### Citations

1035 |
Optimal Filtering
- Anderson, Moore
- 1979
(Show Context)
Citation Context ...w (n)g is known. The Kalman filter order as given by the (d + 1 + q) \Theta (d + 1+ q) state matrix F may be reduced by a suitable transformation of the state-vector or of the observation signal y(n) =-=[2]-=-. Unfortunately, the reduced order Kalman filter needs a more complicated Kalman gain vector update. Thus, we decided to stick to the increased order Kalman filter. In most practical situations, the o... |

403 |
Digital Spectral Analysis with Applications
- Marple
- 1987
(Show Context)
Citation Context ...ique where the block length and overlapping is chosen in accordance to the nonstationarity of the AR signal. The Levinson algorithm is applied to overlapping signal blocks for AR parameter estimation =-=[5]-=-. On the contrary, method II employs a recursive least-squares lattice (LSL) algorithm [4]. The LSL algorithm operates on a sample per sample basis and in caseProc. IX European Signal Proc. Conf., EU... |

105 |
All-pole modeling of degraded speech
- Lim, Oppenheim
- 1987
(Show Context)
Citation Context ...e convergence speed at higher SNRs). The second approach is based on an iterated parameter estimation and is suited for colored noise disturbances [6]. It is an extension of an algorithm described in =-=[1]-=-. No suppression of impulsive noise has been included in the proposed algorithms. The AR parameters are estimated using an iterative procedure, i.e. filtering of a signal block and parameter estimatio... |

68 |
Filtering of colored noise for speech enhancement and coding
- Gibson, Koo, et al.
- 1991
(Show Context)
Citation Context ...sented in [8]. The system can also be easily modified to suppress impulse noise [9]. Our system does not require an extended Kalman filter as used in [7] or an iterated parameter estimation procedure =-=[6]-=-. It is computationally efficient and may be implemented for real-time operation in the audio frequency range using integrated digital signal processors. We first present a computationally efficient s... |

17 |
Adaptive scheme for elimination of broadband noise and impulsivedisturbances from AR and ARMA signals
- Niedzwiecki, Cisowski
- 1996
(Show Context)
Citation Context ...colored noise is an extension to the system we presented in [8]. The system can also be easily modified to suppress impulse noise [9]. Our system does not require an extended Kalman filter as used in =-=[7]-=- or an iterated parameter estimation procedure [6]. It is computationally efficient and may be implemented for real-time operation in the audio frequency range using integrated digital signal processo... |

7 |
Adaptive Filter Theory, chapter 9
- Haykin
- 1991
(Show Context)
Citation Context ... of the AR signal. The Levinson algorithm is applied to overlapping signal blocks for AR parameter estimation [5]. On the contrary, method II employs a recursive least-squares lattice (LSL) algorithm =-=[4]-=-. The LSL algorithm operates on a sample per sample basis and in caseProc. IX European Signal Proc. Conf., EUSIPCO-98, Vol. II, pp. 781-784, Sept. 8-11, 1998, Island of Rhodes, Greece 4 of an exponen... |

3 |
Adaptive filtering prediction and control", Chapter 7
- Goodwin
- 1984
(Show Context)
Citation Context ...rocedure. As a consequence, the estimation error variance is further decreased, and with a large lag length (delay) the performance approaches that of a noncausal Wiener filter for stationary signals =-=[3]-=-. In order to obtain a Kalman fixed-lag smoother we represent the AR signal model (2) in state-space form s(n + 1) = A(n)s(n) + \Gamma u(n); 0; : : : ; 0 \Delta T : (4) In contrast to (2), the d + 1 d... |

3 |
Adaptive filter theory", Chapter 9
- Haykin
- 1987
(Show Context)
Citation Context ... of the AR signal. The Levinson algorithm is applied to overlapping signal blocks for AR parameter estimation [5]. On the contrary, method II employs a recursive least-squares lattice (LSL) algorithm =-=[4]-=-. The LSL algorithm operates on a sample per sample basis and in case Proc. IX European Signal Proc. Conf., EUSIPCO-98, Vol. II, pp. 781-784, Sept. 8-11, 1998, Island of Rhodes, Greece 4 of an exponen... |

3 |
Jr., "Digital Spectral analysis with applications", Chapter 7
- Marple
- 1987
(Show Context)
Citation Context ...ique where the block length and overlapping is chosen in accordance to the nonstationarity of the AR signal. The Levinson algorithm is applied to overlapping signal blocks for AR parameter estimation =-=[5]-=-. On the contrary, method II employs a recursive least-squares lattice (LSL) algorithm [4]. The LSL algorithm operates on a sample per sample basis and in case Proc. IX European Signal Proc. Conf., EU... |

3 |
Adaptive filtering prediction and control”, Chapter 7
- Goodwin
- 1984
(Show Context)
Citation Context ...rocedure. As a consequence, the estimation error variance is further decreased, and with a large lag length (delay) the performance approaches that of a noncausal Wiener filter for stationary signals =-=[3]-=-. In order to obtain a Kalman fixed-lag smoother we represent the AR signal model (2) in state-space form ; T s(n +1)=A(n)s(n) + u(n)� 0�::: �0 : (4) In contrast to (2), the d +1dimensional state vect... |

2 | An adaptive Kalman filter for the enhancement of noisy AR signals
- Doblinger
- 1998
(Show Context)
Citation Context ...hich allows for tracking of nonstationary signals. Furthermore, our system is designed to suppress white or colored noise. The inclusion of colored noise is an extension to the system we presented in =-=[8]-=-. The system can also be easily modified to suppress impulse noise [9]. Our system does not require an extended Kalman filter as used in [7] or an iterated parameter estimation procedure [6]. It is co... |

1 | Adaptive Kalman Smoothing of AR Signals Disturbed by Impulses and Colored Noise
- Doblinger
(Show Context)
Citation Context ...stem is designed to suppress white or colored noise. The inclusion of colored noise is an extension to the system we presented in [8]. The system can also be easily modified to suppress impulse noise =-=[9]-=-. Our system does not require an extended Kalman filter as used in [7] or an iterated parameter estimation procedure [6]. It is computationally efficient and may be implemented for real-time operation... |