• 환경영향평가
• /
• 제8권3호
• /
• pp.77-90
• /
• 1999

The degree of green naturality (DGN) has played a key role for maintaining the environmental quality from inappropriate developments, although the quality and effectiveness of the mapping of DGN has been under debate. In this study, spatial distribution of degree of green naturality was initially estimated from forest stand maps that were produced from the aerial photo interpretation and extensive field survey. Once the boundary of initial classes of DGN were defined, it were overlaid with normalized difference vegetation index (NDVI) data that were derived from the recently obtained Landsat Thematic Mapper data. NDVI was calculated for each pixel from the radiometrically corrected satellite image. There were no significant differences in mean values of vegetation index among the initial DGN classes. However, the satellite derived vegetation index was very effective to delineate the developed and damaged forest lands and to adjust the initial value of DGN according to the distribution of NDVI within each class.

• 한국멀티미디어학회논문지
• /
• 제3권4호
• /
• pp.372-379
• /
• 2000

본 논문에서는 비트패턴을 기반으로 한 고속의 적응적 가변 블록 움직임 예측 알고리즘을 제안한다. 제안된 방법은 블록 내의 평균값을 기준으로 8bit 화소값을 0과 1의 비트패턴으로 변환한 후 블록의 움직임 예측을 수행한다. 비트변환을 통한 영상의 단순화는 움직임 추정의 계산적 부담을 감소시켜 빠른 탐색을 가능하게 한다. 그리고 블록 내의 움직임 정도를 미리 판별하여 이를 기반으로 한 적응적 탐색이 불필요한 탐색을 제거하고 움직임이 큰 블록에서는 정합 과정을 심화시켜 보다 빠르고 정확한 움직임 예측을 수행한다. 본 제안된 방식을 가지고 실험한 결과, 한 프레임 당 적은 수의 블록으로 고정된 크기의 블록을 가진 전역 탐색블록 정합 알고리즘(full search block matching algorithm; FS-BMA)보다 예측 에러를 적게 발생시켜 평균 0.5dB 정도의 PSNR 개선을 가져왔다.

• 대한전기학회논문지:시스템및제어부문D
• /
• 제51권10호
• /
• pp.468-476
• /
• 2002

Digital images are easily corrupted by noise during the data transmission, data capture and data processing. A technical method of noise analyzing and adaptive filtering for reducing of quantum noise in medical radio-fluoroscopy images is presented. By adjusting the characteristics of the filter according to local statistics around each pixel of the image as moving windowing, it is possible to suppress noise sufficiently while preserve edge and other significant information required in diagnosis. We proposed adaptive weighed median(AWM) filters based on local statistics. We showed two ways of realizing the AWM filters. One is a simple type of AWM filter, which is constructed by Homogeneous factor(HF). Homogeneous factor(HF) from the noise models that enables the filter to recognize the local structures of the image is introduced, and an algorithm for determining the HF fitted to the diagnostic systems with various inner statistical properties is proposed. We show by the experimented that the performances of proposed method is superior to these of other filters and models in preserving small details and suppressing the noise at homogeneous region. The proposed algorithms were implemented by Visual C++ language on a IBM-PC Pentium 550 for testing purposes and the effects and results of the filter in the various levels of noise and images were proposed by comparing the values of NMSE(normalized mean square error) with the value of the other existing filtering methods.

• 대한원격탐사학회지
• /
• 제16권4호
• /
• pp.339-345
• /
• 2000

KOMPSAT-1 was launched on 21 December, 1999 and the main mission of the satellite is the cartography to provide the imagery from a remote earth view for the production of maps of Korean territory. For this purpose, the satellite has capability to tilt the spacecraft utmost $\pm$45 degrees to acquire stereo satellite imagery in different paths. This study aims to estimate the three dimensional positioning accuracy of stereo satellite imagery from EOC(electro-optical camera), a payload of KOMPSAT-1 satellite. For this purpose, the ground control points and check points were obtained by GPS surveying. The sensor modeling and the adjustment was performed by PCI software installed in KARI (Korea Aerospace Research Institute), which contained mathematical analysis module for KOMPSAT-1 EOC. The study areas were Taejon and Nonsan, placed in the middle part of Korea. As a result of this study, we found that the RMSE(root mean square error) value of three dimensional positioning KOMPST-1 stereo imagery can be less than 1 pixel (6.6 m) if we can use about 10 GCPs(ground control points). Then, a standarrd of FGDC (Federal Geographic Data Committee) of USA was applied to the result to estimate the three dimensional positioning accuracy of KOMPSAT-1 stereo imagery.

• 한국의학물리학회:학술대회논문집
• /
• 한국의학물리학회 2003년도 제27회 추계학술대회
• /
• pp.45-45
• /
• 2003

The aim of this study was the characterization and performance validation of new prototype avalanche photodiode (APD) arrays for positron emission tomography (PET). Two different APD array prototypes (noted A and B) developed by Radiation Monitoring Device (RMD) have been investigated. Principal characteristics of the two APD array were measured and compared. In order to characterize and evaluate the APD performance, capacitance, doping concentration, quantum efficiency, gain and dark current were measured. The doping concentration that shows the impurity distribution within an APD pixel as a function of depth was derived from the relationship between capacitance and bias voltage. Quantum efficiency was measured using a mercury vapor light source and a monochromator used to select a wavelength within the range of 300 to 700 nm. Quantum efficiency measurements were done at 500 V, for which the APD gain is equal to one. For the gain measurements, a pencil beam with 450 nm in wavelength was illuminating the center of each pixel. The APD dark currents were measured as a function of gain and bias. A linear fitting method was used to determine the value of surface and bulk leakage currents. Mean quantum efficiencies measured at 400 and 450 nm were 0.41 and 0.54, for array A, and 0.50 and 0.65 for array B. Mean gain at a bias voltage of 1700 V, was 617.6 for array A and 515.7 for type B. The values based on linear fitting were 0.08${\pm}$0.02 nA 38.40${\pm}$6.26 nA, 0.08${\pm}$0.0l nA 36.87${\pm}$5.19 nA, and 0.05${\pm}$0.00 nA, 21.80${\pm}$1.30 nA in bulk surface leakage current for array A and B respectively. Results of characterization demonstrate the importance of performance measurement validating the capability of APD array as the detector for PET imaging.

• 대한원격탐사학회지
• /
• 제39권6_3호
• /
• pp.1693-1705
• /
• 2023

국토 면적의 약 90%를 차지하는 농촌은 여러가지 공익적 기능을 수행하는 공간으로서 중요성과 가치가 증가하고 있지만 주거지 인근에 축사, 공장, 태양광패널 등 주민생활에 불편을 미치는 시설들이 무분별하게 들어서면서 농촌 환경과 경관이 훼손되고 주민 삶의 질이 낮아지고 있다. 농촌지역의 무질서한 개발을 방지하고 농촌 공간을 계획적으로 관리하기 위해서는 농촌지역 내 위해시설에 대한 탐지 및 모니터링이 필요하다. 주기적으로 취득 가능하고 전체 지역에 대한 정보를 얻을 수 있는 위성영상을 통해 데이터의 취득이 가능하고, 합성곱 신경망 기법을 통한 영상 기반 딥러닝 기술을 활용하여 효과적인 탐지가 가능하다. 따라서 본 연구에서는 의미적 분할(Semantic segmentation)에서 높은 성능을 보이는 U-Net 모델을 이용하여 농촌 지역에서 잠재적으로 위해시설이 될 수 있는 농촌시설을 분류하는 연구를 수행하였다. 본 연구에서는 2020년에 제작된 공간해상도 0.7 m의 KOMPSAT 정사모자이크 광학영상을 한국항공우주연구원으로부터 제공받아 사용하였으며 축사, 공장, 태양광 패널에 대한 AI 학습용 데이터를 직접 제작하여 학습 및 추론을 진행하였다. U-Net을 통해 학습시킨 결과 픽셀 정확도(pixel accuracy)는 0.9739, mean Intersection over Union (mIOU)은 0.7025의 값을 도출하였다. 본 연구 결과는 농촌 지역의 위험 시설물 모니터링에 활용될 수 있으며, 농촌계획 수립에 있어 기초 자료로 활용될 수 있을 것으로 기대된다.

• 대한원격탐사학회지
• /
• 제27권3호
• /
• pp.329-338
• /
• 2011

The time-series normalized difference vegetation index (NDVI) product has proven to be a powerful tool to investigate the phenological information because it can monitor the change of the forests with very high time-resolution, This study described the application of the DFT analysis over the 9 year MODIS data for the identification of the two types of vegetation cover, Pinus densiflora(Pd) and Querqus mongolica(Qm) which are dominant species of evergreen and broadleaved deciduous forest, respectively, The total number of samples was 5148 reference cycles which consist of 2160 Pd and 2988 Qm. They were extracted from the pixel-based MODIS scenes over the 9 years from 2000 to 2008 of South Korea. The DFT analysis was mainly focused on the 0th and $1^{st}$ harmonic components, each of which represents the mean value and the variation amplitude of the NDVI over the years, respectively. The $0^{th}$ harmonic values of the vegetation Pd and Qm averaged over the 9 years were 0.74 and 0.65, respectively. This implies that Pd has a higher NDVI than Qm. Similarly obtained $1^{st}$ harmonic values of Pd and Qm were 0.19 and 0.27, respectively. This can be intuitively understood considering that the seasonal variation of Qm is much larger than Pd. This distinctive difference of the $1^{st}$ harmonic value has been used to identify evergreen and deciduous forests. Overall agreement between the Fourier analysis-based map and the actal vegetation map has been estimated to be as high as 75%. This study found that the DFT analysis can be a concise and repeatable method to separate and trace the changes of evergreen and deciduous forest using the annual NDVI cycles.

• 한국통신학회논문지
• /
• 제24권8B호
• /
• pp.1480-1489
• /
• 1999

본 논문에서는 블록 기반 DCT 부호화 방식의 단점인 블록화 현상을 제거하기 위하여 MSDS 방법을 개선한 후 처리 기법을 제안한다. MSDS방법의 문제점인 예측된 DCT 계수값의 범위를 제한하기 위하여 입력 영상의 블록 경계 화소차 분포를 규정할 수 있는 OSLD(Overlapped Sub-Laplacian Distribution)를 정의한다. 블록화 현상은 블록간의 기울기를 이용하여 불연속 정도를 측정함으로써 정량화 되고, 정량화 된 값을 최소화하도록 양자화 오류값을 예측한다. OSLD를 이용하여 각 블록들을 네 가지 형태로 분류하고 이를 에지 부류와 평탄 부류로 구분한다. 에지 부류로 판별된 블록에서는 예측된 양자화 오류의 범위가 해당되는 양자화 간격보다 크면 이 간격으로 예측된 양자화 오류를 보정한다. 본 방법을 사용하여 실험 영상에서 블록화 현상을 제거할 때 기존의 MSDS 방법에서 요구하였던 입력 영상에 따라 실험적으로 문턱값을 설정하였던 문제점을 해결하고, PSNR 값을 영상에 따라 0.1∼0.3 dB 정도 향상시키면서 시각적으로 화질을 향상시킬 수 있다.

• 한국수자원학회:학술대회논문집
• /
• 한국수자원학회 2021년도 학술발표회
• /
• pp.136-136
• /
• 2021

In this study, a deep convolutional neural network (DCNN) model is proposed for short-term precipitation forecasting using weather radar-based images. The DCNN model is a combination of convolutional neural networks, autoencoder neural networks, and U-net architecture. The weather radar-based image data used here are retrieved from competition for rainfall forecasting in Korea (AI Contest for Rainfall Prediction of Hydroelectric Dam Using Public Data), organized by Dacon under the sponsorship of the Korean Water Resources Association in October 2020. This data is collected from rainy events during the rainy season (April - October) from 2010 to 2017. These images have undergone a preprocessing step to convert from weather radar data to grayscale image data before they are exploited for the competition. Accordingly, each of these gray images covers a spatial dimension of 120×120 pixels and has a corresponding temporal resolution of 10 minutes. Here, each pixel corresponds to a grid of size 4km×4km. The DCNN model is designed in this study to provide 10-minute predictive images in advance. Then, precipitation information can be obtained from these forecast images through empirical conversion formulas. Model performance is assessed by comparing the Score index, which is defined based on the ratio of MAE (mean absolute error) to CSI (critical success index) values. The competition results have demonstrated the impressive performance of the DCNN model, where the Score value is 0.530 compared to the best value from the competition of 0.500, ranking 16^th out of 463 participating teams. This study's findings exhibit the potential of applying the DCNN model to short-term rainfall prediction using weather radar-based images. As a result, this model can be applied to other areas with different spatiotemporal resolutions.

• 한국방사선학회논문지
• /
• 제16권3호
• /
• pp.281-288
• /
• 2022

본 연구에서는 고관절에 금속삽입물(Metal implant)이 삽입되어 있는 환자를 대상으로 일반적인 CT검사(Before MAR) 영상과 MAR을 사용하여 얻어진(After MAR) 영상을 4개의 알고리즘(Soft, Standard, Detail, Bone)에 적용하여 Noise, SNR, CNR을 비교 분석하여 정량적 평가로 최적의 알고리즘을 알아보고자 하였다. 분석방법으로는 4개의 알고리즘으로 재구성한 영상에 이미지 분석과 영역 및 픽셀값을 계산할 수 있는 Image J 프로그램을 사용하였다. Noise, SNR, CNR을 구하기 위해 측정부위를 영상에서 금속삽입물이 가장 인접해 있는 Bone(궁둥뼈, ischium)을 지정하여 HU mean값과 HU SD값을 구하고 배경잡음(Background)은 주위 근육으로 하였다. 관심영역(region of interest, ROI)은 뼈의 크기를 감안하여 동일하게 15×15 mm로 지정하였으며 SNR과 CNR의 값은 주어진 식에 의거하여 산출하였다. 결과적으로 노이즈는 After MAR, Soft 알고리즘에서 노이즈가 가장 낮게 나타났으며, SNR, CNR은 Before MAR, Soft 알고리즘이 가장 높게 나타났다. 따라서 Soft 알고리즘이 고관절 금속삽입술 CT에 가장 적절한 알고리즘으로 판단된다.