A graphic user interface for effectively browsing object-oriented databases in complex applications is designed and implemented. The rationale behind our design lies in enabling the users of various levels and needs to investigate the database according to their respective interests and desired depths. A novel idea in our design is the introduction of zooming techniques from a logical view, which vidualize the backbone abstraction concepts of the object-oriented data model. These objectives are verified by evaluating the results of its implementation.

A new graph problem is formulated to limit the maximum path length of a general directed graph when a minimal set of vertices together with their incident edges are removed from the graph. An optimal algorithm and a heuristic algorithm are proposed and the proposed heuristic algorithm is shown to be effective through experiments using a collection of graphs obtained from large sequential circuits. The heuristic algorithm is based on a feedback vertex set algorithm based on graph reduction.

In this paper, a independent and structured microassembler was designed for easily changing the system design, and for designing various microarchitecture. When the designer's hardware and microprogramming process were made concurrently, it is needed to easily change or improve the instruction set and executable code format. But this type of developed environment requires a high const and a large software system. A proposed microassembler was designed so the designer directly defines the microinstruction set and format to be executed. And we implemented a module from each part of the software, so it is now possible to use practically and upgrade the function of each part, First, the symbol was separated from the assembler. And then microinstruction was copied into it. The microinstruction format was designed using the defined language that was designed for free microinstruction. This was implemented in an IBM-PC by using the C-language, FLEX,and BISON.

In this paper, we propose an efficient architecture for motion estimation processor which performs one of essential functions in moving picture coding algorithms. Simple control mechanism of data flow in register array which stores pixel data, parallel processing of pixel data and pipelining scheme in arithmetic umit allow this architecture to process a 352*288 pixel image at the frame rate of 30fs, which is compatable with CCITT standard H.261.

Stereo vision is useful to obtain three dimensional depth information from two images taken from different view points. In this paper, we reduce searching area for correspondence by using the intra-scanline constraint, and utilize the inter-scanline constraint and the property of disparity continuity among the neighboring pixels for relaxation. Nodes with 3-D stucture are located on the axes of two views, and have matching possibility of correspondent pixels of two images. A matching is accepted if a node at the intersection of the disparity axes has the greatest matching possibility. Otherwise, the matching possibility of the node is updated by relaxation with the cooperation of neighboring nodes. Further relaxation with competition of two views is applied to a matching possibility of randomly selected node. The consensus of two views increases the confidence of matching, and removes a blurring phenomenon on the discontinuity of object. This approach has been tested with various types of image such as random dot stereogram and aerial image, and the experimental results show good matching performance.

In this paper, we propose a face recognition method using a neuro-fuzzy algorithm. In the preprocessing step, we extract the face part from the background image by tracking face boundaries. Then based on the a priori knowledge of human faces we extract the features such as widths of eyes and mouth, and distances from eye to nose and nose to mouth. In the recognition step. We use a neuro-fuzzy algorithm that employs a fuzzy membership function and modified error backpropagation algorithm. The former absorbs the variation of feature values and the latter shows good learning efficiency. Computer simulation results with 20 persons show that the proposed method gives higher recognition rate than the conventional ones.

We propose a new image coding method which takes into account both statistical redundancy and perceptual irrelevancy of the DCT coefficients so as to provide a high quality of the reconstructed images with a reduced transmission bit rate First, a block of DCT coefficients are decomposed into 16 subvectors so as for a subvector to convey key information about one of the low-pass or the dirctional filtered images. Then, the most significant subvector is selected as the principal edge of the block and then vector quantized. After that, the residuals of the block are computed and then sequentially quantized through aforementioned procedure until the quantization distortion is smaller than the target distortion. The proposed scheme is good at encoding images with a variety of transmission bit rates, especially at very low bit rate coding. In addition, it is another benifit of the proposed scheme that an image can be quantized with a wide range of the transmission bit rates by simply adapting the stopping criterion of the sequential vector quantizer according to the target distortion of the reconstructed image.

New morphological positioning algorithm and inspection method are presented and compared with a method by means of the Hough transform. The positioning algorithm is the process of finding the center and the rotated angle of the surface mounted device (SMD). The inspection method is capable of detecting the location of broken or bent leads. In order to obtain the center and the orientation of the SMD rapidly, the Hough transform method utilizes feature points (concave points) and is executed on a DSP board. The proposed morphological method is implemented by using the morphological skeleton subsets, and an ultimate orientation is decided by the Hit-or-Miss transform (HMT). In the inspection process, two inspection methods also are presented. The first method utilizes the morphological methods, i.e., opening and closing. It is performed before the positioning process and called an initial inspection. The second method follows the positioning process and is performed by an inspection of intersections of rulers and the lead edge (or the skeleton). It is a ruling technique which is referred to as a detailed inspection. We find the morphological approach is preciser and faster than the Hough approach by the comparison of the proposed algorithms.

The error intensity models for the ML estimation of a signal parameter have been developed in a companion paper [1]. While the methods described in [1] are applicable to any estimation problem with continuous parameters, our main application in this paper is the time delay estimation, and comparisons among the models derived in [1] (i.e. LC, LM, and ALM models)have been made. We first consider the case where only additive Gaussian noise is involved, and then the shot noise environment where coherent impulsive noise is also involved in addition to the Gaussian noise. We compare the models in terms of the probability of error, MSE(Mean Squared Error), and the computational complexity, which are the most important performance criteria in the analysis of parameter estimation. In conclusion, the ALM model turned out to be the most adequate model of all from the viewpoints of the criteria mentioned above.

This paper presents the VHSIC Hardware Description Language(VHDL) implementation of the Fixed Point Covariance Lattice(FLAT) algorithm for an Linear Predictive Coding(LPC) analysis and its related algorithms, such as the forth order high pass Infinite Impulse Response(IIR) filter, covariance matrix calculation, and Spectral Smoothing Technique(SST) in the Vector Sum Exited Linear Predictive(VSELP) speech coder that has been Selected as the standard speech coder for the North America and Japanese digital cellular. Existing Digital Signal Processor(DSP) chips used in digital cellular phones are derived from general purpose DSP chips, and thus, these DSP chips may not be optimal and effective architectures are to be designed for the above mentioned algorithms. Then we implemented the VHDL code based on the C code, Finally, we verified that VHDL results are the same as C code results for real speech data. The implemented VHDL code can be used for performing logic synthesis and for designing an LPC Application Specific Integrated Circuit(ASOC) chip and DsP chips. We first developed the C language code to investigate the correctness of algorithms and to compare C code results with VHDL code results block by block.

In this paper, we propose a postprocessing method that neatly removes blocking effect but retains visually important image details and edges. The iterative thresholded lowpass filter is basically a low pass filter whose ouput depends on three variable elements. I.e. iteration number, threshold value and passband width. The threshold value restricts the difference between the output of the proposed filter and the original input independent of the iteration number. With this property, the iterative thresholded lowpass filter can retain most of the image details while smoothing the block boundaries. The other two variable elements, i.e. iteration number and passband width, can determine the convergence speed of the proposed filter. In this paper, we also propose several adaptive filtering techniques based on the iterative thresholded lowpass filter with their simulation results.

The lateral resolution in an ultrasound imaging system is one of the most important factors for quality of the image and is determined by the beam focusing. For the better lateral resolution SDF(Sampled Delay Focusing) capable of digital focusing was proposed. The second-order sampling, one of band-width sampling methods, is suggested as being the best suitable for SDF because it allows total digital processing and is simple and economical. By proving that it introduces too much error, this article shows the second-order sampling is not appropriate for sampling of the wide-band signal generally used in ultrasound imaging systems. Also, this article suggests new sampling methods that maintain the advantages and reduce the unavoidable errors of the second-order sampling method. From computer simulation it is expected that the proposed methods reduce the errors of the second-order sampling method and can be used in real applications.

The diffusion neural network can be cfficiently applied to the Gaussian processing. For example, a difference of two Gaussians(DOG) is performed by this network with ease. In this paper, we model a neural network to perform the function /t(.del.${\Delta}^{2}$G) by using the diffusion neural network. This model is used to detect the edges of moving target in image. By this model not only moving target is separated from stationary background but also their trajectories are obtained using accumulated past information in the diffusion neural network. Furthermore this model needs a small number of connections per cell and the connection weights are fixed-valued. Therefore its hardware can be easily implemented with simple structure.

The diffusion neural network forms a Gaussian distribution by transferring an excitation to the surround. A DOG(difference of two Gaussians) is obtained by the diffusion neural network. This type of the DOG, which can detect the intensity changes of an image, has the same shape as a LOG(Laplacian of a Gaussian:${\Delta}^2$G) and narrow band pass characteristics. In this paper we show that another type of the DOG which has a very narrow Gaussian for the excitatory and a very wide Gaussian for the inhibitory, can be formed by the diffusion process of this network, This type of the DOG has a wide band width in spatial frequency domain and can be used efficiently in detecting special type of edges.

In this paper, a Japanese morphological analyzer for Japanese to Korean Machine Translation System is designed. The analyzer reconstructs the Japanese input sentence into word phrases that include grammatical and dictionary informations. Thus we propose the algorithm to separate morphemes and then connect them by reference to a corresponding Korean word phrases. And we define the connector to control Japanese word phrases It is used in controlling the start and the end point of the word phrase in the Japanese sentence which is without a space. The proposed analyzer uses the analysis dictionary to perform more efficient analysis than the existing analyzer. And we can decrease the number of its dictionary searches. Since the analyzer, proposed in this paper, for Japanese to Korean Machine Translation System processes each word phrase in consideration of the corresponding Korean word phrase, it can generate more accurate Korean expressions than the existing one which places great importance on the generation of the entire sentence structure.

In this paper, we propose the position- and size-independent handwritten on-line Korean character recognition system using multilayer neural networks which are trained with error back-propagation learning algorithm and the features of Hanguel consonants and vowels. Starting point, end point, and three vectors from starting point to end point of each stroke of characters inputted from mouse or tablet are applied as inputs of neural networks. If double consonants and vowels are separated by single consonants and vowels, all consonants and vowels have at most four strokes. Therefore, four neural networks learn the consonants and the vowels having each number of strokes. Also, we propose the algorithm of separating the consonants and vowels and constructing a character.

Maps are one of the most complicated types of drawings. Drawing recognition technology is not yet sophisticated enough for automated map reading. To automatically extract a road map directly form more complicated topographical maps, a very complicated algorithm is needed, simce the image generally involves such complicated patterns as symbols, characters, residential sections, rivers,etc. This paper describes a new feature extraction method based on the human optical neural field. We apply this method to extract complete set of road segments from topographical maps. The proposed method successfully extract road segments from various areas.

본 논문에서는 전문가 시스템의 사용자 인터페이스 모듈을 구현하기 위하여 사용자 인터페이스의 구조적 스키마(description schema)로서 수트(SUIT : Schematic User Interface Tasks)를 정의하였으며, 이를 관리하는 '수트관리기'(SUIMAN : SUIT Manager)를 구현하였다. 수트는 전문가 시스템의 사용자 인터페이스 설계 초기 단계에서, 수트를 사용하여 사용자 인터페이스 구현을 위해 이루어져야 할 행위를 개념적인 상위 수준에서 기술한다. 수트관리기는 수트에서 기술된 것이 언제 실행될지 결정한다. 그래픽 사용자 인터페이스의 실현은 X-윈도우의 MOTIF 라이브러리에 의해 제공되는 함수들을 실행함으로써 이루어진다. 즉, 수트관리기는 수트에 기술된 사용자 인터페이스 명세를 해석하여 실제 그래픽 사용자 인터페이스를 자동으로 실현시킨다. 수트와 수트관리기를 사용함으로써 전문가 시스템 개발에서 사용자 인터페이스 모듈의 보다 신속한 개발이 가능하다. 본 논문의 사용자 인터페이스 기술 방법은 `자동 회전 구성 전문가 시스템`의 사용자 인터페이스에 사용되었다.

In many continuous speech recognition systems, phoneme is used as a basic recognition unit However, the coarticulation generated among neighboring phonemes makes difficult to recognize phonemes consistently. This paper proposes allophone as an alternative recognition unit. We have classified each phoneme into three different allophone groups by the location of phoneme within a syllable. For a recognition algorithm, time-delay neural network(TDNN) has been designed. To recognize all Korean allophones, TDNNs are constructed in modular fashion according to acoustic-phonetic features (e.g. voiced/unvoiced, the location of phoneme within a word). Each TDNN is trained independently, and then they are integrated hierarchically into a whole speech recognition system. In this study, we have experimented Korean plosives with phoneme-based recognition system and allophone-based recognition system. Experimental results show that allophone-based recognition is much less affected by the coarticulation.

A modified Kohonen's self-organizing feature map (SOFM) algorithm which has binary reinforcement function and a constant learning rate is proposed. In contrast to the time-varing adaptaion gain of the original Kohonen's SOFM algorithm, the proposed algorithm uses a constant adaptation gain, and adds a binary reinforcement function in order to compensate for the lowered learning ability of SOFM due to the constant learning rate. Since the proposed algorithm does not have the complicated multiplication, it's digital hardware implementation is much easier than that of the original SOFM.