• 대한원격탐사학회지
• /
• 제20권4호
• /
• pp.275-288
• /
• 2004

이 논문은 다중 센서 원격탐사 화상의 분류를 위해 퍼지 논리 융합과 결합된 relaxation labeling 방법을 제안하였다. 다중 센서 원격탐사 화상의 융합에는 퍼지 논리를, 분광정보와 공간정보의 융합에는 반복적인 relaxation labeling 방법을 적용하였다. 특히 반복적 relaxation labeling 방법은 공간정보의 이용에 따른 분류 화소의 변화양상을 얻을 수 있는 장점이 있다. 토지 피복의 감독 분류를 목적으로 광학 화상과 다중 주파수/편광 SAR 화상에 제안 기법을 적용한 결과, 다중 센서 자료를 이용하고 공간정보를 함께 결합하였을 때 향상된 분류 정확도를 얻을 수 있었다.

• International Journal of Computer Science & Network Security
• /
• 제22권10호
• /
• pp.406-412
• /
• 2022

Scene classification of very high-resolution (VHR) imagery can attribute semantics to land cover in a variety of domains. Real-world application requirements have not been addressed by conventional techniques for remote sensing image classification. Recent research has demonstrated that deep convolutional neural networks (CNNs) are effective at extracting features due to their strong feature extraction capabilities. In order to improve classification performance, these approaches rely primarily on semantic information. Since the abstract and global semantic information makes it difficult for the network to correctly classify scene images with similar structures and high interclass similarity, it achieves a low classification accuracy. We propose a VHR remote sensing image classification model that uses extracts the global feature from the original VHR image using an EfficientNet-V2L CNN pre-trained to detect similar classes. The image is then classified using a multilayer perceptron (MLP). This method was evaluated using two benchmark remote sensing datasets: the 21-class UC Merced, and the 38-class PatternNet. As compared to other state-of-the-art models, the proposed model significantly improves performance.

• 대한원격탐사학회지
• /
• 제27권3호
• /
• pp.289-301
• /
• 2011

In this paper an attempt has been made to classify Asia land cover considering climatic and vegetative characteristics. The sub-class clustering based on the 13 MODIS land cover classes (except water) over Asia was performed with the climate map and the NOVI derived from SPOT 5 VGT D10 data. The unsupervised classification for the sub-class clustering was performed in each land cover class, and total 74 clusters were determined over the study area. Via these clusters, the annual variations (from 1999 to 2007) of precipitation rate and temperature were analyzed as an example by a simple linear regression model. The various annual variations (negative or positive pattern) were represented for each cluster because of the various climate zones and NOVI annual cycles. Therefore, the detailed land cover map as the classification result by the sub-class clustering in this study can be useful information in modelling works for requiring the detailed climatic and vegetative information as a boundary condition.

• 대한원격탐사학회지
• /
• 제34권4호
• /
• pp.601-610
• /
• 2018

The national accuracy of global land cover (GLC) products is of great importance to ecological and environmental research. However, GLC products that are derived from different satellite sensors, with differing spatial resolutions, classification methods, and classification schemes are certain to show some discrepancies. The goal of this study is to assess the accuracy of four commonly used GLC datasets in South Korea, GLC2000, GlobCover2009, MCD12Q1, and GlobeLand30. First, we compared the area of seven classes between four GLC datasets and a reference dataset. Then, we calculated the accuracy of the four GLC datasets based on an aggregated classification scheme containing seven classes, using overall, producer's and user's accuracies, and kappa coefficient. GlobeLand30 had the highest overall accuracy (77.59%). The overall accuracies of MCD12Q1, GLC2000, and GlobCover2009 were 75.51%, 68.38%, and 57.99%, respectively. These results indicate that GlobeLand30 is the most suitable dataset to support a variety of national scientific endeavors in South Korea.

• 한국농공학회논문집
• /
• 제48권5호
• /
• pp.39-48
• /
• 2006

The objective of this research was to find an indirect method to estimate land surface temperature (LST) efficiently, using Landsat images. Agricultural fields including paddy fields have long been known to have multi-functions beneficial to the environment and ecology of the urban surrounding areas. Among these functions, the ambient temperature cooling (ATC) effect is widely acknowledged. However, quantitative and regional assessment of such effect has not been performed. Thermal remote sensing has been used over urban areas to assess the ATC effect, Thermal Island Effect(TIE), and as input for models of urban surface atmosphere exchange. Here, we review the use of thermal remote sensing in the study of paddy fields and urban climates, focusing primarily on the ATC effect. Landsat satellite images were used to determine the surface temperatures of different land cover types of a $44km^{2}$ study area in Cheongiu, Korea. The results show that the ATC is a function of paddy area percentage in Landsat pixels. Landsat pixels with higher paddy area percentage have much more cooling effect. The use of satellite data may contribute to a globally consistent method for analysis of ATC effect.

• 대한원격탐사학회지
• /
• 제23권3호
• /
• pp.181-188
• /
• 2007

The objective of this research was to investigate the optimal land cover classification algorithm for the monitoring of North Korea with MODIS multi-temporal data based on monthly phenological characteristics. Three frequently used land cover classification algorithms, ISODATA1), SMA2), and SOM3) were employed for this study; the land cover categories were forest, grass, agricultural, wetland, barren, built-up, and water body. The outcomes of the study can be summarized as follows. First, the overall classification accuracy of ISODATA, SMA, and SOM was 69.03%, 64.28%, and 73.57%, respectively. Second, ISODATA and SMA resulted in a higher classification accuracy of forest and agricultural categories, but SOM performed better for the built-up area, bare soil, grassland, and water. A possible explanation for this difference would be related to the difference of sensitivity against the vegetation activity. This would be related to the capability of SOM to express all of their values without any loss of data by maintaining the topology between pixels of primitive data after classification, while ISODATA and SMA retain limited amount of data after normalization process. Third, we can conclude that SOM is the best algorithm for monitoring the land cover change of North Korea.

• 대한원격탐사학회지
• /
• 제22권5호
• /
• pp.421-425
• /
• 2006

It can be important framework data to monitor the change of land-use pattern of river boundary in design and management of river. This study analyzed the change of land-use pattern of Gab and Yudeung River using time-series aerial images. To do this, we carried out radiation and geometric correction of image, and estimated land-use changes in inland and floodplain. As the analysis of inland, the ratio of residential, commercial, industrial, educational and public area, that is urbanized element, increases, but that of agricultural area shows a decline on the basis of 1990. Also, Minimum Distance Method, which is a kind of supervised classification method, is applied to extract water-body and sand bar layer in floodplain. As the analysis of land-use, the ratio of level-upped riverside land and water-body increases, but that of sand bar decreases. These time-series land use information can be important decision making data to evaluate the urbanization of river boundary, and especially it gives us goodness in river development project such as the composition of ecological habitat.

• 대한원격탐사학회지
• /
• 제15권3호
• /
• pp.239-251
• /
• 1999

The elimination process of cloud-contaminated pixels is one of important steps before obtaining the accurate parameters of land and ocean surface from AVHRR imagery. We developed a 6step threshold method to detect the cloud-contaminated pixels from NOAA-14/AVHRR datime imagery over land using different combination of channels. This algorithm has two phases : the first is to make a cloud-free characteristic data of land surface using compositing techniques from channel 1 and 5 imagery and a dynamic threshold of brightness temperature, and the second is to identify the each pixel as a cloud-free or cloudy one through 4-step threshold tests. The merits of this method are its simplicity in input data and automation in determining threshold values. The threshold of infrared data is calculated through the combination of brightness temperature of land surface obtained from AVHRR imagery, spatial variance of them and temporal variance of observed land surface temperature. The method detected the could-comtaminated pixels successfully embedded inthe NOAA-14/AVHRR daytime imagery for the August 1 to November 30, 1996 and March 1 to July 30, 1997. This method was evaluated through the comparison with ground-based cloud observations and with the enhanced visible and infrared imagery.

• 대한원격탐사학회지
• /
• 제24권1호
• /
• pp.35-43
• /
• 2008

The purpose of this study is to evaluate the hydrological impact due to temporal land cover change by gradual urbanization of upstream watershed of Pyeongtaek gauging station of Anseong-cheon. WMS HEC-1 was adopted, and OEM with 200 m resolution and hydrologic soil group from 1:50,000 scale soil map were prepared. Land covers of 1986, 1990, 1994 and 1999 Landsat TM images were classified by maximum likelihood method. The watershed showed a trend that forest & paddy areas decreased and urban/residential area gradually increased during the four selected years. The model was calibrated at 2 locations (Pyeonglaek and Gongdo) by comparing observed with simulated discharge results for 5 summer storm events from 1998 to 2001. The watershed average CN values varied from 61.7 to 62.3 for the 4 selected years. To identify the impact of streamflow by temporal area change of a target land use, a simple evaluation method that the CN values of areas except the target land use are unified as one representative CN value was suggested. By applying the method, watershed average CN value was affected in the order of paddy, forest and urban/residential, respectively.

• 대한원격탐사학회지
• /
• 제37권4호
• /
• pp.763-776
• /
• 2021

This paper introduces a method of selecting priority update areas for subdivided land cover maps by training orthoimages and serial cadastral maps in a deep learning model. For the experiment, orthoimages and serial cadastral maps were obtained from the National Spatial Data Infrastructure Portal. Based on the VGG-16 model, 51,470 images were trained on 33 subdivided classifications within the experimental area and an accuracy evaluation was conducted. The overall accuracy was 61.42%. In addition, using the differences in the classification prediction probability of the misclassified polygon and the cosine similarity that numerically expresses the similarity of the land category features with the original subdivided land cover class, the cases were classified and the areas in which the boundary setting was incorrect and in which the image itself was determined to have a problem were identified as the priority update polygons that should be checked by operators.