Efficient tracking of moving objects using a relational database

Tracking uncooperative moving objects by means of radar is a complex task due to clutter and association problems in multi-target scenarios. An approach to solve this problem is probabilistic multiple hypothesis tracking (PMHT). This method combines classical track filtering with a likelihood ratio test for the estimation of the plot-to-track association. The basics of PMHT and similar algorithms have gained much attention recently. However, the efficient implementation of real world applications of this technique still represents a challenging task. Since a common requirement in this context is the reliable storage of track data in a database, an implementation of the tracker's calculation inside a database management system (DBMS) using SQL views is desirable. A naive implementation of PMHT using a commercial DBMS, however, usually leads to performance problems because of the high frequency of measurement updates. In this paper, we propose possible optimizations for solving these performance problems. Their usage leads to a dramatic run-time improvement in our sample case and makes the implementation of PMHT in a database context feasible.     

Online blind speech separation using multiple acoustic speaker tracking and time–frequency masking

Separating speech signals of multiple simultaneous talkers in a reverberant enclosure is known as the cocktail party problem. In real-time applications online solutions capable of separating the signals as they are observed are required in contrast to separating the signals offline after observation. Often a talker may move, which should also be considered by the separation system. This work proposes an online method for speaker detection, speaker direction tracking, and speech separation. The separation is based on multiple acoustic source tracking (MAST) using Bayesian filtering and time–frequency masking. Measurements from three room environments with varying amounts of reverberation using two different designs of microphone arrays are used to evaluate the capability of the method to separate up to four simultaneously active speakers. Separation of moving talkers is also considered. Results are compared to two reference methods: ideal binary masking (IBM) and oracle tracking (O-T). Simulations are used to evaluate the effect of number of microphones and their spacing.     

Personalising speech-to-speech translation: Unsupervised cross-lingual speaker adaptation for HMM-based speech synthesis

In this paper we present results of unsupervised cross-lingual speaker adaptation applied to text-to-speech synthesis. The application of our research is the personalisation of speech-to-speech translation in which we employ a HMM statistical framework for both speech recognition and synthesis. This framework provides a logical mechanism to adapt synthesised speech output to the voice of the user by way of speech recognition. In this work we present results of several different unsupervised and cross-lingual adaptation approaches as well as an end-to-end speaker adaptive speech-to-speech translation system. Our experiments show that we can successfully apply speaker adaptation in both unsupervised and cross-lingual scenarios and our proposed algorithms seem to generalise well for several language pairs. We also discuss important future directions including the need for better evaluation metrics.     

Determining the performance of website-based relationship marketing

Company websites are an important instrument for relationship marketing activities. We present a methodological framework that aligns website performance assessment and marketing intelligence for evaluating the performance of relationship marketing activities. In this context, we develop an extended web mining approach that integrates managerial perspectives in the analyst’s investigation of the customer-website interaction based on historical clickstream data.This approach enables quantification of the moderating effect of a website’s structure and content regarding website-based relationship marketing. The applicability of our approach is demonstrated by clickstream data of 477,471 visitor sessions on a software developer’s website. The results provide detailed insights into the usage behavior on a website and the mechanisms to enhance e-commerce efficiency via website optimization.     

Integrated FDD system for valve stiction in a paperboard machine

The performance of a modern industrial plant can be severely affected by the performance of its key devices, such as valves. In particular, valve stiction can cause poor performance in control loops and can consequently lower the efficiency of the plant and the quality of the product. This paper presents an integrated FDD system for valve stiction which employs various FDD methods in a parallel configuration. A reliability index was integrated into each method in order to estimate their degree of influence in the final diagnosis of the system. Each method and the integrated system were tested using industrial data.     

Parameter Control of Genetic Algorithms by Learning and Simulation of Bayesian Networks — A Case Study for the Optimal Ordering of Tables

Parameter setting for evolutionary algorithms is still an important issue in evolutionary computation. There are two main approaches to parameter setting: parameter tuning and parameter control. In this paper, we introduce self-adaptive parameter control of a genetic algorithm based on Bayesian network learning and simulation. The nodes of this Bayesian network are genetic algorithm parameters to be controlled. Its structure captures probabilistic conditional (in)dependence relationships between the parameters. They are learned from the best individuals, i.e., the best configurations of the genetic algorithm. Individuals are evaluated by running the genetic algorithm for the respective parameter configuration. Since all these runs are time-consuming tasks, each genetic algorithm uses a small-sized population and is stopped before convergence. In this way promising individuals should not be lost. Experiments with an optimal search problem for simultaneous row and column orderings yield the same optima as state-of-the-art methods but with a sharp reduction in computational time. Moreover, our approach can cope with as yet unsolved high-dimensional problems.     

A multi-dimensional histogram technique for cloud classification

An old method (often called graphical) for decomposing a mixed distribution to Gaussian components was generalized for the multidimensional case. The technique approximates an initial histogram by means of a sum of normally distributed components. A special separation algorithm for n-dimensional (nD) histograms was introduced for that purpose. Application of the separation method enables one to replace the customary pixel-by-pixel processing with a cluster-by-cluster procedure in any threshold algorithm. Stability of the algorithm was tested, comparing the decomposition of the radiation histograms produced by means of the Advanced Very High Resolution Radiometer (AVHRR) five channel measurements in the 3D and 5D cases for an area over Europe during nine orbits. The results show that multidimensional histograms are easily separable due to sufficiently large Euclidean distance between basic cloud and surface clusters in the measurement space. Applying the separation scheme in conjunction with a certain threshold technique to process the AVHRR-based histograms enables one to produce an automatic cloud detection algorithm. The algorithm sets necessary thresholds without auxiliary (i.e. beyond AVHRR) information and estimates average cloud amount, cloud top temperature and cloud reflectance at three levels for the histogram area. An example of such an algorithm for determining cloudiness parameters necessary for the Earth's radiation budget monitoring is presented.     

Correction of ozone influence on TOA radiance

An accurate atmospheric correction (AC) of Earth remote-sensing data in the spectral region 450–800 nm has to account for the ozone gas absorption influence. Usual operational AC codes employ a fixed ozone concentration corresponding to a climatologic average for a certain region and season, e.g. the mid-latitude summer atmosphere of the Moderate Resolution Atmospheric Transmission (MODTRAN) code. The reasons for a fixed ozone column are that ozone does not vary rapidly on a spatial and temporal scale, and additionally, the look-up table (LUT) size for AC is already big. This means that another degree of freedom for the ozone parameter would dramatically increase the size of the LUT database and the time required for LUT interpolation. In order to account for this effect, we use already existing LUTs that were calculated for a certain ozone reference level, e.g. an ozone column of g = 330 Dobson Units (DU) for MODTRAN’s mid-latitude summer atmosphere. Then the deviation of the top-of-atmosphere (TOA) radiance ΔL(g) from the reference level L(g = 330) is calculated as a function of solar and view geometries. The calculation is performed for a set of 36 wavelengths in the ozone-sensitive spectrum (450–800 nm) and five ozone columns. The last step computes the linear regression coefficients for each wavelength and geometry. The results are stored in a small table (11 kB). It is shown that the ozone influence is accurately accounted for by multiplying the modelled radiance L(g = 330) with a factor depending on g and wavelength yielding TOA radiance relative errors smaller than 0.5% for a wide range of ozone concentrations between 180 and 500 DU. Selected examples of a sensitivity study of the ozone effect on the retrieval of water constituents demonstrate the need to account for ozone in the AC step.