This work is concerned with recursive procedures in which the data run through sequentially. Two stochastic approximation recursions derived from the EM (expectation-maximization).and SAGE (space-alternating generalized expectation-maximization). algorithms are proposed. We show that under regularity conditions, these recursions lead to strong consistency and asymptotic normality. Although the recursive EM and SAGE algorithm do not have the optimal convergence rate, they are usually easy to implement. As an example, we derive recursive procedures for direction of arrival (DOA) estimation. In numerical experiments both algorithms provide good results for low computational cost.
{"title":"Recursive EM and SAGE algorithms","authors":"Pei-Jung Chung, J. Bohme","doi":"10.1109/SSP.2001.955342","DOIUrl":"https://doi.org/10.1109/SSP.2001.955342","url":null,"abstract":"This work is concerned with recursive procedures in which the data run through sequentially. Two stochastic approximation recursions derived from the EM (expectation-maximization).and SAGE (space-alternating generalized expectation-maximization). algorithms are proposed. We show that under regularity conditions, these recursions lead to strong consistency and asymptotic normality. Although the recursive EM and SAGE algorithm do not have the optimal convergence rate, they are usually easy to implement. As an example, we derive recursive procedures for direction of arrival (DOA) estimation. In numerical experiments both algorithms provide good results for low computational cost.","PeriodicalId":70952,"journal":{"name":"信号处理","volume":"45 1","pages":"540-543"},"PeriodicalIF":0.0,"publicationDate":"2001-08-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"88920730","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
This paper describes a rate-distortion (R-D) optimal scheme for bit-plane based quantisation of complex coefficients, which is suitable for zerotree image coding systems. Most zerotree-type image codecs operate on real-valued wavelet coefficients. The dual-tree complex wavelet transform, which has several advantages over the discrete wavelet transform, produces complex coefficients. Our scheme offers progressive bit-by-bit refinement of coefficient magnitude and phase values. It ensures that refinement decisions always maximise the expected distortion decrease.
{"title":"R-D quantisation of complex coefficients in zerotree coding","authors":"T. H. Reeves, N. Kingsbury","doi":"10.1109/SSP.2001.955327","DOIUrl":"https://doi.org/10.1109/SSP.2001.955327","url":null,"abstract":"This paper describes a rate-distortion (R-D) optimal scheme for bit-plane based quantisation of complex coefficients, which is suitable for zerotree image coding systems. Most zerotree-type image codecs operate on real-valued wavelet coefficients. The dual-tree complex wavelet transform, which has several advantages over the discrete wavelet transform, produces complex coefficients. Our scheme offers progressive bit-by-bit refinement of coefficient magnitude and phase values. It ensures that refinement decisions always maximise the expected distortion decrease.","PeriodicalId":70952,"journal":{"name":"信号处理","volume":"57 1","pages":"480-483"},"PeriodicalIF":0.0,"publicationDate":"2001-08-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"89349555","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
A pipelined algorithm and parallel architecture is under development for real time detection of landmines. Our previous work has dealt with monochromatic images from airborne active infrared scanners and images from a low-altitude aircraft-mounted multi-spectral scanner. Because of the nature of the sensors and the aerial observation platform, the landmines were treated as small, sparse, discrete objects in a large clutter field. Our current work deals with passive infrared imagery obtained from cameras mounted on ground vehicle. In contrast to the previous work, although the targets are still relatively sparse, they are no longer small in the sense of occupying just a few pixels and the signal to noise ratio is considerably worse than in for the airborne active infrared and multi-spectral scanner problems. So significant changes to our detection algorithm are needed. The paper describes the overall algorithm and the particular issues, such as irregular shapes, that need to be dealt with in FLIR imagery. Some early results are presented. In addition, changes in computer processing power and interprocessor communications has led to a rethink of the real-time hardware implementations of the system and these issues are discussed.
{"title":"Towards real-time detection of landmines in FLIR imagery","authors":"M.R. Ito, Sinh Duong, J. McFee, K. Russell","doi":"10.1109/SSP.2001.955245","DOIUrl":"https://doi.org/10.1109/SSP.2001.955245","url":null,"abstract":"A pipelined algorithm and parallel architecture is under development for real time detection of landmines. Our previous work has dealt with monochromatic images from airborne active infrared scanners and images from a low-altitude aircraft-mounted multi-spectral scanner. Because of the nature of the sensors and the aerial observation platform, the landmines were treated as small, sparse, discrete objects in a large clutter field. Our current work deals with passive infrared imagery obtained from cameras mounted on ground vehicle. In contrast to the previous work, although the targets are still relatively sparse, they are no longer small in the sense of occupying just a few pixels and the signal to noise ratio is considerably worse than in for the airborne active infrared and multi-spectral scanner problems. So significant changes to our detection algorithm are needed. The paper describes the overall algorithm and the particular issues, such as irregular shapes, that need to be dealt with in FLIR imagery. Some early results are presented. In addition, changes in computer processing power and interprocessor communications has led to a rethink of the real-time hardware implementations of the system and these issues are discussed.","PeriodicalId":70952,"journal":{"name":"信号处理","volume":"36 1","pages":"154-157"},"PeriodicalIF":0.0,"publicationDate":"2001-08-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"81119889","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
This paper addresses the blind source separation (BSS) problem in the context of "heavy-tailed", or "impulsive" source signals, characterized by the nonexistence of finite second (or higher) order moments. We consider Pham's (1997) quasi-maximum likelihood (QML) approach, a modification of the maximum likelihood (ML) approach, applied using some presumed distributions of the sources. We introduce a related family of suboptimal estimators, termed restricted QML (RQML). A theoretical analysis of the asymptotic performance of RQML is presented. The analysis is used for showing that the variance of the optimal (non-RQML) estimator's error must decrease at a rate faster than 1/T (where T is the number of independent observations). This surprising property, sometimes called super-efficiency, has been observed before (in the BSS context) only for finite-support source distributions. Simulation results illustrate the good agreement with theory.
{"title":"Super-efficiency in blind signal separation of symmetric heavy-tailed sources","authors":"Yoav Shereshevski, A. Yeredor, H. Messer","doi":"10.1109/SSP.2001.955226","DOIUrl":"https://doi.org/10.1109/SSP.2001.955226","url":null,"abstract":"This paper addresses the blind source separation (BSS) problem in the context of \"heavy-tailed\", or \"impulsive\" source signals, characterized by the nonexistence of finite second (or higher) order moments. We consider Pham's (1997) quasi-maximum likelihood (QML) approach, a modification of the maximum likelihood (ML) approach, applied using some presumed distributions of the sources. We introduce a related family of suboptimal estimators, termed restricted QML (RQML). A theoretical analysis of the asymptotic performance of RQML is presented. The analysis is used for showing that the variance of the optimal (non-RQML) estimator's error must decrease at a rate faster than 1/T (where T is the number of independent observations). This surprising property, sometimes called super-efficiency, has been observed before (in the BSS context) only for finite-support source distributions. Simulation results illustrate the good agreement with theory.","PeriodicalId":70952,"journal":{"name":"信号处理","volume":"46 1","pages":"78-81"},"PeriodicalIF":0.0,"publicationDate":"2001-08-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"81127989","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Short-time spectral attenuation is a common form of audio signal enhancement in which a time-varying filter, or suppression rule, is applied to the frequency-domain transform of a corrupted signal. The suppression rule (see Ephraim, Y. and Malah, D., IEEE Trans. on Acoustics, Speech and Signal Proc., vol.ASSP-32, no.6, p.1109-21, 1984) for speech enhancement is both optimal in the minimum mean-square error sense and well-known for its associated colourless residual noise; however, it requires the computation of exponential and Bessel functions. We show that, under the same modelling assumptions, alternative Bayesian approaches lead to suppression rules exhibiting almost identical behaviour. We derive three such rules and show that they are efficient to implement and yield a more intuitive interpretation.
短时频谱衰减是音频信号增强的一种常见形式，其中时变滤波器或抑制规则应用于损坏信号的频域变换。抑制规则(见Ephraim, Y. and Malah, D.， IEEE译)。声学，语音与信号处理，vol. 32, no。6, p.1109- 21,1984)的语音增强在最小均方误差意义上是最优的，并且以其相关的无色残余噪声而闻名;然而，它需要计算指数函数和贝塞尔函数。我们表明，在相同的建模假设下，替代贝叶斯方法导致抑制规则表现出几乎相同的行为。我们推导了三个这样的规则，并表明它们是有效的实现和产生更直观的解释。
{"title":"Simple alternatives to the Ephraim and Malah suppression rule for speech enhancement","authors":"Patrick J. Wolfe, S. Godsill","doi":"10.1109/SSP.2001.955331","DOIUrl":"https://doi.org/10.1109/SSP.2001.955331","url":null,"abstract":"Short-time spectral attenuation is a common form of audio signal enhancement in which a time-varying filter, or suppression rule, is applied to the frequency-domain transform of a corrupted signal. The suppression rule (see Ephraim, Y. and Malah, D., IEEE Trans. on Acoustics, Speech and Signal Proc., vol.ASSP-32, no.6, p.1109-21, 1984) for speech enhancement is both optimal in the minimum mean-square error sense and well-known for its associated colourless residual noise; however, it requires the computation of exponential and Bessel functions. We show that, under the same modelling assumptions, alternative Bayesian approaches lead to suppression rules exhibiting almost identical behaviour. We derive three such rules and show that they are efficient to implement and yield a more intuitive interpretation.","PeriodicalId":70952,"journal":{"name":"信号处理","volume":"63 1","pages":"496-499"},"PeriodicalIF":0.0,"publicationDate":"2001-08-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"83213298","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
We describe methods for applying Monte Carlo filtering and smoothing for estimation of unobserved states in a nonlinear state space model. By exploiting the statistical structure of the model, we develop a Rao-Blackwellised particle smoother. The suggested algorithm is tested with real speech and audio data and the results are shown and compared with those generated using the generic particle smoother and the extended Kalman filter. It is found that the suggested algorithm gives better results.
{"title":"Monte Carlo smoothing with application to audio signal enhancement","authors":"W. Fong, S. Godsill, A. Doucet, M. West","doi":"10.1109/SSP.2001.955211","DOIUrl":"https://doi.org/10.1109/SSP.2001.955211","url":null,"abstract":"We describe methods for applying Monte Carlo filtering and smoothing for estimation of unobserved states in a nonlinear state space model. By exploiting the statistical structure of the model, we develop a Rao-Blackwellised particle smoother. The suggested algorithm is tested with real speech and audio data and the results are shown and compared with those generated using the generic particle smoother and the extended Kalman filter. It is found that the suggested algorithm gives better results.","PeriodicalId":70952,"journal":{"name":"信号处理","volume":"4 1","pages":"18-21"},"PeriodicalIF":0.0,"publicationDate":"2001-08-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"78181612","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
We propose a new algorithm for blind source separation (BSS), in which independent component analysis (ICA) and beamforming are combined to resolve the low-convergence problem through optimization in ICA. The proposed method consists of the following two parts: frequency-domain ICA with direction-of-arrival (DOA) estimation, and null beamforming based on the estimated DOA. The alternation of learning between ICA and beamforming can realize fast- and high-convergence optimization. The results of the signal separation experiments reveal that the signal separation performance of the proposed algorithm is superior to that of the conventional ICA-based BSS method.
{"title":"Fast-convergence algorithm for ICA-based blind source separation using array signal processing","authors":"H. Saruwatari, T. Kawamura, K. Shikano","doi":"10.1109/SSP.2001.955323","DOIUrl":"https://doi.org/10.1109/SSP.2001.955323","url":null,"abstract":"We propose a new algorithm for blind source separation (BSS), in which independent component analysis (ICA) and beamforming are combined to resolve the low-convergence problem through optimization in ICA. The proposed method consists of the following two parts: frequency-domain ICA with direction-of-arrival (DOA) estimation, and null beamforming based on the estimated DOA. The alternation of learning between ICA and beamforming can realize fast- and high-convergence optimization. The results of the signal separation experiments reveal that the signal separation performance of the proposed algorithm is superior to that of the conventional ICA-based BSS method.","PeriodicalId":70952,"journal":{"name":"信号处理","volume":"232 1","pages":"464-467"},"PeriodicalIF":0.0,"publicationDate":"2001-08-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"76217908","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
This work is concerned with extension techniques of finite signals for subband processing using tree-structured filter banks. In many applications it is desirable that the selected extension defines an orthogonal transform. Although it is clear that periodization solves this problem, some complications arise when using this technique: spurious high frequencies or artificial discontinuities appear in the transform vector. Considering AR processes as input signals, the solution of this problem is an algorithm for the generation of alternative orthogonal signal extensions which do not introduce artificial discontinuities in the subband signals. Experimental results that illustrate the effectiveness of the proposed design method are discussed briefly.
{"title":"Orthogonal extensions of AR processes without artificial discontinuities for size-limited filter banks","authors":"M. Jiménez, N. G. Prelcic","doi":"10.1109/SSP.2001.955353","DOIUrl":"https://doi.org/10.1109/SSP.2001.955353","url":null,"abstract":"This work is concerned with extension techniques of finite signals for subband processing using tree-structured filter banks. In many applications it is desirable that the selected extension defines an orthogonal transform. Although it is clear that periodization solves this problem, some complications arise when using this technique: spurious high frequencies or artificial discontinuities appear in the transform vector. Considering AR processes as input signals, the solution of this problem is an algorithm for the generation of alternative orthogonal signal extensions which do not introduce artificial discontinuities in the subband signals. Experimental results that illustrate the effectiveness of the proposed design method are discussed briefly.","PeriodicalId":70952,"journal":{"name":"信号处理","volume":"20 1","pages":"579-582"},"PeriodicalIF":0.0,"publicationDate":"2001-08-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"75793825","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
We present new algorithms that aim at estimating the "static" parameters of a latent variable process in an on-line manner. This new class of on-line algorithms is inspired by Monte Carlo Markov chain (MCMC) methods whose use has been mainly restricted to static problems, i.e., for which the set of observations is fixed. The main interest of this new class of algorithms is that it combines MCMC and particle filtering techniques, for which extensive know-how and literature are now available.
{"title":"Recursive Monte Carlo algorithms for parameter estimation in general state space models","authors":"C. Andrieu, A. Doucet","doi":"10.1109/SSP.2001.955210","DOIUrl":"https://doi.org/10.1109/SSP.2001.955210","url":null,"abstract":"We present new algorithms that aim at estimating the \"static\" parameters of a latent variable process in an on-line manner. This new class of on-line algorithms is inspired by Monte Carlo Markov chain (MCMC) methods whose use has been mainly restricted to static problems, i.e., for which the set of observations is fixed. The main interest of this new class of algorithms is that it combines MCMC and particle filtering techniques, for which extensive know-how and literature are now available.","PeriodicalId":70952,"journal":{"name":"信号处理","volume":"11 1","pages":"14-17"},"PeriodicalIF":0.0,"publicationDate":"2001-08-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"74690679","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Subband arrays have been proposed as a useful means to realize joint spatio-temporal domain equalization in digital mobile communications. They are used to mitigate channel impairment problems caused by inter-symbol interference (ISI) and co-channel interference (CCI). We propose normalized subband array and locally orthogonalized subband array techniques for channel equalization. The least square mean (LMS) algorithm is used for adaptation. The convergence performance of the proposed techniques is analyzed and compared with that of conventional space-time adaptive processing (STAP) techniques. It is shown that subband decompositions provide great flexibility in implementing spatio-temporal equalization. Both analytical and numerical simulation results demonstrate that the proposed subband array techniques substantially improve the convergence performance without significant additional computations.
{"title":"Convergence performance of subband arrays for spatio-temporal equalization","authors":"Yimin D. Zhang, Kehu Yang, M. Amin","doi":"10.1109/SSP.2001.955343","DOIUrl":"https://doi.org/10.1109/SSP.2001.955343","url":null,"abstract":"Subband arrays have been proposed as a useful means to realize joint spatio-temporal domain equalization in digital mobile communications. They are used to mitigate channel impairment problems caused by inter-symbol interference (ISI) and co-channel interference (CCI). We propose normalized subband array and locally orthogonalized subband array techniques for channel equalization. The least square mean (LMS) algorithm is used for adaptation. The convergence performance of the proposed techniques is analyzed and compared with that of conventional space-time adaptive processing (STAP) techniques. It is shown that subband decompositions provide great flexibility in implementing spatio-temporal equalization. Both analytical and numerical simulation results demonstrate that the proposed subband array techniques substantially improve the convergence performance without significant additional computations.","PeriodicalId":70952,"journal":{"name":"信号处理","volume":"12 1","pages":"544-547"},"PeriodicalIF":0.0,"publicationDate":"2001-08-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"85223146","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：