Extracting objects from range and radiance images

In this paper, we present a pipeline and several key techniques necessary for editing a real scene captured with both cameras and laser range scanners. We develop automatic algorithms to segment the geometry from range images into distinct surfaces, register texture from radiance images with the geometry, and synthesize compact high-quality texture maps. The result is an object-level representation of the scene which can be rendered with modifications to structure via traditional rendering methods. The segmentation algorithm for geometry operates directly on the point cloud from multiple registered 3D range images instead of a reconstructed mesh. It is a top-down algorithm which recursively partitions a point set into two subsets using a pairwise similarity measure. The result is a binary tree with individual surfaces as leaves. Our image registration technique performs a very efficient search to automatically find the camera poses for arbitrary position and orientation relative to the geometry. Thus, we can take photographs from any location without precalibration between the scanner and the camera. The algorithms have been applied to large-scale real data. We demonstrate our ability to edit a captured scene by moving, inserting, and deleting objects     

Simultaneous modelling of rock freezing and water seepage and its practical applications

All existing models of the rock freezing process are able to simulate the conductive heat transfer, but, however the convective terms are dominant in many cases, when water seepage also occurs. The origin of the majority of failures were caused by the effect of water seepage. A new integrated finite difference simulation technique has been developed to model rock freezing and water seepage simultaneously. The mathematical model, the base of numerical algorithm, the input and output data, the verification of the model using analytical solutions and a rather sophisticated practical case history are discussed in the paper. Finally, the first practical application of the model to support decisions on the proper system of freezing and on the proper combination with depressurization of water bearing layers is presented as well.     

Stability of linear time‐periodic delay‐differential equations via Chebyshev polynomials

This paper presents a new technique for studying the stability properties of dynamic systems modeled by delay‐differential equations (DDEs) with time‐periodic parameters. By employing a shifted Chebyshev polynomial approximation in each time interval with length equal to the delay and parametric excitation period, the dynamic system can be reduced to a set of linear difference equations for the Chebyshev expansion coefficients of the state vector in the previous and current intervals. This defines a linear map which is the ‘infinite‐dimensional Floquet transition matrix U’. Two different formulas for the computation of the approximate U, whose size is determined by the number of polynomials employed, are given. The first one uses the direct integral form of the original system in state space form while the second uses a convolution integral (variation of parameters) formulation. Additionally, a variation on the former method for direct application to second‐order systems is also shown. An error analysis is presented which allows the number of polynomials employed in the approximation to be selected in advance for a desired tolerance. An extension of the method to the case where the delay and parametric periods are commensurate is also shown. Stability charts are produced for several examples of time‐periodic DDEs, including the delayed Mathieu equation and a model for regenerative chatter in impedance‐modulated turning. The results indicate that this method is an effective way to study the stability of time‐periodic DDEs. Copyright © 2004 John Wiley & Sons, Ltd.     

Yield forecasting for wheat and corn in Hungary by satellite remote sensing

We have developed an advanced version of our yield estimation method [Ferencz et al., 2004 Cs, Ferencz, Bognár, P., Lichtenberger, J., Hamar, D., Gy, Tarcsai, Timár, G., Molnár, G., Sz, Pásztor, Steinbach, P., Székely, B., Ferencz, O.E. and Ferencz-Árkos, I. 2004. Crop yield estimation by satellite remote sensing. International Journal of Remote Sensing, 25: 4113–4149.  [Google Scholar], Crop yield estimation by satellite remote sensing. International Journal of Remote Sensing, 25, pp. 4113–4149], that is able to provide reliable forecasts for corn and wheat, several weeks before the harvest. The forecasting method is based on the data of the Advanced Very High Resolution Radiometer (AVHRR) instruments of the National Oceanic and Atmospheric Administration's (NOAA) Polar Orbiting Environmental Satellites (POES). The method was applied to Hungary between the years 1996 and 2000. The forecasted yield values are all within 5% reliability with respect to the actual yield data produced by classic (non-satellite based) methods and provided by the Hungarian Statistical Office, with the exception of 1997, where the absolute error is about 8%.     

Improved upper bounds on synchronizing nondeterministic automata

We show that i-directable nondeterministic automata can be i-directed with a word of length O(2ⁿ) for i=1,2, where n stands for the number of states. Since for i=1,2 there exist i-directable automata having i-directing words of length Ω(2ⁿ), these upper bounds are asymptotically optimal. We also show that a 3-directable nondeterministic automaton with n states can be 3-directed with a word of length , improving the previously known upper bound O(2ⁿ). Here the best known lower bound is .     

The Personal Interaction Panel – a Two-Handed Interface for Augmented Reality

This paper describes the introduction of a new interaction paradigm to augmented reality applications. The everyday tool handling experience of working with pen and notebooks is extended to create a three dimensional two-handed interface, that supports easy-to-understand manipulation tasks in augmented and virtual environments. In the design step we take advantage from the freedom, given by our very low demands on hardware and augment form and functionality to this device. On the basis of examples from object manipulation, augmented research environments and scientific visualization we show the generality of applicability. Although being in the first stages implementation, we consider the wide spectrum of suitability for different purposes.     

Semi-On-line Scheduling on Two Parallel Processors with an Upper Bound on the Items

We study a variant of the on-line scheduling problem on two parallel processors. The size of the items is unknown and, as soon as an item is released, it must be immediately assigned to a processor and the assignment cannot be changed later. Optimal algorithms (with respect to competitive ratio) are known for some variants of this problem, where some partial information is given on the instance: the sum of the items is known, or a buffer is available to store a finite number of items. In these cases the best possible competitive ratio of the algorithms is 4/3. In this paper we assume that the sum of items is known in advance (supposed to equal 2) and also that the size of items does not exceed a fixed upper bound < 1. We provide, for all the possible values of , a lower bound for the competitive ratio of any algorithm and propose different algorithms, for different ranges of the upper bound, for which a worst-case analysis is provided. The proposed algorithms are optimal for &frac; \le \le 3/5, =&frac; and 16/17 \le < 1.     

Module-Based Reinforcement Learning: Experiments with a Real Robot

The behavior of reinforcement learning (RL) algorithms is best understood in completely observable, discrete-time controlled Markov chains with finite state and action spaces. In contrast, robot-learning domains are inherently continuous both in time and space, and moreover are partially observable. Here we suggest a systematic approach to solve such problems in which the available qualitative and quantitative knowledge is used to reduce the complexity of learning task. The steps of the design process are to:i) decompose the task into subtasks using the qualitative knowledge at hand; ii) design local controllers to solve the subtasks using the available quantitative knowledge and iii) learn a coordination of these controllers by means of reinforcement learning. It is argued that the approach enables fast, semi-automatic, but still high quality robot-control as no fine-tuning of the local controllers is needed. The approach was verified on a non-trivial real-life robot task. Several RL algorithms were compared by ANOVA and it was found that the model-based approach worked significantly better than the model-free approach. The learnt switching strategy performed comparably to a handcrafted version. Moreover, the learnt strategy seemed to exploit certain properties of the environment which were not foreseen in advance, thus supporting the view that adaptive algorithms are advantageous to non-adaptive ones in complex environments.     

3D shape estimation in video sequences provides high precision evaluation of facial expressions

Person independent and pose invariant estimations of facial expressions and action unit (AU) intensity estimation are important for situation analysis and for automated video annotation. We evaluated raw 2D shape data of the CK+ database, used Procrustes transformation and the multi-class SVM leave-one-out method for classification. We found close to 100% performance demonstrating the relevance and the strength of details of the shape. Precise 3D shape information was computed by means of constrained local models (CLM) on video sequences. Such sequences offer the opportunity to compute a time-averaged ‘3D personal mean shape’ (PMS) from the estimated CLM shapes, which – upon subtraction – gives rise to person independent emotion estimation. On CK+ data PMS showed significant improvements over AU0 normalization; performance reached and sometimes surpassed state-of-the-art results on emotion classification and on AU intensity estimation. 3D PMS from 3D CLM offers pose invariant emotion estimation that we studied by rendering a 3D emotional database for different poses and different subjects from the BU 4DFE database. Frontal shapes derived from CLM fits of the 3D shape were evaluated. Results demonstrate that shape estimation alone can be used for robust, high quality pose invariant emotion classification and AU intensity estimation.