In the well known paper by Wax and Kailath [2], the AIC and MDL criteria for determining the number of signals in a multichannel time-series are presented. An essential element of these criteria is the number of degre...
详细信息
In the well known paper by Wax and Kailath [2], the AIC and MDL criteria for determining the number of signals in a multichannel time-series are presented. An essential element of these criteria is the number of degrees of freedom in the model. We propose a different number for the degrees of freedom and show that the resulting MDL criterion performs noticeably better than the previous criterion. We also show that the same criteria are appropriate for either real or complex data.
Wavelets are a relative newcomer to signal decomposition, and offer a flexible analog transform to provide multiresolution signal decomposition. Their advantages over Fourier and short-time Fourier transforms (STFTs) ...
详细信息
Wavelets are a relative newcomer to signal decomposition, and offer a flexible analog transform to provide multiresolution signal decomposition. Their advantages over Fourier and short-time Fourier transforms (STFTs) are significant, and the linkages and practical commonalities of these two transform techniques have generated interdisciplinary research activities among mathematicians, physicists and electrical engineers. This article presents the fundamentals of wavelet transform theory. In this context, the subject's mathematical rigor is avoided. We discuss the differences between the conventional STFT and wavelet transforms from a time-frequency "tiling" point of view. Then, we highlight the significant role of discrete-time filter banks in wavelet theory, and assess the practicality of wavelets in signalprocessing applications.< >
In this correspondence, we present a fast thresholded linear convolution representation of morphological operations. The thresholded linear convolution representation of dilation and erosion is first proposed. A compa...
详细信息
In this correspondence, we present a fast thresholded linear convolution representation of morphological operations. The thresholded linear convolution representation of dilation and erosion is first proposed. A comparison of the efficiency of the direct implementation of morphological operations and the thresholded linear convolution representation of morphological operations is subsequently conducted.
In this paper, we formulate and develop an approach which integrates different modules (feature extractor, matching and interpolation) involved in stereo. We study the integration process at the finest resolution when...
详细信息
In this paper, we formulate and develop an approach which integrates different modules (feature extractor, matching and interpolation) involved in stereo. We study the integration process at the finest resolution when (i) the precomputed edge map is the only line filed driving the model, (ii) the line field are computed interactively by the feature extracting module of the model, and (iii) when both the interactive line field computation module and the precomputed line field modules are present. The integration process being computationally intensive, we develop a multiresolution stereo integration approach. The energy function for each module at different resolutions is constructed and minimized in an integrated manner yielding a dense disparity map. A new energy function for the matching module is proposed. Experimental results are presented to illustrate our approach.
We describe the parallel implementation of fully recurrent neural networks (RRN) on a transputer-based multiprocessor system. To train the RNN, the real-time recurrent learning (RTRL) algorithm was used. The computati...
详细信息
We describe the parallel implementation of fully recurrent neural networks (RRN) on a transputer-based multiprocessor system. To train the RNN, the real-time recurrent learning (RTRL) algorithm was used. The computationally intensive sequential RTRL algorithm has been transformed to an equivalent parallel algorithm, realized in a ring topology that can be matched to a variety of target architectures, ranging from application specific VLSI arrays to general purpose multiprocessor systems. A ring array of up to 19 T800 transputers was programmed to efficiently perform the parallel RTRL algorithm. The speedup of the transputer implementation was estimated both analytically and through simulations, and the effect of the communication overhead is discussed. It is shown that as more neuron units are allocated to the same processor the efficiency is increased.
In order to solve the speed problem and shallow reasoning problem met in current research in fault diagnosis expert system, this paper presents a model based parallel fault diagnosis expert system for energy managemen...
In this brief paper, the Real Time Recurrent Learning (RTRL) algorithm for training fully recurrent neural networks in real time, is extended for the case of a recurrent neural network whose inputs, outputs, weights a...
详细信息
In this brief paper, the Real Time Recurrent Learning (RTRL) algorithm for training fully recurrent neural networks in real time, is extended for the case of a recurrent neural network whose inputs, outputs, weights and activation functions are complex. A practical definition of the complex activation function is adopted and the complex form of the conventional RTRL algorithm is derived. The performance of the proposed algorithm is demonstrated with an application in complex communication channel equalization.
In this correspondence, we investigate the invertibility of the morphological representation of binary images. A criteria for the invertibility of the morphological representation of binary images is proposed. Necessa...
详细信息
In this correspondence, we investigate the invertibility of the morphological representation of binary images. A criteria for the invertibility of the morphological representation of binary images is proposed. Necessary and sufficient conditions for the exact reconstruction of a binary image from its morphological representation are provided.
In this paper we address the problem of reconstruction of signals from their nonequally spaced samples. Exploiting the close-to-band structure of the composing matrix, a two-stage procedure for the recovery of uniform...
详细信息
In this paper we address the problem of reconstruction of signals from their nonequally spaced samples. Exploiting the close-to-band structure of the composing matrix, a two-stage procedure for the recovery of uniform samples from nonuniform samples has been suggested by Plotkin and Swamy (1987). In order to reduce the computational complexity, a special procedure of partitioning the composing matrix into a set of overlapping submatrices was used. Then the error in the estimate was reduced by applying an iterative procedure. The present paper is an extension of results presented by Plotkin and Swamy (1987). In this we propose a modification to their procedure, so as to recover an equal number of uniformly spaced samples as are those in the nonuniform set. We show that the iterative algorithm converges conditionally and the conditions are weak and may be implemented easily. computer simulation results have been presented which show that the proposed technique performs well even for deviations of nonuniform sample positions well beyond the corresponding uniform positions. The proposed method is attractive from computational point of view also.
Pyramid image coding is a natural coding scheme for applications where progressive transmission is desired. In this kind of coder, versions of the original image at several resolution levels are formed by successive f...
详细信息
暂无评论