Pub Date : 1996-09-01DOI: 10.1109/DSPWS.1996.555538
G. Moustakides
The recursive least squares (RLS) algorithm is one of the most well known algorithms used for adaptive filtering and system identification. We consider the convergence properties of the forgetting factor RLS algorithm in a stationary data environment. We study the dependence of the speed of convergence of RLS with respect to the initialization of the input sample covariance matrix and with respect to the observation noise level. By obtaining estimates of the settling time we show that RLS, in a high SNR environment, when initialized with a matrix of small norm, has a very fast convergence. The convergence speed decreases as we increase the norm of the initialization matrix. In a medium SNR environment the optimum convergence speed of the algorithm is reduced, but the RLS becomes more insensitive to initialization. Finally in a low SNR environment it is preferable to start the algorithm with a matrix of large norm.
{"title":"Performance of the forgetting factor RLS during the transient phase","authors":"G. Moustakides","doi":"10.1109/DSPWS.1996.555538","DOIUrl":"https://doi.org/10.1109/DSPWS.1996.555538","url":null,"abstract":"The recursive least squares (RLS) algorithm is one of the most well known algorithms used for adaptive filtering and system identification. We consider the convergence properties of the forgetting factor RLS algorithm in a stationary data environment. We study the dependence of the speed of convergence of RLS with respect to the initialization of the input sample covariance matrix and with respect to the observation noise level. By obtaining estimates of the settling time we show that RLS, in a high SNR environment, when initialized with a matrix of small norm, has a very fast convergence. The convergence speed decreases as we increase the norm of the initialization matrix. In a medium SNR environment the optimum convergence speed of the algorithm is reduced, but the RLS becomes more insensitive to initialization. Finally in a low SNR environment it is preferable to start the algorithm with a matrix of large norm.","PeriodicalId":131323,"journal":{"name":"1996 IEEE Digital Signal Processing Workshop Proceedings","volume":"123 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1996-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124191166","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}
Pub Date : 1996-09-01DOI: 10.1109/DSPWS.1996.555561
M. Prandini, M. Campi, R. Leonardi
This paper deals with the problem of recovering the input signal applied to a linear time-invariant system from the measures of its output and the a-priori knowledge of the input statistics (blind equalization). Under the assumption of an i.i.d. non-Gaussian input sequence, a new iterative procedure based on phase sensitive high-order cumulants for adjusting the coefficients of a transversal equalizer is introduced. The main feature of the proposed technique is that it realizes the automatic selection of the equalization delay so as to improve the equalization performance.
{"title":"A new algorithm for the automatic search of the best delay in blind equalization","authors":"M. Prandini, M. Campi, R. Leonardi","doi":"10.1109/DSPWS.1996.555561","DOIUrl":"https://doi.org/10.1109/DSPWS.1996.555561","url":null,"abstract":"This paper deals with the problem of recovering the input signal applied to a linear time-invariant system from the measures of its output and the a-priori knowledge of the input statistics (blind equalization). Under the assumption of an i.i.d. non-Gaussian input sequence, a new iterative procedure based on phase sensitive high-order cumulants for adjusting the coefficients of a transversal equalizer is introduced. The main feature of the proposed technique is that it realizes the automatic selection of the equalization delay so as to improve the equalization performance.","PeriodicalId":131323,"journal":{"name":"1996 IEEE Digital Signal Processing Workshop Proceedings","volume":"578 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1996-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124973640","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}
Pub Date : 1996-09-01DOI: 10.1109/DSPWS.1996.555483
D. Darlington, Douglas R. Campbell
An adaptive noise cancellation scheme for speech processing is described, in which adaptive filters operate in frequency-limited sub-bands. Previously the filters had been distributed in a linear fashion in the frequency domain. This work investigates the effects of spacing the filters more in sympathy with the signal power and spectral characteristics. It is found that improvements in signal-to-noise ratio of processed noisy speech signals may be obtained in certain cases when the sub-bands are spaced according to a published cochlear function.
{"title":"The effect of modified filter distribution on an adaptive, sub-band speech enhancement method","authors":"D. Darlington, Douglas R. Campbell","doi":"10.1109/DSPWS.1996.555483","DOIUrl":"https://doi.org/10.1109/DSPWS.1996.555483","url":null,"abstract":"An adaptive noise cancellation scheme for speech processing is described, in which adaptive filters operate in frequency-limited sub-bands. Previously the filters had been distributed in a linear fashion in the frequency domain. This work investigates the effects of spacing the filters more in sympathy with the signal power and spectral characteristics. It is found that improvements in signal-to-noise ratio of processed noisy speech signals may be obtained in certain cases when the sub-bands are spaced according to a published cochlear function.","PeriodicalId":131323,"journal":{"name":"1996 IEEE Digital Signal Processing Workshop Proceedings","volume":"35 2","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1996-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131810451","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}
Pub Date : 1996-09-01DOI: 10.1109/DSPWS.1996.555496
M. A. Shcherbakov
A parallel-serial implementation for an adaptive nonlinear Volterra filter, which exploits the symmetric properties of kernels in the frequency domain, is presented. The realization is based on the successive generation of the m-th order frequency domain of the kernel definition in terms of lower order domains. An efficient implementation structure of the frequency-domain adaptive filters is presented using a specific array of identical processor elements. A great degree of modularity and regularity ensures suitability of the proposed architecture for fast implementation using VLSI technologies.
{"title":"A parallel architecture for adaptive frequency-domain Volterra filtering","authors":"M. A. Shcherbakov","doi":"10.1109/DSPWS.1996.555496","DOIUrl":"https://doi.org/10.1109/DSPWS.1996.555496","url":null,"abstract":"A parallel-serial implementation for an adaptive nonlinear Volterra filter, which exploits the symmetric properties of kernels in the frequency domain, is presented. The realization is based on the successive generation of the m-th order frequency domain of the kernel definition in terms of lower order domains. An efficient implementation structure of the frequency-domain adaptive filters is presented using a specific array of identical processor elements. A great degree of modularity and regularity ensures suitability of the proposed architecture for fast implementation using VLSI technologies.","PeriodicalId":131323,"journal":{"name":"1996 IEEE Digital Signal Processing Workshop Proceedings","volume":"11 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1996-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116460945","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}
Pub Date : 1996-09-01DOI: 10.1109/DSPWS.1996.555537
K. Mio, E. Moisan
We address the problem of noise cancellation with bilinear filters. These filters have infinite memory and degree. It is shown how to modify the equation error bilinear filter to use it in the context of noise cancellation. This truncated filter has good properties of convergence but theoretically yield biased results. These results are compared with that of output error bilinear filters (which are unbiased but may converge to local minima). Both of them prove much better than linear noise cancellation in a sonar real case study.
{"title":"Bilinear noise cancellation","authors":"K. Mio, E. Moisan","doi":"10.1109/DSPWS.1996.555537","DOIUrl":"https://doi.org/10.1109/DSPWS.1996.555537","url":null,"abstract":"We address the problem of noise cancellation with bilinear filters. These filters have infinite memory and degree. It is shown how to modify the equation error bilinear filter to use it in the context of noise cancellation. This truncated filter has good properties of convergence but theoretically yield biased results. These results are compared with that of output error bilinear filters (which are unbiased but may converge to local minima). Both of them prove much better than linear noise cancellation in a sonar real case study.","PeriodicalId":131323,"journal":{"name":"1996 IEEE Digital Signal Processing Workshop Proceedings","volume":"75 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1996-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132759830","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}
Pub Date : 1996-09-01DOI: 10.1109/DSPWS.1996.555491
C. Dumontier, F. Luthon, J. Charras
The main concern in image processing is computation cost. Markov random fields (MRF) based algorithms particularly require a significant computation cost. Most of implementations of this kind of algorithms are made on parallel machines. This paper investigates an original solution for real time implementation of a robust MRF-based motion detection algorithm. A PC board, based on a pipeline architecture using a single powerfull DSP and FPGA components, is developed. The algorithm and the board are described. A processing rate of 15 images per second is achieved, showing the validity of this approach.
{"title":"Real time implementation of an MRF-based motion detection algorithm on a DSP board","authors":"C. Dumontier, F. Luthon, J. Charras","doi":"10.1109/DSPWS.1996.555491","DOIUrl":"https://doi.org/10.1109/DSPWS.1996.555491","url":null,"abstract":"The main concern in image processing is computation cost. Markov random fields (MRF) based algorithms particularly require a significant computation cost. Most of implementations of this kind of algorithms are made on parallel machines. This paper investigates an original solution for real time implementation of a robust MRF-based motion detection algorithm. A PC board, based on a pipeline architecture using a single powerfull DSP and FPGA components, is developed. The algorithm and the board are described. A processing rate of 15 images per second is achieved, showing the validity of this approach.","PeriodicalId":131323,"journal":{"name":"1996 IEEE Digital Signal Processing Workshop Proceedings","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1996-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130989532","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}
Pub Date : 1996-09-01DOI: 10.1109/DSPWS.1996.555535
F. Lombardini, P. Lombardo
Conventional SAR interferometers derive the surface height from an estimate of the phase difference between the SAR processed echoes received by two displaced phase centers. This paper introduces a maximum likelihood algorithm to process the SAR data from an array of K phase centers. The accuracy of the new technique is derived and compared to the conventional one. It is shown that this new approach to SAR interferometry provides better accuracy, together with adaptivity to the look angle and reduced phase ambiguity.
{"title":"Maximum likelihood array SAR interferometry","authors":"F. Lombardini, P. Lombardo","doi":"10.1109/DSPWS.1996.555535","DOIUrl":"https://doi.org/10.1109/DSPWS.1996.555535","url":null,"abstract":"Conventional SAR interferometers derive the surface height from an estimate of the phase difference between the SAR processed echoes received by two displaced phase centers. This paper introduces a maximum likelihood algorithm to process the SAR data from an array of K phase centers. The accuracy of the new technique is derived and compared to the conventional one. It is shown that this new approach to SAR interferometry provides better accuracy, together with adaptivity to the look angle and reduced phase ambiguity.","PeriodicalId":131323,"journal":{"name":"1996 IEEE Digital Signal Processing Workshop Proceedings","volume":"14 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1996-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133040764","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}
Pub Date : 1996-09-01DOI: 10.1109/DSPWS.1996.555505
O. Ikeda
A CW laser array imaging system designed to obtain precise images of the object, where the eigenvector of the largest eigenvalue is derived from the array data collected and is then computationally beamsteared, has been presented by the author. However, only two-dimensional imaging of the object is achievable with the system. Both the data acquisition system and the algorithm are modified, aimed at achieving three-dimensional imaging. A target is placed near the apertures as part of the object, and multiple eigenvectors with the largest eigenvalues are derived to estimate, through a Fourier analysis, the three-dimensional positions of the representative object points. First, the spatial heterodyne detection process is described, second, the signal processing is explained, and, third, some results of computer simulation are given.
{"title":"Three-dimensional imaging of the object using multiple eigenvectors in the CW laser array imaging system","authors":"O. Ikeda","doi":"10.1109/DSPWS.1996.555505","DOIUrl":"https://doi.org/10.1109/DSPWS.1996.555505","url":null,"abstract":"A CW laser array imaging system designed to obtain precise images of the object, where the eigenvector of the largest eigenvalue is derived from the array data collected and is then computationally beamsteared, has been presented by the author. However, only two-dimensional imaging of the object is achievable with the system. Both the data acquisition system and the algorithm are modified, aimed at achieving three-dimensional imaging. A target is placed near the apertures as part of the object, and multiple eigenvectors with the largest eigenvalues are derived to estimate, through a Fourier analysis, the three-dimensional positions of the representative object points. First, the spatial heterodyne detection process is described, second, the signal processing is explained, and, third, some results of computer simulation are given.","PeriodicalId":131323,"journal":{"name":"1996 IEEE Digital Signal Processing Workshop Proceedings","volume":"12 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1996-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114296297","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}
Pub Date : 1996-09-01DOI: 10.1109/DSPWS.1996.555456
B. Zhu, M. D. Swanson, A. Tewfik
We propose a novel scheme to embed an invisible signature into an image to check image integrity and measure its distortion. The technique is based on the pseudo-noise sequences and visual masking effects. The values of an image are modified by a pseudo-noise signature which is shaped by the perceptual thresholds from masking effects. The method is robust and can gauge errors accurately up to half of the perceptual thresholds. It also readily identifies large image distortion. Experimental results after applying JPEG and white noise to the image are also reported.
{"title":"Transparent robust authentication and distortion measurement technique for images","authors":"B. Zhu, M. D. Swanson, A. Tewfik","doi":"10.1109/DSPWS.1996.555456","DOIUrl":"https://doi.org/10.1109/DSPWS.1996.555456","url":null,"abstract":"We propose a novel scheme to embed an invisible signature into an image to check image integrity and measure its distortion. The technique is based on the pseudo-noise sequences and visual masking effects. The values of an image are modified by a pseudo-noise signature which is shaped by the perceptual thresholds from masking effects. The method is robust and can gauge errors accurately up to half of the perceptual thresholds. It also readily identifies large image distortion. Experimental results after applying JPEG and white noise to the image are also reported.","PeriodicalId":131323,"journal":{"name":"1996 IEEE Digital Signal Processing Workshop Proceedings","volume":"143 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1996-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116280115","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}
Pub Date : 1996-09-01DOI: 10.1109/DSPWS.1996.555469
W. Zheng, Z. Quan
In MPEG, video compression is achieved by motion compensation, transformation, quantization and entropy coding. We follow the same path by using wavelet based modules in the coding chain. The prediction error signal is obtained by using variable block size motion compensation, and then, discrete wavelet transform (DWT) is applied to the prediction error signal to obtain wavelet coefficients. Some new schemes are introduced. First, a new and effective hybrid coder using quadtree and lattice vector quantization (QTLVQ) is introduced. Second, a new rate allocation scheme is introduced which can flexibly control the coder's output rate and the quality of reconstructed video signal.
{"title":"Image sequence coding using wavelet transform","authors":"W. Zheng, Z. Quan","doi":"10.1109/DSPWS.1996.555469","DOIUrl":"https://doi.org/10.1109/DSPWS.1996.555469","url":null,"abstract":"In MPEG, video compression is achieved by motion compensation, transformation, quantization and entropy coding. We follow the same path by using wavelet based modules in the coding chain. The prediction error signal is obtained by using variable block size motion compensation, and then, discrete wavelet transform (DWT) is applied to the prediction error signal to obtain wavelet coefficients. Some new schemes are introduced. First, a new and effective hybrid coder using quadtree and lattice vector quantization (QTLVQ) is introduced. Second, a new rate allocation scheme is introduced which can flexibly control the coder's output rate and the quality of reconstructed video signal.","PeriodicalId":131323,"journal":{"name":"1996 IEEE Digital Signal Processing Workshop Proceedings","volume":"151 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1996-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132127273","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：