• Conservation Science in Museum
• /
• v.31
• /
• pp.55-70
• /
• 2024

Style transfer algorithms are currently undergoing active research and are used, for example, to convert ordinary images into classical painting styles. However, such algorithms have yet to produce appropriate results when applied to Korean cultural heritage images, while the number of cases for such applications also remains insufficient. Accordingly, this study attempts to develop a style transfer algorithm that can be applied to styles found among Korean cultural heritage. The algorithm was produced by improving data comprehension by enabling it to learn meaningful characteristics of the styles through representation learning and to separate the cultural heritage from the background in the target images, allowing it to extract the style-relevant areas with the desired color and texture from the style images. This study confirmed that, by doing so, a new image can be created by effectively transferring the characteristics of the style image while maintaining the form of the target image, which thereby enables the transfer of a variety of cultural heritage styles.

• Journal of Broadcast Engineering
• /
• v.27 no.1
• /
• pp.104-123
• /
• 2022

Style transfer based on neural network provides very high quality results by reflecting the high level structural characteristics of images, and thereby has recently attracted great attention. This paper deals with the problem of resolution limitation due to GPU memory in performing such neural style transfer. We can expect that the gradient operation for style transfer based on partial image, with the aid of the fixed size of receptive field, can produce the same result as the gradient operation using the entire image. Based on this idea, each component of the style transfer loss function is analyzed in this paper to obtain the necessary conditions for partitioning and padding, and to identify, among the information required for gradient calculation, the one that depends on the entire input. By structuring such information for using it as auxiliary constant input for partition-based gradient calculation, this paper develops a recursive algorithm for super high-resolution image style transfer. Since the proposed method performs style transfer by partitioning input image into the size that a GPU can handle, it can perform style transfer without the limit of the input image resolution accompanied by the GPU memory size. With the aid of such super high-resolution support, the proposed method can provide a unique style characteristics of detailed area which can only be appreciated in super high-resolution style transfer.

• 대한두경부종양학회:학술대회논문집
• /
• 2002.11a
• /
• pp.235-235
• /
• 2002

• Proceedings of the IEEK Conference
• /
• 2001.09a
• /
• pp.805-808
• /
• 2001

음성인식, 합성 및 분석과 같은 음성신호처리 분야에 있어서 기본주파수 즉, 피치를 정확히 검출하는 것은 중요하나 만일 음성신호의 기존주파수를 정확히 검출할 수 있다면 음성인식에 있어서 화자에 따른 영향을 줄일 수 있기 때문에 인식의 정확도를 높일 수 있고, 음성합성 시에 자연성과 개성을 쉽게 변경하거나 유지할 수 있다. 또한 분석시 피치에 동기 시켜 분석하면 성문의 영향이 제거된 정확한 성도 파라미터를 얻을 수 있다. 따라서 제안한 논문에서는 시간영역처리에서 시간영역파형에서 먼저 성도성분이 제거된 성문특성 즉 피치주기 성분을 강조하기 위해 음성신호의 기울기를 이용한 Positive 센터 클리핑를 수행하고 주파수 영역에서는 원신호의 스펙트럼과 센터클립된 신호의 스펙트럼과의 Peak-Fiting을 수행하고 선형 인터폴래이션(스무딩)을 통해 평탄화된 스펙트럼을 얻었다. 결과적으로 유성음 구간과 음성이 변하는 전이구간에서 G-peak가 강조된 더욱 정확한 Pitch를 검출할 수 있었다.

• Korean Journal of Optics and Photonics
• /
• v.15 no.2
• /
• pp.177-182
• /
• 2004

We have observed the saturated absorption spectrum of the P(16) line of the V$_1$＋V$_3$ band of $^{13}$ C$_2$H$_2$ molecule by using the external cavity spectroscopy method. The frequency of laser has been stabilized to the saturated absorption spectrum. The relative contrast of the saturation spectrum is about 7 % with respect to the linear absorption and the linewidth is about 1.8 MHz. The frequency fluctuation of the stabilized LD is about $\pm$20 KHz during the sampling time 100 S.

• Annual Conference on Human and Language Technology
• /
• 2008.10a
• /
• pp.162-166
• /
• 2008

본 연구는 말뭉치로부터 추출된 정보를 사용하여 대화 시스템에 필요한 과정들을 통합 처리하는 시스템을 제안한다. 기존 연구는 영역 확장 시 대화 시스템의 각 과정들을 위해 많은 노력이 필요하였지만, 제안하는 방법은 말뭉치를 사용하여 각 과정들을 통합적으로 업데이트함으로서 이 문제를 해결하고자 한다. 사용자 입력문장과 말뭉치의 각 문장들 간의 유사도 계산을 통하여 의미적으로 가장 유사한 말뭉치 문장의 정보를 이용하고, 시스템 응답에 필요한 정보를 선택한다. 또한, 문맥에 관련된 정보를 자동으로 추출하여 대화 관리를 위한 대화 전이 네트워크(network)를 생성한다. 따라서, 제안 시스템은 말뭉치의 추가 및 수정만으로 새로운 영역 확장과 관리에 용이한 구조를 갖는다. 실험으로 관찰한 제안된 시스템의 성능은 유사도 계산 만족도 약 77%, 시스템 응답의 적절성 약 84%로 충분히 작업 수행이 가능한 점수를 보여주었다.

• Proceedings of the Korean Society of Propulsion Engineers Conference
• /
• 2010.11a
• /
• pp.807-808
• /
• 2010

To insert payload to the orbit over the 200km-altitude using launch vehicle which has 300sec the Isp, multi staging technique for launch is necessary. The range between the sea-level to the transfer orbit about 200~250km is for operation of 1st and 2nd rocket engines and the higher altitude is for propulsion system of the acceleration block and satellite. The upper stage rocket engine should have the high technology for entering the payload into the orbit precisely more than the performance for high thrust level. With this investigation of the upper stage rocket engines which have been used, we want to understand their development trend and prospect which is going to be references for the development of ours.

• Journal of KIISE:Computing Practices and Letters
• /
• v.13 no.7
• /
• pp.488-498
• /
• 2007

This paper presents a novel approach for representing the spatio-temporal topology of the camera network with overlapping and non-overlapping fields of view (FOVs) in Ubiquitous Smart Space (USS). The topology is determined by tracking moving objects and establishing object correspondence across multiple cameras. To track people successfully in multiple camera views, we used the Merge-Split (MS) approach for object occlusion in a single camera and the grid-based approach for extracting the accurate object feature. In addition, we considered the appearance of people and the transition time between entry and exit zones for tracking objects across blind regions of multiple cameras with non-overlapping FOVs. The main contribution of this paper is to estimate transition times between various entry and exit zones, and to graphically represent the camera topology as an undirected weighted graph using the transition probabilities.

• KSBB Journal
• /
• v.10 no.1
• /
• pp.9-14
• /
• 1995

Continuous production of fructooligosaccharides from sucrose by a dual immobilized enzyme system of fructosyltransferase and glucose isomerase was studied in a column reactor. The optimal temperature and pH of the immobilized fructosyltransferase were $65^{\circ}C$ and 5.5, respectively. The activity of glucose isomerase was favorable as temperature and pH were increased within the ranges examined. However, both the immobilized enzymes were thermally unstable over $5^{\circ}C$, suggesting that long-term operation of the dual immobilized enzyme column should be carried out below $50^{\circ}C$. The optimum packing ratio of fructosyltransferase to glucose isomerase was found to be around 5/3. Under the optimized reaction conditions, the dual enzyme column was successfully operated for 40 days without any loss of initial enzyme activities, yielding 66% of fructooligosaccharides. Furthermore, the relative sweetness of fructooligosaccharides produced by a dual emzyme system was enhanced by 6% compared with that of fructosyltransferase alone.

• Proceedings of the Korean Vacuum Society Conference
• /
• 2012.08a
• /
• pp.335-335
• /
• 2012

III-V 족 반도체 물질 중, GaN는 넓은 밴드갭을 가지고 있어 발광 다이오드나 레이저 다이오드, 트랜지스터, 스핀트로닉스 등의 응용에 유용한 물질이다 [1]. 실시간 성장 제어 및 최적화된 특정 소자 응용을 위해서는 GaN의 다양한 온도에 대한 유전율 함수 정보가 필수적이다. 편광분석법을 이용한 상온에서의 hexagonal GaN 유전율 함수는 이미 여러 연구에서 보고되었고, 80~650 K 사이의 온도 범위에 대한 언구도 수행되었다 [2,3]. 그러나, 온도변화에 대한 GaN 유전율 함수와 $E_0$ 전이점에 대한 해석은 부정확하다. 따라서 본 연구에서는 사파이어 기판 위에 분자살박막증착장치를 이용하여 c-축 방향 (0001)으로 성장 시킨 hexagonal GaN를 0.74~6.42 eV 에너지 구간에서 보다 확장된 온도 영역(26~693 K)의 유전율 함수를 편광분석법을 이용하여 측정하였다. 측정된 GaN의 유전율 함수를 회기분석법을 통한 2차 미분 표준해석법을 이용해 분석 하였고, 그 결과 $E_0$와 excitonic $E_0$ 전이점을 명확히 얻을 수 있었다. 온도가 감소함에 따라 격자상수 및 전자-포논 상호작용이 감소하여 전자 전이점이 청색천이 하고, 그 구조가 명확해 지는 결과를 얻었다. 본 연구의 결과는 GaN 유전율 함수의 온도 의존성에 대한 데이터베이스를 제공함은 물론, 실시간 모니터링과 GaN를 기반으로 하는 광소자 제작 등에 유용할 것이다.