• 음성과학
• /
• 제1권
• /
• pp.39-53
• /
• 1997

As the segmental phonetic output is derived from its underlying form, the phonetic surface of intonation could also be derived from its underlying tone melody. In order to show clearly the phonological processes (in fact, we need more than just phonological processes) involved in the generation of intonational surface, we need to formalize the description of those processes. This paper firstly examines different types of sentence-final intonation contour in Korean, and then attempt to formalize the intonational behavior of those contours. In this attempt, we will investigate what kinds of linguistic information participate in deciding the shapes of the. contours and what kinds of tonological processes the underlying tone melody undergoes before it takes the surface shape. In this analysis of intonation contours, we focus on the linguistic structure rather than the acoustic property, adopting just two tones L and H as phonological tones, with four phonetic pitches.

• 한국정보처리학회논문지
• /
• 제6권7호
• /
• pp.1962-1969
• /
• 1999

활성 윤곽선은 내부 에너지와 외부에너지에 의해 조절되는 형태 변형이 가능한 에너지 최소화 곡선이다. 내부 에너지는 곡선을 부드럽게 유지하기 위한 제약 조건이고, 외부 지는 곡선을 영상 특징 쪽으로 이끈다. 활성 윤곽선이 제어 점에 의한 B-스플라인 표현은 많은 장점을 갖는다. Mentet[3] 등은 유한 차분 법에 의한 활성 윤곽선이 B-스플라인 표현을 제안하였다. 본 논문에서는 활성 윤곽선을 구간별 3차 B-스플라인으로 표현하고, 이 모델의 에너지를 최소로 하는 제어 점을 찾기 위한 방법으로 동적 프로그래밍을 사용한 방법을 제안한다. 제안된 방법은 유한 차분 법에 의한 B-스플라인 방법에 비해 간단하고 효과적이다.

• 대한원격탐사학회지
• /
• 제11권1호
• /
• pp.47-64
• /
• 1995

In this paper, an algorithm that finds man-made structures in a praylevel aerial images is proposed to perform stereo matching. An extracted contour of buildings must have a high accuracy in order to get a good feature-based stereo matching result. Therefore this study focuses on the use of edge following in the original image rather than use of ordinary edge filters. The Algorithm is composed of two main categories; one is to find candidate regions in the whole image and the other is to extract exact contours of each building which each candidate region.. The region growing method using the centroid linkage method of variance value is used to find candidate regions of building and the contour line tracing algorithm based on an adge following method is used to extract exact contours. The result shows that the almost contours of building composed of line segments are extracted.

• 대한전자공학회논문지SP
• /
• 제44권5호
• /
• pp.35-44
• /
• 2007

넓은 시야각을 갖는 전방향(omnidirectional) 카메라 시스템은 적은 수의 영상으로도 주변 장면에 대해 많은 정보를 취득할 수 있는 장점으로 카메라 교정(calibration), 공간의 3차원 재구성(reconstruction) 등에 널리 응용되고 있다. 실 세계에 존재하는 직선 성분들은 전방향 카메라 모델에 의해 컨투어로 사영(projection)되기 때문에, 영상간에 대응되는 컨투어 성분은 카메라의 회전 및 이동 등의 추정에 효과적으로 활용될 수 있다. 본 논문에서는 전방향 카메라의 변환 파라미터를 추정하기 위한 2단계 최소화 알고리즘이 제안된다. 제안된 알고리즘은 컨투어를 이루는 대응점에 대한 에피폴라(epipolar) 평면과 3차원 벡터간의 각도 오차함수 및 사영된 컨투어의 거리 오차를 단계별로 최소화하는 카메라 파라미터를 계산한다. 등거리(equidistance) 사영된 합성영상과 어안렌즈(fisheye lens)로 취득한 실제 영상을 대상으로 제안된 알고리즘이 카메라의 위치 정보를 정확하게 추정함을 확인하였다.

• 한국CDE학회논문집
• /
• 제18권4호
• /
• pp.275-282
• /
• 2013

This paper proposes a method for human identification using teeth contours extracted from dental images that are captured from the frontal views of subjects each of who opens his or her mouth slightly. Each dental image has a black-colored region containing the subject's teeth contours which are usually different from subject to subject. This means that this black-colored region has bio-mimetic information useful for human identification. The basic idea of the method is to extract the upper and lower teeth contours from the dental image of each subject and to encode their geometric patterns using a back-propagation neural network model. After acquiring 400 teeth images form 10 university students, we used 300 images for the training data of the neural network model and 100 images for its verification. Experimental results have shown that the proposed neural network-based method can be used as an alternative solution for identification among a small group of humans with a low cost and simple setup.

• 음성과학
• /
• 제11권1호
• /
• pp.105-116
• /
• 2004

The fundamental frequency of a vowel is known to be affected by the identity of the preceding consonant. The general agreement is that strong consonants trigger higher F0 than weak consonants. However, there has been a disagreement on the shape of this segmentally affected F0 contours. Some studies report that shapes of contours are differentiated based on the consonant type, but others regard this observation as misleading. This research attempts to resolve this controversy by investigating shapes and slopes of F0 contours of Korean word level speech data produced by four male speakers. Instead of entirely relying on traditional human intuition and judgment, I employed an automatic F0 contour analysis technique known as tilt parameterisation (Taylor 2000). After necessary manipulation of an F0 contour of each data token, various parameters are collapsed into a single tilt value which directly indicates the shape of the contour. The result, in terms of statistical inference, shows that it is not viable to conclude that the type of consonant is significantly related to the shape of F0 contour. A supplementary measurement is also made to see if the slope of each contour bears meaningful information. Unlike shapes themselves, slopes are suspected to be practically more practical for consonantal differentiation, although confirmation is required through further refined experiments.

• 한국경영과학회:학술대회논문집
• /
• 대한산업공학회/한국경영과학회 2004년도 춘계공동학술대회 논문집
• /
• pp.44-47
• /
• 2004

This paper proposes an approach for composite surface reconstruction from 2D serial cross-sections, where the number of contours varies from section to section. In a triangular surface-based approach taken in most reconstruction methods, a triangular $G^{1}$ surface is constructed by stitching triangular patches over a triangular net generated from the compiled contours. In the proposed approach, the resulting surface is a composite $G^{1}$ surface consisting of three kinds of surfaces: skinned, surface is first represented by a B-spline surface approximating the serial contours of the skinned region and then serial contours of the skinned region and then transformed into a mesh of rectangular Bezier patches. On branched and capped regions, triangular $G^{1}$ surfaces are constructed so that the connections between the triangular surfaces and their neighboring surfaces are $G^{1}$ continuous. Since each skinned region is represented by an approximated rectangular $G^{2}$ surface instead of an interpolated triangular $G^{1}$ surface, the proposed approach can provide more visually pleasing surfaces and realize more efficient data reduction than the triangular surface-based approach. Some experimental results demonstrate its usefulness and quality.

• 한국정보처리학회논문지
• /
• 제5권1호
• /
• pp.236-248
• /
• 1998

인간의 시각계는 가상 윤곽(Illusory Contours)을 인식한다. 가상 윤곽이란 물리적으로는 윤곽이 없는 곳에서 주변의 패턴의 영향에 의해 윤곽이 있는 것처럼 느끼는 현상으로서, 이와 같은 현상은 물체 인식에 있어서 일부가 숨어 있는 물체를 그 주위의 정보를 이용해서 보충하면서 인식해 가는 능력과 관계가 있다. 본 논문은 가상 윤곽을 인식하기 위한 것으로, 가상 윤곽에서 추출된 유도 자극으로부터 가상표면을 생성하는 신경회로망 모델을 제안하고 실험을 통해 성능을 확인한다.

• 음성과학
• /
• 제14권3호
• /
• pp.77-89
• /
• 2007

This paper examined the intonation contours of English sentences produced by the Korean students. Thirty students participated in the speaking tasks made up of three parts: an oral interview, picture description and a conversational text reading. Their pronunciations in the parts were recorded on a minidisk. Then, two native instructors evaluated their proficiency level focusing on general intelligibility and suprasegmental aspects of the speech. Based on the results of evaluation they were divided into two groups: high and low proficiency groups. The pitch contours of three sentences produced by both the Korean students and a native speaker were compared to find any similarities and differences in the students' intonation patterns using Praat. Results showed that there was a moderate correlation in the proficiency scores of the students by the two native speakers. Secondly, students who earned high scores in the proficiency level matched better the native model. Thirdly, the high group students knew more on the pitch contour and tried to carefully realize them while fewer students in the low group answered positively on the questionnaire. In conclusion, English learners need to know the proper intonation patterns and to practice them consciously and sufficiently to realize correct intonation contours. Further studies would be desirable on the students' pronunciation focused on discourse structure.

• 한국정보처리학회논문지
• /
• 제7권3호
• /
• pp.943-952
• /
• 2000

The surface reconstruction problem from a set of wire-frame contours is very important in diverse fields such as medical imaging or computer animation. In this paper, surface triangulation method is proposed for solving the problem. Generally, many optimal triangulation techniques suffer from the large computation time but heuristic approaches may produce very unnatural surface when contours are widely different in shape. To compensate the disadvantages of these approaches, we propose a new heuristic triangulation method which iteratively decomposes the surface generation problem from a band (a pair of vertices chain) into tow subproblems from two sub-bands. Generally, conventional greedy heuristic contour triangulation algorithm, suffer from the drastic error propagation during surface modeling when the adjacent contours are different in shape. Our divide-and-conquer algorithm, called band partitioning algorithm, processes eccentric parts of the contours first with more global information. Consequently, the resulting facet model becomes more stable and natural even though the shapes are widely different. An interesting property of our method is hat it supports multi-resolution capability in surface modeling time. According to experiments, it is proved to be very robust and efficient in many applications.