http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
A Hybrid Fractal-Based Image Coding
최종주,Han, Sung Hyun 대한전자공학회 1994 ISPACS:Intelligent Signal Processing and Communica Vol.1 No.1
The complexity of the fractal encoding is grooving very quickly with the image size. In this paper, we present a novel algorithm which reduces the encoding time of fractal-based image coding. To begin with, the simplest type of image pyramid is constructed. In searching a optimal domain block for each range block, this structure allows to initially search low-resolution version of an image. After decoding in the next higher resolution, distortion measures are computed between the decoded range block and the block of the same position in the same resolution of the original image. For every range block such that the distortion is above the threshold, searching is performed in that resolution again. In the original resolution, we consider two-level square partition. Simulation results show good visual quality of decoded images with a lower computational cast.
최종주,박병권 서강전문대학 1994 論文集 Vol.13 No.-
A reconfigurable bus system is defined to be a bus system whose configuration can be dynamically changed by establishing the local switch of eath processor. A processor array connected to a reconfigurable bus system is defined as a processor array with a reconfigurable bus system. Two constant time algorithms are presented for linking the non-zero terms in a binary sequence and for computing the histogram of an image on a processor array with a reconfigurable bus system, respectively. Then, these two algorithms can be used to compute n angles of the Hough transform of an image with N edge pixels in constant time using O(nN³) processors.
최종주,최성복 서강정보대학 1993 論文集 Vol.12 No.-
Skeletons have been largely used as descriptors of shape in the field of image processing. Only binary pictures, however, have been considered so far. In this paper a grey-weighted skeleton is defined for grey-valued continuous and quantized image. In order to extend the invertibility property of the skeleton to the grey case, a transformation is defined, which is a generalization of both direct and inverse binary skeleton transformations. By taking advantage of the properties of this transformation, a procedure for the inversion of a grey-weighted skeleton is finally proposed.
싶다' 구문의 격과 부가어 : HPSG의 확대 논항 구조적 분석 An Extended Argument Structure Analysis in HPSG
최종주,이남근 대한언어학회 2000 언어학 Vol.8 No.1
Choe, Jong-Joo and Lee, Nam-Geun. 2000. Case and Adjunct in the Siphta Construction: An Extended Argument Structure Analysis in HPSG. Linguistics 8-1, 133-152. The siphta construction in Korean shows some intriguing properties such as the case alternation in the complement NP of the lower verb and the optional - but not arbitrary - case marking on the adjunct. We will show that the 'argument structure' alone is not enough to deal with the case marking on the adjunct simply because the adjunct cannot be contained in the argument structure. Therefore, in the framework of Head-driven Phrase Structure Grammar (HPSG for short), we will take advantage of the new feature DEPS (or dependendy structure) proposed by Bouma, Malouf and Sag (1999). The extended argument structure DEPS will enable us, with the help of Case Principle and the notion of structural/lexical case, to take care of the case alternation and case marking on the adjunct in the siphta construction. (Chosun University)
최종주,박병권 서강정보대학 1996 論文集 Vol.15 No.-
To detect shapes in noisy data, the fuzzy Hough transform is introduced. This technique finds shapes by approximately fitting the data points, which avoids the spurious shapes detected when using the conventional Hough transform. An efficient implementation of this method is found in detecting lines and circles.
최종주 서강정보대학 1998 論文集 Vol.17 No.-
Neural net has a difficultly in recognizing images which have much noise. Therefore we retrieve the discernible images. using hough transform. We construct the knowledge base of indiscernible images using images learned by neural net and made many kinds of inputted images recognized. Neural net used in this paper is composed of input layers, hidden layers and output layers. Since images used in learning are inputted regardless of their size, decline and image shapes, normalization of them is essential. Since this reduces the learning time, it is necessary to adjust the number of input layers appropriately. The less input layers, the lower system's cognitive faculty. The more input layers, the longer the learning time. The backpropagation algorithm used in this paper is composed of three layers. Input layers and hidden layers include content layers. Output units of output layer are connected to themselves, hidden units and other output units. Generally, in classifying patterns using backpropagation algorithm, hidden units are learned so that they can maximize informations delivered from input patterns. Output units also are learned so that they can separate informations delivered from hidden units. It is important to deliver more informations from lower units to output units, since it improve the separation faculty and the generalization faculty. We found that images not recognized by backpropagation algorithm rightly are the very damaged or noisy images. And we found that backpropagation algorithm recognition system recognizes images regardless of center point, decline or size of inputted images.
최종주 서강정보대학 1995 論文集 Vol.14 No.-
An important aspect of any scientific discipline is the objective and independent comparison of algorithms which perform common tasks. In image analysis this problem has been much neglected. in this paper we present the results and conclusions of a comparison of four Hough Transform, HT, based line finding algorithms on a range of realistic images from the industrial domain. The Hough transform is a robust technique which is useful in detection straight lines in an edge-enhanced picture. It can be used to handle partially occluded and noise corrupted images.