Papers by Κωνσταντίνος Διαμαντάρας
Greek named entity recognition using support vector machines, maximum entropy and onetime
2006 14th European Signal Processing Conference, 2006
In this paper we present a novel second-order statistics method for the blind deconvolution of a ... more In this paper we present a novel second-order statistics method for the blind deconvolution of a real signal propagating through a complex channel. The method is computationally very efficient since it involves only one SVD computation for the time-delayed output covariance matrix. No other optimization is involved. The subspaces corresponding to the left and right singular value matrices can be used to “deflate” the channel: projecting the output signal on either subspace reduces the filter length to 1, thus source reconstruction is simplified. The method is suitable for any channel length and it offers performance improvement compared to well established methods.
European Journal of Applied Mathematics, 2017
This paper proposes a neural network architecture for solving systems of non-linear equations. A ... more This paper proposes a neural network architecture for solving systems of non-linear equations. A back propagation algorithm is applied to solve the problem, using an adaptive learning rate procedure, based on the minimization of the mean squared error function defined by the system, as well as the network activation function, which can be linear or non-linear. The results obtained are compared with some of the standard global optimization techniques that are used for solving non-linear equations systems. The method was tested with some well-known and difficult applications (such as Gauss–Legendre 2-point formula for numerical integration, chemical equilibrium application, kinematic application, neuropsychology application, combustion application and interval arithmetic benchmark) in order to evaluate the performance of the new approach. Empirical results reveal that the proposed method is characterized by fast convergence and is able to deal with high-dimensional equations systems.
A novel rigid object segmentation method based on multiresolution 3-D motion and luminance analysis
Proceedings 2000 International Conference on Image Processing (Cat. No.00CH37101), 2000
We present a new image sequence segmentation method which combines both spatial and temporal info... more We present a new image sequence segmentation method which combines both spatial and temporal information in a multiresolution framework. A region growing technique in a multiresolution scheme outputs an over-segmented partition of the image scene. Pure temporal information is collected for each region using a feature extraction/feature tracking technique. Motion information is further processed in a pyramidal robust motion estimation
Computers in Cardiology 1996
Analyzing data into quantized components
2014 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2014
ABSTRACT Signals in various applications are often generated by linear combinations of quantized ... more ABSTRACT Signals in various applications are often generated by linear combinations of quantized components. The analysis of data into such components is treated here as a matrix analysis problem. We first show that the component alphabet can always be normalized to the levels 0, ..., M-1, without loss of generality. Then we study certain conditions under which the decomposition is possible. In particular, we present an analytical algorithm based on the differences of the observed points and the recursive estimation of the quantized components when the number of unique observed points is sufficiently large.
IEEE International Conference on Image Processing 2005, 2005
A new method for the blind separation of linear image mixtures is presented in this paper. Such m... more A new method for the blind separation of linear image mixtures is presented in this paper. Such mixtures often occur, when, for example, we photograph a scene through a semireflecting medium (windshield or glass). The proposed method requires two mixtures of two scenes captured under different illumination conditions. We show that the boundary values of the ratio of the two mixtures can lead to an accurate estimation of the separation matrix. The technique is very simple, fast, and reliable, as it does not depend on iterative procedures. The method effectiveness is tested on both artificially mixed images and real images.
been shown to solve the instantaneous BSS (Blind Source Separation) problem for temporally colore... more been shown to solve the instantaneous BSS (Blind Source Separation) problem for temporally colored sources. In t h i s paper we show that arbitrary temporal filtering combined w i t h models associated to the extension of standard P C A known as Oriented P C A ( O P C A ) provide a solution to the problem that is based on second order statistics and requires no prewhitening of the observation signals. Furthermore the issue of the optimal t e m p o r a l filter is addressed for filters of length 2 a n d 3 although t h e design of the universally optimal filter is still an o p e n question. Earlier neural O P C A networks are used to demonstrate the validity of t,he m e t h o d on artificially generated datasets.
Signal Processing: Image Communication, 2000
Data quantization is an essential step before digitization, i.e. prior to representing real numbe... more Data quantization is an essential step before digitization, i.e. prior to representing real numbers by bits for further digital processing. In this paper we show how a statistically non-optimal quantizer (e.g. a uniform quantizer) can be improved by a simple scaling operation before reconstructing the original value. The scale factor depends on the statistics of the input and it is not constant, except for the case of the optimal Lloyd}Max quantizer where it is always equal to 1. However in other cases, for example in the common uniform quantizer we report improvement of the order of 0.1}0.6 db depending on the bit-rate. The proposed method can be used in any application where uniform (or other non-optimal) quantizers are used. In particular, it can be used for the quantization scheme of the JPEG image coding standard, and elsewhere.
Support Vector Machines Content-Based Video Retrieval based solely on Motion Information
... Markos Zampoglou, Theophilos Papadimitriou, Member, IEEE, and Konstantinos I. Diamantaras, Me... more ... Markos Zampoglou, Theophilos Papadimitriou, Member, IEEE, and Konstantinos I. Diamantaras, Member IEEE ... Inspired by their work, we decided to apply the PMES video shot descriptor to large-scale content-based video classification, for the purpose of retrieval. ...
Separating two binary sources from a single nonlinear mixture
In this paper we present a novel blind method for separating two binary sources from a single, ar... more In this paper we present a novel blind method for separating two binary sources from a single, arbitrary nonlinear mixture. The method is analytical and does not involve nonlinear optimization. Our approach proceeds by linearizing the problem and extending known, clustering-based results from the linear binary BSS case to the nonlinear case. The proposed algorithm is computationally efficient. Due to
Channel Shortening of Multi-Input Multi-Output Convolutive Systems with Binary Sources
ABSTRACT In this paper we treat the problem of channel shortening of convolutive multi-input mult... more ABSTRACT In this paper we treat the problem of channel shortening of convolutive multi-input multi-output (MIMO) systems with binary source signals. Our approach exploits the constellation properties of the distant successor values. In the noiseless case the Magged successor of each output prototype form a characteristic finite set of clusters (lagged successor constellation). The shape of this constellation is invariant of the predecessor value and it only depends on the I last filter taps. Consequently, the localization of the Magged successors constellation can lead to the annihilation of I filter taps - a process we call channel shortening. Next we show that if the observation dataset is rich enough, the proposed method can transform the convolutive MIMO system into an instantaneous one, which is a much simpler system for blind separation. Furthermore, the treatment of the system in the presence of noise is described using data clustering techniques.
Blind Multichannel Deconvolution Using Subspace-Based Single Delay Channel Deflation
In this contribution we present a new subspace based technique for the separation of convolutive ... more In this contribution we present a new subspace based technique for the separation of convolutive complex mixtures. We show that using the SVD analysis of the delayed covariance mixture matrix, we can construct a subspace projector. The mixture signals when projected in that subspace lose their memory, yielding a set of instantaneous mixtures of the same sources. The instantaneous mixture
Histogram Based Blind Identification and Source Separation from Linear Instantaneous Mixtures
The paper presents a new geometric method for the blind identification of linear instantaneous MI... more The paper presents a new geometric method for the blind identification of linear instantaneous MIMO systems driven by multi-level inputs. The number of outputs may be greater than, equal to, or even less than the number of sources. The sources are then extracted using the identified system parameters. Our approach is based on the fact that the distribution of the distances between the cluster centers of the observed data cloud reveals the mixing vectors in a simple way. In the noiseless case the method is deterministic, non-iterative and fast: it suffices to calculate the histogram of these distances. In the noisy case, the core algorithm must be combined with efficient clustering methods in order to yield satisfactory results for various SNR levels.
Neurocomputing, 2009
Principal component analysis is often thought of as a preprocessing step for blind source separat... more Principal component analysis is often thought of as a preprocessing step for blind source separation (BSS). Although second order methods have been proposed for BSS in the past, these approaches cannot be easily implemented by neural models. In this paper we demonstrate that PCA is more than a preprocessing step and, in fact, it can be used directly for solving the BSS problem in combination with very simple temporal filtering process. We also demonstrate that a PCA extension called oriented PCA (OPCA) can be also used for the same purpose without prewhitening the observed data. Both approaches can be implemented using efficient neural models that are shown to successfully extract the hidden sources.
Journal of Signal Processing Systems, 2008
As the quantity of publicly available multimedia material becomes larger and larger, automatic in... more As the quantity of publicly available multimedia material becomes larger and larger, automatic indexing becomes increasingly important in accessing multimedia databases. In this paper, a novel set of low-level descriptors is presented for the aim of content-based video classification. Concerning temporal features, we use a modified PMES descriptor for the spatial distribution of local motion and a Dominant Direction Histogram we have developed to represent the temporal distribution of camera motion. Concerning color, we present the Weighted Color Histogram we have designed in order to model color distribution. The histogram models the H parameter of the HSV color space, and we combine it with weighted means for the S and V parameters. For the selection of key-frames from which to extract the spatial descriptors we use a modified version of a simple efficient method. We then proceed to evaluate our descriptor set on a database of video shots resulting from the temporal segmentation of the archive of a real-world TV station. Results demonstrate that our approach can achieve high success rates on a wide range of semantic classes.
International Journal of Medical Informatics, 1998
The most widely used signal in clinical practice is the ECG. ECG conveys information regarding th... more The most widely used signal in clinical practice is the ECG. ECG conveys information regarding the electrical function of the heart, by altering the shape of its constituent waves, namely the P, QRS, and T waves. Thus, the required tasks of ECG processing are the reliable recognition of these waves, and the accurate measurement of clinically important parameters measured from the temporal distribution of the ECG constituent waves. In this paper, we shall review some current trends on ECG pattern recognition. In particular, we shall review non-linear transformations of the ECG, the use of principal component analysis (linear and non-linear), ways to map the transformed data into n-dimensional spaces, and the use of neural networks (NN) based techniques for ECG pattern recognition and classification. The problems we shall deal with are the QRS/PVC recognition and classification, the recognition of ischemic beats and episodes, and the detection of atrial fibrillation. Finally, a generalised approach to the classification problems in n-dimensional spaces will be presented using among others NN, radial basis function networks (RBFN) and non-linear principal component analysis (NLPCA) techniques. The performance measures of the sensitivity and specificity of these algorithms will also be presented using as training and testing data sets from the MIT-BIH and the European ST-T databases.
IEEE Transactions on Signal Processing, 2008
We present a novel method for the blind identification of linear, single-input multiple-output (S... more We present a novel method for the blind identification of linear, single-input multiple-output (SIMO) finiteimpulse-response (FIR) systems, based on second-order statistics. Our approach, called the truncated transfer matrix method (TTM) proceeds in two major steps: first, the SVD analysis of the lagged covariance matrix gives the subspace of the clipped system transfer matrix and second, the block-Toeplitz structure of the transfer matrix gives extra constraints that allow us to reconstruct the matrix through the solution of a linear system of equations. The proposed TTM method is analytical (no optimization procedure involved), and it is robust to noise. We find that the method comes with an increased computational cost but it significantly outperforms state of the art second-order methods in low signal-to-noise ratio (SNR) situations.
Multi-Input Single-Output Nonlinear Blind Separation of Binary Sources
IEEE Transactions on Signal Processing, 2013
ABSTRACT The problem of blindly separating multiple binary sources from a single nonlinear mixtur... more ABSTRACT The problem of blindly separating multiple binary sources from a single nonlinear mixture is addressed through a novel clustering approach without the use of any optimization procedure. The method is based on the assumption that the source probabilities are asymmetric in which case the output probability distribution can be expressed as a linear mixture of the sources. We are then able to solve the problem by using a known linear Multiple-Input Single-Output (MISO) blind separation method. The overall procedure is very fast and, in theory, it works for any number of independent binary sources and for a wide range of nonlinear functions. In practice, the accuracy of the method depends on the estimation accuracy of the output probabilities and the cluster centers. It can be quite sensitive to noise especially as the number of sources increases or the number of data samples is reduced. However, in our experiments we have been able to demonstrate successful separation of up to four sources.
IEEE Transactions on Circuits and Systems for Video Technology, 2000
The estimation of rigid-body 3-D motion parameters using point correspondences from a pair of ima... more The estimation of rigid-body 3-D motion parameters using point correspondences from a pair of images under perspective projection is, typically, very sensitive to noise. In this paper, we present a novel robust method combining two approaches: 1) the SVD analysis of a linear operator resulting from the feature points and the displacement vectors and 2) a modified version of the well-known weighted least-squares method proposed by Huber in the context of robust statistics. We give a detailed rank analysis of the involved linear operator and study the effects of noise. We also propose a robust method guided by the structure of this operator, using weighted least squares and data partitioning. The method has been tested on artificial data and on real image sequences showing a remarkable robustness, even in the presence of up to 50% outliers in the data set.
Uploads
Papers by Κωνσταντίνος Διαμαντάρας