS). HLS must efficiently handle both dataflow-dominated and controlflow-dominated designs as well as designs of a mixed nature. In the past efficient tools for the former type have been developed but so far HLS of conditional behaviors lags behind. To bridge this gap an efficient scheduling heuristic for conditional behaviors is presented. Our heuristic and the techniques it utilizes are based on a unifying design representation appropriate for both types of behavioral descriptions, enabling the proposed heuristic to exploit under the same framework several well-established techniques (chaining, multicycling) as well as conditional resource sharing and speculative execution which are essential in efficiently scheduling conditional behaviors. Preliminary experiments confirm the effectiveness of our approach and prompted the development of the CODESIS HLS tool for further experimentation.

When performed using an exhaustive search, the maximum likelihood (ML) joint detection of all users in a multicarrier code-division multiple-access (MC-CDMA) system has a prohibitive complexity, growing exponentially with the number of users and the number of bits in each modulation symbol. In this paper, a novel ML multiuser detection algorithm is proposed, the complexity of which is a polynomial function of the number of users and is independent of the modulation size. The MC-CDMA system is modeled as a sphere packing lattice and a low-complexity optimum lattice decoder, the sphere decoder, is applied to jointly detect all users. Suboptimum simplifications, based on the orthogonal projection of the received signal on a facet of the lattice constellation, are also proposed to further decrease the complexity. Simulation results are shown with up to 64 users transmitting 16-quadrature-amplitude modulation symbols.

We present a class of image fusion techniques to automatically combine images of a scene captured under different illumination. Beyond providing digital tools for artists for creating surrealist images and videos, the methods can also be used for practical applications. For example, the non-realistic appearance can be used to enhance the context of nighttime traffic videos so that they are easier to understand. The context is automatically captured from a fixed camera and inserted from a day-time image (of the same scene). Our approach is based on a gradient domain technique that preserves important local perceptual cues while avoiding traditional problems such as aliasing, ghosting and haloing. We presents several results in generating surrealistic videos and in increasing the information density of low quality nighttime videos.