• Journal of Advanced Navigation Technology
• /
• v.27 no.2
• /
• pp.182-189
• /
• 2023

Depth estimation is a key technology in 3D map generation for autonomous driving of vehicles, robots, and drones. The existing sensor-based method has high accuracy but is expensive and has low resolution, while the camera-based method is more affordable with higher resolution. In this study, we propose self-attention-based unsupervised monocular depth estimation for UAV camera system. Self-Attention operation is applied to the network to improve the global feature extraction performance. In addition, we reduce the weight size of the self-attention operation for a low computational amount. The estimated depth and camera pose are transformed into point cloud. The point cloud is mapped into 3D map using the occupancy grid of Octree structure. The proposed network is evaluated using synthesized images and depth sequences from the Mid-Air dataset. Our network demonstrates a 7.69% reduction in error compared to prior studies.

• Journal of Internet Computing and Services
• /
• v.13 no.3
• /
• pp.49-59
• /
• 2012

Throughout developing digital technology, reproduction of image is growing better day by day. And at the same time, diverse image editing softwares are developed to manage images easily. In the process of editing images, those programs could delete or modify EXIF files which have the original image information; therefore images without the origin source are widely spread on the web site after editing. This matter could affect analysis of images due to the distortion of originality. Especially in the court of law, the source of evidence should be expressed clearly; therefore digital image EXIF file without deletion or distortion could not be the objective evidence. In this research, we try to trace the identification of a digital camera in order to solve digital images originality, and also we focus on lens distortion correction algorism which is used in digital image processing. Lens distortion correction uses mapping algorism, and at this moment it also uses interpolation algorism to prevent aliasing artifact and reconstruction artifact. At this point interpolation shows the similar mapping pattern; therefore we want to find out the interpolation evidence. We propose a minimum filter algorism in order to detect interpolation pattern and adjust the same minimum filter coefficient in two areas; one has interpolation and the second has no interpolation. Throughout DFT, we confirm frequency character between each area. Based on this result, we make the final detection map by using differences between two areas. In other words, thereby the area which has the interpolation caused by mapping is adjusted using minimum filter for detection algorism; the second area which has no interpolation tends to different frequency character.

• 한국HCI학회:학술대회논문집
• /
• 2006.02a
• /
• pp.541-546
• /
• 2006

3차원 관상동맥을 분석하기 위해서는 혈관의 분기점, 극단점, 혈관의 계층적 구조 관계를 함축적으로 표현하는 것이 매우 중요하다. 본 논문에서는3차원 CT 혈관 조영 영상으로부터 관상동맥의 3차원 골격을 자동으로 추출하는 방법을 개발하였다. 먼저, CT혈관 조영술에 의해 획득된 슬라이스 이미지로부터 3차원 조작 및 수술 시뮬레이션 등을 위하여 혈관의 3차원 표면에 대한 메쉬 모델을 생성한다. 생성된 메쉬 모델이 임의로 변형된 후에도 자동으로 골격을 쉽게 추출할 수 있도록 메쉬 모델을 복셀화하는 단계를 거친다. 이렇게 얻어진 복셀 모델로부터 표면복셀을 결정하고 표면 복셀로부터 객체 복셀까지의 유클리드 거리값를 계산하여 유클리드 거리맵(EDM)을 계산한다. 계산된 EDM 으로부터 객체 복셀이 가지게 되는 최대 내접 구를 계산하여 Discrete Medial Surface을 생성하게 되는데 이것은 골격의 후보가 된다. 골격의 후보집합 복셀에 대하여 Dijkstra 최단 경로 결정 알고리즘을 적용하여 골격을 자동으로 추출하게 된다. 이렇게 추출된 3차원 골격은 관상동맥 수술 시뮬레이션 등의 다양한 형상 분석에 유용하게 사용될 수 있다.

• Annual Conference of KIPS
• /
• 2005.05a
• /
• pp.1661-1664
• /
• 2005

본 논문에서는 computer-aided analysis 의 일환으로 X-ray 심혈관 조영도를 이용하여 관상동맥의 구조를 보여주는 방법에 대해 제시하고자 한다. 관상동맥 폐색증 환자들에게 시술되는 스텐트 삽입 시술이나 관상동맥 우회로 시술을 할 때에는 X-ray 의 조영 영상이 매우 중요한 시술의 기준이 되고 있으며, 조영 영상에서 혈관을 빠르고 정확하게 인식하는 것은 시술의 필수 조건이다. 이러한 시술중의 혈관구조 인식을 돕기 위하여 본 논문에서는 심혈관 조영 영상으로부터 관상동맥의 골격을 추출하기 위한 방법을 제안한다. 본 논문에서는 혈관 구조 추출을 위하여 3 단계 알고리즘을 제시한다. 첫번째 단계에서는 조영도에서 잡음을 제거하기 위하여 동질영역을 블러링할 수 있는 speckle reducing anisotropic diffusion 을 이용한 이미지 필터링을 수행한다. 이 필터링은 영상내 잡음을 제거하고 혈관의 경계선을 강화하여 정확한 영상인식을 가능하게 한다. 두번째 단계에서는 영상 내에서 보여지는 주요 혈관을 분할하는 것이다. 이 영상분할에는 canny edge detection 과 개선된 영역확장법(adaptive region growing)을 동시에 이용하는 복합적 분할기법이 수행된다. 세번째 단계에서는 형태학적 기법(Morphology)을 이용하여 분할결과의 부족한 부분을 보완하고 골격화를 수행하여 정확한 혈관 구조를 추출해낸다. 실험을 위해서는 정상인의 관상동맥 영상 뿐 아니라 혈관이 가늘어지는 폐색이 관찰되는 환자의 영상에 대해서도 실험하였다. 또한 논문에서 제시한 알고리즘에 대한 검증을 위하여 실험 결과들은 의료진의 감수를 거쳤다.

• Culinary science and hospitality research
• /
• v.13 no.2
• /
• pp.275-291
• /
• 2007

The growth of foodservice industry has caused keen competition in the family restaurant industry. For a differentiation strategy to be competitive, on the preferential basis, it is necessary for each family restaurant business to analyze its position in the market. Therefore, this study is to certify customers' perceived servicescape and image similarity and analyzes positioning of seven family restaurants. For this research, multidimensional scaling(MDS) is performed to determine how they are positioned relative to competitors. According to the result of the positioning map(ALSCAL and PROFIT map), it is found that Outback steakhouse, T.G.I. Friday's, Bennigans' have more strong competitiveness than VIPS, Marche and Sizzler in servicescape factors and restaurant selection attribute except menu diversity. But the preference order does not follow the order of competitiveness. This result from the positioning map means that each business needs various marketing strategies that cannot be copied easily.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.23 no.2
• /
• pp.93-99
• /
• 2013

Fractal theory has been adopted as an effective tool for modelling complex and irregular natural phenomena facing in the fields of Computer Science, Engineering, Medical, Climatology and so on. In this paper, we presents an algorithm which enhances the performance of the triangular prism method(TPM) which has been widely used for fractal dimension extraction of natural terrains and images. For this, existing sampling methods are analyzed and a new sampling method which takes their merits is proposed. The effectiveness of the proposed algorithm is tested on fractal terrain maps and its performance is compared with that of other methods.

• The Journal of Korean Association of Computer Education
• /
• v.9 no.2
• /
• pp.101-109
• /
• 2006

We propose an adaptive digital watermarking algorithm using HVS(human visual system) and SOM(Self-Organizing Map) among neural networks. This method adjusts adaptively the strength of the watermark which is embedded in different blocks according to block classification in DCT(Discrete Cosine Transform) domain. All blocks in 3 classes out of 4 are selected to embed a watermark. Watermark sequences are embedded in 6 lowest frequency coefficients of each block except the DC component. The experimental results are excellent.

• FLOWER RESEARCH JOURNAL
• /
• v.17 no.4
• /
• pp.336-341
• /
• 2009

This paper is to propose creating brand of Korean floricultural industry due to decreasing income of farm, cultivation area, market volume. Japanese flower brands have been setting in market in form of chain store or franchise. But in Korea, there is no indigenous image and style of floricultural brands yet. However, The successful brand of floricultural distribution market and flower production, for example, the 10 cm pot plant, the parents' day carnation, and the wreath of condolence or congratulation will be expected revival in the future of Korean floricultural industry.

• Journal of IKEEE
• /
• v.25 no.2
• /
• pp.337-343
• /
• 2021

Deep reinforcement learning is an artificial intelligence algorithm that enables learners to select optimal behavior based on raw and, high-dimensional input data. A lot of research using this is being conducted to create an optimal movement path of a mobile robot in an environment in which obstacles exist. In this paper, we selected the Dueling Double DQN (D3QN) algorithm that uses the prioritized experience replay to create the moving path of mobile robot from the image of the complex surrounding environment. The virtual environment is implemented using Webots, a robot simulator, and through simulation, it is confirmed that the mobile robot grasped the position of the obstacle in real time and avoided it to reach the destination.

• The Journal of the Korea institute of electronic communication sciences
• /
• v.17 no.6
• /
• pp.1187-1192
• /
• 2022

Gangwon-do is a representative leisure resort. However, the provision of tourism information has not escaped the existing image- and text-oriented simplicity. Therefore, a linkage strategy for the content business is needed to revitalize tourist attractions. In this paper, first, we provide a 360-degree rotating VR content authoring function. Second, we propose a dynamic platform including a function that allows users to easily author tour scheduling after registering VR content on a map. Finally, the proposal system provides pre-tourism customers with an opportunity to increase satisfaction by providing online pre-experience.