摘要:In the coding of video sequences, the statistical character of motion compensated frame differential images is quite different from that of natural images.Discrete Cosine Transform (DCT) can be regarded as the suboptimal transform for still natural images,however,its coding effectiveness for differential images is somewhat lower.In this paper,the authors make a theoretical analysis on this problem and then prove it with some experiments so as to show that there are some potents in improving the effectiveness of video sequence coding.
摘要:In this paper,a new image compression coding algorithm based on Self-Organizing neural network,VQ+DPCM+DCT algorithm,is proposed.Experiment results show that the compression ratio is 31.8∶1,the PSNR is 35. 82dB(Lenna luminance image),and subjective image quality is very good.
摘要:Transition region is a special region located between the object and background in real images.This paper describes the properties of transition region and the method for determining the transition region.This paper also proposes a new automatic segmentation algorithm based on transition region and compares it with some common employed threshold selection algorithms.The analysis and experimental results show that this new algorithm is computational effciency and robust against disturbance.
关键词:Image analysis;transition region;Effective average gradient;Image segmentation;threshold selection
摘要:A partially adaptive approach based on orthogonal projection was proposed in this paper.Unlike these approaches based on eigenstructure or output power minimization,knowledge of jammer enviroment,eigenvalue decomposition or no-linear optimisation computation are not required.Thus, its algorithm is robust ,simple and fast.Simulations illustrate the effectiveness of this approach.
摘要:In this paper,it is proved that τ (G) can be sued to describe the connectivity of a graph.The properties of the toughness of a graph, such as the range of value,enumeration,computing, Are studied.
关键词:Toughness of a graph;connectivity;Undirect graph
摘要:This paper presents a new image compression and decompression scheme which uses quickly Local Iteration Function System[LIFS].We obtain 25 to 1 compression as Peak SNR of 30dB.
摘要:Based on the recently developed chaos and fractal theories,this paper introduces new methods for speech signal processing.A novel phase space reconstruction algorithms is proposed for speech signal,the distributions of the maximum Lyapunov exponent and the fractal dimension of speech signal are tested and analyzed statistically,and a new low bit rate speech coding method based on fractal code excited linear prediction is presented.The results of this study indicate that chaos and fractal theories have great potentials in the field of speech signal processing.
关键词:chaos theory;fractal theory;speech signal processing
摘要:Some results about estimation of the domain of attraction of memory patterns and exponential convergence rate of the network trajectories to memory patterns for Hopfield’s continuous-time associative memory are obtained. These results can be used not only for evaluation of errorcorrection capability of Hopfield contimuous-time associative memories,but also for the synthesis of efficient continuous-time associative memory neural networks.
关键词:neural networks;Fundamental constraints;domain of attraction;Exponential convergence rate
摘要:A few more new results about the period distribution of block codes are proved.For example the relationship between period distributions and ranks of some special matrices are discovered;the exact values of period distributions for a few block codes are calculated.At the end of this paper some open problems are proposed.
摘要:From the viewpoint of mutual information,correlations between a nonlinear combining function and combinations of part of its variables ae studied,and the effect of such correlations on the security of stream ciphers is investigated.Using Walsh transform, correlations between a Bent function and part of its variables are analysed,and a class of Boolean functions is introduced which hold the same asymptotic spectrum properties as Bent functions.
关键词:stream cipher;Boolean function;correlation analysis;Walsh transform;Bent function
摘要:With the concept of characteristic function,a limit distribution theorem is obtained for sum of independent random variables in a finite field. As applications of the theorem,the construction of nonlinear combining functions is addressed in finite fields,and a class of nonlinear combining functions is introduced which has small correlations with linear functions.
摘要:We put forward a novel system of two channels free-space optical communication with quantum optical filters as critical instruments.When we use this system to transmit voice and pictures at the background of strong sunlights,the signal to noise ratio is improved by 2~3 orders.
摘要:In this paper,the performance of the DMPSK/SSMA systems with trellis-coded modulation over indoor channel is evaluated. The Gaussian approximation is employed to analyze the multipath interference. The pairwise error probability is obtained by means of the Laplac transform and the residue theorem.The bit error probability of the coded SSMA systems are presented and compared with that of the uncoded DMPSK/SSMA systems.
摘要:A novel Two-Stage Least Square(TS-LS) algorithm has been proposed based on the oretical derivation for nonstationary dynamic systems with input slow time-varying disturbance,also,its performance has been theoretically analysised in detail. Numerical simulation results show its efficiency and some important properties.
关键词:Two-Stage Least Square;Slow time-varying disturbance;Correlativity;system identification
摘要:This paper analyzes the performance of noncoherent Hybrid direct sequence(DS)slow frequency hopping (SFH) spread spectrum multiple access (SSMA) communications which are used over Rayleigh multipath fading channel. Based on the assumption that the sum of the multipath interference,multiple access interference and the channel noise is a Gaussian random variable,the approximation for the average error probability of systerms with equal gain combining (EGC)and selective maximum (SM) multipath time diversity algorithms are developed.Analysis and numerical results indicate that multipath time diversity can be used to significantly improve the performance of DS-SFH SSMA system over Rayleigh multipath fading channel and that systems with EGC diversity have improved the performance compared to SC diversity.
摘要:Based on wavelet transform,a new method for detecting transient signals has been proposed.A class of wavelet basis functions has been introduced to detect transient signals with unknown waveforms and arrival times.By using multiscale wavelet transform, the proposed detector works well under low SNR.The detection statistic based on the wavelet coefficients has been given and the performance of the detector has been evaluated.Theoretical analysis and computer simulations have shown that the proposed scheme is effective.
摘要:Simulated Annealing(SA)is a powerful statistical computational technique for obtaining approximate solutions to combinatorial optimization problems.An efficient simulated annealing algorithm is employed to design good phase coding waveforms,which include binary sequences,binary sequence sets,and polyphase sequence.Some of the designed results and their properties are presented in the paper.Such phase coding waveforms may be applicable to radar and communication areas.
摘要:This paper reviews the research works on the statistical characteristics and modeling of variable bit rate video coding in ATM environment.It gives some measures for characterizing the burstiness of video sources and investigates several statistical models.The paper also shows the applications of statistical models and their influences on ATM networks.After summarizing prevous research works,it indicates that some problems should further be studied and solved.
关键词:Video sources;Variable bit rate coding;Statistical characteristics;Modeling
摘要:This paper proposed that the design of fast search path for image matching should be done right in the feature space of the reference image. By rearranging the feature space into an orderly structure,we can have a global grasp of its distribution.In this way,we are no longer restricted to the local information that traditional fast search algorithms can only rely on,hence the problem of local minima is avoided.In the orderly feature space,binary search can be adopted as the core of a new fast search algorithm.Results from simulated experiments have shown that the new algorithm can yield excellent searching speed and very good matching accuracy.
摘要:Based on duality theory of linear programming,a neural network approach for least absolute value problems is proposed in this paper.The approach consists of two phases,at first.an approximate duality solution is obtained by using LSSM neural network,then the prime solution is given by using T-H neural network.It is demonstrated that the asymptotically stable point of T-H networkis the exact prime solution. Simulation results are given to show the validity of the proposed neural network approach.
摘要:In this paper,Finite-State Radial Basis Function (FSRBF) network structure is proposed,which can be trained layer-by-layer by means of k-means clustering and LMS algorithm.The properties of FSRBF net and subnet are studied by Chinese digit recognition experiments. The results suggest that FSRBF is suitable for time series signal processing and can be generalized easily.
关键词:radial basis function;speech recognition;neural networks;Finite state
摘要:The stability test theorems in frequency domain for two-dimensional (2-D) discrete systems have been presented in the paper,which can be used to 2-D discrete systems with arbitrary order. The test algorithm can avoid loop integral so that the computation amount is reduced and the test exactness has been improved.
摘要:The paper investigates the adaptive IIR filter realized with cascade-form,which provides robust stability monitoring with less complexity than the direct-form realization.The adaptive algorithm presented in the paper overcome the problems, such as,initial state and multiple-repeated poles.The convergence performance of adaptive IIR filter illustrated by simulation is improved significantly.
摘要:A new Learning algorithm for fuzzy associative memories (FAMs) with Max-Min composition units is presented in this paper.Under some conditions,multiple fuzzy pattern pairs can be encoded in a single FAM by the proposed algorithm and the perfect association of stored pairs can be achieved.
摘要:The nonlinear intermodulation distortion in active filter is analyzed using the intermodulation balance technique.The computed results agree well with the measurement.
摘要:This paper presents a fuzzy tone recognition approach for Chinese speech. One level clipper,reducing sampling rate and linear interpolation form the fast pitch detecion. Using the theory of fuzzy sets the membership function is adopted for the four tone pattern classification.This method is characterized for its simple,reliable,fast real-time performance and tolerance.
摘要:This paper proposes a method of realizing a media transform from Chinese text to a front mouth image.First we construct a lip prounciation model which comes from an elastic deformable model.Then using texture mapping technique and basic phonemic parameters of a front mouth image of various syllables. As an example we demonstrate a sequence of images for pronouncing"Xi an dian zi ke ji da xue".