摘要:This paper presents a novel and efficient real-time multi-target imaging technique.First,the motion parameters of each target is estimated by applying a new time-frequency technique(the modified short-time Fourier’transform)to the received echoes. Then ,motion compensation is performed to get the image of the targets.Although the image is poor,the contour of target is distinguishable one from another.So we can get the signal of individual target by windowing. After applying a new ISAR motion compensation technique which called scattering centroid tracking to the windowed signal,we finally obtain the high-quality image of every target. Compared with the parameterbased (such as MLE)compensation, the proposed technique requires much less computation and is easy for implementation. Computer simulations have shown the effectiveness of the technique.
摘要:In this paper,we intreduce a range image segmentation method based on the basic operations of mathematical morphology. First the peak image containing jump edges and valley image containing roof edges are obtained using the mathematical morphological operators. Then the segmentation is accomplised by controlled region growing.This technique has been applied successfully to many real range images.In comparison with traditional methods, the proposed scheme has higher speed and better noise immunity.
摘要:Based on seeing the process of muiti-layered neural network learning to get optimal weight set as the self-organizing and self-learning parameter identification of non-linear dynamic system model ,the GBP adaptive recursive identification algorithm and medel test methed for time-invariant or time-variant dynamic systems are intreduced in this paper. GBP recursive algorithm includes dual-direction identification processes.One is at the sampling interval ,the other is at the time series.Also it is extension of the existing algorithm for multi-layered NN. Simulation study shows its efficlency.
关键词:Multi-layered feed forward neural network;Nonlinear dynamic system;Adaptive identification;Model test
摘要:A new decision-directed cost function (criterion) for blind equalization proposed by Daoben Li and Proakis is firstly introduced in this paper. The cost function can simultaneously force the error probability as small as possible and the overall channel’s impluse response function as close as possible to a desired one(full or partial response).The least-square algorithm which adopts the above cost function is also intreduced. Simulation results demonstrate the effectiveness of the algorithm.
摘要:A modified VQ algorithm based on"Structure Constraint Matrix"is presented in this paper. After image decomposition,a multiresolution pyramidial representaion is yielded and the geometric image structure is preserved.We use Vector Quantization to code the wavelet coefficients."Structure Constraint Matrix" is proposed tO describe the correlationship caused by the space-frequency localization property of wavelet transform ,and to reduce the computational load in VQ algorithm.Experimental results are given to show its better perfomance.
摘要:Some SOFM-based image VQ codebook design algorithms have been obtained by arranging output nodes of Kohonen network in 1,2,and 8 dimensional space respectively.The experimental results show that the quantization performance of SOFM-based VQ is similar to that of LBG-based VQ ,while the former has better ability of resisting channel errors.
摘要:Some adjacent image blocks may have the same address when ceding them with vector quantization(VQ),especially in the stationary and uniform regions of an image. This paper proposes a correlation vector quantization scheme to decorrelate the addresses of adjacent coded blocks.Correlation codebook and improved self-organizing feature maps(ISOFM) codebook are sued to encode the four adjacent blocks simutaneously in a window.For a typical image"Lenna",the computational quantity during the coding process is reduced by a factor of 2 and the bit rate is reduced by about 40% over the memoryless VQ. Since preferential measures are taken to the vectors belonging to the edge classes in the training process of Kohonen’s self-organization neural network,the subjective quality of the decoded image is significatly improved.
摘要:We analyse some properties of morphological filtered signals. We derive the conditions in which a signal is invariant to morphological filters. Then we point out that morphological filtered signals are locally monotonic in certain regions and the monotonicity can be predicted.
摘要:In this paper,the guided scrambling cede,a new line cede,is studied more deeply.It is proved that the output characteristics of guided scrambling code is similar to that of white noise in some conditions. The two technical metheds of hardware and software are set up and successful. The experiment shows that the results are believable.
摘要:An algorithm for determing the punctured distance of a generalized TMR cede has been derived,which provides a general method for designing generalized TMR codes with arbitrary byte lenth.
摘要:A method to get the solution of the Chinese Remainder Theorem by solving a differential equation is proposed. Stable equillibrium points and error of the optimal solution of the nonlinear differential equation with residue reduction operation coefficients are discussed. The nonlinear differential equation formulated in this paper can converge to the solution of the Chinese Remainder Theorem with arbitrarily small error.The differential equation can be solved in real-time by means of artifitial neural networks. This network can be used to realize residue-to-decimal conversion. The results of our computer simulation demonstrate that the proposed theory is reasonable.
摘要:we put forward the implementation with VLSI structure systolic array.we present four kinds of procesing cells based on paralleled algorithm and recurrence formula of triangle.So we inferenced four kinds of systolic array.One is two-dimensional,the others are one-dimensional.And WE also test and verify the function of the arrays on Transputer.These arrays are greatly parralleled and very simple,regular and modular.So these structures are simple with VLSI structur impementation.
摘要:The mixed traffic including bursty and random is modeled as bursty source and Bernoulli source respectively. Partial buffer sharing scheme is modified to meet the needs of different QOS of two classes of traffic. The effects of mean bursty length,buffer threshold,traffic load and channel grouping on system performance are investigated.
摘要:This paper discusses the performance of spread spectrum slotted ALOHA of integrated packet voice/data of satellite.On the condition of improved GAUSSIAN MAI(Multiple Access Interfence)and finite user in a certain slot,the formula of system throughput and average delay are obtained. These formula are more accurate than those on the condition of standard GAUSSLAN MAI and infinite user.The simulation results indicat that the protocol performance is superior to ordinary protocol.
摘要:This paper proposes a new recognition approach Which can recognize a 3-D object from a single perspective projected line drawing. First we present some new methods of recognizing spacial polygons,then introduce a distributed appearance representation with ordered connection for model polyhedron,and propose a gradually characteristic-expending recognition algorithm.The algorithm has been demonstrated by computer simulations.
摘要:M/G/1 queueing system with repairable service station is further analyzed in this paper.some new reliability quantities of the service station are obtained,for example,the joint probability distribution of the busy period and the service station failure number in this busy period. Further,the asymptotic expressions for calculating the transient probability that the service station fails at the time t and the expected failure number in(0,t],and the calculating example for showing the validity of these asymptotic expressions are given.
摘要:A fast algorithm with the simple structure is presented for the computation of all four types of the radix-2 discrete W transform(DWT). Experimetal results have shown that the subroutines of proposed fast W transform are more efficient than any other published ones. In addition ,it has smaller error because of using cosine multipliers.
摘要:The introduction of the asynchronous transfer mode(ATM) concept has significantly influenced the coding of video services.This paper discussses the advantages and problems of variable bit rate(VBR) coding. It briefly introduces various cell loss protection strategies and ,in particular,elaborates on layered coding techniques.It also reviews different VBR layered coding algorithms and gives some recommendations for further study.
关键词:Variable bit rate coding;ATM;Cell loss;Layered coding
摘要:In this paper,the approach to identify three-dimensional objects by moments is presented.We have used moment invariants and standard moments to identify five different aircraft,and compared the performance of these two techniques. We have also studied the effects of the number of feature elements on classification results ,and the invariant property to scale change of moments under the discrete situation. Experiment results show that the order three standard moments with the variance balancing technique can classify aircraft effectively.
摘要:A neural network soft-decision deceder of linear block cedes is structured and a neural network soft-decision decoding algorithm for cyclic codes is presented. The computer simulation for the Golay cede’s soft-decision deceding shows that the new algorithm has both very fast decoding speed and good performance.
摘要:A new soft decision decoding algorithm for (24, 12,8) extended Golay code is presented,and the complexity of it is at most 507 real additions,which is,to our knowlege,the best among the existing algorihtms. It is proved to be a generalized minimum distance (GMD)algorithm.Computer simulation shows that its performance is almost the same as that of Maximum likelihood decoding(MLD).
摘要:In this paper’we propose a scheme for antomatic seal imprint identification using moment invariants as attribute.A major advantage of our scheme is that our cheme needs not rotate ,locate and match ,high speed identification ,small internal memory space ,and unrestrained for shape of seals. Experimental results confirm that the proposed scheme has certain ability of identification and can be used as a scheme of the seal identification.
摘要:In this paper,we point out that the conventional orthogonal condition of complex signals,which is stated in most references,is not a necessary condition of the orthogonality of the complex envelopes of bandpass signals in communication systems.Based on the equivalent lowpass method,the author derives the equivalent orthogonal condition of the complex envelopes of two real bandpass signals using the complex crosscorrelation function of the complex envelopes of real bandpass signals. It can be widely used for the orthogonality analysis of real bandpass signals in communication systems. The analysis of OQAM signal orthogonality is completely emphasized in both sides of same subchannel and adjacent subchannels.
关键词:Equivalent orthogonal condition;Complex envelope;OQAM signal
摘要:The basic CELP structure is introduced in this paper.A constrained-stochastic excitation linear predictive coding scheme is suggested.It links the amount of stochastic excitation codewords injected into the synthetic filter to a performance index of adaptive codebcok. The noisy effects of the excitation on synthetic speech are efficiently reduced. The computer simulation results show the subjective improvement of speech quality with this approach.
关键词:speech processing;Code-excited linear predictive;Codebook
摘要:This paper discusses the problem of estimating the frequencies of sinusoidal signals in colored non-Gaussian ARMA noise.Combining third-order cumulants with autocorrelations,we propose a new method in which AR parameters of noise model are first estimated ;then the measurement sequence is prefiltered ;finally , the parameters of signal medel are estimated. Simulation examples are presented to demonstrate the well frequency estimation performance of this method.
摘要:This paper advances a three-dimensional space interpolation method of grey/depth image sequences. The way breaks away from the limit of original photographing route. Pictures can cruise at will in space. By space sparse sampling, great memory capacity can be saved and reproduced scenes can be controlled.Fast and practical algorithm of scattered space lattice and that of "Warp" algorithm with proper depth are also studied. Results of simulated experiments show that the new method is absolutely feasible.
关键词:Grey/depth image;three-dimensional space interpolation;Computer image synthesis;algorithm