• 대한원격탐사학회지
• /
• 제38권4호
• /
• pp.327-341
• /
• 2022

Cloud removal is often required to construct time-series sets of optical images for environmental monitoring. In regression-based cloud removal, the selection of an appropriate regression model and the impact analysis of the input images significantly affect the prediction performance. This study evaluates the potential of Gaussian process (GP) regression for cloud removal and also analyzes the effects of cloud-free optical images and spectral bands on prediction performance. Unlike other machine learning-based regression models, GP regression provides uncertainty information and automatically optimizes hyperparameters. An experiment using Sentinel-2 multi-spectral images was conducted for cloud removal in the two agricultural regions. The prediction performance of GP regression was compared with that of random forest (RF) regression. Various combinations of input images and multi-spectral bands were considered for quantitative evaluations. The experimental results showed that using multi-temporal images with multi-spectral bands as inputs achieved the best prediction accuracy. Highly correlated adjacent multi-spectral bands and temporally correlated multi-temporal images resulted in an improved prediction accuracy. The prediction performance of GP regression was significantly improved in predicting the near-infrared band compared to that of RF regression. Estimating the distribution function of input data in GP regression could reflect the variations in the considered spectral band with a broader range. In particular, GP regression was superior to RF regression for reproducing structural patterns at both sites in terms of structural similarity. In addition, uncertainty information provided by GP regression showed a reasonable similarity to prediction errors for some sub-areas, indicating that uncertainty estimates may be used to measure the prediction result quality. These findings suggest that GP regression could be beneficial for cloud removal and optical image reconstruction. In addition, the impact analysis results of the input images provide guidelines for selecting optimal images for regression-based cloud removal.

• 대한원격탐사학회지
• /
• 제38권6_1호
• /
• pp.1357-1369
• /
• 2022

구름 제거는 식생 모니터링, 변화 탐지 등과 같은 광학 영상이 필요한 모든 작업에서 필수적인 영상 처리 과정이다. 이 논문에서는 조건부 생성적 적대 신경망(conditional generative adversarial networks, cGANs)과 회귀 기반 보정을 결합하여 구름이 없는 시계열 광학 영상 세트를 구성하는 2단계의 구름 제거 기법을 제안하였다. 첫 번째 단계에서는 광학 영상과 synthetic aperture radar 영상 간 정량적 관계를 이용하는 cGANs을 이용하여 초기 예측 결과를 생성한다. 두 번째 단계에서는 구름이 아닌 영역에서 예측 결과와 실제 값과의 관계를 random forest 기반 회귀 모델링을 통해 정량화한 후에 cGANs 기반 예측 결과를 보정한다. 제안 기법은 김제의 벼 재배지에서 Sentinel-2 영상과 COSMO-SkyMed 영상을 이용한 구름 제거 실험을 통해 적용 가능성을 평가하였다. cGAN 모델은 구름 영역에서 지표면 상태의 급격한 변화가 발생하는 논 재배지를 대상으로 반사율 값을 효과적으로 예측할 수 있었다. 또한 두 번째 단계의 회귀 기반 보정은 예측 대상 영상에서 시간적으로 떨어진 보조 영상을 이용하는 회귀 기반 구름 제거 기법에 비해 예측 정확도를 향상시킬 수 있었다. 이러한 실험 결과는 구름이 없는 광학 영상을 환경 모니터링에 이용할 수 없는 경우 제안된 방법이 구름 오염 지역을 복원하는데 효과적으로 적용될 수 있음을 나타낸다.

• 대한원격탐사학회지
• /
• 제39권5_1호
• /
• pp.507-519
• /
• 2023

Multi-temporal optical images have been utilized for time-series monitoring of croplands. However, the presence of clouds imposes limitations on image availability, often requiring a cloud removal procedure. This study assesses the applicability of various machine learning algorithms for effective cloud removal in optical imagery. We conducted comparative experiments by focusing on two key variables that significantly influence the predictive performance of machine learning algorithms: (1) land-cover types of training data and (2) temporal variability of land-cover types. Three machine learning algorithms, including Gaussian process regression (GPR), support vector machine (SVM), and random forest (RF), were employed for the experiments using simulated cloudy images in paddy fields of Gunsan. GPR and SVM exhibited superior prediction accuracy when the training data had the same land-cover types as the cloud region, and GPR showed the best stability with respect to sampling fluctuations. In addition, RF was the least affected by the land-cover types and temporal variations of training data. These results indicate that GPR is recommended when the land-cover type and spectral characteristics of the training data are the same as those of the cloud region. On the other hand, RF should be applied when it is difficult to obtain training data with the same land-cover types as the cloud region. Therefore, the land-cover types in cloud areas should be taken into account for extracting informative training data along with selecting the optimal machine learning algorithm.

• 한국BIM학회 논문집
• /
• 제10권4호
• /
• pp.32-39
• /
• 2020

In recent years, the construction industry is getting bigger and more complex, so it is becoming difficult to acquire point cloud data for construction equipments and workers. Point cloud data is measured using a drone and MMS(Mobile Mapping System), and the collected point cloud data is used to create a 3D digital map. In particular, the construction site is located at outdoors and there are many irregular terrains, making it difficult to collect point cloud data. For these reasons, adopting a noise reduction algorithm suitable for the characteristics of the construction industry can affect the improvement of the analysis accuracy of digital maps. This is related to various environments and variables of the construction site. Therefore, this study reviewed and analyzed the existing research and techniques on the noise reduction algorithm. And based on the results of literature review, performance evaluation of major noise reduction algorithms was conducted for digital maps of construction sites. As a result of the performance evaluation in this study, the voxel grid algorithm showed relatively less execution time than the statistical outlier removal algorithm. In addition, analysis results in slope, space, and earth walls of the construction site digital map showed that the voxel grid algorithm was relatively superior to the statistical outlier removal algorithm and that the noise removal performance of voxel grid algorithm was superior and the object preservation ability was also superior. In the future, based on the results reviewed through the performance evaluation of the noise reduction algorithm of this study, we will develop a noise reduction algorithm for 3D point cloud data that reflects the characteristics of the construction site.

• 대기
• /
• 제17권2호
• /
• pp.195-205
• /
• 2007

The influences of ice microphysical processes on urban heat island-induced convection and precipitation are numerically investigated using a cloud-resolving model (ARPS). Both warm- and cold-cloud simulations show that the downwind upward motion forced by specified low-level heating, which is regarded as representing an urban heat island, initiates moist convection and results in downwind precipitation. The surface precipitation in the cold-cloud simulation is produced earlier than that in the warm-cloud simulation. The maximum updraft is stronger in the cold-cloud simulation than in the warm-cloud simulation due to the latent heat release by freezing and deposition. The outflow formed in the boundary layer is cooler and propagates faster in the cold-cloud simulation due mainly to the additional cooling by the melting of falling hail particles. The removal of the specified low-level heating after the onset of surface precipitation results in cooler and faster propagating outflow in both the warm- and cold-cloud simulations.

• 한국대기환경학회:학술대회논문집
• /
• 한국대기환경학회 2006년도 추계학술대회 논문집
• /
• pp.47-50
• /
• 2006

• Journal of Korean Society for Atmospheric Environment
• /
• 제24권E1호
• /
• pp.44-53
• /
• 2008

This study was undertaken to measure the properties of individual mineral particles in an artificially saturated atmosphere at a vertical extinct mine with 430 m height. By synchrotron radiation X-ray fluorescence (SR-XRF) microprobe analysis, it was possible to determine the elemental composition of residual insoluble particles on individual cloud droplet replicas formed on the Collodion film. The XRF visualized elemental maps enabled us not only to presume the chemical mixing state of particles retained in cloud droplet, but also to estimate their source. Details about the individual mineral particles captured by artificial cloud droplets should be helpful to understand about the removal characteristics of dust particles such as interaction with clouds. Nearly all individual particles captured in cloud droplets are strongly enriched in Fe. Mass of Fe is ranged between 41 fg and 360 fg with average 112 fg. There is a good agreement between single particle analysis by SR-XRF and bulk particle analysis by PIXE.

• 대한원격탐사학회지
• /
• 제37권5_1호
• /
• pp.1149-1161
• /
• 2021

컴퓨터 비전 기술이 위성영상에 적용되면서, 최근 들어 딥러닝 영상인식을 이용한 구름 탐지가 관심을 끌고 있다. 본연구에서는 SPARCS (Spatial Procedures for Automated Removal of Cloud and Shadow) Cloud Dataset과 영상자료증대 기법을 활용하여 U-Net 구름탐지 모델링을 수행하고, 10폴드 교차검증을 통해 객관적인 정확도 평가를 수행하였다. 512×512 화소로 구성된 1800장의 학습자료에 대한 암맹평가 결과, Accuracy 0.821, Precision 0.847, Recall 0.821, F1-score 0.831, IoU (Intersection over Union) 0.723의 비교적 높은 정확도를 나타냈다. 그러나 구름그림자 중 14.5%, 구름 중 19.7% 정도가 땅으로 잘못 예측되기도 했는데, 이는 학습자료의 양과 질을 보다 더 향상시킴으로써 개선 가능할 것으로 보인다. 또한 최근 각광받고 있는 DeepLab V3+ 모델이나 NAS(Neural Architecture Search) 최적화 기법을 통해 차세대중형위성 1, 2, 4호 등의 구름탐지에 활용 가능할 것으로 기대한다.

• International Journal of Fuzzy Logic and Intelligent Systems
• /
• 제13권1호
• /
• pp.67-72
• /
• 2013

Cloud detection and analysis from satellite images has been a topic of research in many atmospheric and environmental studies; however, it still is a challenging task for many reasons. In this paper, we propose a new method for cloud-type classification using fuzzy logic. Knowing that visible-light images of clouds contain thickness related information, while infrared images haves height-related information, we propose a two-layered fuzzy logic based on the input source to provide us with a relatively clear-cut threshold in classification. Traditional noise-removal methods that use reflection/release characteristics of infrared images often produce false positive cloud areas, such as fog thereby it negatively affecting the classification accuracy. In this study, we used the color information from source images to extract the region of interest while avoiding false positives. The structure of fuzzy inference was also changed, because we utilized three types of source images: visible-light, infrared, and near-infrared images. When a cloud appears in both the visible-light image and the infrared image, the fuzzy membership function has a different form. Therefore we designed two sets of fuzzy inference rules and related classification rules. In our experiment, the proposed method was verified to be efficient and more accurate than the previous fuzzy logic attempt that used infrared image features.

• 한국건축시공학회:학술대회논문집
• /
• 한국건축시공학회 2020년도 봄 학술논문 발표대회
• /
• pp.40-41
• /
• 2020

As the capacity of the 3d scanner developed, the reverse engineering using the 3d scanner is emphasized in the construction industry to obtain the 3d geometric representation of buildings. However, big size of the indoor point cloud data acquired by the 3d scanner restricts the efficient process in the reverse engineering. In order to solve this inefficiency, several pre-processing methods simplifying and denoising the raw point cloud data by the rough standard are developed, but these non-standard methods can cause the inaccurate recognition and removal the key-points. This paper analyzes the correlation between the accuracy of wall recognition and the density of the data, thus proposes the proper method for the raw point cloud data. The result of this study could improve the efficiency of the data processing phase in the reverse engineering for indoor point cloud data.