• Journal of the Korea Society of Computer and Information
• /
• v.26 no.12
• /
• pp.45-52
• /
• 2021

In this paper, we propose a prediction model for osteopenia and osteoporosis based on a convolutional neural network(CNN) using computed tomography(CT) images. In a single CT image, CNN had a limitation in utilizing important local features for diagnosis. So we propose a compound model which has two identical structures. As an input, two different texture images are used, which are converted from a single normalized CT image. The two networks train different information by using dissimilarity loss function. As a result, our model trains various features in a single CT image which includes important local features, then we ensemble them to improve the accuracy of predicting osteopenia and osteoporosis. In experiment results, our method shows an accuracy of 77.11% and the feature visualize of this model is confirmed by using Grad-CAM.

• Journal of the Korea institute for structural maintenance and inspection
• /
• v.27 no.2
• /
• pp.58-66
• /
• 2023

In this study, cable tension was measured using the vibration method, and a vision-based system was applied as a sensor to measure the displacement response of a cable in a non-contact method. In the vision-based system, the camera is installed in a location that considers the target structure and the field of view of the camera. However, it can be difficult to recognize the control points required to measure the displacement response of a structure as the target structure and other structures such as vehicles may be included in the image at the intended installation location. In this study, a distorted image including a vehicle shows inaccurate results in image analysis due to the installation position of the vision-based system. Accordingly, the image including the vehicle was eliminated by calculating the similarity between the two images. To verify the validity of the method of estimating the cable tension of cable-stayed bridges using the proposed method, the vibration method was applied to cable-stayed bridges in service to measure the tension.

• Science of Emotion and Sensibility
• /
• v.7 no.3
• /
• pp.1-6
• /
• 2004

The algorithm developed in this paper allows us to generate or synthesize a large amount of data sets using only a small amount of signal features obtained from the original data set. Because the simulated density profiles of yarns retain the original features without a significant loss of information on the location of imperfections, the resulting fabric images are likely to resemble the original images. The data expansion system developed could generate a large area of fabric images by combining the Monte Carlo simulation and the wavelet sub-band exchange algorithm developed. The system has proven effective for simulating realistic fabric images by retaining the location of imperfections such as neps, thin and thick places.

• Journal of the Korea Society of Computer and Information
• /
• v.13 no.1
• /
• pp.117-124
• /
• 2008

We suggest CBIR(Content Based Image Retrieval) method using color and shape information. Using just one feature information may cause inaccuracy compared with using more than two feature information. Therefore many image retrieval system use many feature informations like color, shape and other features. We use two feature, HSI color information especially Hue value and CSS(Curvature Scale Space) as shape information. We search candidate image form DB which include feature information of many images. When we use two features, we could approach better result.

• Proceedings of the Korea Multimedia Society Conference
• /
• 2002.11b
• /
• pp.776-779
• /
• 2002

본 논문에서는 OpenGL을 이용한 삼차원 뷰어(Viewer)를 개발한다. 3D Max와 같은 3D 개발 툴에서 침대, 소파 등 3D 모델 라이브러리를 제작한 후 일어들이는 렌더링 모듈을 개발한다. 3D 모델 라이브러리에 직물 이미지 라이브러리를 매핑하는 모듈도 개발하여 직물 이미지와 유사한 효과를 낼 수 있도록 한다. 3D max에서 광원과 질감을 포함하여 렌더링한 후 개발되는 프로그램으로 데이터 손실없이 읽어들이는 것을 목표로 한다. 3차원 모델에 설계된 직물 이미지를 사용하여 3D 매핑된 후. 사용자가 확대, 축소, 이동, 회전 등의 효과를 줄 수 있게 하는 사용자 인터페이스 제공 및 3D 애니메이션 기능을 구현한다.

• Proceedings of the KAIS Fall Conference
• /
• 2009.05a
• /
• pp.407-410
• /
• 2009

본 논문에서는 인덱싱 카메라나, 감시 시스템의 네트워크로 부터 입력받은 영상들에서 사용자가 원하는 오브젝트를 추출해내고, 카메라의 위치와 시간으로 정렬되어진 데이터베이스상의 이미지들과 비교 분석하여 원하는 객체와의 유사도를 측정하고, 이것을 바탕으로 동일한 오브젝트들을 가진 이미지들의 집합을 시간과 공간데이터를 이용하여 객체의 이동경로를 파악하는 시스템을 제시한다.

• Proceedings of the Korean Society for Information Management Conference
• /
• 2001.08a
• /
• pp.149-152
• /
• 2001

이용자 중심의 인터페이스 개발은 현재 검색시스템에서 가장 중요한 이슈가 되고 있다. 인터페이스에서 메타포의 사용은 바로 이용자 편의성을 높이는 방안이라 할 수 있다. 본 연구는 대학도서관과 공공도서관의 검색인터페이스에서 사용하고 있는 메타포의 유용성을 평가하였다. 그 결과 현재 사용되고 있는 메타포의 문제점은 비슷비슷하다는 점이며 한 이미지가 유일한 의미를 갖고 있지 못하여 혼동을 야기 시킨다는 점이며, 이미지와 텍스트간의 의미적 관계 또한 명료하지 않게 나타났다. 따라서 메타포 개발시 이미지에 담겨진 의미가 모든 이용자와 같을 수 있도록 논리적 유사점을 찾아내어야 할 것이다.

• Cartoon and Animation Studies
• /
• s.51
• /
• pp.391-411
• /
• 2018

The purpose of this thesis is to examine ontology of digital photo image based on a Simulacre concept of Gilles Deleuze & Jean Baudrillard. Traditionally, analog image follows the logic of reproduction with a similarity with original target. Therefore, visual reality of analog image is illuminated, interpreted, and described in a subjective viewpoint, but does not deviate from the interpreted reality. However, digital image does not exist physically but exists as information that is made of mathematical data, a digital algorithm. This digital image is that newness of every reproduction, that is, essence of subject 'once existing there' does not exist anymore, and does not instruct or reproduce an outside target. Therefore, digital image does not have the similarity and does not keep the index instruction ability anymore. It means that this digital image is converted into a virtual area, and this is not reproduction of already existing but display of not existing yet. This not-being of digital image changes understanding of reality, existence, and imagination. Now, dividing it into reality and imagination itself is meaningless, and this does not make digital image with technical improvement but is a new image that is basically completely different from existing image. Eventually, digital image of the day passes step to visualize an existent target, nonexistent things have been visualized, and reality operates virtually. It means that digital image does not reproduce our reality but reproduces other reality realistically. In other words, it is a virtual reproduction producing an image that is not related to a target, that is to say Simulacre. In the virtually simulated world, reality has an infinite possibility, and it is not a picture of the past and present and has a possibility as the infinite virtual that is not fixed, is infinitely mutable, and is not actualized yet.

• Annual Conference of KIPS
• /
• 2019.10a
• /
• pp.868-871
• /
• 2019

본연구는 커플들의 이미지 빅 데이터를 분석하여 각각 얼굴과 패션에 따라 유사한 유형 끼리 클러스터링 하여 새로운 사람 이미지가 주어졌을 때 해당 사람이 어느 유형에 속하는지 찾아내고 해당 유형의 사람들은 어떤 유형의 이성과 잘 맞는지 찾아 추천해주는 플랫폼이다. 빅 데이터를 수집하기 위하여 SNS상에서 커플들의 이미지를 크롤링하여 저장한다. 수집된 커플들의 이미지를 AI 머신 러닝으로 나이, 성별을 분석하여 미리 설정한 나이대의 이성 커플들의 이미지 만을 추려내서 각각 남, 여의 이미지를 분리하여 저장한다. 해당 이미지들로 비슷한 얼굴, 패션 유형의 사람들을 같은 클러스터로 모으고 CNN 으로 학습 시켜서 새로운 이미지가 들어올 경우 효율적으로 해당 이미지가 어느 클러스터에 속하는지 찾아낼 수 있도록 한다. 특정 이미지가 속하는 클러스터를 찾아내면 해당 클러스터에 속하는 사람들의 연인들이 어느 클러스터에 가장 많이 포함되어 있는지 찾아서 해당 클러스터 유형의 이성을 추천해준다. 웹과 어플리케이션으로 이루어진 플랫폼 서비스이며, 커플 매칭 기능 뿐만 아니라 매칭된 회원 간 연락 기능, 실제 커플의 이미지로 두 사람의 매칭도 확인 등의 부가적 기능 또한 인공 지능 서비스로 제공된다.

• Journal of IKEEE
• /
• v.27 no.3
• /
• pp.280-287
• /
• 2023

Deep learning advancements can solve computer vision problems, but large-scale datasets are necessary for high accuracy. In this paper, we propose an image generation technique using object bounding boxes and image edge components. The object bounding boxes are extracted from the images through object detection, and image edge components are used as input values for the image generation model to create new image data. As results of experiments, the images generated by the proposed method demonstrated similar image quality to the source images in the image quality assessment, and also exhibited good performance during the deep learning training process.