Pub Date : 1998-10-12DOI: 10.1109/ICOSP.1998.770804
W. Lie, H. Hsieh
This research discusses the problems of lips localization and feature point detection which have gained much attention in applications such as lip-reading, visual-assisted speech recognition, audio/video synchronization, etc. Our method is primarily based upon morphological processing of time-difference images which highlight lip motions. Experiments show that acceptable results can be obtained.
{"title":"Lips detection by morphological image processing","authors":"W. Lie, H. Hsieh","doi":"10.1109/ICOSP.1998.770804","DOIUrl":"https://doi.org/10.1109/ICOSP.1998.770804","url":null,"abstract":"This research discusses the problems of lips localization and feature point detection which have gained much attention in applications such as lip-reading, visual-assisted speech recognition, audio/video synchronization, etc. Our method is primarily based upon morphological processing of time-difference images which highlight lip motions. Experiments show that acceptable results can be obtained.","PeriodicalId":145700,"journal":{"name":"ICSP '98. 1998 Fourth International Conference on Signal Processing (Cat. No.98TH8344)","volume":"387 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1998-10-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126735582","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 : 1998-10-12DOI: 10.1109/ICOSP.1998.770227
B. Xu, T. B. Vu
This paper presents a blind MMSE near-far resistant receiver for a DS/CDMA system by combining reduced-length space-time MMSE (STMMSE) receiver with blind constant-modulus-aIgorithm (CMA) approaches. The reduced-length STMMSE receiver with training sequences is studied first. Two CMA approaches which are capable of combating both ICI and MAI without stringent power control are proposed to generate training sequences for the STMSE receiver.
{"title":"Blind MAI and CCI suppression using adaptive array antennas in DS/CDMA mobile communications systems","authors":"B. Xu, T. B. Vu","doi":"10.1109/ICOSP.1998.770227","DOIUrl":"https://doi.org/10.1109/ICOSP.1998.770227","url":null,"abstract":"This paper presents a blind MMSE near-far resistant receiver for a DS/CDMA system by combining reduced-length space-time MMSE (STMMSE) receiver with blind constant-modulus-aIgorithm (CMA) approaches. The reduced-length STMMSE receiver with training sequences is studied first. Two CMA approaches which are capable of combating both ICI and MAI without stringent power control are proposed to generate training sequences for the STMSE receiver.","PeriodicalId":145700,"journal":{"name":"ICSP '98. 1998 Fourth International Conference on Signal Processing (Cat. No.98TH8344)","volume":"66 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1998-10-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122584933","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 : 1998-10-12DOI: 10.1109/ICOSP.1998.770749
Jiwi Huang, Y. Shi
Postfiltering of block artifacts in decoded images has drawn extensive attention. But less improvements have been made for staircase noise. In this paper a space and frequency hybrid algorithm is presented. First, the edge information in the decoded images is extracted. Based on an analysis of artifacts in the frequency domain, the block effect in the edge image is removed by using a notch filter. The filtered edge image is used for a neighborhood analysis. A space-variant filter, consisting of a 2D LPF and a 1D directional LPF, is then applied to the decoded image to reduce artifacts. Which of the filters is chosen depends on the neighborhood analysis. Finally, the contrast is enhanced. The simulation results demonstrate that the proposed algorithm improves both the subjective visual quality and PSNR of decoded images effectively.
{"title":"A space and frequency hybrid algorithm to remove block artifacts","authors":"Jiwi Huang, Y. Shi","doi":"10.1109/ICOSP.1998.770749","DOIUrl":"https://doi.org/10.1109/ICOSP.1998.770749","url":null,"abstract":"Postfiltering of block artifacts in decoded images has drawn extensive attention. But less improvements have been made for staircase noise. In this paper a space and frequency hybrid algorithm is presented. First, the edge information in the decoded images is extracted. Based on an analysis of artifacts in the frequency domain, the block effect in the edge image is removed by using a notch filter. The filtered edge image is used for a neighborhood analysis. A space-variant filter, consisting of a 2D LPF and a 1D directional LPF, is then applied to the decoded image to reduce artifacts. Which of the filters is chosen depends on the neighborhood analysis. Finally, the contrast is enhanced. The simulation results demonstrate that the proposed algorithm improves both the subjective visual quality and PSNR of decoded images effectively.","PeriodicalId":145700,"journal":{"name":"ICSP '98. 1998 Fourth International Conference on Signal Processing (Cat. No.98TH8344)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1998-10-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116091480","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 : 1998-10-12DOI: 10.1109/ICOSP.1998.770153
F. Yang, M. Paindavoine, H. Abdi
In order to localize the face in an image, our approach consists of approximating the face oval shape with an ellipse and to compute coordinates of the center of the ellipse. For this purpose, we explore a new version of the Hough transformation: the fuzzy generalized Hough transformation. To reduce the computation time, we present also a parallel implementation of the algorithm on 2 digital signal processors and we show that an acceleration of a factor of 1.62 has been obtained.
{"title":"Parallel implementation on DSPs of a face detection algorithm","authors":"F. Yang, M. Paindavoine, H. Abdi","doi":"10.1109/ICOSP.1998.770153","DOIUrl":"https://doi.org/10.1109/ICOSP.1998.770153","url":null,"abstract":"In order to localize the face in an image, our approach consists of approximating the face oval shape with an ellipse and to compute coordinates of the center of the ellipse. For this purpose, we explore a new version of the Hough transformation: the fuzzy generalized Hough transformation. To reduce the computation time, we present also a parallel implementation of the algorithm on 2 digital signal processors and we show that an acceleration of a factor of 1.62 has been obtained.","PeriodicalId":145700,"journal":{"name":"ICSP '98. 1998 Fourth International Conference on Signal Processing (Cat. No.98TH8344)","volume":"19 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1998-10-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116158738","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 : 1998-10-12DOI: 10.1109/ICOSP.1998.770325
Pei-Min Chen
Depth-first (DF) and breath-first (BF) linear quadtree representations have the advantages in the performance of image operations and the storage requirements respectively. In this paper, the general guidelines and algorithms for code transformation between BF and DF linear quadtrees are proposed. Thus, we can exploit the advantages of DF and BF representations without suffering their drawbacks. Here, one typical DF linear quadtree scheme (Gargantini, 1982), and two BF linear quadtree schemes with different styles, FBLQ (Chang et al., 1994) with variant length and CBLQ (Lin), with constant length, are selected to depict the guidelines and algorithms. Each algorithm has time complexity O(l), where l is the length of the transformed code. This complexity is much less than that of the BF or DF encoding process for the corresponding image, which is at least of the same order as the number of pixels in this image.
深度优先(DF)和呼吸优先(BF)线性四叉树表示分别在图像操作性能和存储要求方面具有优势。本文提出了BF和DF线性四叉树之间代码转换的一般准则和算法。因此，我们可以利用DF和BF表示的优点，而不会遭受它们的缺点。本文选择一种典型的DF线性四叉树方案(Gargantini, 1982)和两种不同风格的BF线性四叉树方案，变长度的FBLQ (Chang et al.， 1994)和定长度的CBLQ (Lin)来描述准则和算法。每种算法的时间复杂度为O(l)，其中l为转换后的代码长度。这种复杂度远低于对应图像的BF或DF编码过程，其复杂度至少与该图像中的像素数相同。
{"title":"Code transformations for variant linear quadtrees","authors":"Pei-Min Chen","doi":"10.1109/ICOSP.1998.770325","DOIUrl":"https://doi.org/10.1109/ICOSP.1998.770325","url":null,"abstract":"Depth-first (DF) and breath-first (BF) linear quadtree representations have the advantages in the performance of image operations and the storage requirements respectively. In this paper, the general guidelines and algorithms for code transformation between BF and DF linear quadtrees are proposed. Thus, we can exploit the advantages of DF and BF representations without suffering their drawbacks. Here, one typical DF linear quadtree scheme (Gargantini, 1982), and two BF linear quadtree schemes with different styles, FBLQ (Chang et al., 1994) with variant length and CBLQ (Lin), with constant length, are selected to depict the guidelines and algorithms. Each algorithm has time complexity O(l), where l is the length of the transformed code. This complexity is much less than that of the BF or DF encoding process for the corresponding image, which is at least of the same order as the number of pixels in this image.","PeriodicalId":145700,"journal":{"name":"ICSP '98. 1998 Fourth International Conference on Signal Processing (Cat. No.98TH8344)","volume":"15 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1998-10-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123777538","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 : 1998-10-12DOI: 10.1109/ICOSP.1998.770239
F. Asano
A method of speech enhancement using microphone-array signal processing based on the 4th-order spatial cross-cumulant matrix is proposed. The results of the simulation show approximately 5 dB of improvement in noise reduction obtained for a vowel portion of speech.
{"title":"Speech enhancement using array signal processing with 4th-order cumulant","authors":"F. Asano","doi":"10.1109/ICOSP.1998.770239","DOIUrl":"https://doi.org/10.1109/ICOSP.1998.770239","url":null,"abstract":"A method of speech enhancement using microphone-array signal processing based on the 4th-order spatial cross-cumulant matrix is proposed. The results of the simulation show approximately 5 dB of improvement in noise reduction obtained for a vowel portion of speech.","PeriodicalId":145700,"journal":{"name":"ICSP '98. 1998 Fourth International Conference on Signal Processing (Cat. No.98TH8344)","volume":"29 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1998-10-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121418090","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 : 1998-10-12DOI: 10.1109/ICOSP.1998.770166
M. Wintermantel, E. Luder
For the discrete convolution with a Toeplitz coefficient matrix, a general algorithm with minimum number of multiplications is derived by means of a linear transformation. In order to keep the results applicable to long convolutions with limited wordlength, modulo arithmetic and block-partitioning is introduced. The resulting algorithms reveal small complexity and generate no roundoff noise. The same holds for linear and cyclic convolution algorithms derived from the presented algorithms of the more general Toeplitz convolution. The main advantages of the new algorithms compared to similar algorithms based on number theoretic transforms are a simpler and more general derivation and far less restrictions for the convolution length.
{"title":"Reducing the complexity of discrete convolutions by a linear transformation and modulo arithmetic","authors":"M. Wintermantel, E. Luder","doi":"10.1109/ICOSP.1998.770166","DOIUrl":"https://doi.org/10.1109/ICOSP.1998.770166","url":null,"abstract":"For the discrete convolution with a Toeplitz coefficient matrix, a general algorithm with minimum number of multiplications is derived by means of a linear transformation. In order to keep the results applicable to long convolutions with limited wordlength, modulo arithmetic and block-partitioning is introduced. The resulting algorithms reveal small complexity and generate no roundoff noise. The same holds for linear and cyclic convolution algorithms derived from the presented algorithms of the more general Toeplitz convolution. The main advantages of the new algorithms compared to similar algorithms based on number theoretic transforms are a simpler and more general derivation and far less restrictions for the convolution length.","PeriodicalId":145700,"journal":{"name":"ICSP '98. 1998 Fourth International Conference on Signal Processing (Cat. No.98TH8344)","volume":"16 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1998-10-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121507935","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 : 1998-10-12DOI: 10.1109/ICOSP.1998.770187
Sun Shenghe, Wang Wei
Taking advantage of the threshold decomposition property of the stack filter, a kind of weighted median filter is presented based on a threshold decomposition architecture. It allows parallel processing and VLSI implementation, and is effective in the application of image processing.
{"title":"Weighted median filter based on threshold decomposition","authors":"Sun Shenghe, Wang Wei","doi":"10.1109/ICOSP.1998.770187","DOIUrl":"https://doi.org/10.1109/ICOSP.1998.770187","url":null,"abstract":"Taking advantage of the threshold decomposition property of the stack filter, a kind of weighted median filter is presented based on a threshold decomposition architecture. It allows parallel processing and VLSI implementation, and is effective in the application of image processing.","PeriodicalId":145700,"journal":{"name":"ICSP '98. 1998 Fourth International Conference on Signal Processing (Cat. No.98TH8344)","volume":"16 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1998-10-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121512089","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 : 1998-10-12DOI: 10.1109/ICOSP.1998.770221
Zhang Linrang, B. Zheng, Liao Guisheng
In many radar/sonar and other application areas, the white noise is an inadequate model for measured data. In this paper, we study the method and performance of adaptive beamforming in a colored noise environment, and analyze the principle of beam distortion. A new adaptive beamforming method is proposed to remedy beam distortion completely, which is verified by computer simulation.
{"title":"Adaptive beamforming in colored noise environment","authors":"Zhang Linrang, B. Zheng, Liao Guisheng","doi":"10.1109/ICOSP.1998.770221","DOIUrl":"https://doi.org/10.1109/ICOSP.1998.770221","url":null,"abstract":"In many radar/sonar and other application areas, the white noise is an inadequate model for measured data. In this paper, we study the method and performance of adaptive beamforming in a colored noise environment, and analyze the principle of beam distortion. A new adaptive beamforming method is proposed to remedy beam distortion completely, which is verified by computer simulation.","PeriodicalId":145700,"journal":{"name":"ICSP '98. 1998 Fourth International Conference on Signal Processing (Cat. No.98TH8344)","volume":"3 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1998-10-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125298155","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 : 1998-10-12DOI: 10.1109/ICOSP.1998.770298
Yao Kaisheng, C. Zhigang
In real world applications, additive noise will contaminate input speech features for speech recognition and representation when speech recognition systems are working in real environments. There have been many attempts made to find a robust speech feature. In this paper, we propose a robust speech feature, the perceptive scalogram, for speech representation and recognition. The new feature is based on some propositions which state that a human's perception of speech is a perception of specific components of sounds, and the components have a specific changing rate of their short-time spectrum. The proposed perceptive scalogram also takes consideration of the fact that speech is non-stationary, and uses wavelets as its signal analysis tool. Simulation results show the robustness of the perceptive scalogram against additive Gaussian noise.
{"title":"A robust speech feature-perceptive scalogram based on wavelet analysis","authors":"Yao Kaisheng, C. Zhigang","doi":"10.1109/ICOSP.1998.770298","DOIUrl":"https://doi.org/10.1109/ICOSP.1998.770298","url":null,"abstract":"In real world applications, additive noise will contaminate input speech features for speech recognition and representation when speech recognition systems are working in real environments. There have been many attempts made to find a robust speech feature. In this paper, we propose a robust speech feature, the perceptive scalogram, for speech representation and recognition. The new feature is based on some propositions which state that a human's perception of speech is a perception of specific components of sounds, and the components have a specific changing rate of their short-time spectrum. The proposed perceptive scalogram also takes consideration of the fact that speech is non-stationary, and uses wavelets as its signal analysis tool. Simulation results show the robustness of the perceptive scalogram against additive Gaussian noise.","PeriodicalId":145700,"journal":{"name":"ICSP '98. 1998 Fourth International Conference on Signal Processing (Cat. No.98TH8344)","volume":"6 6","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1998-10-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114000698","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：