A parallel implementation of algorithms to achieve viewpoint invariant target recognition from laser radar images is described. The active/passive CO/sub 2/ laser radar sensor used provides pixel-registered thermal, visual, relative range and artifacts and detects edges. Viewpoint invariance obtained using the translation invariant property of the Fourier transform transforms 2-D rotations and scale changes into translations. This scene representation provides 2-D viewpoint invariance and is used as input to a neural network that has been trained to recognize representative views of the targets of interest. The system is implemented on the Geometric Arithmetic Parallel Processor (GAPP), a massively parallel single-instruction-multiple-data (SIMD) computer that contains a 216*384 array of processing elements. Results obtained by processing real lasers radar images are presented.<>
{"title":"Parallel algorithms for automatic target recognition using Co/sub 2/ laser radar images","authors":"D. Sullivan, A. Forman","doi":"10.1109/MDSP.1989.97026","DOIUrl":"https://doi.org/10.1109/MDSP.1989.97026","url":null,"abstract":"A parallel implementation of algorithms to achieve viewpoint invariant target recognition from laser radar images is described. The active/passive CO/sub 2/ laser radar sensor used provides pixel-registered thermal, visual, relative range and artifacts and detects edges. Viewpoint invariance obtained using the translation invariant property of the Fourier transform transforms 2-D rotations and scale changes into translations. This scene representation provides 2-D viewpoint invariance and is used as input to a neural network that has been trained to recognize representative views of the targets of interest. The system is implemented on the Geometric Arithmetic Parallel Processor (GAPP), a massively parallel single-instruction-multiple-data (SIMD) computer that contains a 216*384 array of processing elements. Results obtained by processing real lasers radar images are presented.<<ETX>>","PeriodicalId":340681,"journal":{"name":"Sixth Multidimensional Signal Processing Workshop,","volume":"35 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1989-09-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115772818","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 hybrid image predictive coding method is presented. The intraframe predictor is an adaptive finite impulse response (FIR) filter using the well-known least-mean-square (LMS) algorithm to track continuously the spatial local characteristics of the intensity. The interframe predictor is motion-adaptive, using a pel-recursive method estimating the displacement vector. A weight coefficient is adapted continuously in order to favor the prediction mode that performs better between intraframe and only-motion-compensation mode. A crucial problem in predictive coding, particularly with adaptive techniques, is that of sensitivity to transmission errors. A method ensuring the autoadjustment of the decoder in the presence of isolated transmission errors is proposed for the intraframe mode. Neither overhead information nor error-correcting code is needed.<>
{"title":"A hybrid image coder: adaptive intra-interframe prediction using motion compensation","authors":"G. Tziritas, J. Pesquet","doi":"10.1109/MDSP.1989.97135","DOIUrl":"https://doi.org/10.1109/MDSP.1989.97135","url":null,"abstract":"A hybrid image predictive coding method is presented. The intraframe predictor is an adaptive finite impulse response (FIR) filter using the well-known least-mean-square (LMS) algorithm to track continuously the spatial local characteristics of the intensity. The interframe predictor is motion-adaptive, using a pel-recursive method estimating the displacement vector. A weight coefficient is adapted continuously in order to favor the prediction mode that performs better between intraframe and only-motion-compensation mode. A crucial problem in predictive coding, particularly with adaptive techniques, is that of sensitivity to transmission errors. A method ensuring the autoadjustment of the decoder in the presence of isolated transmission errors is proposed for the intraframe mode. Neither overhead information nor error-correcting code is needed.<<ETX>>","PeriodicalId":340681,"journal":{"name":"Sixth Multidimensional Signal Processing Workshop,","volume":"422 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1989-09-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115929285","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}
Summary form only given. A feature extraction method using differential angles is discussed. A pointer follows a connected path along a skeletonized image. As it moves along, angles of the pointer movement from reference are recorded. From this, a differential angle vector, whose element is obtained by subtracting the previous angle from the current angle, is obtained. The differential angle vector is processed in such a way that isolated pairs of (-45 degrees ,45 degrees ), (45 degrees ,-45 degrees ), (-90 degrees ,-90 degrees ), (90 degrees ,-90 degrees ) (135 degrees ,-135 degrees ), (-35 degrees ,135 degrees ) are removed; there is no effect on the final decision. A string of zeros in the differential angle vector indicates the existence of a straight line. The differential angle vector is compressed by eliminating all zeros in a string of zeros. When the point reaches an end point and no further advancement is possible, it moves backward until reaching an untraversed segment of image.<>
{"title":"Compressed differential angles as a feature in handwritten digit recognition","authors":"S.Y. Kang","doi":"10.1109/MDSP.1989.97034","DOIUrl":"https://doi.org/10.1109/MDSP.1989.97034","url":null,"abstract":"Summary form only given. A feature extraction method using differential angles is discussed. A pointer follows a connected path along a skeletonized image. As it moves along, angles of the pointer movement from reference are recorded. From this, a differential angle vector, whose element is obtained by subtracting the previous angle from the current angle, is obtained. The differential angle vector is processed in such a way that isolated pairs of (-45 degrees ,45 degrees ), (45 degrees ,-45 degrees ), (-90 degrees ,-90 degrees ), (90 degrees ,-90 degrees ) (135 degrees ,-135 degrees ), (-35 degrees ,135 degrees ) are removed; there is no effect on the final decision. A string of zeros in the differential angle vector indicates the existence of a straight line. The differential angle vector is compressed by eliminating all zeros in a string of zeros. When the point reaches an end point and no further advancement is possible, it moves backward until reaching an untraversed segment of image.<<ETX>>","PeriodicalId":340681,"journal":{"name":"Sixth Multidimensional Signal Processing Workshop,","volume":"142 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1989-09-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114748595","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}
Summary form only given. A new class of orthogonal basis functions that can be relevant to signal processing has recently been introduced. These bases are constructed from a single smooth bandpass function psi (t), the wavelet, by considering its translates and dilates on a dyadic grid 2/sup n/, 2/sup n/m of points, psi /sub n,m/(t)=2/sup -n/2/ psi (2/sup -n/t-m). It is required that psi (t) be well localized in both the time and frequency domain, without violating the uncertainty principle. Any one-dimensional signal can be represented by the bidimensional set of its expansion coefficients. Multidimensional signals can also be expanded in terms of wavelet bases. An algorithm for computing the expansion coefficients of a signal in terms of wavelet bases has been found, the structure of which is that of a pruned-tree quadrature mirror multirate filter bank. The construction of wavelet bases and their relation to filter banks, together with several design techniques for wavelet generating quadrature mirror filters and examples, are reviewed.<>
{"title":"Orthogonal wavelet transforms and filter banks","authors":"G. Evangelista","doi":"10.1109/MDSP.1989.97053","DOIUrl":"https://doi.org/10.1109/MDSP.1989.97053","url":null,"abstract":"Summary form only given. A new class of orthogonal basis functions that can be relevant to signal processing has recently been introduced. These bases are constructed from a single smooth bandpass function psi (t), the wavelet, by considering its translates and dilates on a dyadic grid 2/sup n/, 2/sup n/m of points, psi /sub n,m/(t)=2/sup -n/2/ psi (2/sup -n/t-m). It is required that psi (t) be well localized in both the time and frequency domain, without violating the uncertainty principle. Any one-dimensional signal can be represented by the bidimensional set of its expansion coefficients. Multidimensional signals can also be expanded in terms of wavelet bases. An algorithm for computing the expansion coefficients of a signal in terms of wavelet bases has been found, the structure of which is that of a pruned-tree quadrature mirror multirate filter bank. The construction of wavelet bases and their relation to filter banks, together with several design techniques for wavelet generating quadrature mirror filters and examples, are reviewed.<<ETX>>","PeriodicalId":340681,"journal":{"name":"Sixth Multidimensional Signal Processing Workshop,","volume":"8 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1989-09-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122052169","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}
Summary form only given. The basic theory and some recent developments in the theory of multiple-window methods for array data are reviewed. Applied to small samples or nonstationary data, this method has numerous advantages over conventional techniques. It is a small sample theory, essentially an inverse method applied to the finite Fourier transform; its statistical efficiency is typically a factor of two to three higher than that of conventional methods with the same degree of bias protection; and it separates the continuous part of the spectrum from line components. In addition, it has the major advantage that underlying assumptions can be tested. However, because higher-dimensional problems are more delicate than univariate ones, robustness and diagnostics become far from critical. Such diagnostics are illustrated by the application of multiple-window methods to analysis of data from a linear array of three-axis magnetometers.<>
{"title":"An introduction to multiple-window analysis of array data","authors":"D. Thomson","doi":"10.1109/MDSP.1989.97062","DOIUrl":"https://doi.org/10.1109/MDSP.1989.97062","url":null,"abstract":"Summary form only given. The basic theory and some recent developments in the theory of multiple-window methods for array data are reviewed. Applied to small samples or nonstationary data, this method has numerous advantages over conventional techniques. It is a small sample theory, essentially an inverse method applied to the finite Fourier transform; its statistical efficiency is typically a factor of two to three higher than that of conventional methods with the same degree of bias protection; and it separates the continuous part of the spectrum from line components. In addition, it has the major advantage that underlying assumptions can be tested. However, because higher-dimensional problems are more delicate than univariate ones, robustness and diagnostics become far from critical. Such diagnostics are illustrated by the application of multiple-window methods to analysis of data from a linear array of three-axis magnetometers.<<ETX>>","PeriodicalId":340681,"journal":{"name":"Sixth Multidimensional Signal Processing Workshop,","volume":"12 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1989-09-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124749369","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. Willsky, K. C. Chou, A. Benveniste, M. Basseveille
Summary form only given. It has been shown that wavelet transforms and multiscale representations lead naturally to the study of stochastic processes indexed by nodes on lattices and trees, where different depths in the tree or lattice correspond to different spatial scales or resolutions in representing the signal. This framework has been used to develop a theory of modeling for multiscale stochastic processes that leads to a highly nontrivial generalization of Levinson's algorithm involving recursive generation of models of increasing order, in which the direction of recursion is from coarse to fine resolutions. A theory of optimal estimation for multiresolution stochastic models has been developed. These models lead naturally to several algorithmic structures, one reminiscent of the Laplacian pyramid, one that can be viewed as a multigrid relaxation algorithm, and one that is a generalization of the Rauch-Tung-Striebel algorithm for optimal smoothing of state space models.<>
{"title":"Multiresolution stochastic models and multiscale estimation algorithms","authors":"A. Willsky, K. C. Chou, A. Benveniste, M. Basseveille","doi":"10.1109/MDSP.1989.97060","DOIUrl":"https://doi.org/10.1109/MDSP.1989.97060","url":null,"abstract":"Summary form only given. It has been shown that wavelet transforms and multiscale representations lead naturally to the study of stochastic processes indexed by nodes on lattices and trees, where different depths in the tree or lattice correspond to different spatial scales or resolutions in representing the signal. This framework has been used to develop a theory of modeling for multiscale stochastic processes that leads to a highly nontrivial generalization of Levinson's algorithm involving recursive generation of models of increasing order, in which the direction of recursion is from coarse to fine resolutions. A theory of optimal estimation for multiresolution stochastic models has been developed. These models lead naturally to several algorithmic structures, one reminiscent of the Laplacian pyramid, one that can be viewed as a multigrid relaxation algorithm, and one that is a generalization of the Rauch-Tung-Striebel algorithm for optimal smoothing of state space models.<<ETX>>","PeriodicalId":340681,"journal":{"name":"Sixth Multidimensional Signal Processing Workshop,","volume":"34 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1989-09-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130772788","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}
Summary form only given. A fundamental issue in the problem of finding an efficient algorithm for estimation of 3D orientation is how 3D orientation should be represented. A representation is regarded as suitable if it meets the three basic requirements of uniqueness, uniformity, and polar separability. A tensor representation suitable in the above sense has been obtained. The uniqueness requirement implies a mapping that maps all pairs of 3D vectors x and -x onto the same tensor T. Uniformity implies that the mapping implicitly carries a definition of distance between 3D planes (and lines) that is rotation invariant and monotone with the angle between the planes. Polar separability means that the norm of the representing tensor T is rotation invariant. One way to describe the mapping is that it maps a 3D sphere into 6D in such a way that the surface is uniformly stretched and all pairs of antipodal points map onto the same tensor. It has been demonstrated that the above mapping can be implemented by sampling the 3D space using a specific class of symmetrically distributed quadrature filters.<>
{"title":"Spatio-temporal analysis using tensors","authors":"H. Knutsson, G. Granlund","doi":"10.1109/MDSP.1989.96989","DOIUrl":"https://doi.org/10.1109/MDSP.1989.96989","url":null,"abstract":"Summary form only given. A fundamental issue in the problem of finding an efficient algorithm for estimation of 3D orientation is how 3D orientation should be represented. A representation is regarded as suitable if it meets the three basic requirements of uniqueness, uniformity, and polar separability. A tensor representation suitable in the above sense has been obtained. The uniqueness requirement implies a mapping that maps all pairs of 3D vectors x and -x onto the same tensor T. Uniformity implies that the mapping implicitly carries a definition of distance between 3D planes (and lines) that is rotation invariant and monotone with the angle between the planes. Polar separability means that the norm of the representing tensor T is rotation invariant. One way to describe the mapping is that it maps a 3D sphere into 6D in such a way that the surface is uniformly stretched and all pairs of antipodal points map onto the same tensor. It has been demonstrated that the above mapping can be implemented by sampling the 3D space using a specific class of symmetrically distributed quadrature filters.<<ETX>>","PeriodicalId":340681,"journal":{"name":"Sixth Multidimensional Signal Processing Workshop,","volume":"507 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1989-09-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129031403","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}
Summary form only given, as follows. A computing system can be regarded as a collection on nonlinear gates, within which signals must interact, and interconnections between those gates and between groups of gates for communications. The use of optics at an interconnect technology is addressed. The appropriateness of optics as an interconnect solution varies through the hierarchy of levels of interconnect present in computing. At the machine-to-machine level there is no argument about the success of optics as an interconnect medium. At the lowest level, gate-to-gate, it can be shown that optics will probably not be competitive with electronic solutions. Somewhere between the highest level and the lowest level there is a level where optics becomes noncompetitive. Just where this level may be is a subject of much interest and research. From the knowledge now in hand, it appears likely that optics will provide a competitive advantage at those levels where terminated transmission lines would otherwise be required. However, the relatively short lifetimes (less than 100 h) of laser diodes will limit the number of such devices that can be used within a single machine and therefore will influence how low in the interconnect hierarchy optics can penetrate.<>
{"title":"Optical interconnections and their impact on computing","authors":"J. Goodman","doi":"10.1109/MDSP.1989.97114","DOIUrl":"https://doi.org/10.1109/MDSP.1989.97114","url":null,"abstract":"Summary form only given, as follows. A computing system can be regarded as a collection on nonlinear gates, within which signals must interact, and interconnections between those gates and between groups of gates for communications. The use of optics at an interconnect technology is addressed. The appropriateness of optics as an interconnect solution varies through the hierarchy of levels of interconnect present in computing. At the machine-to-machine level there is no argument about the success of optics as an interconnect medium. At the lowest level, gate-to-gate, it can be shown that optics will probably not be competitive with electronic solutions. Somewhere between the highest level and the lowest level there is a level where optics becomes noncompetitive. Just where this level may be is a subject of much interest and research. From the knowledge now in hand, it appears likely that optics will provide a competitive advantage at those levels where terminated transmission lines would otherwise be required. However, the relatively short lifetimes (less than 100 h) of laser diodes will limit the number of such devices that can be used within a single machine and therefore will influence how low in the interconnect hierarchy optics can penetrate.<<ETX>>","PeriodicalId":340681,"journal":{"name":"Sixth Multidimensional Signal Processing Workshop,","volume":"263 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1989-09-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122161670","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}
Summary form only given. A robust method of 2-D spectral estimation of signals in additivbe white noise whose distribution is the so-called outlier contaminated Gaussian process was investigated. The term robustness refers here to insensitivity to small deviation in the underlying Gaussian noise assumption. Robust spectral estimation methods are known to be computationally feasible only when the number of parameters to be estimated is small, and recent approaches to 2-D robust spectral estimation require very extensive computation. In the work reported the 2-D spectral estimation problem was converted into a set of 1-D independent problems using the Radon transform. The 2-D array data were transformed into a set of 1-D sequences (projections), and each projection was modeled as a 1-D autoregressive (AR) process. A robust technique based on the Huber's minimax approach was utilized to estimate the AR parameters. The 2-D spectrum was finally obtained on a polar raster. This method is highly amenable to parallel processing.<>
{"title":"Robust 2-D spectrum estimation using Radon transform","authors":"N. Srinivasa, D.D. Lee, R. Kashyap","doi":"10.1109/MDSP.1989.97046","DOIUrl":"https://doi.org/10.1109/MDSP.1989.97046","url":null,"abstract":"Summary form only given. A robust method of 2-D spectral estimation of signals in additivbe white noise whose distribution is the so-called outlier contaminated Gaussian process was investigated. The term robustness refers here to insensitivity to small deviation in the underlying Gaussian noise assumption. Robust spectral estimation methods are known to be computationally feasible only when the number of parameters to be estimated is small, and recent approaches to 2-D robust spectral estimation require very extensive computation. In the work reported the 2-D spectral estimation problem was converted into a set of 1-D independent problems using the Radon transform. The 2-D array data were transformed into a set of 1-D sequences (projections), and each projection was modeled as a 1-D autoregressive (AR) process. A robust technique based on the Huber's minimax approach was utilized to estimate the AR parameters. The 2-D spectrum was finally obtained on a polar raster. This method is highly amenable to parallel processing.<<ETX>>","PeriodicalId":340681,"journal":{"name":"Sixth Multidimensional Signal Processing Workshop,","volume":"77 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1989-09-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116193884","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}
Summary form only given. The use of the generalized expectation maximization (GEM) algorithm for image reconstruction from projections and restoration from broad point spread functions is proposed. A GEM algorithm has been developed for maximum a posteriori (MAP) estimation using Markov random field prior distributions for a set of Poisson data whose mean is related to the unknown image by a linear transformation. This method is applicable in emission tomography (PET and SPECT) and to the restoration of radioastronomical images. The EM algorithm is applicable to problems in which there is a more natural formulation of the estimation problem in terms of a set of complete unobserved data which is related to the incomplete observed data by a known many-to-one transformation. Applying this approach to the MAP image reconstruction problem results in a two-step iterative algorithm. The resulting computational costs are significantly lower than those for the coordinate descent algorithms. The algorithm does not guarantee convergence to a global maximum, but will converge to a stationary point of the posterior density for the image conditional on the observed data.<>
{"title":"3D Bayesian image reconstruction using the generalized EM algorithm","authors":"R. Leahy, T. Hebert","doi":"10.1109/MDSP.1989.97123","DOIUrl":"https://doi.org/10.1109/MDSP.1989.97123","url":null,"abstract":"Summary form only given. The use of the generalized expectation maximization (GEM) algorithm for image reconstruction from projections and restoration from broad point spread functions is proposed. A GEM algorithm has been developed for maximum a posteriori (MAP) estimation using Markov random field prior distributions for a set of Poisson data whose mean is related to the unknown image by a linear transformation. This method is applicable in emission tomography (PET and SPECT) and to the restoration of radioastronomical images. The EM algorithm is applicable to problems in which there is a more natural formulation of the estimation problem in terms of a set of complete unobserved data which is related to the incomplete observed data by a known many-to-one transformation. Applying this approach to the MAP image reconstruction problem results in a two-step iterative algorithm. The resulting computational costs are significantly lower than those for the coordinate descent algorithms. The algorithm does not guarantee convergence to a global maximum, but will converge to a stationary point of the posterior density for the image conditional on the observed data.<<ETX>>","PeriodicalId":340681,"journal":{"name":"Sixth Multidimensional Signal Processing Workshop,","volume":"7 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1989-09-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116674320","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：