• Proceedings of the Korean Information Science Society Conference
• /
• 2004.04a
• /
• pp.913-915
• /
• 2004

시술 중 제공되는 2D영상은 실시간으로 환자와 시술도구의 상태정보를 제공해주지만 환부의 입체적ㆍ해부학적 파악이 어렵다. 따라서 긴 촬영시간으로 시술 전 획득되는 3D영상과 시술 중 얻어지는 2D영상간 정합영상은 영상 유도술에 있어서 유용한 정보를 제공한다. 이를 위해 본 논문에서는 볼륨영상으로부터 혈관모델을 추출하고 이를 평면으로 투영하였다. 두 2D영상에서 정차대상이 되는 혈관골격을 추출한 후 혈관의 분기특성을 고려 한 초기정합을 수행하였다. 크기와 초기 위치를 맞춘 혈관골격을 골격간 거리가 최소가 되도록 반복적으로 혈관을 기하변환시키고 최종 변환된 혈관골격을 시술 중 제공되는 2D영상에 겹쳐 가시화 하였다. 이로써 시술시간 경감과 시술성공률 향상을 유도할 수 있는 시술경로맵을 제시하고자 하였다.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2003.05a
• /
• pp.591-594
• /
• 2003

본 논문에서는 hopfield 신경망을 사용한 다양한 부분적인 패턴 복원에 관하여 연구하였다. 여섯 개의 $32{\times}32$ 비트맵 훈련패턴들은 한글자음 ㄱ, ㅁ, ㅂ, ㅇ, ㅊ, ㅍ, 그리고 남자와 여자 이미지로 구성되어 있다. 그리고 부분패턴들의 크기, 범위, 방향의 효과를 알아보기 위해서 훈련패턴에서 여덟 가지 형태의 테스트 패턴을 만든다. 한글 자음의 경우 유사 패턴이 많기 때문에 완전히 복원되지 못하였으나, 400회 정도 수렵된 후에는 테스트패턴들이 견본패턴과 비슷한 모양으로 복원되었다. 이 유사도를 측정하기 위해 해밍거리 (Hamming distance)를 이용하였다. 유사도를 측정하여 해밍거리가 가장 적은 것으로 본래의 이미지들 복원하였다.

• Proceedings of the Korean Information Science Society Conference
• /
• 2005.07b
• /
• pp.46-48
• /
• 2005

기존의 다차원 색인 기법들이 고차원의 특징 벡터를 갖는 멀티미디어 컨텐츠 검색 분야에서 만족할 만한 성능을 보이지 못하므로, 이를 해결하기 위해 VA-File, LPC-File 등의 벡터 근사 방법이 개발 되었다. 이러한 방법들은 데이터의 접근에 소요되는 시간이 전체 검색시간의 대부분을 차지하는 경우에 효과적으로 사용할 수 있다. 그러나 고차원의 멀티미디어 데이터 검색에서 객체간의 거리 계산 시간은 데이터 접근 시간에 비해 무시할 만큼 작지 않으므로 이 방법들을 그대로 적용하기는 어렵다. 본 논문에서는 객체간의 거리 계산 시간을 줄이기 위한 새로운 색인 기법을 제안하고 실험을 통해 이 방법이 기존의 방법들에 비해 우수한 검색 성능을 가진다는 것을 보인다.

• Journal of Broadcast Engineering
• /
• v.26 no.3
• /
• pp.283-295
• /
• 2021

This paper proposes a method for generating dense depth map using information of color images and depth map generated based on lidar based on self-organizing map. The proposed depth map upsampling method consists of an initial depth prediction step for an area that has not been acquired from LiDAR and an initial depth filtering step. In the initial depth prediction step, stereo matching is performed on two color images to predict an initial depth value. In the depth map filtering step, in order to reduce the error of the predicted initial depth value, a self-organizing map technique is performed on the predicted depth pixel by using the measured depth pixel around the predicted depth pixel. In the process of self-organization map, a weight is determined according to a difference between a distance between a predicted depth pixel and an measured depth pixel and a color value corresponding to each pixel. In this paper, we compared the proposed method with the bilateral filter and k-nearest neighbor widely used as a depth map upsampling method for performance comparison. Compared to the bilateral filter and the k-nearest neighbor, the proposed method reduced by about 6.4% and 8.6% in terms of MAE, and about 10.8% and 14.3% in terms of RMSE.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.40 no.9
• /
• pp.1856-1864
• /
• 2015

In this paper, the localization algorithm for improving the accuracy of the positioning using the Wi-Fi fingerprint using the normal distribution probability and the built-in typed accelerometer sensor, the gyroscope sensor of smartphone in the indoor environment is proposed. The experiments for analyzing the performance of the proposed algorithm were carried out at the region of the horizontal and vertical 20m * 10m in the engineering school building of our university, and the performance of the proposed algorithm is compared with the fingerprint and the DR (dead reckoning) while user is moving according to the assigned region. As a result, the maximum error distance in the proposed algorithm was decreased to 2cm and 36cm compared with two algorithms, respectively. In addition to this, the maximum error distance was also less than compared with two algorithms as 16.64cm and 36.25cm, respectively. It can be seen that the fingerprint map searching time of the proposed algorithm was also reduced to 0.15 seconds compared with two algorithms.

• Journal of KIISE:Computing Practices and Letters
• /
• v.16 no.2
• /
• pp.186-200
• /
• 2010

In order to render global illumination realistically, we need to accurately compute the direct and indirect illumination that represents the light information incoming through complex light paths. In this process, the indirect illumination at given point is greatly affected by surrounding geometries. Harmonic mean distance is a mathematical tool which is often used as a metric indicating the distance from a surface point to its visible objects in 3D space, and plays a key role in such advanced global illumination algorithms as irradiance/radiance caching and ambient occlusion. In this paper, we analyze the accuracy of harmonic mean distance estimated against various environments in the final gathering and photon mapping methods. Based on the experimental results, we discuss our experiences and future directions that may help develop an effective harmonic mean distance computation method in the future.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2016.10a
• /
• pp.559-560
• /
• 2016

Modern was also the era that can go quickly and easily, as well as overseas travel domestic travel. Whereby the two grew more and more people enjoying the trip. In the first place, but if you been easy to get lost overseas will receive help if not the ability to speak local languages and also to develop a tool that is easy to find your way in the first place because been difficult. Previously, became the drone that was used for military purposes now enjoy easy and convenient operation and safety of the public. Drones are using this time because of the wide application field width should also try to develop a drone navigation to find your way. If you set the destination using Google Maps it aims to make the user want to go to fly toward the destination while keeping the user with a certain distance.

• Journal of Korea Multimedia Society
• /
• v.15 no.6
• /
• pp.794-804
• /
• 2012

For the geometric computations on the protein molecules, the proximity queries, such as computing the minimum distance from an arbitrary point to the molecule or detecting the collision between a point and the molecule, are essential. For the proximity queries, the efficiency of the computation time can be different according to the data structure used for the molecule. In this paper, we present the data structures and algorithms for applying proximity queries to a molecule with GPU acceleration. We present two data structures, a voxel map and a sphere tree, where the molecule is represented as a set of spheres, and corresponding algorithms. Moreover, we show that the performance of presented data structures are improved from 3 to 633 times compared to the previous data structure for the molecules containing 1,000~15,000 atoms.

• Journal of the Korea Society of Computer and Information
• /
• v.27 no.8
• /
• pp.135-142
• /
• 2022

In this paper, we propose a graph-based user route control for rapidly conducting virus inspections in emergency situations (eg, COVID-19) and a framework that can simulate this on a city map. A^* and navigation mesh data structures, which are widely used pathfinding algorithms in virtual environments, are effective when applied to CS(Computer science) problems that control Agents in virtual environments because they guide only a fixed static movement path. However, it is not enough to solve the problem by applying it to the real COVID-19 environment. In particular, there are many situations to consider, such as the actual road traffic situation, the size of the hospital, the number of patients moved, and the patient processing time, rather than using only a short distance to receive a fast virus inspection.

• Journal of the Korea Convergence Society
• /
• v.12 no.10
• /
• pp.55-61
• /
• 2021

In this paper, we proposed a DeepLabv3+ based encoder-decoder model utilizing an attention mechanism for precise semantic segmentation. The DeepLabv3+ is a semantic segmentation method based on deep learning and is mainly used in applications such as autonomous vehicles, and infrared image analysis. In the conventional DeepLabv3+, there is little use of the encoder's intermediate feature map in the decoder part, resulting in loss in restoration process. Such restoration loss causes a problem of reducing segmentation accuracy. Therefore, the proposed method firstly minimized the restoration loss by additionally using one intermediate feature map. Furthermore, we fused hierarchically from small feature map in order to effectively utilize this. Finally, we applied an attention mechanism to the decoder to maximize the decoder's ability to converge intermediate feature maps. We evaluated the proposed method on the Cityscapes dataset, which is commonly used for street scene image segmentation research. Experiment results showed that our proposed method improved segmentation results compared to the conventional DeepLabv3+. The proposed method can be used in applications that require high accuracy.