• Korean Journal of Remote Sensing
• /
• v.21 no.3
• /
• pp.229-242
• /
• 2005

Photogrammetry is a current method of GIS data acquisition. However, as a matter of fact, a large manpower and expenditure for making detailed 3D spatial information is required especially in urban areas where various buildings exist. There are no photogrammetric systems which can automate a process of spatial information acquisition completely. On the other hand, LiDAR has high potential of automating 3D spatial data acquisition because it can directly measure 3D coordinates of objects, but it is rather difficult to recognize the object with only LiDAR data, for its low resolution at this moment. With this background, we believe that it is very advantageous to integrate LiDAR data and stereo CCD images for more efficient and automated acquisition of the 3D spatial data with higher resolution. In this research, the automatic urban object recognition methodology was proposed by integrating ultra highresolution stereo images and LiDAR data. Moreover, a method to enable more reliable and detailed stereo matching method for CCD images was examined by using LiDAR data as an initial 3D data to determine the search range and to detect possibility of occlusions. Finally, intellectual DSMs, which were identified urban features with high resolution, were generated with high speed processing.

• Investigative Magnetic Resonance Imaging
• /
• v.23 no.2
• /
• pp.81-99
• /
• 2019

Recently, deep learning methods have shown great potential in various tasks that involve handling large amounts of digital data. In the field of MR imaging research, deep learning methods are also rapidly being applied in a wide range of areas to complement or replace traditional model-based methods. Deep learning methods have shown remarkable improvements in several MR image processing areas such as image reconstruction, image quality improvement, parameter mapping, image contrast conversion, and image segmentation. With the current rapid development of deep learning technologies, the importance of the role of deep learning in MR imaging research appears to be growing. In this article, we introduce the basic concepts of deep learning and review recent studies on various MR image processing applications.

• Smart Structures and Systems
• /
• v.29 no.1
• /
• pp.153-166
• /
• 2022

Identifying fine cracks in steel bridge facilities is a challenging task of structural health monitoring (SHM). This study proposed an end-to-end crack image segmentation framework based on a one-step Convolutional Neural Network (CNN) for pixel-level object recognition with high accuracy. To particularly address the challenges arising from small object detection in complex background, efforts were made in loss function selection aiming at sample imbalance and module modification in order to improve the generalization ability on complicated images. Specifically, loss functions were compared among alternatives including the Binary Cross Entropy (BCE), Focal, Tversky and Dice loss, with the last three specialized for biased sample distribution. Structural modifications with dilated convolution, Spatial Pyramid Pooling (SPP) and Feature Pyramid Network (FPN) were also performed to form a new backbone termed CrackDet. Models of various loss functions and feature extraction modules were trained on crack images and tested on full-scale images collected on steel box girders. The CNN model incorporated the classic U-Net as its backbone, and Dice loss as its loss function achieved the highest mean Intersection-over-Union (mIoU) of 0.7571 on full-scale pictures. In contrast, the best performance on cropped crack images was achieved by integrating CrackDet with Dice loss at a mIoU of 0.7670.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2022.05a
• /
• pp.535-538
• /
• 2022

근감소증은 영양부족, 운동량 감소 그리고 노화 등으로 정상적인 근육의 양과 근력 및 근 기능이 감소하는 질환을 말한다. 근감소증은 보편적으로 유럽 근감소증 실무그룹분석(EWGSOP)에서 정의한 측정 방법을 따른다. 본 논문에서는 근감소증 진단을 위한 영상 분할 모델을 개발하고 외부검증하는 방법에 대해서 제안한다. 우리는 CT 영상에서 L3 영역을 선별하여 자동으로 근육, 피하지방, 내장지방을 분할할 수 있는 인공지능 모델을 U-Net을 사용하여 개발하였다. 또한 모델의 성능을 평가하기 위해서 분할영역의 IOU(Intersection over Union)를 계산하여 내부검증을 진행하였으며, 타 병원의 데이터를 이용하여 같은 방법으로 외부검증을 진행한 결과를 보인다. 검증 결과를 토대로 문제점과 해결방안에 대해서 고찰하고 보완하고자 했다.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2020.11a
• /
• pp.113-117
• /
• 2020

Images were taken under various weather such as rain, haze, snow often show low visibility, which can dramatically decrease accuracy of some tasks in computer vision: object detection, segmentation. Besides, previous work to enhance image usually downsample the image to receive consistency features but have not yet good upsample algorithm to recover original size. So, in this research, we jointly implement removal streak in heavy rain image and super resolution using a deep network. We put forth a 2-stage network: a multi-model network followed by a refinement network. The first stage using rain formula in the single image and two operation layers (addition, multiplication) removes rain streak and noise to get clean image in low resolution. The second stage uses refinement network to recover damaged background information as well as upsample, and receive high resolution image. Our method improves visual quality image, gains accuracy in human action recognition task in datasets. Extensive experiments show that our network outperforms the state of the art (SoTA) methods.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2021.11a
• /
• pp.577-579
• /
• 2021

의료영상기반의 인공지능 연구는 질환의 조기진단 및 예측 분야에 눈부신 기술발전이 되어왔다. 근감소증 질환은 다양한 기저질환을 기반으로 발생하며, 특히 60대 이상은 30%의 유병율을 갖는다. 해당 질환은 임상적인 진단 방법의 발달과 임상 결과가 알려지면서 관심이 증가하고 있다. 최근 근감소증 진단방법 중의 하나로 CT 또는 MR 의료영상을 통한 진단방법이 제시되었다. 본 논문에서는 인공지능을 기반으로 하여, 근감소증을 진단하기 위해 척추부위 중 Lumbar 3 영역의 근육, 지방 영역의 영상분할 모델을 제시하고자 한다. 이를 위해 인공지능 영상분할 모델을 개발하는 과정과 그 근육과 지방의 영상분할 결과를 보인다. 본 논문에서 제시한 영상분할모델을 통해 근감소증을 빠르게 진단할 수 있을 것으로 기대한다.

• Proceedings of the Korean Institute of Building Construction Conference
• /
• 2021.11a
• /
• pp.219-220
• /
• 2021

It is considered that the construction industry is one of the pivotal players in the national economy in terms of Gross Domestic Production (GDP) and employment. Behind the positive role of this industrial sector to the national economy, the construction industry generates approximately 50 % of the total waste generation from all the industrial sectors. There are several measures to mitigate the adverse impacts of the construction waste such as reduce, reuse and recycle. Recycling would be one of the effective strategies for waste minimisation, which would be able to reduce the demand upon new resources as well as enhance reusing the construction materials on sites. The automated construction waste classification system would make it possible not only to reduce the amount of labour input but also mitigate the possibility of errors during the manual classification process. In this study, we proposed an automated waste segmentation and classification system for recycling the construction and demolition waste in the real construction site context. Since the practical application to the real-world construction sites was one of the significant factors to develop the system, a YOLACT (You Only Look At CoefficienTs) algorithm was chosen to conduct the study. In this study, it is expected that the proposed system would make it possible to enhance the productivity as well as the cost efficiency by reducing the manpower for the construction and demolition waste management at the construction site.

• Investigative Magnetic Resonance Imaging
• /
• v.3 no.2
• /
• pp.179-187
• /
• 1999

Purpose : This paper introduces a computationally inexpensive context-dependent classification of multi-echo MRI with Bayes compound decision model. In order to produce accurate region segmentation especially in homogeneous area and along boundaries of the regions, we propose a classification method that uses contextual information of local enighborhood system in the image. Material and Methods : The performance of the context free classifier over a statistically heterogeneous image can be improved if the local stationary regions in the image are disassociated from each other through the mechanism of the interaction parameters defined at he local neighborhood level. In order to improve the classification accuracy, we use the contextual information which resolves ambiguities in the class assignment of a pattern based on the labels of the neighboring patterns in classifying the image. Since the data immediately surrounding a given pixel is intimately associated with this given pixel., then if the true nature of the surrounding pixel is known this can be used to extract the true nature of the given pixel. The proposed context-dependent compound decision model uses the compound Bayes decision rule with the contextual information. As for the contextual information in the model, the directional transition probabilities estimated from the local neighborhood system are used for the interaction parameters. Results : The context-dependent classification paradigm with compound Bayesian model for multi-echo MR images is developed. Compared to context free classification which does not consider contextual information, context-dependent classifier show improved classification results especially in homogeneous and along boundaries of regions since contextual information is used during the classification. Conclusion : We introduce a new paradigm to classify multi-echo MRI using clustering analysis and Bayesian compound decision model to improve the classification results.

• Journal of Digital Contents Society
• /
• v.7 no.4
• /
• pp.287-294
• /
• 2006

Mass-customization is fast growing a segment of the apparel market. 2D Virtual wearing system is one of visual support tools that make possible to sell apparel before producing and reduce the time and costs related to product development and manufacturing in the world of apparel mass-customization. This paper is related to fabric color mapping method for 2D image-based virtual wearing system. In proposed method, clothing shape section of interest is segmented from a clothes model image using a region growing method, and then mapping a new fabric color selected by user into it based on its intensity difference map is processed. With the proposed method in 2D virtual wearing system, regardless of color or intensity of model clothes, it is possible to virtually change the fabric color with holding the illumination and shading properties of the selected clothing shape section, and also to quickly and easily simulate, compare, and select multiple fabric color combinations for individual styles or entire outfits.

• The Journal of the Korea institute of electronic communication sciences
• /
• v.8 no.6
• /
• pp.805-814
• /
• 2013

This paper proposes a robust hand segmentation method using view-invariant characteristic of a wrist-mounted camera, and deals with a hand shape recognition system based on segmented hand information. We actively utilize the advantage of the proposed camera device that provides view-invariant images physically, and segment hand region using a Bayesian rule based on adaptive histograms. We construct HSV histograms from RGB histograms, and update HSV histograms using hand region information from a current image. We also propose a user adaptation method by which hand models gradually approach user-dependent models from user-independent models as the user uses the system. The proposed method was evaluated using 16 Korean manual alphabet, and we obtained increases of 27.91% in recognition success rate.