This paper presents an adaptive algorithm to design optimal structuring element of morphological filters. The adaptation process is similar to the classical Least Mean Square (LMS) algorithm. At first, the optimization formulas of the structuring element are derived in the case of erosion and dilation. Then, adaptation formulas for an arbitrary composition of erosions and dilations are obtained. The simulation experimental results show that the method improves the performances of morphological filters and has properties of easy design and good practicability.

This paper combines the methods of multiresolution analysis with the approaches of Kalman filtering using the tool of wavelet transforms, the measurement multiresolutional preprocessing method for signal denoising is proposed. Because the wavelet transform possesses the characteristic of lowpass, it can effectively restrain noises can increase the signal to noise ratio, thereby, by use of wavelet transform the good estimation results for signal processing are obtained. Computer simulation shows that the algorithm is feasible and effective.

Synchronous run-length DSGM-SC compression technique is presented in this paper for one-dimensional random signal with local fast variation and expansive slow variation. First, run length DSGM detection is implemented to detect fast variation part, then the fast part’s spectrum is compressed with interpolation as optimal times principle to improve integrative correlation, and transform coding is adopted to get high compression ratio. Data files, which include high and low spectrum information, are synchronously compressed. Position information is prerecorded to fulfill time recover without distortion. The compression ratio with the method can reach 32.4 to electrocardiogram data recorded by multi-functional bedside instrument of NEC, BSM8300 while the fidelity of medical treatment is high. The research has application in one-dimensional data analyzing, recording, transmitting and integrated service.

A new method for designing the prototype filter for cosine-modulated filter banks with near-perfect-reconstruction is proposed. The prototype filter is formulated as the form A(z²)B(z) and indirectly designed by varying the passband edge of the FIR filter A(z) with low order, where B(z) is a maximally flat FIR filter. A design example is presented to show that filter banks with high stop attenuation can be achieved using this method.

A new kind of bilinear time-frequency distribution-adaptive Gaussian-kernel distribution (AGD) is proposed according to multi-component linear frequency-modulated (LFM) signals, and a new efficient kernel estimation method is presented. With the AGD as an example, the influences of the adaptive kernel to the signalterms and crossterms are discussed in detail, which from another point of view indicates its advantages over the fixed kernel. Finally, a common adaptive kernel design method based on ambiguity domain for LFM signals is generalized. Computer simulations confirm the performance improvement of AGD and its corresponding analysis.

Two algorithms are discussed for estimating the power of the signal component from a noisy signal. They have simple structures, thus are computationally very efficient. They can be designed to be predictive. The bias and variance of the power estimation can be minimized or reduced by optimizing the FIR subsystems. It is shown by simulations that employing them to the power control loop in a CDMA mobile communication system results in higher control abilities and lower BERs when signal-to-noise ratio is low.

By using 1-dimensional DOA searching rather than resort to multidimensional parametric searching, an algorithm for DOA estimation in the presence of mutual coupling is got. A necessary condition for identifiability is proposed. The ambiguous peaks appeared in certain cases must be eliminated. Detailed analysis of ambiguous problem is presented and is illuminated by numerical examples.

Infrared radiation and infrared bicolor ratio are studied in the paper. The feature of infrared bicolor quasi-imaging recognition is investigated. Robust iterative sub-block gray scale bilevel is studied for infrared bicolor quasi image with space blur and low SNR, then fuzzy spatial segmentation and feature parameter recognition are studied for 2-bit image of infrared quasi image. Infrared quasi image frame accumulation and original dragging are studied to fulfill precise target detection with lower false and omission possibility. Successive segmentation, recognition and locked tracking with adaptive movement compensation are studied for sparse quasi images. The experiment reveals that infrared bicolor quasi image tracking is adaptive to terminal guiding of stationary, slow, and fast infrared targets.

This paper analyzes the statistics of speckle in SAR. And based on wavelet transform, this paper puts forwards a new method for speckle suppression. After wavelet transform, Lee filter is adopted for HH part and CFAR (Constant False Alarm Ratio) is used for HG, GH, HH. This method not only improves the ENL of images, but also keeps the important targets and edge. Compared with Lee algorithm that was generally used in SAR speckle filtering, this new method has the great advantage in speckle suppression and edge keeping. The experimental results prove the effectiveness of the method.

In this paper, the method to determine the bearings and ranges to three highlight sections of a ship is proposed and studied by means of a small rhombic acoustic intensity array. The simulation results show that the method is correct. An experimental facility is designed and manufactured. The scaledown experiment determining the bearings and ranges to three highlight sections of a ship is completed in serni-anechoic room. The experimental results are consistent with simulation results.

This paper presents a model identification approach of nonlinear-systems. Through structure identification (learning) and parameter identification (learning), a fuzzy neural network is constructed. Having adjusted the weights of the fuzzy neural network, a precise fuzzy model is obtained. Finally, the effectiveness of the proposed technique is confirmed by the simulation results of two nonlinear systems.

A novel fair queuing algorithm LL-DRR(Low Latency Deficit Round Robin)is proposed in this paper, which is based on DRR (Deficit Round Robin) algorithm and suitable to schedule variable length packets. The simulation and theoretical analysis show that the delay performance of LL-DRR is much better than DRR, and the maximum packet delay of a session in LL-DRR is independent of the number of sessions. LL-DRR can support real time service. LL-DRR inherits the fairness of DRR on average throughput. LL-DRR is simple to implement and applicable to high speed networks.

A generalized discrete Fourier transfom is used to give a fast algorithm for determing the linear complexity of a pseudo-random sequence with periodic n2^v, and a new conclusion of Hasse derivatives of polynoimial on GF(2) are proposed.

The study and realization of elliptic curve cryptography have become to be the mainstream of the study on public cryptography, and the selection of secure elliptic curves in cipher and their base points is the basis of the implementing of the elliptic curves cryptography. In this paper, the selection algorithm of the secure elliptic curves and their base point over large prime number is discussed, further more, the algorithm is implemented successfully in C language using the MIRACL system.

Optimal design for C/S system is studied using system analysis method. Firstly, C/S system with multiple type servers based on queuing theory and analytic performance results for the system are presented. A mathematical model for system optimal design and relative algorithm under different conditions are then obtained. Finally, an example is followed to illustrate the computing process. This method is important to analyze the bottleneck of C/S system and optimize the system programming.

Starting from homogeneous scalar Helmholtz’s equations associated with Borgnis potential function in a cylindrical coordinate system, and based on the standing wave concept of slow-wave introduced in the paper, the analytic expressions of the dispersion relation for azimuthally symmetric transverse magnetic modes in the split cavity with three cavities are derived, by using boundary conditions for Borgnis potential function in conjunction with field matching conditions at the common interface between the adjacent subregions. Besides, the effect of the high frequency structure dimensions of split cavity with three cavities on its resonant frequency is analyzed in detail.

The input impedance of antenna array that consists of dual-cylindrical units is investigated. In this paper, the antenna array is separated into two parts: feeding sub-system and radiant sub-system. Simultaneously, the radiant sub-system is equivalent to a N-port network. First, a set of voltage equations is presented according to the transmission line theory, where the voltages, currents, and impedances for each port of network are deduced. Secondly, the relations between total input impedance and that of each port are given. It is indicated that the input impedance of the antenna can be matched by moving the two pistons at the ends. The measured data of the antenna sample verify the results of the calculation.

An algorithm for correcting probe position errors in planar near-field measurements is presented and simulated. The calculation results show that the algorithm has fast convergence rate, good numerical stability, high precision of calculation and is a practical and effective method for correcting probe position errors in planar near-field measurements.

Based on black box analysis, the unharmed decryption of the encrypted programmable logic device has been realized successfully with Local Exhaustion Approach(LEA) which is proposed to deal with enormous data acquisition and processing in PC. With the development tendency of object oriented method, an improved method for the unharmed decryption of the encrypted programmable logic device is given from object oriented analysis and object oriented design.

This paper studies mainly different properties of non-stationary signals and different kinds of noise under wavelet transformation, and presents a novel algorithm based on wavelet transformation and order statistic filter to attenuate noises. The algorithm attenuates the noise by choosing local threshold through the relativity property between different wavelet transformation with different scales, the different properties of different noises under wavelet transform and the order statistical theory. The simulation experiments show that that the algorithm has good property for different kinds of noise.

Three families of authentication codes with arbitration are constructed from unitary geometry, the parameters of the codes are computed. Assuming the encoding rules of the transmitter and the decoding rules of the receiver are chosen according to a uniform probability distribution, the probabilities of each kind of deceptions are also computed.

A new nonlinear dynamical model is presented and numerically studied. It is shown that the model has a positive Lyapunov exponent, it also has pseudo-random waveform and continuous power spectra, so it is a chaos generator. Since the conditional Lyapunov exponents of its linear subsystem are all negative, the linear subsystem is synchronizable and the new chaos generator can be used for chaotic communication. Two kinds of piecewise-linear circuit realization based on the model are given.

In this paper, a modified Chua s circuit is presented, its properties are studied by numerical simulation and the Lyapunov exponents of the modified Cliua s circuit are calculated. A simple circuit realization method for the modified Cliua s circuit is also given.