In the multistage imaging proc ssing for SAR digital imaging and the applications of SAR image, it is often required to extract the luminance edge features of the SAR imagery. It has been studied to extract the luminance edge features for the ordinary imagery-produced by visible or infrared imaging system The methods using gradient are effectivery for the extraction of the luminance edge features of the ordinary imagery and commonly used. Because the serious noise of coherent speckle exists in the SAR imagery, some people believe that edge extraction using gradient for SAR imagery gives poor results. However in this paper, a rather ideal method is derived for the extraction of luminance edge features for SAR imagery with the consideration of the characteristics of SAR imagery. This method uses the relative average gradient and combines detection with tracking.

Householder transform is used to triangularize the data matrix which is based on the linear prediction error equation. It is proved that the sum of squared residuals for each AR order can be obtained by the main diagonal elements of upper triangular matrix, so the column by column procedure can be used to develop a recursive algorithm for AR modeling and spectral estimation. In the most cases, the presented algorithm yield the same results as the covariance method or modified covariance method does. But in some special cases where the numericall ill-conditioned problems are so serious that the covariance method and inodified covariance method fail to estimate AR spectrum, the presented algorithm still tends to keep good performance. The typical computational results are given finally.

The capabilities of decoding beyond the bound of the. complex-rotary codes and its dual codes are analysed. It is obtained that the complex-rotary codes with t = (p+1)/2 can correct C^t+1_p²+p(p-1)-p²C^t_p+1 errors of (t +1) and its dual codes can correct C^t₁+1_p²+2t1p-2tpC^t₁+1_p+1 errors of (t₁+1), where t₁= [(p+1)/2]-1 and p is a prime.

The problems about the periods, the circle numbers, the distribution properties of the linear recurring sequences over Z/(m) and the polynomials in Z/(m) [x] are discussed.

A numerical method of direct search optimization is presented for calculating parameters of a reciprocal lossless two-port network. Some examples are given. It is pointed out that the results obtained by different methods agree with one another.

The eigenfields in finline are expanded in terms of the hybrid modes. The propagation constants of the dominant mode and the higer modes as well as the associated eigenfields in the spectral domain are then determined by using the spectral domain method. Finally the hybrid-mode scattering characteristics of finline impedance discontinuities are analysed in combination with the mode-matching technique. The propagation constants of the first 30 modes in unilateral finline and the first 20 modes in bilateral finline as well as the scattering parameters of the impedance discontinuities are computed. The computed results agree well with the published data. The problems about the numerical convergence are also discussed.

According to the model of equivalent dipoles for active molecules, the analysis of inelastic EM scattering by active molecules embedded in a sphere is given by the method of dyadic Green s functions at first, and rhen, based on the theory of elastic muptiple scattering by randomly distributed spherical scatterers, a new theory for analysis of inelastic multiple scattering by active molecules embedded in randomly distributed spherical scatterers is developed. This theory gives the expansions of the multiple scattering fields in all space region in terms of vector spherical wave functions in which the expansion coefficients can be solved from a set of coupled linear equations.

Based on the strong fluctuation theory, the calculating method for effective electromagnetic parameters of srochastically oriented granular media is derived. Applying the theory to study absorbed materials, the optimum shape of absorbed particles in antiradar coat-multiple shape ferrite absorbed materials is ober effective permittivity and permeability of multiple shape ferrite absorbed materials is obtained. The measured results are in good agree-ment with the calculating ones.

A new method for analysing the stabilities of analog electronic neural networks is presented. The energy functions with clear and definite physical meaning are derived by introducing the static equivalent circuit models, which has expanded the Tellegen Theorem for application on circuit analysis. The method used to derive the energy functions of nets from first order differential equations is valid for all first order continuous autonomous systems. The stability analysis of cellular neural networks is made by the use of stationary cocontent theorem. Some results are instructive for the network implementation on circuits.

A new method of 3-D reconstruction of space object from its space continual series images is introduced based on the principle of sampling. Compared with old methods, this method has many advantages such as higher precision and fewer requested known points. The result of reconstruction with this method is an t.ntity which involves the exterior shape and inte-tior construction information of the object simultaneously.

A simple method of analysing sensitivities of the transfer functions and the errors between the transfer functions and the corresponding direct transfer functions to parasitic effects, by using the relations between the transfer function, return difference, null return difference and sensitivity, is presented. Also, an example to illustrate how to analyze circuits, containing operational amplifiers is given.

An approach for caculating sensitivity of electronic circuits is presented. The approach is based on the incremental network, and some improvements are obtained. The main advantage is that it can calculate sensitivities of all network outputs with respect to all network parameters by solving a system of linear equations only once. A simple example is given to illustrate its application.

On the basis of experiments, a new conductive expression of photoconductive switch is proposed. According to this expression, the generation and sampling of high speed pulses, as well as the characteristics of photoelectron correlation measurement are analysed. The analysis agrees with experimental results and will be useful to the electronic measurement by photoelectron correlation method. A rather simple processing of devices is adopted. The rise time of photoconductive switch is 100 ps. It is sure that the rise time may obtain a ten-fold improvement only with ion-bombardment to the substrate.

Statistical analyses on the quantized error of pulse-counting in digitized measurement are made. A Probability distribution of pulse-counting in random measurement is obtamed The probable applications in measurement are studied.

An electrostatic wiggler FEL can use an electrostatic wiggler instead of a mag-netostic wiggler to generate coherent radiations. A new kind of electrostatic wiggler is presented and studied. The results show that a circularly-polarized electrostatic wiggler field with a period from 3 cm to 5 mm and a relatively arge amplitude is available. It is sure that this wiggler will further promote the studies of FELs.

The characteristics and principles of a new current regulator are analysed- As A result of adopting the current regulator as the current supply and realizing the optirrium temperature compensation, a very stable output current have been obtained. The device not only has large dynamic impedance, current temperature coefficient less than or equal to 10^-4/℃. but also has the advantage of continuous adjustability of output current from 1 mA to 500 mA.

This paper is intented to explain that Molecular Electronic Device (MED) as the third generation devices following vacuum valves and semiconductors will take part in information science on accounting of the development not only of devices but also of new computing systems based on evolution principle. Some models of associative memories and networks built from MED are proposed. The objects and ways to develop MED in the near future are also discussed.