• The Transactions of the Korea Information Processing Society
• /
• v.3 no.7
• /
• pp.1763-1772
• /
• 1996

The conventional training set selection methods for the satellite image classification usually depend on the manual selection using data from the direct measurements of the ground or the ground map. However this task takes much time and cost, and some feature values vary in wide ranges even if they are in the same class. Such feature values can increase the robustness of the neural net but learning time becomes longer. In this paper,we propose anew training set selection algorithm using a rule-based method. By the technique proposed, the SPOT multispectral Imagery is classified in 3 bands, and the pixels which satisfy the rule are employed as the training sets for the neutralist classifier. The experimental results show faster initial convergence and almost the same or better classification accuracy. We also showed an improvement of the classification accuracy by using texture features and NDV1.

• Journal of KIISE:Computer Systems and Theory
• /
• v.31 no.7
• /
• pp.414-420
• /
• 2004

Among the desirable properties of a piecewise linear parameterization, guaranteeing a one-to-one mapping (i.e., no triangle flips in the parameter plane) is often sought. A one-to-one mapping is accomplished by non-negative coefficients in the affine transformation. In the Floater's method, the coefficients were computed after the 3D mesh was flattened by geodesic polar-mapping. But using this geodesic polar map introduces unnecessary local distortion. In this paper, a simple variant of the original shape-preserving mapping technique by Floater is introduced. A new simple method for calculating barycentric coordinates by using straightest geodesics is proposed. With this method, the non-negative coefficients are computed directly on the mesh, reducing the shape distortion introduced by the previously-used polar mapping. The parameterization is then found by solving a sparse linear system, and it provides a simple and visually-smooth piecewise linear mapping, without foldovers.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.10 no.10
• /
• pp.1842-1847
• /
• 2006

This paper is the environmental impact assessment of at road design in the light of the sense for the real from the virtual reality. For In this papers, This study developed 3D-model and virtual reality contents by suggesting the environmental impact assessment based on GIS in the road design. Ant through this process, it's possible to visualize the environmental impact assessment by constructing the 3D-model and simulation. The 3D-model can be a method to show the road effectively by maximizing the road's shape visually after the construction. The main construction which composes polyhedron model that is generated from digital map and aerial photo is built by mapping the real texture, so the Sense for the Real was more heightened. Through this study, it must be made to shorten a long time exhausting period of conference and construct more real road after due scene consideration by specific and various low-cost strategy in the environmental impact assessment afterwards.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.14 no.3
• /
• pp.1086-1103
• /
• 2020

The automatic identification and classification of image-based weld defects is a difficult task due to the complex texture of the X-ray images of the weld defect. Several depth learning methods for automatically identifying welds were proposed and tested. In this work, four different depth convolutional neural networks were evaluated and compared on the 1631 image set. The concavity, undercut, bar defects, circular defects, unfused defects and incomplete penetration in the weld image 6 different types of defects are classified. Another contribution of this paper is to train a CNN model "RayNet" for the dataset from scratch. In the experiment part, the parameters of convolution operation are compared and analyzed, in which the experimental part performs a comparative analysis of various parameters in the convolution operation, compares the size of the input image, gives the classification results for each defect, and finally shows the partial feature map during feature extraction with the classification accuracy reaching 96.5%, which is 6.6% higher than the classification accuracy of other existing fine-tuned models, and even improves the classification accuracy compared with the traditional image processing methods, and also proves that the model trained from scratch also has a good performance on small-scale data sets. Our proposed method can assist the evaluators in classifying pipeline welding defects.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.34 no.11B
• /
• pp.1283-1288
• /
• 2009

In general, a 3D computer graphic model is being used to generate a digital hologram as theinput information because the 3D information of an object can be extracted from a 3D model, easily. The 3D information of a real scene can be extracted by using a depth camera. The 3D information, point cloud, corresponding to real scene is extracted from a taken image pair, a gray texture and a depth map, by a depth camera. The extracted point cloud is used to generate a digital hologram as input information. The digital hologram is generated by using the coherent holographic stereogram, which is a fast digital hologram generation algorithm based on segmentation. The generated digital hologram using the taken image pair by a depth camera is reconstructed by the Fresnel approximation. By this method, the digital hologram corresponding to a real scene or a real object could be generated by using the fast digital hologram generation algorithm. Furthermore, experimental results are satisfactory.

• Economic and Environmental Geology
• /
• v.37 no.5
• /
• pp.521-532
• /
• 2004

The aim of this study is to evaluate the susceptibility of landslides at Kangneung area, Korea, using a Geographic Information System (GIS) and remote sensing. Landslide locations were identified from interpretation of satellite image and field surveys. The topographic, soil, forest, geologic, lineament and land cover data were collected, processed and constructed into a spatial database using GIS and remote sensing data. Using frequency ratio model which is one of the probability model, the relationships between landslides and related factors such as slope, aspect, curvature and type of topography, texture, material, drainage and effective thickness of soil, type, age, diameter and density of wood, lithology, distance from lineament and land cover were calculated as frequency ratios. Then, the frequency ratio were summed to calculate a landslide susceptibility indexes and the landslide susceptibility maps were generated using the indexes. The results of the analysis were verified and cross-validated using actual landslide location data. The verification results showed satisfactory agreement between the susceptibility map and the existing data on landslide locations.

• Korean Journal of Food Science and Technology
• /
• v.25 no.3
• /
• pp.270-276
• /
• 1993

Leaf lettuce (Lactuca sativa L., var. crispa) packed in $20{\mu}m$ LDPE (perforated), $20{\mu}m$ LDPE, $30{\mu}m$ HDPE, and $40{\mu}m$ LDPE pouches was stored at $4^{\circ}C,\;10^{\circ}C\;and\;20^{\circ}C$. The quality of leaf lettuce during the storage was investigated in terms of weight loss, color, soluble solids content, texture and sensory properties. Leaf lettuce exhibited the highest storage stability at $4^{\circ}C$. Shelf-life of the packed leaf lettuce was prolonged approximately $10{\sim}15$ times in comparision with that of the unpacked under all temperature conditions. Among the packed leaf lettuce, self-life in $20{\mu}m$ LDPE (perforated) and $20{\mu}m$ LDPE pouches was relatively short mainly due to discoloration and deterioration. On the other hand, the quality of the leaf lettuce packed in $30{\mu}m$ HDPE and $40{\mu}m$ LDPE pouches was kept well during the storage. Particularly, $40{\mu}m$ LDPE pouches conferred the shelf-life superior to other film pouches.

• Proceedings of the Korean Geotechical Society Conference
• /
• 2002.10a
• /
• pp.499-506
• /
• 2002

The purpose of this study is to develop landslide susceptibility analysis techniques using artificial neural networks and to apply the newly developed techniques for assessment of landslide susceptibility to the study area of Yongin in Korea. Landslide locations were identified in the study area from interpretation of aerial Photographs and field survey data, and a spatial database of the topography, soil type and timber cover were constructed. The landslide-related factors such as topographic slope, topographic curvature, soil texture, soil drainage, soil effective thickness, timber age, and timber diameter were extracted from the spatial database. Using those factors, landslide susceptibility and weights of each factor were analyzed by two artificial neural network methods. In the first method, the landslide susceptibility index was calculated by the back propagation method, which is a type of artificial neural network method. Then, the susceptibility map was made with a GIS program. The results of the landslide susceptibility analysis were verified using landslide location data. The verification results show satisfactory agreement between the susceptibility index and existing landslide location data. In the second method, weights of each factor were determinated. The weights, relative importance of each factor, were calculated using importance-free characteristics method of artificial neural networks.

• The Research Journal of the Costume Culture
• /
• v.11 no.6
• /
• pp.835-851
• /
• 2003

This study analyses patent information of apparel design using computer technology and researches the trend of patent application focused on International Patent Classification. In terms of trend by filling data, Patent application started first in 1974 and increased sharply in 1993 with 14 cases and increased to 25 cases in 2000. In case of Korea, they began somewhat late in 1996, but reached a similar level with the leading country in 2000. In terms of trend by applicant, Gerber Garment Technology, Inc. filed 7 cases TORAY IND INC, filed 6 cases Levi Strauss & Co. filed 4 cases, NEC HOME ELECTRONICS LTD filed 3 cases, TOYOBO CO LTD filed 3 cases. Japanese companies occupied 52% and United States's companies occupied 48%. In terms of trend by country, foreigner occupied 47% of the patents filed by United State. Japanese take up 10% of total patent of United States. Korean occupied 84% of total patent of Korea and foreigner, american occupied 16% of the patents filed by Korea. In regared to International Patent Classification, in the section level G filed 92 cases(53%). In class level, G06 marked the first place in United States, Japan, and Korea. In subclass level, G06F marksed the first place with 74 cases. G06T and A61B were regarded as the new technologies. The new technologies are representing the dimensions of garment or computer-rendered model, providing the virtual reality through the texture mapping, digital dressing room or virtual dressing, and performing or retriving display on a screen for the result of changing pattern ao dress design, The technologies of core patent are designing or producing custom manufactured item, providing or prealtering the data for pattern making and visually displaying, interactively generating or previewing of various articles.

• Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography
• /
• v.22 no.3
• /
• pp.303-309
• /
• 2004

In this paper, an adaptive Bayesian approach to image segmentation was developed for boundary detection. Both image intensities and texture information were used for obtaining better quality of the image segmentation by using the C programming language. Fuzzy c-mean clustering was applied fer the conditional probability density function, and Gibbs random field model was used for the prior probability density function. To simply test the algorithm, a synthetic image (256$\times$256) with a set of low gray values (50, 100, 150 and 200) was created and normalized between 0 and 1 n double precision. Results have been presented that demonstrate the effectiveness of the algorithm in segmenting the synthetic image, resulting in more than 99% accuracy when noise characteristics are correctly modeled. The algorithm was applied to the Antarctic mosaic that was generated using 1963 Declassified Intelligence Satellite Photographs. The accuracy of the resulting vector map was estimated about 300-m.