The Experts below are selected from a list of 10665 Experts worldwide ranked by ideXlab platform

Robert Krupinski - One of the best experts on this subject based on the ideXlab platform.

  • The Use of Generalized Gaussian Distribution in Vibroacoustic Detection of Power Transformer Core Damage
    Energies, 2020
    Co-Authors: Robert Krupinski, Eugeniusz Kornatowski
    Abstract:

    Vibroacoustic diagnostics (VM—Vibroacoustic Method) is one of the methods for diagnosing the active part of power transformers. Measurement technologies have been refined over the past several years, but the methods of analyzing data obtained in VM diagnostics are still in development. In most cases, they are based on a simple frequency spectrum analysis, and the diagnostic conclusions are subjective and depend on the expert’s professional experience. The article presents an objective method for the detection of transformer unit core damage, based on the analysis of the statistical properties of the vibration signal registered on the surface of the tank of an unloaded transformer in the steady state of vibrations (VM). The algorithm for proceeding further is: FFT analysis of the vibroacoustic signal, with the determination of the relative changes in vibration power as a function of frequency P r ( f ) and, finally, the determination of the statistic properties of the dataset P r ( f ) . The Generalized Gaussian Distribution (GGD) is used to describe the P r ( f ) set. The detector output values are the λ and p parameters of the GGD Distribution. These two numerical values form the basis for the classification of the technical condition of the transformer unit core. The correctness of the described solution was verified on the example of ten pieces of 16 MVA power transformers with different operating times and degrees of wear.

  • binarization of degraded document images with Generalized Gaussian Distribution
    International Conference on Computational Science, 2019
    Co-Authors: Robert Krupinski, Piotr Lech, Mateusz Teclaw, Krzysztof Okarma
    Abstract:

    One of the most crucial steps of preprocessing of document images subjected to further text recognition is their binarization, which influences significantly obtained OCR results. Since for degrades images, particularly historical documents, classical global and local thresholding methods may be inappropriate, a challenging task of their binarization is still up-to-date. In the paper a novel approach to the use of Generalized Gaussian Distribution for this purpose is presented. Assuming the presence of distortions, which may be modelled using the Gaussian noise Distribution, in historical document images, a significant similarity of their histograms to those obtained for binary images corrupted by Gaussian noise may be observed. Therefore, extracting the parameters of Generalized Gaussian Distribution, distortions may be modelled and removed, enhancing the quality of input data for further thresholding and text recognition. Due to relatively long processing time, its shortening using the Monte Carlo method is proposed as well. The presented algorithm has been verified using well-known DIBCO datasets leading to very promising binarization results.

  • ICCS (5) - Binarization of Degraded Document Images with Generalized Gaussian Distribution.
    Lecture Notes in Computer Science, 2019
    Co-Authors: Robert Krupinski, Piotr Lech, Mateusz Tecław, Krzysztof Okarma
    Abstract:

    One of the most crucial steps of preprocessing of document images subjected to further text recognition is their binarization, which influences significantly obtained OCR results. Since for degrades images, particularly historical documents, classical global and local thresholding methods may be inappropriate, a challenging task of their binarization is still up-to-date. In the paper a novel approach to the use of Generalized Gaussian Distribution for this purpose is presented. Assuming the presence of distortions, which may be modelled using the Gaussian noise Distribution, in historical document images, a significant similarity of their histograms to those obtained for binary images corrupted by Gaussian noise may be observed. Therefore, extracting the parameters of Generalized Gaussian Distribution, distortions may be modelled and removed, enhancing the quality of input data for further thresholding and text recognition. Due to relatively long processing time, its shortening using the Monte Carlo method is proposed as well. The presented algorithm has been verified using well-known DIBCO datasets leading to very promising binarization results.

  • Generating Augmented Quaternion Random Variable With Generalized Gaussian Distribution
    IEEE Access, 2018
    Co-Authors: Robert Krupinski
    Abstract:

    There is a need to generate quaternion valued processes with a given Distribution ranging from super-Gaussian, Gaussian to sub-Gaussian with different degrees of properness for numerous practical applications, such as modeling the signal source, testing nonlinear adaptive filter models, watermarking, generating $\mathbb {Q}$ -proper and $\mathbb {Q}$ -improper processes, the synthesis of color images, and testing the algorithms of step detection. Therefore, in this paper, the alternative quaternion Generalized Gaussian probability density function is derived for an augmented quaternion valued variable, and the procedure for generating the augmented quaternion valued random variables is defined for this Distribution. Additionally, the 3-D Generalized Gaussian Distribution probability density function parameterized by the shape parameter $p$ and the covariance matrix $C$ is revised.

  • modeling quantized coefficients with Generalized Gaussian Distribution with exponent 1 m m 2 3 ldots
    International Conference on Man–Machine Interactions, 2017
    Co-Authors: Robert Krupinski
    Abstract:

    The different types of signals in the image and signal processing applications can be modeled with Generalized Gaussian Distribution (GGD). When limiting to the special cases, then the closed form equations can be determined. The special cases with the exponents \(p=2\) (Gaussian Distribution), \(p=1\) (Laplacian Distribution), \(p=1/2\) and \(p=1/3\) are considered in literature. In the article, more general approach for the exponents 1 / m, \(m=2,3,\ldots \) is analyzed, which are related to the peaky shapes of GGD. The maximum likelihood method for a discrete random variable is derived for this subclass of Distributions.

Norman C. Beaulieu - One of the best experts on this subject based on the ideXlab platform.

Tomoya Takatani - One of the best experts on this subject based on the ideXlab platform.

  • speech kurtosis estimation from observed noisy signal based on Generalized Gaussian Distribution prior and additivity of cumulants
    International Conference on Acoustics Speech and Signal Processing, 2012
    Co-Authors: Ryo Wakisaka, Hiroshi Saruwatari, Kiyohiro Shikano, Tomoya Takatani
    Abstract:

    In this paper, we propose a new method for stable estimation of the kurtosis of a speech power spectrum. Speech kurtosis can be used for the prediction of speech recognition accuracy as reported in recent studies. However, the conventional estimation method is very unstable owing to the high sensitivity of higher-order statistics. To overcome this problem, we introduce the Generalized Gaussian Distribution prior in order to avoid the calculation of higher-order statistics, and construct a kurtosis table that directly represents the relationship among the kurtosis of speech, noise, and their mixture in the power spectrum domain. Speech kurtosis can be estimated stably from observable data by looking up values in the table. An experimental evaluation confirms the efficacy of the proposed method.

  • ICASSP - Speech kurtosis estimation from observed noisy signal based on Generalized Gaussian Distribution prior and additivity of cumulants
    2012 IEEE International Conference on Acoustics Speech and Signal Processing (ICASSP), 2012
    Co-Authors: Ryo Wakisaka, Hiroshi Saruwatari, Kiyohiro Shikano, Tomoya Takatani
    Abstract:

    In this paper, we propose a new method for stable estimation of the kurtosis of a speech power spectrum. Speech kurtosis can be used for the prediction of speech recognition accuracy as reported in recent studies. However, the conventional estimation method is very unstable owing to the high sensitivity of higher-order statistics. To overcome this problem, we introduce the Generalized Gaussian Distribution prior in order to avoid the calculation of higher-order statistics, and construct a kurtosis table that directly represents the relationship among the kurtosis of speech, noise, and their mixture in the power spectrum domain. Speech kurtosis can be estimated stably from observable data by looking up values in the table. An experimental evaluation confirms the efficacy of the proposed method.

Ryo Wakisaka - One of the best experts on this subject based on the ideXlab platform.

  • speech kurtosis estimation from observed noisy signal based on Generalized Gaussian Distribution prior and additivity of cumulants
    International Conference on Acoustics Speech and Signal Processing, 2012
    Co-Authors: Ryo Wakisaka, Hiroshi Saruwatari, Kiyohiro Shikano, Tomoya Takatani
    Abstract:

    In this paper, we propose a new method for stable estimation of the kurtosis of a speech power spectrum. Speech kurtosis can be used for the prediction of speech recognition accuracy as reported in recent studies. However, the conventional estimation method is very unstable owing to the high sensitivity of higher-order statistics. To overcome this problem, we introduce the Generalized Gaussian Distribution prior in order to avoid the calculation of higher-order statistics, and construct a kurtosis table that directly represents the relationship among the kurtosis of speech, noise, and their mixture in the power spectrum domain. Speech kurtosis can be estimated stably from observable data by looking up values in the table. An experimental evaluation confirms the efficacy of the proposed method.

  • ICASSP - Speech kurtosis estimation from observed noisy signal based on Generalized Gaussian Distribution prior and additivity of cumulants
    2012 IEEE International Conference on Acoustics Speech and Signal Processing (ICASSP), 2012
    Co-Authors: Ryo Wakisaka, Hiroshi Saruwatari, Kiyohiro Shikano, Tomoya Takatani
    Abstract:

    In this paper, we propose a new method for stable estimation of the kurtosis of a speech power spectrum. Speech kurtosis can be used for the prediction of speech recognition accuracy as reported in recent studies. However, the conventional estimation method is very unstable owing to the high sensitivity of higher-order statistics. To overcome this problem, we introduce the Generalized Gaussian Distribution prior in order to avoid the calculation of higher-order statistics, and construct a kurtosis table that directly represents the relationship among the kurtosis of speech, noise, and their mixture in the power spectrum domain. Speech kurtosis can be estimated stably from observable data by looking up values in the table. An experimental evaluation confirms the efficacy of the proposed method.

Hiroshi Saruwatari - One of the best experts on this subject based on the ideXlab platform.

  • Blind Speech Extraction Based on Rank-Constrained Spatial Covariance Matrix Estimation With Multivariate Generalized Gaussian Distribution
    IEEE ACM Transactions on Audio Speech and Language Processing, 2020
    Co-Authors: Yuki Kubo, Norihiro Takamune, Daichi Kitamura, Hiroshi Saruwatari
    Abstract:

    In this article, we propose a new blind speech extraction (BSE) method that robustly extracts a directional speech from background diffuse noise by combining independent low-rank matrix analysis (ILRMA) and efficient rank-constrained spatial covariance matrix (SCM) estimation. To achieve more accurate BSE than ILRMA, which assumes each source to be a point source (rank-1 spatial model), the proposed method restores the lost spatial basis for the full-rank SCM of diffuse noise. We adopt the multivariate complex Generalized Gaussian Distribution (GGD) as the statistical generative model to express various types of observed signal. To estimate the model parameters for an arbitrary shape parameter of the multivariate GGD, we derive a new inequality for rank-constrained SCMs. Also, we propose new acceleration methods to accomplish much faster extraction than conventional blind source separation methods. In BSE experiments using simulated and real recorded data, we confirm that the proposed method achieves more accurate and faster speech extraction than conventional methods.

  • speech kurtosis estimation from observed noisy signal based on Generalized Gaussian Distribution prior and additivity of cumulants
    International Conference on Acoustics Speech and Signal Processing, 2012
    Co-Authors: Ryo Wakisaka, Hiroshi Saruwatari, Kiyohiro Shikano, Tomoya Takatani
    Abstract:

    In this paper, we propose a new method for stable estimation of the kurtosis of a speech power spectrum. Speech kurtosis can be used for the prediction of speech recognition accuracy as reported in recent studies. However, the conventional estimation method is very unstable owing to the high sensitivity of higher-order statistics. To overcome this problem, we introduce the Generalized Gaussian Distribution prior in order to avoid the calculation of higher-order statistics, and construct a kurtosis table that directly represents the relationship among the kurtosis of speech, noise, and their mixture in the power spectrum domain. Speech kurtosis can be estimated stably from observable data by looking up values in the table. An experimental evaluation confirms the efficacy of the proposed method.

  • ICASSP - Speech kurtosis estimation from observed noisy signal based on Generalized Gaussian Distribution prior and additivity of cumulants
    2012 IEEE International Conference on Acoustics Speech and Signal Processing (ICASSP), 2012
    Co-Authors: Ryo Wakisaka, Hiroshi Saruwatari, Kiyohiro Shikano, Tomoya Takatani
    Abstract:

    In this paper, we propose a new method for stable estimation of the kurtosis of a speech power spectrum. Speech kurtosis can be used for the prediction of speech recognition accuracy as reported in recent studies. However, the conventional estimation method is very unstable owing to the high sensitivity of higher-order statistics. To overcome this problem, we introduce the Generalized Gaussian Distribution prior in order to avoid the calculation of higher-order statistics, and construct a kurtosis table that directly represents the relationship among the kurtosis of speech, noise, and their mixture in the power spectrum domain. Speech kurtosis can be estimated stably from observable data by looking up values in the table. An experimental evaluation confirms the efficacy of the proposed method.