• Korean Journal of Radiology
• /
• v.23 no.2
• /
• pp.226-236
• /
• 2022

Objective: This study aimed to explore the myelin volume change in patients with mild traumatic brain injury (mTBI) with post-concussion syndrome (PCS) using a multidynamic multiecho (MDME) sequence and automatic whole-brain segmentation. Materials and Methods: Forty-one consecutive mTBI patients with PCS and 29 controls, who had undergone MRI including the MDME sequence between October 2016 and April 2018, were included. Myelin volume fraction (MVF) maps were derived from the MDME sequence. After three dimensional T1-based brain segmentation, the average MVF was analyzed at the bilateral cerebral white matter (WM), bilateral cerebral gray matter (GM), corpus callosum, and brainstem. The Mann-Whitney U-test was performed to compare MVF and myelin volume between patients with mTBI and controls. Myelin volume was correlated with neuropsychological test scores using the Spearman rank correlation test. Results: The average MVF at the bilateral cerebral WM was lower in mTBI patients with PCS (median [interquartile range], 25.2% [22.6%-26.4%]) than that in controls (26.8% [25.6%-27.8%]) (p = 0.004). The region-of-interest myelin volume was lower in mTBI patients with PCS than that in controls at the corpus callosum (1.87 cm³ [1.70-2.05 cm³] vs. 2.21 cm³ [1.86-3.46 cm³]; p = 0.003) and brainstem (9.98 cm³ [9.45-11.00 cm³] vs. 11.05 cm³ [10.10-11.53 cm³]; p = 0.015). The total myelin volume was lower in mTBI patients with PCS than that in controls at the corpus callosum (0.45 cm³ [0.39-0.48 cm³] vs. 0.48 cm³ [0.45-0.54 cm³]; p = 0.004) and brainstem (1.45 cm³ [1.28-1.59 cm³] vs. 1.54 cm³ [1.42-1.67 cm³]; p = 0.042). No significant correlation was observed between myelin volume parameters and neuropsychological test scores, except for the total myelin volume at the bilateral cerebral WM and verbal learning test (delayed recall) (r = 0.425; p = 0.048). Conclusion: MVF quantified from the MDME sequence was decreased at the bilateral cerebral WM in mTBI patients with PCS. The total myelin volumes at the corpus callosum and brainstem were decreased in mTBI patients with PCS due to atrophic changes.

• The KIPS Transactions:PartB
• /
• v.17B no.5
• /
• pp.363-374
• /
• 2010

In this paper, we propose the method for extracting Homogeneity Threshold($H_T$) and for segmenting homogeneous regions by USRG(Unseeded Region Growing) with $H_T$. The $H_T$ is a criterion to distinguish homogeneity in neighbor pixels and is computed automatically from the original image by proposed method. Theoretical background for proposed method is based on the Otsu's single level threshold method. The method is used to divide a small local part of original image int o two classes and the sum($\sigma_c$) of standard deviations for the classes to satisfy special conditions for distinguishing as different regions from each other is used to compute $H_T$. To find validity for proposed method, we compare the original image with the image that is regenerated with only the segmented homogeneous regions and show up the fact that the difference between two images is not exist visually and also present the steps to regenerate the image in order the size of segmented homogeneous regions and in order the intensity that includes pixels. Also, we show up the validity of proposed method with various results that is segmented using the homogeneity thresholds($H^*_T$) that is added a coefficient ${\alpha}$ for adjusting scope of $H_T$. We expect that the proposed method can be applied in various fields such as visualization and animation of natural image, anatomy and biology and so on.

• The Journal of the Acoustical Society of Korea
• /
• v.20 no.3
• /
• pp.24-33
• /
• 2001

this paper describes a baseline for an implementation of a corpus-based Korean TTS system. The conventional TTS systems using small-sized speech still generate machine-like synthetic speech. To overcome this problem we introduce the corpus-based TTS system which enables to generate natural synthetic speech without prosodic modifications. The corpus should be composed of a natural prosody of source speech and multiple instances of synthesis units. To make a phone level synthesis unit, we train a speech recognizer with the target speech, and then perform an automatic phoneme segmentation. We also detect the fine pitch period using Laryngo graph signals, which is used for prosodic feature extraction. For break strength allocation, 4 levels of break indices are decided as pause length and also attached to phones to reflect prosodic variations in phrase boundaries. To predict the break strength on texts, we utilize the statistical information of POS (Part-of-Speech) sequences. The best triphone sequences are selected by Viterbi search considering the minimization of accumulative Euclidean distance of concatenating distortion. To get high quality synthesis speech applicable to commercial purpose, we introduce a domain specific database. By adding domain specific database to general domain database, we can greatly improve the quality of synthetic speech on specific domain. From the subjective evaluation, the new Korean corpus-based TTS system shows better naturalness than the conventional demisyllable-based one.

• Journal of Korean Society for Geospatial Information Science
• /
• v.21 no.2
• /
• pp.93-98
• /
• 2013

Multi-temporal high resolution satellite images are essential data for efficient city analysis and monitoring. Yet even when acquired from the same location, identical sensors as well as different sensors, these multi-temporal images have a geometric inconsistency. Matching points between images, therefore, must be extracted to match the images. With images of an urban area, however, it is difficult to extract matching points accurately because buildings, trees, bridges, and other artificial objects cause shadows over a wide area, which have different intensities and directions in multi-temporal images. In this study, we analyze a shadow effect on image matching of high resolution satellite images in urban area using Scale-Invariant Feature Transform(SIFT), the representative matching points extraction method, and automatic shadow extraction method. The shadow segments are extracted using spatial and spectral attributes derived from the image segmentation. Also, we consider information of shadow adjacency with the building edge buffer. SIFT matching points extracted from shadow segments are eliminated from matching point pairs and then image matching is performed. Finally, we evaluate the quality of matching points and image matching results, visually and quantitatively, for the analysis of shadow effect on image matching of high resolution satellite image.

• Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography
• /
• v.27 no.5
• /
• pp.619-628
• /
• 2009

LiDAR has capability of rapid data acquisition and provides useful information for reconstructing surface of the Earth. However, Extracting information from LiDAR data is not easy task because LiDAR data consist of irregularly distributed point clouds of 3D coordinates and lack of semantic and visual information. This thesis proposed methods for automatic extraction of buildings and 3D detail modeling using airborne LiDAR data. As for preprocessing, noise and unnecessary data were removed by iterative surface fitting and then classification of ground and non-ground data was performed by analyzing histogram. Footprints of the buildings were extracted by tracing points on the building boundaries. The refined footprints were obtained by regularization based on the building hypothesis. The accuracy of building footprints were evaluated by comparing with 1:1,000 digital vector maps. The horizontal RMSE was 0.56m for test areas. Finally, a method of 3D modeling of roof superstructure was developed. Statistical and geometric information of the LiDAR data on building roof were analyzed to segment data and to determine roof shape. The superstructures on the roof were modeled by 3D analytical functions that were derived by least square method. The accuracy of the 3D modeling was estimated using simulation data. The RMSEs were 0.91m, 1.43m, 1.85m and 1.97m for flat, sloped, arch and dome shapes, respectively. The methods developed in study show that the automation of 3D building modeling process was effectively performed.

• Spatial Information Research
• /
• v.21 no.6
• /
• pp.91-98
• /
• 2013

Government introduces a Total Maximum Daily Loads(TMDL) which can be implemented for total pollutant amounts in 2004. Normally, the local governments have been calculated the amounts of pollutant discharge of each watershed using a water quality model. However, among the input data to use the water quality model, creating a schematic diagram of the stream or the modeling usually requires considerable amount of time and efforts due to the manual work. Therefore, this study tried to develop an algorithm which automates the creation of a schematic diagram for water quality modeling using the Korean Reach File capable of river network analysis. Further, this study creates a schematic diagram with the shape of a stream utilizing GIS capabilities. The diagram can be easily analyzed with overlapping various spatial information such as pollution sources and discharge points. This study mainly has automated element segmentation algorithm to divide streamflows into equal distance using line graphic data of Koran Reach File. Also, automated attribute input algorithm has also been developed to enable to insert element order and type into elements using point graphic data of Korean Reach File. For the verification of the developed algorithm, the algorithm was applied to kyungan stream basin to see the acceptable results. To conclude, it was possible to automate generating of schematic diagram of water quality model and it is expected to be able to save time and cost required for the water modeling. In future study, it is necessary to develop an automatic creation system of various types of input data for water quality modeling and this will lead to relatively easier and simple water quality modeling.

• Journal of Korean Tunnelling and Underground Space Association
• /
• v.24 no.6
• /
• pp.583-598
• /
• 2022

As human-based tunnel inspections are affected by the subjective judgment of the inspector, making continuous history management difficult. There is a lot of deep learning-based automatic crack detection research recently. However, the large public crack datasets used in most studies differ significantly from those in tunnels. Also, additional work is required to build sophisticated crack labels in current tunnel evaluation. Therefore, we present a method to improve crack detection performance by inputting existing datasets into a deep learning model. We evaluate and compare the performance of deep learning models trained by combining existing tunnel datasets, high-quality tunnel datasets, and public crack datasets. As a result, DeepLabv3+ with Cross-Entropy loss function performed best when trained on both public datasets, patchwise classification, and oversampled tunnel datasets. In the future, we expect to contribute to establishing a plan to efficiently utilize the tunnel image acquisition system's data for deep learning model learning.

• Korean Journal of Remote Sensing
• /
• v.38 no.5_1
• /
• pp.597-608
• /
• 2022

Remotely sensed data have been used in various fields, such as disasters, agriculture, urban planning, and the military. Recently, the demand for the multitemporal dataset with the high-spatial-resolution has increased. This manuscript proposed an automatic image matching algorithm using a deep learning technique to utilize a multitemporal remotely sensed dataset. The proposed deep learning model was based on High Resolution Net (HRNet), widely used in image segmentation. In this manuscript, denseblock was added to calculate the correlation map between images effectively and to increase learning efficiency. The training of the proposed model was performed using the multitemporal orthophotos of the National Geographic Information Institute (NGII). In order to evaluate the performance of image matching using a deep learning model, a comparative evaluation was performed. As a result of the experiment, the average horizontal error of the proposed algorithm based on 80% of the image matching rate was 3 pixels. At the same time, that of the Zero Normalized Cross-Correlation (ZNCC) was 25 pixels. In particular, it was confirmed that the proposed method is effective even in mountainous and farmland areas where the image changes according to vegetation growth. Therefore, it is expected that the proposed deep learning algorithm can perform relative image registration and image matching of a multitemporal remote sensed dataset.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.23 no.3
• /
• pp.199-206
• /
• 2023

This study utilizes deep learning algorithms to automatically classify construction waste into three categories: wood waste, plastic waste, and concrete waste. Two models, VGG-16 and ViT (Vision Transformer), which are convolutional neural network image classification algorithms and NLP-based models that sequence images, respectively, were compared for their performance in classifying construction waste. Image data for construction waste was collected by crawling images from search engines worldwide, and 3,000 images, with 1,000 images for each category, were obtained by excluding images that were difficult to distinguish with the naked eye or that were duplicated and would interfere with the experiment. In addition, to improve the accuracy of the models, data augmentation was performed during training with a total of 30,000 images. Despite the unstructured nature of the collected image data, the experimental results showed that VGG-16 achieved an accuracy of 91.5%, and ViT achieved an accuracy of 92.7%. This seems to suggest the possibility of practical application in actual construction waste data management work. If object detection techniques or semantic segmentation techniques are utilized based on this study, more precise classification will be possible even within a single image, resulting in more accurate waste classification

• Korean Journal of Remote Sensing
• /
• v.39 no.2
• /
• pp.183-192
• /
• 2023

The remotely sensed data, such as satellite imagery and aerial photos, can be used to extract and detect some objects in the image through image interpretation and processing techniques. Significantly, the possibility for utilizing digital map updating and land monitoring has been increased through automatic object detection since spatial resolution of remotely sensed data has improved and technologies about deep learning have been developed. In this paper, we tried to extract plastic greenhouses into aerial orthophotos by using fully convolutional densely connected convolutional network (FC-DenseNet), one of the representative deep learning models for semantic segmentation. Then, a quantitative analysis of extraction results had performed. Using the farm map of the Ministry of Agriculture, Food and Rural Affairsin Korea, training data was generated by labeling plastic greenhouses into Damyang and Miryang areas. And then, FC-DenseNet was trained through a training dataset. To apply the deep learning model in the remotely sensed imagery, instance norm, which can maintain the spectral characteristics of bands, was used as normalization. In addition, optimal weights for each band were determined by adding attention modules in the deep learning model. In the experiments, it was found that a deep learning model can extract plastic greenhouses. These results can be applied to digital map updating of Farm-map and landcover maps.