• Journal of Electrical Engineering and Technology
• /
• v.13 no.1
• /
• pp.38-50
• /
• 2018

Spatio-temporal load forecasting (STLF) is a foundation for building the prediction-based power map, which could be a useful tool for the visualization and tendency assessment of urban energy application. Constructing one point-forecasting model for each electricity cell in the geographic space is possible; however, it is unadvisable and insufficient, considering the aggregation features of electricity cells and uncertainties in input variables. This paper presents a new STLF method, with a data-driven framework consisting of 3 subroutines: multi-level clustering of cells considering their aggregation features, load regression for each category of cells based on SLS-SVRNs (sparse least squares support vector regression networks), and interval forecasting of spatio-temporal load with sampled blind number. Take some area in Pudong, Shanghai as the region of study. Results of multi-level clustering show that electricity cells in the same category are clustered in geographic space to some extent, which reveals the spatial aggregation feature of cells. For cellular load regression, a comparison has been made with 3 other forecasting methods, indicating the higher accuracy of the proposed method in point-forecasting of spatio-temporal load. Furthermore, results of interval load forecasting demonstrate that the proposed prediction-interval construction method can effectively convey the uncertainties in input variables.

• Korean Journal of Remote Sensing
• /
• v.32 no.1
• /
• pp.25-38
• /
• 2016

With the increasing need for high temporal resolution satellite imagery for monitoring land surfaces, this study evaluated the temporal resolution of the NDVI composites from Geostationary Ocean Color Imager (GOCI) data. The GOCI is the first geostationary satellite sensor designed to provide continuous images over a $2,500{\times}2,500km^2$ area of the northeast Asian region with relatively high spatial resolution of 500 m. We used total 2,944 hourly images of the GOCI level 1B radiance data obtained during the one-year period from April 2011 to March 2012. A daily NDVI composite was produced by maximum value compositing of eight hourly images captured during day-time. Further NDVI composites were created with different compositing periods ranging from two to five days. The cloud coverage of each composite was estimated by the cloud detection method developed in study and then compared with the Moderate Resolution Imaging Spectroradiometer (MODIS) Aqua cloud product and 16-day NDVI composite. The GOCI NDVI composites showed much higher temporal resolution with less cloud coverage than the MODIS NDVI products. The average of cloud coverage for the five-day GOCI composites during the one year was only 2.5%, which is a significant improvement compared to the 8.9%~19.3% cloud coverage in the MODIS 16-day NDVI composites.

• The Bulletin of The Korean Astronomical Society
• /
• v.42 no.2
• /
• pp.81.4-82
• /
• 2017

We investigate velocity oscillations in the active region plage by using the high-spatial, high-spectral and high-temporal resolution spectral data acquired by the Interface Region Imaging Spectrograph (IRIS). From the Mn I $2801.907{\AA}$ (lower chromosphere), C II (lower transition region) and Si IV (middle transition region) lines, we measure the line of sight Doppler velocity at different atmospheric layers, and present results of wavelet analysis of the plage region with a range of periods from 2 to 8 minutes. In addition, we present correlations of the oscillations from the lower chromosphere to the middle transition region. Finally, we will discuss the regional dependence of the oscillation properties on physical properties such as temperature and magnetic field inclination.

• The Bulletin of The Korean Astronomical Society
• /
• v.38 no.2
• /
• pp.90.1-90.1
• /
• 2013

To better understand the physics underlying the eruption of prominences in solar active regions, we studied eruption processes of two active prominences located in the active region NOAA 11261 using multi-wavelength observational data with high temporal and spatial resolution. Specifically, we examined (1) the temporal variation of morphology and plasma properties of the two active prominences, (2) magnetic fields and their evolution on the photospheric surface underneath the prominences, and (3) the time profiles and locations of radio, EUV, and soft/hard X-ray emissions produced by the M9.3 flare related to the prominence eruption. As a result, we found that: (1) a prominence F1 began to erupt and expand as the abrupt and intense EUV brightening occurred in the localized region underneath the western part of F1 at 03:45 UT prior to the peak time of the M9.3 flare, (2) F1 split into two parts: i.e., the western part asymmetrically erupted by producing the M9.3 flare with microwave source motions along the magnetic polarity inversion line between the two flare ribbons, while the eastern part coalesced into a pre-existing prominence F2, (3) F2 became unstable due to the coalescence with the eastern part of F1, and then it partially erupted with clockwise untwisting motions.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.16 no.7
• /
• pp.2345-2358
• /
• 2022

This paper presents a method to generate stained-glass animation from video inputs. The method initially segments an input video volume into several regions considered as fragments of glass by mean-shift segmentation. However, the segmentation predominantly results in over-segmentation, causing several tiny segments in a highly textured area. In practice, assembling significantly tiny or large glass fragments is avoided to ensure architectural stability in stained glass manufacturing. Therefore, we use low-frequency components in the segmentation to prevent over-segmentation and subdivide segmented regions that are oversized. The subdividing must be coherent between adjacent frames to prevent temporal artefacts, such as flickering and the shower door effect. To temporally subdivide regions coherently, we obtain a panoramic image from the segmented regions in input frames, subdivide it using a weighted Voronoi diagram, and thereafter project the subdivided regions onto the input frames. To render stained glass fragment for each coherent region, we determine the optimal match glass fragment for the region from a dataset consisting of real stained-glass fragment images and transfer its color and texture to the region. Finally, applying lead came at the boundary of the regions in each frame yields temporally coherent stained-glass animation.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2020.11a
• /
• pp.237-240
• /
• 2020

The fine-tuned neural network (NN) model for a whole temporal portion in a video does not always yield the best quality (e.g., PSNR) performance over all regions of each frame in the temporal period. For certain regions (usually homogeneous regions) in a frame for super-resolution (SR), even a simple bicubic interpolation method may yield better PSNR performance than the fine-tuned NN model. When there are multiple NN models available at the receivers where each NN model is trained for a group of images having a specific category of image characteristics, the performance of Quality enhancement can be improved by selectively applying an appropriate NN model for each image region according to its image characteristic category to which the NN model was dedicatedly trained. In this case, it is necessary to signal which NN model is applied for each region. This is very advantageous for image restoration and quality enhancement (IRQE) applications at user terminals with limited computing capabilities.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2018.05a
• /
• pp.154-154
• /
• 2018

To interpret the climate projections for the future as well as present, recognition of the consequences of the climate internal variability and quantification its uncertainty play a vital role. The Korean Peninsula belongs to the Far East Asian Monsoon region and its rainfall characteristics are very complex from time and space perspective. Its internal variability is expected to be large, but this variability has not been completely investigated to date especially using models of high temporal resolutions. Due to coarse spatial and temporal resolutions of General Circulation Models (GCM) projections, several studies adopted dynamic and statistical downscaling approaches to infer meterological forcing from climate change projections at local spatial scales and fine temporal resolutions. In this study, stochastic downscaling methodology was adopted to downscale daily GCM resolutions to hourly time scale using an hourly weather generator, the Advanced WEather GENerator (AWE-GEN). After extracting factors of change from the GCM realizations, these were applied to the climatic statistics inferred from historical observations to re-evaluate parameters of the weather generator. The re-parameterized generator yields hourly time series which can be considered to be representative of future climate conditions. Further, 30 ensemble members of hourly precipitation were generated for each selected station to quantify uncertainty. Spatial map was generated to visualize as separated zones formed through K-means cluster algorithm which region is more inconsistent as compared to the climatological norm or in which region the probability of occurrence of the extremes event is high. The results showed that the stations located near the coastal regions are more uncertain as compared to inland regions. Such information will be ultimately helpful for planning future adaptation and mitigation measures against extreme events.

• Korean Journal of Remote Sensing
• /
• v.39 no.5_1
• /
• 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.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.46 no.5
• /
• pp.40-47
• /
• 2009

A motion adaptive video noise reduction scheme is proposed by cascading a temporal filter and a spatial filter. After a noise-robust motion detection is performed with a pre-spatial filter, the strength of the motion adaptive temporal filter is controlled by the amount of temporal movement. In order to fully utilize the temporal correlation of video signal, noisy input image is processed first by the temporal filter, therefore, image details of temporally stationary region are quite well preserved while undesired noises are suppressed. In contrast to the pre-spatial filter used for the robust motion detection, the cascaded post-spatial filter removes the remained noises by considering the strength of the temporal filter and the spatial self-similarity search results obtained from the pre-spatial filter.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.48 no.2
• /
• pp.24-31
• /
• 2011

This paper proposes spatio-temporal denoising algorithm based on nonlocal means. Though the conventional denoising algorithms based on nonlocal means have good performance in noise removal, it is difficult to implement them into the hardware system due to much computational load and the need for several frame buffers. Therefore we adopted infinite impulse response temporal noise reduction algorithm in the proposed algorithm. Proposed algorithm shows less artificial denoised result in the motionless region. In the motion region, spatial filter based on efficiently improved nonlocal means algorithm conduct noise removal with less motion blur. Experimental results including comparisons with conventional algorithms for various noise levels and test images show the proposed algorithm has a good performance in both visual and quantitative criteria.