• 한국정보통신학회논문지
• /
• 제25권11호
• /
• pp.1477-1485
• /
• 2021

최근 이미지 초해상도는 정수배율만 가능한 모델에만 집중적으로 연구되고 있다. 하지만 관심 객체 인식, 디스플레이 화질 개선 등 실제 초해상도 기술의 대표 적용 분야에서는 소수 배율을 포함하는 임의배율 확대 필요성이 대두되고 있다. 본 논문에서는 기존 정수배율 모델의 가중치를 활용하여 임의배율을 실행할 수 있는 모델을 제안한다. 이 모델은 정수배율에 의해 우수한 성능을 가진 초해상도 결과를 DCT 스펙트럼 도메인으로 변환하여 임의배율을 위한 공간을 확장한다. DCT 스펙트럼 도메인에 의한 확장으로 인해 발생하는 이미지의 고주파 정보 손실 문제를 줄이기 위해 고주파 스펙트럼 정보를 적절히 복원할 수 있는 모델인 고주파 집중 네트워크를 제안한다. 제안된 네트워크는 고주파 정보를 제대로 생성하기 위해서 RGB 채널간의 상관관계를 학습하는 레이어인 channel attention을 활용하고, 잔차 학습 구조를 통해 모델을 깊게 만들어 성능을 향상시켰다.

• 대한원격탐사학회지
• /
• 제16권1호
• /
• pp.87-98
• /
• 2000

상업위성에서 공급되는 고해상도영상의 활용을 증대하기 위한 영상합성에 대한 관심이 증가하고 있다. 합성에 사용된 고해상도 흑백영상과 저해상도 다중분광영상은 항공기탑재 다중분광 주사기에 의해 촬영된 네 밴드의 영상을 이용하여 모의 제작하였다. 모의 합성된 2rl 해상도의 흑백 영상과 Bnl 해상도의 네 밴드 영상에 대하여 다섯 가지 합성방법(MWD, ItIS, PCA, HPF, CN, PCA) 을 적용하였다. 합성된 영상에 대해서 원래 영상들이 가지고 있던 공간해상도와 분광정보 측면의 특성을 분석하고자, 육안판독, 통계치비교, semivariogram, 분광반사특성 등을 비교하였다. MWD 변환방법에 의하여 합성된 영상이 공간해상도 및 분광정보 측면에서 모두 합성에 사용된 원래 영상과 근접한 결과를 보였다.

• 대한원격탐사학회:학술대회논문집
• /
• 대한원격탐사학회 2005년도 Proceedings of ISRS 2005
• /
• pp.249-252
• /
• 2005

Microphytobenthos bloom from late January to early March in Korean tidal flats. KOMPSAT-2 will provide multi-spectral images with a spatial resolution of 4 m comparable with IKONOS. Using IKONOS and Landsat data, algal mat detection was tested in the Saemangeum area~ Micro-benthic diatoms are abundant and a major primary product in the tidal flats. A linear spectral unmixing (LSU) method was applied to the test data. LSU was effective to detect algal mat and the classified algal mat fraction well correlated with NDVI image. Fine grained upper tidal flats are generally known to be the best environment for algal mat. Algal mat thriving in coarse grained lower tidal flats as well as upper tidal flats were reported in this study. A high resolution multi-spectral sensor in KOMPSAT-2 will provide useful data for long-term monitoring of microphytobenthos in tidal flats.

• ETRI Journal
• /
• 제39권6호
• /
• pp.832-840
• /
• 2017

Recently, deep recurrent neural networks have achieved great success in various machine learning tasks, and have also been applied for sound event detection. The detection of temporally overlapping sound events in realistic environments is much more challenging than in monophonic detection problems. In this paper, we present an approach to improve the accuracy of polyphonic sound event detection in multichannel audio based on gated recurrent neural networks in combination with auditory spectral features. In the proposed method, human hearing perception-based spatial and spectral-domain noise-reduced harmonic features are extracted from multichannel audio and used as high-resolution spectral inputs to train gated recurrent neural networks. This provides a fast and stable convergence rate compared to long short-term memory recurrent neural networks. Our evaluation reveals that the proposed method outperforms the conventional approaches.

• 대한원격탐사학회:학술대회논문집
• /
• 대한원격탐사학회 2003년도 Proceedings of ACRS 2003 ISRS
• /
• pp.732-734
• /
• 2003

In the middle -resolution remote sensing, the Ground Sampled Distance(GSD) sensed and sampled by the detector is generally larger than the size of objects(or materials) of interest, in which case several objects are embedded in a single pixel and cannot be detected spatially. This study is intended to solve this problem of a hyperspectral data with high spectral resolution. We examined the detection algorithm, Linear Spectral Mixing Model, and also made a test on the Hyperion data. To find class Endmembers, we applied two methods, Spectral Library and Geometric Model, and compared them with each other.

• Mass Spectrometry Letters
• /
• 제9권3호
• /
• pp.73-76
• /
• 2018

The number of matched peaks (NMP) is estimated as the spectral similarity measure in tandem mass spectral library searches of small molecules. In the high resolution mode, NMP provides the same reliable identification as in the case of a common dot-product function. Corresponding true positive rates are ($94{\pm}3$) % and ($96{\pm}3$) %, respectively.

• 대한기계학회논문집B
• /
• 제22권7호
• /
• pp.1030-1040
• /
• 1998

One of the main unresolved issues in large-eddy simulation(LES) of wall-bounded turbulent flows is the requirement of high spatial resolution in the near-wall region, especially in the spanwise direction. Such high resolution required in the near-wall region is generally used throughout the computational domain, making simulations of high Reynolds number, complex-geometry flows prohibitive. A grid-embedding strategy using a nonconforming spectral domain-decomposition method is proposed to address this limitation. This method provides an efficient way of clustering grid points in the near-wall region with spectral accuracy. LES of transitional and turbulent channel flow has been performed to evaluate the proposed grid-embedding technique. The computational domain is divided into three subdomains to resolve the near-wall regions in the spanwise direction. Spectral patching collocation methods are used for the grid-embedding and appropriate conditions are suggested for the interface matching. Results of LES using the grid-embedding strategy are promising compared to LES of global spectral method and direct numerical simulation. Overall, the results show that the spectral domain-decomposition grid-embedding technique provides an efficient method for resolving the near-wall region in LES of complex flows of engineering interest, allowing significant savings in the computational CPU and memory.

• 대한원격탐사학회:학술대회논문집
• /
• 대한원격탐사학회 2003년도 Proceedings of ACRS 2003 ISRS
• /
• pp.820-824
• /
• 2003

One of the most important research areas on remote sensing is spectral unmixing of hyper-spectral data. For spectral unmixing of hyper spectral data, accurate land cover information is necessary. But obtaining accurate land cover information is difficult process. Obtaining land cover information from high-resolution data may be a useful solution. In this study spectral signature of endmembers on ASTER acquired in October was calculated from land cover information on IKONOS acquired in September. Then the spectral signature of endmembers applied to ASTER images acquired on January and March. Then the result of spectral unmxing of them evauateted. The spectral signatures of endmembers could be applied to different seasonal images. When it applied to an ASTER image which have similar zenith angle to the image of the spectral signatures of endmembers, spectral unmixing result was reliable. Although test data has different zenith angle from the image of spectral signatures of endmembers, the spectral unmixing results of urban and vegetation were reliable.

• 천문학회보
• /
• 제39권1호
• /
• pp.42.2-42.2
• /
• 2014

We present high-resolution optical spectra of newly selected candidates of intermediate-mass black holes. The sample was selected based on the variability and spectral shape in X-ray. The spectra was taken with Magellan 6.5 m Clay Telescope and cover the rest-frame region 3500-10000A. The high spectral resolution (R~4000) of the spectrum allows us to estimate BH masses of the sources. Interestingly, the majority of the sample appears to have broad emission lines. Using this dataset, we will estimate the BH masses and Eddington ratio in order to understand their physical properties.

• 대한공간정보학회지
• /
• 제18권2호
• /
• pp.79-86
• /
• 2010

일반적으로, 영상 융합은 서로 다른 특징을 가지는 2개 이상의 영상을 이용하여 각 영상의 장점 및 특징을 모두 가지는 하나의 영상으로 재구성하는 것을 의미한다. 특히, 원격탐사 분야에서의 영상융합은 멀티스펙트럴 영상의 공간해상도를 향상시키는 것을 의미하며 이러한 이유로 인하여 Pan-sharpening 기술로도 불리어진다. 특히, 융합영상은 변화탐지, 영상 지도 제작, 도시 분석 등 다양한 분야에 적용 가능하기 때문에 중요성이 증대되고 있다. 그러나, 기존에 제안된 알고리즘들은 멀티스펙트럴 영상의 분광정보를 왜곡시키거나, 융합 영상의 공간해상도가 흑백영상의 공간해상도에 비하여 저하되는 문제를 지닌다. 이를 위해 본 논문에서는 멀티스펙트럴 영상의 분광 및 공간특성을 고려한 새로운 융합 방법론을 제안하였다. 본 알고리즘의 평가를 위해서 KOMPSAT-2, QuickBird 위성영상에 알고리즘을 적용을 하였으며, 기존의 영상융합 알고리즘에 비하여 공간적/분광적인 측면에서 모두 향상된 결과를 보임을 확인할 수 있었다.