• Korean Journal of Remote Sensing
• /
• v.35 no.6_1
• /
• pp.999-1009
• /
• 2019

The importance of high-resolution sea ice maps of the Arctic Ocean is increasing due to the possibility of pioneering North Pole Routes and the necessity of precise climate prediction models. In this study,sea ice classification algorithms for two deep learning models were examined using Sentinel-1 A/B SAR data to generate high-resolution sea ice classification maps. Based on current ice charts, three classes (Open Water, First Year Ice, Multi Year Ice) of training data sets were generated by Arctic sea ice and remote sensing experts. Ten sea ice classification algorithms were generated by combing two deep learning models (i.e. Simple CNN and Resnet50) and five cases of input bands including incident angles and thermal noise corrected HV bands. For the ten algorithms, analyses were performed by comparing classification results with ground truth points. A confusion matrix and Cohen's kappa coefficient were produced for the case that showed best result. Furthermore, the classification result with the Maximum Likelihood Classifier that has been traditionally employed to classify sea ice. In conclusion, the Convolutional Neural Network case, which has two convolution layers and two max pooling layers, with HV and incident angle input bands shows classification accuracy of 96.66%, and Cohen's kappa coefficient of 0.9499. All deep learning cases shows better classification accuracy than the classification result of the Maximum Likelihood Classifier.

• Journal of Korean Society for Geospatial Information Science
• /
• v.13 no.3 s.33
• /
• pp.33-40
• /
• 2005

The aim of remotely sensed data classification is to produce the best accuracy map of the earth surface assigning each pixel to its appropriate category of the real-world. The classification of satellite multi-spectral image data has become tool for generating ground cover map. Many classification methods exist. In this study, MLC(Maximum Likelihood Classification), ANN(Artificial neural network), SVM(Support Vector Machine), Naive Bayes classifier algorithms are compared using IKONOS image of the part of Dalsung Gun, Daegu area. Two preprocessing methods are performed-PCA(Principal component analysis), ICA(Independent Component Analysis). Boosting algorithms also performed. By the combination of appropriate feature selection pre-processing and classifier, the best results were obtained.

• Korean Journal of Remote Sensing
• /
• v.28 no.6
• /
• pp.611-622
• /
• 2012

This study proposed two multisensor fusion methods for segment-based image classification utilizing a region-growing segmentation. The proposed algorithms employ a Gaussian-PDF measure and an evidential measure respectively. In remote sensing application, segment-based approaches are used to extract more explicit information on spatial structure compared to pixel-based methods. Data from a single sensor may be insufficient to provide accurate description of a ground scene in image classification. Due to the redundant and complementary nature of multisensor data, a combination of information from multiple sensors can make reduce classification error rate. The Gaussian-PDF method defines a regional measure as the PDF average of pixels belonging to the region, and assigns a region into a class associated with the maximum of regional measure. The evidential fusion method uses two measures of plausibility and belief, which are derived from a mass function of the Beta distribution for the basic probability assignment of every hypothesis about region classes. The proposed methods were applied to the SPOT XS and ENVISAT data, which were acquired over Iksan area of of Korean peninsula. The experiment results showed that the segment-based method of evidential measure is greatly effective on improving the classification via multisensor fusion.

• Korean Journal of Remote Sensing
• /
• v.25 no.6
• /
• pp.465-474
• /
• 2009

Convective/stratiform radar echo classification schemes by Steiner et al. (1995) and Biggerstaff and Listemaa (2000) are examined on a monsoonal front during the summer monsoon-Changma period, which is organized as a cloud cluster with mesoscale convective complex. Target radar is S-band with wavelength of 10cm, spatial resolution of 1km, elevation angle interval of 0.5-1.0 degree, and minimum elevation angle of 0.19 degree at Jindo over the Korean Peninsula. For verification of rainfall amount retrieved from the echo classification, ground-based rain gauge observations (Automatic Weather Stations) are examined, converting the radar echo grid data to the station values using the inverse distance weighted method. Improvement from the echo classification is evaluated based on the correlation coefficient and the scattered diagram. Additionally, an optimal use method was designed to produce combined rainfalls from the radar echo and Tropical Rainfall Measuring Mission Precipitation Radar (TRMM/PR) data. Optimal values for the radar rain and TRMM/PR rain are inversely weighted according to the error variance statistics for each single station. It is noted how the rainfall distribution during the summer monsoon frontal system is improved from the classification of convective/stratiform echo and the use of the optimal use technique.

• Proceedings of the Korean Society of Computer Information Conference
• /
• 2021.07a
• /
• pp.313-315
• /
• 2021

Timely and accurate diagnosis of lung diseases using Chest X-ray images has been gained much attention from the computer vision and medical imaging communities. Although previous studies have presented the capability of deep convolutional neural networks by achieving competitive binary classification results, their models were seemingly unreliable to effectively distinguish multiple disease groups using a large number of x-ray images. In this paper, we aim to build an advanced approach, so-called Ensemble Knowledge Distillation (EKD), to significantly boost the classification accuracies, compared to traditional KD methods by distilling knowledge from a cumbersome teacher model into an ensemble of lightweight student models with parallel branches trained with ground truth labels. Therefore, learning features at different branches of the student models could enable the network to learn diverse patterns and improve the qualify of final predictions through an ensemble learning solution. Although we observed that experiments on the well-established ChestX-ray14 dataset showed the classification improvements of traditional KD compared to the base transfer learning approach, the EKD performance would be expected to potentially enhance classification accuracy and model generalization, especially in situations of the imbalanced dataset and the interdependency of 14 weakly annotated thorax diseases.

• Geomechanics and Engineering
• /
• v.37 no.6
• /
• pp.539-554
• /
• 2024

This study explores development of prediction model for seismic site classification through the integration of machine learning techniques with horizontal-to-vertical spectral ratio (HVSR) methodologies. To improve model accuracy, the research employs outlier detection methods and, synthetic minority over-sampling technique (SMOTE) for data balance, and evaluates using seven machine learning models using seismic data from KiK-net. Notably, light gradient boosting method (LGBM), gradient boosting, and decision tree models exhibit improved performance when coupled with SMOTE, while Multiple linear regression (MLR) and Support vector machine (SVM) models show reduced efficacy. Outlier detection techniques significantly enhance accuracy, particularly for LGBM, gradient boosting, and voting boosting. The ensemble of LGBM with the isolation forest and SMOTE achieves the highest accuracy of 0.91, with LGBM and local outlier factor yielding the highest F1-score of 0.79. Consistently outperforming other models, LGBM proves most efficient for seismic site classification when supported by appropriate preprocessing procedures. These findings show the significance of outlier detection and data balancing for precise seismic soil classification prediction, offering insights and highlighting the potential of machine learning in optimizing site classification accuracy.

• Explosives and Blasting
• /
• v.28 no.1
• /
• pp.63-70
• /
• 2010

This study is investigated the extent of the noise and ground vibration in an adjacent zone of a cattle pen and an antiquated housing structures for judgement of the spot applicability on the extents of the noise and ground vibration of the rock-splitting method by an oil pressure. It is studied by measuring and analysing in an adjacent position the extents of the noise and ground vibration according to the work process of the rock-splitting method, such as drilling, rock-splitting, arranging rock, loading and by being compared with the permitted level on the noise and ground vibration fixed at the spot. To the results, it is identified that the influence to the noise has to be considered, even if the rock-splitting method is applied as an excavation method to lower a ground vibration by the classification on blasting method of the ministry of land, transport and marine affairs.

• Proceedings of the Korean Geotechical Society Conference
• /
• 2009.09a
• /
• pp.660-670
• /
• 2009

Neighboring construction becomes mainstream of Ground excavation in downtown area. This causes the displacement, deformation, stress condition, etc of the ground surroundings. Therefore Neighboring construction have an effect on Neighboring structure. All these years a lot of Neighboring construction carried out, and the accumulation of technology also get accomplished. But earth retaining structure collapse happens yet. Types of earth retaining structure collapse are 12. 1. Failure of anchor or strut system, 2. Insufficiency of penetration, 3. H-pile Failure on excessive bending moment, 4. Slope sliding failure, 5. Excessive settlement of the back, 6. Deflection of H-pile, 7. Joint failure of coupled H-pile, 8. Rock failure when H-pile penetration is rock mass, 9. Plane arrangement of support systems are mechanically weak, 10. Boiling, 11. Heaving, 12. Over excavation. But field collapses are difficult for classification according to the type, because collapse process are complex with various types. When we consider the 12 collapse field, insufficient recognition of ground condition is 4 case. Thorough construction management prevents from fault construction. For limitations of soil survey, It is difficult to estimate ground condition exactly. Therefore, it should estimate the safety of earth retaining system, plan for necessary reinforcement, according to measurement and observation continuously.

• Journal of the Korea Computer Graphics Society
• /
• v.24 no.5
• /
• pp.31-39
• /
• 2018

In this paper, we propose an automated method for the ground-glass nodule(GGN) classification using GGN-Net based on intensity, texture, and shape-enhanced images in chest CT images. First, we propose the utilization of image that enhances the intensity, texture, and shape information so that the input image includes the presence and size information of the solid component in GGN. Second, we propose GGN-Net which integrates and trains feature maps obtained from various input images through multiple convolution modules on the internal network. To evaluate the classification accuracy of the proposed method, we used 90 pure GGNs, 38 part-solid GGNs less than 5mm with solid component, and 23 part-solid GGNs larger than 5mm with solid component. To evaluate the effect of input image, various input image set is composed and classification results were compared. The results showed that the proposed method using the composition of intensity, texture and shape-enhanced images showed the best result with 82.75% accuracy.

• 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.