K-emans alsor/thm was used to classify cloud-type that is low, mix and cumuionimbus Tnitiat ciustercenters and K parameter is given in this paper by coatse computins and Fisher’s alsorithm. Results indicate that performance index is minimized and mix cloud is well clallified.

Texture provides an important acurce of information about the threedicensfornarry information of visible surface particulary for stationary conccular views. To recover three dicmensinoary information, the distorging effects of pro jection must be distinguished from properties of the texture on which the distrortion acts. In this paper, we show an approximated maximum likelihood estimation method by which we find surface oriemtation of the visible surface in gaussian sphere using local analysis of the texture, In addition assuming that an orthographic projection and a circle is an image formation system and a texel(texture element)respectively we derive the surface orientation from the distribution of variation by means of orthographic pro jemction of a tangent directon which exstis regulary in the are length of a circle we present the orientation parameters of textured surface with saint and tilt and also the surface normal of the resvlted surface orimentation as needle map. This algorithm was applied to geograghic contour and synthetic textures.

Some modification of laplacian pyramidal coding have been done and tested for a test image, From our studies, its apperar that 3-D redictor which exploits inter and intra plane redundancy simultaneously somewhat imporves the image quality, We also have tried to reduce the bit rate by only sending zeroth plane image values that correstpond to estimated dege points which can be obtained from the first plane For both cases the feedback of quantization errof in the previous plane has been in the reconstruction of each plane. Subjective and SNR tests show the better performance of the studied methods oner the conventional one.

A new efficientcoding system which can transmit video conferenceof viedeophone signals at a 64kbps is proposed. In addition to the interframe and CRC (Conditional Repleni shment Coding) system, BTVQ (Binary Tree searched Vector Quantizer)and RLC (Run Length Coding) methods are incorporated. Couble buffer memory is used for simple comtrol of channel symbol transmission and memory underflow And also buffer memory onerfolw is easily controlled by the thresholds of a MAD (Moving Area Betector)

The purpose of this paper is to compare a class of vector quantization techniques which include GVQ(Genera VQ) MSVQ(Mean separated VQ) and DCT_VQ The VQ techniques are applied to six images and both subjective and objective performance comparison are made The results indicate that the transform domain approach(DCT_VQ) yields more syable results than the spatial domain method (GVQ, MSVQ)

In this paper we describe the syrucure of vector trellis comding system for image data coression. And the simuatlion results are compared with those of the genera VQ(vector quantlzation)on the basis of SNR search time and memory requiement

In this paper a target tracking algorithm of the centroid extraction from moving edges is proposed, It aims to avoid the difficulty of imahe segmentation in case of the centroid extraction from one frame. The performance of the proposed algorithmfor noisy and occluded images is discussed Finally it is also applied to a real time target tracker.

In the difference picture(DP) which is obtatined from two subsequent images we detect edge intersection points(EIP) and estimate the directions in which edges disappear at those points, Then we group the DP regions which the motion of the object makes and we extract the moving object.

In this paper we propose a segmentiation algorithm to get binary images by using edge positional information We use the global and local properties of an image in this algorithm. Computer simulation shows that its segmentation results are better than thyose of conventional algorthm using only global characteristics.

In tihis paper an improved matching method using subtemplate of moving region and 3-step search algorithm is proposed It reduces heavy computational load of the conventional method. The proposed method is applied to an automatic target tracker using high speed 16 bit microprocessor in order to track one moving target in real time. Expericmental results show that the proposed method has better performance over the conventional methodin spite of greately reducing the computational load, even in came complex background.

In this study the moving information extraction techniques of moving objects are processed digital imaqe data by sampling three frames in a fixed-bacqround two-dimensional line sequence image the brightness of interframe are compared to extract difference image and difference image are two level formed and neighber averged From neigbber averaged image the parameters for recoqnition of the object are the number of contorur pixels, the number of vertex points and the distance between the vertex points Agtercomparing the same object the moving distance obtained from the coordinate which is constructed by the bit processing of the digital data and the moving velocity is obtained from the moving distance and the time interval between the first andsecond sampled frames.

In this paper a fast matching method using subtemplate and search and down technique to reduce very heavy computational load of the conventional matching method, is presented The proposed method is spplied to an automatic target tracker in order to track one moving object in comparatively simple backgoriund. Experimental results show that istperformanced is not so degraded in spite of high computational reduction as that of the conventional matching method.

No 1A ESS must operate in real time. Call processing capacity is the maximm number of originating plus incoming calls which the processor can process in some fixed interval of real time shile all service criteria have been satisfied. In this paper. we discuss the postcutower method of determining call processing capacity of No 1A ESS which can be applicable to an in service office and explain the usuage mithod of estimated call processing capacity of No 1A ESS in service

565Mb/s multiplexer/Demulitiplexer have been developed as a part of the 565Mb/s Optical Transmission System We describetheir functions and configurations with the design philosophy.

The basic fuctions design necessity and basic structure of a computerized systed which keeps records on the insided and outside plant facilities mechanizes the assigment of loop and exchange facilities maintains short jumpers on main distribution frame digtributes and controls the service order and upgates the exchange datasutomatically are considered in this paper.

The traffic characteristicsof digitalswich network depend on the structure blocking probability, path searching method and etc. This paper presents the study of TDX-1 swicth network traffic capacity by considering some decisive factors such as call processing software, switch network structrure and control schme Conclusively the study shows that the switch network of TDX-1 can handle approximately up to 1650 Erlang.

This paper describes the technology for the improvement of subxdriber loop utilization. Nearly all customer loops are now carried at voice frequency on individual wire pairs. Reenct digital technology improvements together with the increasing cost of wire, have resulted in some penetration of carrier systems into the loop plant. Moreover the increasing use of local digital switching systems should permit digital carrier systems to become an important factor in the loop plant in the future.

We implement an universal signaling equipment including many kinds of singling devices just as R2 MFC seder.receiver DTMF sender/receiver Tone sender sued in Digital Switching System the universal signaling equipment has two parts sender part and receiver part. The sender part is implemnted with ROM-LOOK-UP Table method and the receiver part is implemnted that the equipment satisfies the Recomandations of CCITT. Also the capability of the equipment is evaluated by the LCS(Local Call Simulatior)Test in the TDX-1A system.

In this paper the data link control procedures for the TDX-1 ESS had been designed and implemented in accordance with the CCITT recommandation X.25 LAPB After a general discussion on intermational standards of link montrol procedures the TDX-1 data link control procedures are described in detail.

본 논문에서는 유한 축적용량과 단일 도기 서버를 갖는 패킷 스위칭 노드로서 처리량이 향상될 수 있도록 하는 버퍼 운용방법으로 패킷 교체 수용 큐잉 시스템을 제시하였다. 시스템의 해석은 J.F CMANG 과 R.F CHAMG 이 제안한 시간 미소 단위 근사법을 이용하였다. 위의 시스템과 기존적인 패킷 수용 방법으로서 패킷이 시스템에 도착하는 순서에 따라 수용되어 처리 되는 J.F CHANG 과 R.F CHANG 이 해석한 큐잉 시스템에 대한 성능을 처리량과 시험옥텟의 Rjocking 확률로 비교한 결과 패킷 교체 수용 큐잉시스템의 성능이 더 향상되었음을 보였으며, 이 해석적 결과로 컴퓨터 시뮬레이션으로 검증하였다.

In this paper mimimm desperaioms of atep index single mode sptical fibers manufzctrured by the phase madulatiem method.From them we find that the caloulated data agree closely with the resulte measured at the wavelength around 1.3 um and that a method to shift sere dispersion wavelength at the wavelenghth around 1.55 um is to control refractive index profile and relative refractive difference

The method of Conjugate Gradient(C.G.M) is applied to the solution of current distribution from body of revolution (B.O.R).Using the C.G.M it is possible to analyze electrically large arbtrarily oriented B.O.R. The fundamental differences between C.G.M and moment method are outlined. This method converges for any initial guess and this techniqe gurantees a monotonic convergence. Numerical results are presented for electromagnetic case which show good agreement with moment solution.

In this paper the real-time analysis of vibration objects using projection moir\ulcorner method is realized. By measuring the vibration of two different speakers using three grationgs with different pitch. it is possible to analyze the vibrating speakers, so the fundamental frequency mode of speakers is obserbed at 550Hz and 105Hz respectively and phase lag is shown nonlinear characteristics. Also it is possible to measure the vibration as to nonperiodic signals as well as peoriodic signals.

When a 160QAM receiver in the band of spread spectrum system is interfered with a pseudo-random frequency-hopp-ing signal the interference signal has the form of the pulse amplitude probability density function. Each interfrering pulse amplitude probability density function Each interfering pulse amplitude is dependent upon theprobability density function of hopping frequency and the selectivity characteristic of 16QAM system in this paper the error rate equation of a system performance has been derived in the above condition which are interfrerence form and impulsive noise environmensts. The results of system analysis show us that the system performance is de-graded by impulsive noise and interference power, which improved as the frequency hopping bandsidth increases whit respect to receiver bandwidth

In this paper A NET-CUE System is designed and implemented for tramsmission of the network cuing informations with using data packet broad casting techniques. This system is composed with encoder and decoder. To show the performance of this system An eye height and bit error rate are cheched. The eye height is greater than 74% and The bit error rate is less than 4.6 * 10 The exper imental results show that this system provides a good quality of the operation.

In FH-SS systems when the bopping band is enough wide to onerlap with conventional band limited communication cethod (CBM). The portion of suchacts as an interfering signal. Thus it is gard to use them all together. This paper presents how the frequency gopping systems can simultaneously share the same band with CBM. The proposed mithod is that the frequency gopping band can arbitraily controlled by setting the specific input bith of digital frequency synthesizer to logical zero state We realized this by putting the hopping band Controller between pseudeo random generater and frequency synthesizer.

본 논문에서는 입력 신호인 정현파의 위상을 PGZC(positive going zero corssing)에 대해 한주기당 2회씩 추적하도록 하는 디지털 위상도기 회로(DPLL)을 제시한다. 제시된 루루프는 한 주기당 2번의 표본화를 갖도록 함으로써 한 주기당 하나의 표본을 취하는 기존의 DPLLqhek 정상 상태에서의 위상오차 변동범위가 전체적으로 1/2로 감소되었고, 연속 표본들간의 오차와 양자화 준위의 선택에 따라서 루우프의 천이 응답이 좋아짐을 알 수 있었다. 그 해석적 결과를 실제적으로 요구되는 조건류에 대하여 컴퓨터 시뮬레이션을 행함으로써 검증하였다.

The error rate equations of digital modulatuon techniques(M-ary PSK MSK Multi-level QAM) have been derived in the environment of Qaussian noise and Jamming And the results are shown in the figures as parameters of carrier power to noise power ration(CNR) carrier power to Jamming power ration(CJR) Jamming power to carrier power ration(JCR) alteration figure of envelope of Jamming(M) As the results of analysis the system performance is more degreeded by Qau-ssian Famming than Tone Famming In case that envelope of Famming has m-distribution, the s/stem parformance is desreeded in the deep alteration of envelope

A Semi-cusion VLSI Digital Filter Design for TDM/FDM tran-smultiplexer is decribed. Using the polyphase network approach a filter bank composed of only all-pasdigital filter sections was designed. The use of all-pass filters as basic building blocks is shown to provide a Transmultiplexer structure that has low computational requirements low quanization noise and hign modularity. A design of 1st order 2nd All pass filter is done using COMS 2um double metal.

The error rate eqations of digital modulated signals transmitted through the fading cdannel have been derived in the Gaussian/Impulsive noise environments Whing the derived equations for the error drobadillties of ASK, QAM, FSK, MSK, PSK, and DPSK signais, the error tate performance of digital modulation systems have been evaluated and represented in the graghes as parameters of carrier to \ulcornernoise power ratio (CNR) and fading figures The results show that in the fading environenet the error is occurred more frequently by gaussian noise in the deep fading Howerer in the shallow fading lmpulsive noise is more domiant than gaussian nosie in occurring the error

A Dual Disital Phase Locked Loop is analyzeddesigned and tested. Two specific confisurations are considered generations second and thisrd order DPLL’s and it is found using a computer simulation and verified therretically . As a result of computer simulation the characteristcof designed I-Dullis better than the at of P-DPLL or C-Dull