摘要:In this paper,some sufficient conditions for unconditioned stability of cellular neural networks with delay are obtained by means of Liapunov functional approach. These conditions can be used to design globally stable networks and thus have important significance in both theory and applications.
关键词:Cellular neural networks with delay (DCNN);Liapunov functional;Unconditioned stability
摘要:A symbolic analysis method is introduced for analyzing PWM switching power converters,and the principle of the method is described in detail. The symbolic expressions of ripple and small-signal model are easily obtained. It is shown that the method offers the advantages of both simplicity and accuracy.
关键词:switching power converter;Power electronics;Time-yaring system;symbolic analysis;Analytical method.
摘要:In this paper a new signal estimating method-wavelet transformation is given. It’s also proved by theoretical and emulation that spectrum estimation based on wavelet transformation has better distinguishing rate than that of FFT. It also discussed the algorithm of spectrum estimating,the dual-sine emulative result is given at last.
摘要:In ATM (Asynchronous Transfer Mode) communication networks,various kinds of traffic will be transferred. Among them,a kind of traffic is composed of continuous bit-stream. The arriving time of its transferrence cells is periodic. In this paper,a queueing model for transport of periodic cell stream through an ATM multiplexer is analyzed. The probability distribution of the queueing delay,the autocorrelation and power spectrum of the cell delay jitter are obtained.
摘要:As a phenomenon found in non-linear dynamic system,Chaos has been paid attention to. Researchers are looking for its possible application in communication. This paper analyzes the statistic characteristic of chaotic sequences generated by a widely studied discretetime dynamic system,Logistic-Map. We find that their statistic characteristics is identical with those of white noise. We also propose a digital hardware implementation for Logistic-Map and the conception of digital chaotic sequence. The simulation results show that Logistic-Map digital chaotic sequences have good correlation property and their number is large,suitable for acting as spread-spectrum address sequences.
摘要:We design a diversity structure of CDMA multiuser detector for frequency selective Rayleigh fading channels. The multiuser detection algorithms for Gaussian channel can be effectively employed in the Rayleigh fading channel through the structure. Simulation results show that the diversity combining of multiuser detection will significantly improve the probability of error in the fading channel,and its capability for dealing with the multi-access interference will not be affected by multiuser detection algorithms. Furthermore,the comparison is made between different combiners and different detection algorithms.
关键词:Code-division multiple access;Multiuser signal detection;Multipath fading;diversity reception
摘要:This paper proposes a fault-tolerant switching network for ATM switching systems. Based on baseline network,the proposed network can provide redundant paths by adding some extra elements between stages. It has the advantages of high performance and simple routine algrotihm. Compared with other fault-tolerant networks which have the same hardware complexty,the network can tolerate more faults and the effect is increased when network size is larger.
摘要:A new idea of the adaptive RLS grey equalizer which is suited for time-varying channels has been presented. The convergence perfermance of the RLS grey equalizer is better than that of the classical RLS equalizer. Its symbol error performace is not only much better than that of the RLS blind equalizer whose derivation is based on the literatures[1]and[2],but also better than that of the classical RLS equalizer. The analysis and simulation indicate that the RLS grey equalizer is a very good equalization method on time-varying channels.
摘要:It is essential for realizing color constancy to acquire color descriptors that are not affected by illumination and geometry conditions. To combine dichromatic reflection model with the finite dimensional linear model to derive the normalized deviation body reflectance, has been demonstrated to be an efficient method to realize color constancy. In the present work,we studied the properties of the normalized deviation body reflectance,its computational method,and utilize this kind of feature in color image segmentation and color inspecction.
关键词:color constancy;Dichromatic reflection model;Deviation body reflectance
摘要:Calculating eigenvalues and eigenvectors of a matrix is widely encountered in science and technology. In this paper we turn the eigenproblem of a real symmetric matrix into a nonlinear equation problem. Then we construct a neural network of continuous-time dynamical system to explore the above nonlinear equation. The stability and convergence of the network are investigated by using Lyapunov function under rather mild conditions.
关键词:Eigenproblem;neural network;stability;convergence;Lyapunov function
摘要:This paper proposes an alogrithm for fusing motion and stereo cues to get robust estimates of both motion and structure. Our algorithm can compute a MAP estimate of the displacement at each pixel and an associated confidence measure by calculating the instantaneous FOE. Using the displacement estimates we calculate a relative depth map from one of the image sequences. This depth map is then used to derive a prior probability distribution for disparity that is used in matching the two frames of stereo paris. This method is used to estimate the disparity at each pixel independently and does not need any assumptions about smoothness.
关键词:Binocular motion stereo;MAP estimates;Prior probability distribution;Relative depth map
摘要:In this paper a simple chaotic moderation and correlative detection method for image classification and recognition is presented. its characteristic is also analyzed and discussed by experiments.
摘要:This paper will discuss matching tactics of common motion in detecting optical flow by clustered algorithm. The tactics is a basic in matching operation. If the tactics are correct, it will be a way that can be used detecting nonrigied body optical flow field more accurately. So this research is a very important for optical flow detect of matching clustered algorithm.
关键词:Clustered algorithm;Optical flow of nonrigid body;Matching tactics of common motion
摘要:Based on the fact that the projection of the Leech lattice on GF(4) is the (6, 3, 4)Hexacode,a new bounded distance decoding algorithm is presented. its complexity is at most 1151, at least 575,and 911 real operations on the average,as compared to the fastest BD algorithm shown in[10),whose complexity is at most 1007,at least 911 and 959 on the average. The computer simulation shows that the loss in coding gain of our algorithm relative to maximum likelihood decoding algorithm is only about 0. 02dB,as compared to 0. 17dB for the algorithm of [10]
摘要:This paper studied on real-time computation of generalized time-frequency representations (GTFRs )and its optimizing. From the conception of generalized local autocorrelation,we discuss the problem that real-time computation of GTFRs can be performed by means of short time processing. Furthermore,this paper proposes an algorithm optimized in such two aspects as: (1) reducing the computational complexity more than twice by taking advantage of conjugated symmetry of the kernel function for some GTFRs ; (2) deducing a time domain recursive algorithm for the computation of analytic signals accurately with any frame shift interval.
Pan Su, Feng guangzeng (Nanjing university of posts and telecommunications Nanjing 210003)
Issue 1, (1997)
摘要:This paper is devoted to an investigation for the realization of GSM Half-Rate Coder with the CELP method. It is focused on the fast adaptive codebook search algorithm which is proposed by us. The computational complexity is reduced and no harm is introduced into the synthesis voice by this algorithm. The results of computer simulation show that the bit-rate of this coder version is 5. 8-6. 5kb/s and the MOS(mean opinion score)of the synthesis voice is not less than 3. 8.
摘要:The number of effective accumulated frames (NEAF) is defined in this paper. It is proved that SNR in the average accumulated image is increased by M times within NEAF. Further,the techniques of adaptable threshold detection are studied based on image accumulation. the rule of minimal errer probabi lity (ROMEP) is defined and the equation of initlal threshold and the detection probability,false alarm probability as well as error probability are deduced under this rule. The post-processing project is also set up so as to decrease still further false alarm.
摘要:in computer vision symmetry is one of important geometric features of 3-D objects. We can use it to recover the structure of objects and estimate their positions in 3-D space. However,it is difficUlt to obtain perfect symmetry shapes due to occlusion,self-occlusion or noise. This paper describes a new algorithm which recovers a whole closed contour from its partial curve ba$ed on the skewed symmetry. First we find two axes of skewed symmetry. Then all skewed symmetry axes of the closed contour are calculated. The closed contour can be obtained through the linear geometric transformation and is inverse transformation. The experimental results implemented with the algorithm are very satisfactory.
摘要:Absolute error measures such as MSE are used in conventional VQ coding schemes. But they don’t suit for human visual perception. In this paper,one coding error measure based on human visual perceptivity is defined. Furthermore one novel VQ coding scheme is proposed. Experimental results prove that decoded images using it have better subjective quality.
摘要:A fast processing system of black form is presented based on machine learning. By means of learning sample form,the discriminat parameter and the geometric structure information of background can be automatically obtained. On the guide of above information,this system can continuously process same kind of forms while image is scanned.
摘要:In this paper,a new definition of similarity measure between image and template is given. Combining with Distance transformation(DT),a new matching method is presented. Experiments proved that in many cases our matching algorithm is faster than MAD,NCC. And it is robust to grey level change,edge movement and edge discontinuity.
摘要:A new method of analysis for threshold selectionis presented. It is based on the wavelet transform which provides a multiscale analysis of the information of the histogram of an image. By the detection of the zero-crossings of the wavelet transform of the histogram,a sequence thresholds of the histogram is abtained,which are used to describe the image edge features. The complete method uses a coarse-to-fine multiresolution threshold stategy.
Li Shenghong, Wang Shuxun, Zhang Yan, Wei Xiaoli (Jilin University of Technology Changchun 130022)
Issue 1, (1997)
摘要:So far,the problem of the harmonic retrieval with quadratic phase coupling in non-Gaussian noise has not been resolved well. Based on Hilbert transform and cumulant,this paper proposes a new method named HBSTCH-SVD-TLS. Simulation results show that the method not only resolves the problem of the harmonic retrieval with quadratic phase coupling in non Gaussian noise,but also has wide application.
摘要:Several works have concluded that the number of cells per frame of VBR video follows a gamma distribution. But few model meets this character. In this paper we propose a novel video model, a first-order autoregressive (AR) gamma sequence, in which random variables are marginally distributed as gamma variables. We give the definition of GAR model and show some characters of GAR(1). The parameter estimation method is studied too. Simulation results show that our model can reflect the characters of VBR video better than the former models. It is more easy and accurate to create simulation data sequences using this new model.
关键词:ATM network;VBR video traffic;Gamma autoregressive (GAR) model
摘要:In this paper,a method of generating M sequences is introduced. The method is based on the principle of cycle joining and a counting formula is given. A simple algorithm of generating M sequences is given,pairs of nonsingular boolean functions generate unqiue M sequence, then the number of generated M sequences of period 2n+1 is 22n-1-1.
摘要:A new block design(dyadic design)and dyadic addition set is defined. With the relationship among dyadic designs .dyadic addition sets and dyadic codes,a new approach is found for the design of dyadic codes.
摘要:C is a binary regular [n,k,d ] linear quasi-constant weight code. We show that(1)if n=2d,then C is equivalent to the first order Reed-Muller code RM (k - 1, 1 ), (2)if n≠2d, and2k-1 - 1 is a prime number,then C is equivalent to the punctured Reed-Muller code RM* (k-1, 1 )(deleting the first coordinate of RM (k - 1, 1 ) ). Furthermore, we present a new simple proof for Theorem 1 in[1] by using the well known MacWilliams identity in coding theory.
关键词:Regular linear codes;Constant weight codes;Quasi-constant weight codes;Simplex codes;Reed-Muller codes
摘要:We star t form the photon number fluctuation in EDFA’S output,and get the analytical expression of signal to noise ratio(SNR) of EDFA’s output. This method is valid when EDFA works in saturation condition. Then the SNR in optical fiber communication systems using cascaded EDFAs is analyzed and our results are compared with previous work.
关键词:Optical amplifier;Signal to noise ration;Shot noise;Beat noise
摘要:In this paper,the properties of the best code are studied specifically. The results show that the best codes not only have correlation orthogonality and duality,but also their runs and weights are strictly limited. On the other hand,it is proved that a code whose head and tail satisfy certain demands is not the best code. Based on this,a new sieve is presented and an example is given to test its effectiveness.
摘要:A modified nonlinear least-square algorithm is derived from the approach appending a regularization term to the conventional nonlinear least-square criterion,ineluding the batch and recursive versions. Training multilayer feed forward neural networks using its recursive algorithm,the storage and computational requirements are reduced,and also it can be applicable to the ill-conditioned cases. Simulation demonstrates the superior convergence performance of its recur sive algorithm compared with the back propagation routine