• Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography
• /
• v.37 no.3
• /
• pp.199-208
• /
• 2019

As the number of available satellites increases and technology advances, image information outputs are becoming increasingly diverse and a large amount of data is accumulating. In this study, we propose a change detection method for high-resolution satellite images that uses transfer learning and a deep learning network to overcome the limit caused by insufficient training data via the use of pre-trained information. The deep learning network used in this study comprises convolutional layers to extract the spatial and spectral information and convolutional long-short term memory layers to analyze the time series information. To use the learned information, the two initial convolutional layers of the change detection network are designed to use learned values from 40,000 patches of the ISPRS (International Society for Photogrammertry and Remote Sensing) dataset as initial values. In addition, 2D (2-Dimensional) and 3D (3-dimensional) kernels were used to find the optimized structure for the high-resolution satellite images. The experimental results for the KOMPSAT-3A (KOrean Multi-Purpose SATllite-3A) satellite images show that this change detection method can effectively extract changed/unchanged pixels but is less sensitive to changes due to shadow and relief displacements. In addition, the change detection accuracy of two sites was improved by using 3D kernels. This is because a 3D kernel can consider not only the spatial information but also the spectral information. This study indicates that we can effectively detect changes in high-resolution satellite images using the constructed image information and deep learning network. In future work, a pre-trained change detection network will be applied to newly obtained images to extend the scope of the application.

• Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography
• /
• v.37 no.6
• /
• pp.491-498
• /
• 2019

This study analyzed how the accuracy of co-registration varies depending on the convergence angles between two KOMPSAT-3·3A images. Most very-high-resolution satellite images provide initial coordinate information through metadata. Since the search area for performing image co-registration can be reduced by using the initial coordinate information, in this study, the mutual information method showing high matching reliability in the small search area is used. Initial coarse co-registration was performed by using multi-spectral images with relatively low resolution, and precise fine co-registration was conducted centering on the region of interest of the panchromatic image for more accurate co-registration performance. The experiment was conducted by 120 combination of 16 KOMPSAT-3·3A 1G images taken in Daejeon area. Experimental results show that a correlation coefficient between the convergence angles and fine co-registration errors was 0.59. In particular, we have shown the larger the convergence angle, the lower the accuracy of co-registration performance.

• Journal of the Korean Geographical Society
• /
• v.37 no.4
• /
• pp.371-384
• /
• 2002

It is very important to predict the environmental changes in the land-surface as a way of prevention of sustainable nature. This study investigated the difference between the predicted and actual data of Anmyeon-do from 1981 to 2000 through a fuzzy logic operation using multi-spectral image. According to literature survey, maps, and ground truth data, the types of land-use have changed due primarily to shore reclamation or wild land and grassland fostering before the eighties. After the mid-eighties, however, a number of private residents and commercial stores quickly have spreaded throughout beach resorts and quasi-agricultural and forest areas. Moreover, shore and community regions were severely damaged in the nineties with increased farmland, due to the development of tour places and expansion of city area. The predicted result of the environmental changes in the land-surface using the fuzzy logic operation was almost similar to the state of Anmyeon-do obtained through the satellite image. Particularly, the flat lands near the shore was predicted to change slightly. This area is largely under development, thereby raising concerns on the shore environment. Thus, this method is applicable to conducting research on the change in the land-surface.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2006.05a
• /
• pp.462-466
• /
• 2006

Close-to-nature stream evaluation is one of the processing to make the streams over in order to keep them natural. It is integral to evaluate and make an accurate analysis of them on the purpose of maintaining streams healthy. For many instances, there are, stream organization evaluation for restoration by German government, evaluation for ecosystem protection in natural preserves by New Zealand government, and stream-view evaluation for restoration by Britain government so on. In case of the country there are analysis and evaluation of stream physical organization by Cho, Yong-hyun, Close-to-nature stream evaluation for restoration by Kim, Dong-chan, evaluation of stream properties in korea by Park, Bong-jin. Close-to-nature evaluation by Lim, Chan-uk, that is advanced version of Park, Bong-jin's, shows form of stream including waterway curve, sand bar, diversity of flow, river bed material, diversity of minor bed, minor bed bank protection works, bank protection material. It also does environment of stream including side of minor bed vegetation, width of surface of the water/width of the river etc.. By the way, this evaluation does not have free access to apply those details above in the field, it often happens that you get various outcome from the one spot. so you must need more realistic testing method to obtain more accurate data. Remote sensing method is highly recommended because this is very useful for collecting realistic data of vegetation index. what is more, it can not only scan even the minimum area within its resolving power but also do obtain data anytime. Vegetation index indicates Ratio vegetation index, Normalized difference vegetation index, Soil adjusted vegetation index, Atmospherically resistant vegetation index etc.. The research is focusing on Cheokgwa stream which is the branch of Taehwa river and shows 19 sectioned Close-to-nature stream performed according to the method by Lim, chan-uk. Besides let you know vegetation index came from image data of satellite landsat 7 with the variation of buffering area, of the day 9. may. 2003. Of all, the outcome 0.758 at 200m buffer-zone of NDVI was the best we have got so far.

• Journal of Korean Society of Environmental Engineers
• /
• v.38 no.10
• /
• pp.551-557
• /
• 2016

Fusing satellite images and site-specific observations have potential to improve a predictive quality of environmental properties. However, the effect of the utilization of satellite images to predict soil properties in a wetland is still poorly understood. For the reason, block kriging and regression kriging were applied to a natural wetland, Water Conservation Area-2A in Florida, to compare the accuracy improvement of continuous models predicting total phosphorus in soils. Field observations were used to develop the soil total phosphorus prediction models. Additionally, the spectral data and derived indices from Landsat ETM+, which has 30 m spatial resolution, were used as independent variables for the regression kriging model. The block kriging model showed $R^2$ of 0.59 and the regression kriging model showed $R^2$ of 0.49. Although the block kriging performed better than the regession kriging, both models showed similar spatial patterns. Moreover, regression kriging utilizing a Landsat ETM+ image facilitated to capture unique and complex landscape features of the study area.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.13 no.12
• /
• pp.5931-5936
• /
• 2012

This study was to evaluate the clinical usefulness of the PRESS technique based on the correlation between PRESS technique and biopsy results by applying 3.0T high magnetic field MRS technique for evaluation of non-alcoholic fatty liver disease patients. This experiment were carried out using a 3.0T magnetic resonance imaging equipment. The part data of each spectrum is taken by peak area integration. The part data of resonance peak was used to calculate relative ratio. MR spectral peak in patients with non-alcoholic fatty liver disease is from 0.9 to 1.6 ppm. According to MRS method study result, Patients with NAFLD were obtained with 94% sensitivity and 80% specificity(p=0.000). When compared to normal based on MRS and Biopsy results was valid correlation(r=0.79, p=0.04). Results for NAFLD(r=0.89, p=0.002) also showed a correlation. Therefore, PRESS technique to evaluate patients with non-alcoholic fatty liver disease, the distribution difference between normal liver and fatty liver part is significantly distinguished. Biopsy and MRS fatty liver peak ratio(%) proves high lipid over grade(r = 0.7).

• Journal of Sensor Science and Technology
• /
• v.1 no.1
• /
• pp.5-12
• /
• 1992

We have fabricated a-Si:H multilayer for contact-type linear image sensor by means of RF glow discharge decomposition method. The ITO/i-a-Si:H/Al structure has relatively high dark current due to indium diffusion and carrier injection from both electrodes, resulting in low photocurrent to dark current. To suppress the dark current and to enhance interface electric field between ITO and i-a-Si:H film we have fabricated ITO/insulator/i-a-S:H/p-a-S:H/Al multilayer film with blocking structure. The photocurrent of ITO/$SiO_{2}(300{\AA})$/i-a-Si:H/p-a-Si:H($1500{\AA}$)/Al multilayer sensor with 5V bias voltage became saturated at about 20nA under $20{\mu}W/cm^{2}$ light intensity, while the dark current was less than 0.1nA. To increase the light generation efficiency we have adopted ITO/$SiO_{x}N_{y}(300{\AA})$/i-a-Si:H/p-a-Si:H($1500{\AA}$)/Al structure, showing photocurrent of 30nA and dark current of 0.08nA with 5V bias voltage. Also the spectral photosensitivity of the multilayer was enhanced for short wavelength visible region of 560nm, compared with that of the a-Si:H monolayer of 630nm. And its photoresponse time was about 0.3msec with the film homogeneity of 5% deviation.

• Journal of Korean Society of Forest Science
• /
• v.88 no.2
• /
• pp.157-168
• /
• 1999

This paper analyzed vegetation and land use classification, slope and permanent preservation and legal reserves on the farm Jangada and Jamaica-Mato Grosso do Sul, Brazil, using satellite image for assisting the planning of planting Eucalyptus spp. This part of the State of Mato Grosso do Sul represents an important geopolitcal area, since it is located on the borders of Bolivia and Paraguay. Also exportation of goods can be achieved through hydrovias extending to Buenos Aires, Argentina-through the Paraguay River. Also there are road and railroad connection which link the soutreastern part of Brazil to the Andean countries. The vegetation map from sheet SF 21-Campo Grande of the RADAMBRASIL Project was used as the basis for the preliminary interpretation of coverage, and complemented by a visit of the field. After the initial interpretation of the image, definition of classes of use and land occupation were made, and files of spectral signatures were created. On the farms Jamaica and Jangada Open Arboreal Savanna and Grass Savanna are the predominant physiognomies occupying 68% of total area. In spite of the results being satisfactory at the present moment, the development of this project should be revised and adjusted based on the evaluations already made, including a greater detailing of environmental components, principally with respect to soil and topography.

• Journal of radiological science and technology
• /
• v.40 no.3
• /
• pp.443-451
• /
• 2017

Mammography is commonly used for screening early breast cancer. However, mammographic images, which depend on the physical properties of breast components, are limited to provide information about whether a lesion is malignant or benign. Although a dual-energy subtraction technique decomposes a certain material from a mixture, it increases radiation dose and degrades the accuracy of material decomposition. In this study, we simulated a breast phantom using attenuation characteristics, and we proposed a technique to enable the accurate material decomposition by applying weighting factors for the dual-energy mammography based on a photon-counting detector using a Monte Carlo simulation tool. We also evaluated the contrast and noise of simulated breast images for validating the proposed technique. As a result, the contrast for a malignant tumor in the dual-energy weighted subtraction technique was 0.98 and 1.06 times similar than those in the general mammography and dual-energy subtraction techniques, respectively. However the contrast between malignant and benign tumors dramatically increased 13.54 times due to the low contrast of a benign tumor. Therefore, the proposed technique can increase the material decomposition accuracy for malignant tumor and improve the diagnostic accuracy of mammography.

• Journal of Korea Spatial Information System Society
• /
• v.12 no.1
• /
• pp.66-75
• /
• 2010

The authors argues that the current road updating system based on traditional aerial photograph or multi-spectral satellite image appears to be non-user friendly due to lack of the frequent cartographic representation for the new construction sites. Google Earth are currently being emerged as one of important places to extract road features since the RGB satellite image with high multi-temporal resolution can be accessed freely over large areas. This paper is primarily intended to evaluate optimal threshold of RGB pixel values to extract road features from Google Earth. An empirical study for five experimental sites was conducted to confirm how a RGB picture provided Google Earth can be used to extact the road feature. The results indicate that optimal threshold of RGB pixel values to extract road features was identified as 126, 125, 127 for manual operation which corresponds to 25%, 30%, 19%. Also, it was found that display scale difference of Google Earth was not very influential in tracking required RGB pixel value. As a result the 61cm resolution of Quickbird RGB data has shown the potential to realistically identified the major type of road feature by large scale spatial precision while the typical algorithm revealed successfully the area-wide optimal threshold of RGB pixel for road appeared in the study area.