• 한국의상디자인학회지
• /
• 제9권3호
• /
• pp.19-34
• /
• 2007

The purpose of this research was to investigate the image dimension depending on the coloration of the Korean jacket of cromatic colors and the Korean skirt of acromatic color and to elucidate the image difference depending on the tone variation of the Korean jacket and the Korean skirt. The experimental materials used for this study were sets of stimulus and response scales(7 point semantic). The stimuli manipulated by computer simulation were 48 color pictures with various combinations of colors of jackets and skirts. The subjects were 576 female undergraduates living in Jinju city. This experiment was based on the $3{\times}4{\times}4$ factorial designs: jacket color(red, yellow and green), jacket tone(vivid, light, dull and dark), and skirt tone(N9, N7, N4 and N2). Image factor of the stimuli consisted of 4 different dimensions(youthfulness and activity, gracefulness, visibility and tenderness). Among them, the youthfulness and activity, and the gracefulness were important. According to the tone variation of Korean jacket of cromatic colors and Korean skirt of acromatic color, the images for a wearer were expressed diversely and showed the difference in image dimensions.

• 스마트미디어저널
• /
• 제8권2호
• /
• pp.46-57
• /
• 2019

In recent research, deep learning-based methods have achieved state-of-the-art performance in various computer vision tasks. However, these methods are commonly supervised, and require huge amounts of annotated data to train. Acquisition of data demands an additional costly effort, particularly for the tasks where it becomes challenging to obtain large amounts of data considering the time constraints and the requirement of professional human diligence. In this paper, we present a data level synthetic sampling solution to learn from small and imbalanced data sets using Generative Adversarial Networks (GANs). The reason for using GANs are the challenges posed in various fields to manage with the small datasets and fluctuating amounts of samples per class. As a result, we present an approach that can improve learning with respect to data distributions, reducing the partiality introduced by class imbalance and hence shifting the classification decision boundary towards more accurate results. Our novel method is demonstrated on a small dataset of 2789 tomato plant disease images, highly corrupted with class imbalance in 9 disease categories. Moreover, we evaluate our results in terms of different metrics and compare the quality of these results for distinct classes.

• Journal of information and communication convergence engineering
• /
• 제19권2호
• /
• pp.79-83
• /
• 2021

This paper proposes a novel image classification method based on few-shot learning, which is mainly used to solve model overfitting and non-convergence in image classification tasks of small datasets and improve the accuracy of classification. This method uses model structure optimization to extend the basic convolutional neural network (CNN) model and extracts more image features by adding convolutional layers, thereby improving the classification accuracy. We incorporated certain measures to improve the performance of the model. First, we used general methods such as setting a lower learning rate and shuffling to promote the rapid convergence of the model. Second, we used the data expansion technology to preprocess small datasets to increase the number of training data sets and suppress over-fitting. We applied the model to 10 monkey species and achieved outstanding performances. Experiments indicated that our proposed method achieved an accuracy of 87.92%, which is 26.1% higher than that of the traditional CNN method and 1.1% higher than that of the deep convolutional neural network ResNet50.

• 전기전자학회논문지
• /
• 제26권2호
• /
• pp.292-298
• /
• 2022

역광 조명에서 촬영된 영상은 한 장면에 지나치게 밝은 부분과 어두운 부분이 혼재되어 있어서 이를 전역적인 처리로 화질을 개선하는데는 한계가 있다. 본 논문은 역광 촬영된 사진을 각각 어두운 영역과 밝은 영역을 개선하는 두 장의 가상 영상으로 만들어 이를 원본 영상과 함께 라플라시안 피라미드로 융합하여 사진의 품질을 개선하는 방안에 대해 소개한다. 제안된 기법은 두 장의 가상 영상을 만들 때 LUT로 단순화할 수 있는 히스토그램 스트레칭과 감마변환을 활용하여 연산 부담을 저감하였다. 또한 색상 강화된 영상을 얻기 위해 HSV 좌표계를 사용하여 휘도에 대해서만 명암 변환을 실시하였다. 제안된 기법은 표준 영상 데이터 세트를 사용하여 몇 가지의 NIQA 지표를 산출하여 그 효용성을 보였다.

• ETRI Journal
• /
• 제44권1호
• /
• pp.10-23
• /
• 2022

Despite the significant progress in camera pose estimation and structure-from-motion reconstruction from unstructured images, methods that exploit a priori information on camera arrangements have been overlooked. Conventional state-of-the-art methods do not exploit the geometric structure to recover accurate camera poses from a set of patch images in an array for mosaic-based imaging that creates a wide field-of-view image by sewing together a collection of regular images. We propose a camera pose estimation framework that exploits the array-structured image settings in each incremental reconstruction step. It consists of the two-way registration, the 3D point outlier elimination and the bundle adjustment with a constraint term for consistent rotation vectors to reduce reprojection errors during optimization. We demonstrate that by using individual images' connected structures at different camera pose estimation steps, we can estimate camera poses more accurately from all structured mosaic-based image sets, including omnidirectional scenes.

• 디자인학연구
• /
• 제19권2호
• /
• pp.313-322
• /
• 2006

본 논문에서는 내용 기반 영상 검색을 이용하여 사용자가 원하는 영상을 쉽게 찾아내고, 이를 자동 재구성함으로써, 독창적인 영상 언어라 일컬어지는 몽타주를 사용자 중심의 관점에서 구현하고자 한다. 본 논문에서 제안하는 실시간 몽타주 시스템은 이산 푸리에 변환(Discrete Fourier Transform)을 이용해 사용자가 선택한 영상의 특징을 찾고, 유클리디안 거리(Euclidean Distance)를 이용해 데이터베이스에 있는 영상과 유사도를 비교함으로써, 빠르고 효과적으로 사용자가 원하는 영상을 검색할 수 있다. 또한 카메라 트래킹(Camera Tracking)에 의해 실시간으로 사용자의 움직임 영상을 취득하고, 취득된 영상을 검색된 사용자의 영상과 함께 자동 재구성함으로써, 손쉽게 사용자의 의도에 맞춘 영상 재구성을 하게 된다. 본 시스템은 사용자를 즐겁게 참여시킬 수 있는 뉴미디어 영상 디자인 툴(엔터테인먼트)이다. 일방적으로 영상을 시청하는 소극적 영상의 수용자에서 벗어나 영상 재생산의 적극적 주체가 되는, 사용자 중심의 새로운 영화(미디어기반 엔터테인먼트)의 토대가 될 것으로 기대된다.

• 방송공학회논문지
• /
• 제23권6호
• /
• pp.747-759
• /
• 2018

본 논문에서는 실시간 고화질 360 영상 전송을 위해 사용자 시점에 기반한 타일 분할 기법을 적용하고 화질 평가를 위해 360 영상을 가상현실 기기 화면에 표현하는 구현을 설명한다. 사용자 시점에 기반한 고화질 360 영상 전송을 위한 방안으로, 움직임 참조 문제를 해결하기 위한 MCTS (Motion Constrained Tile Sets) 기술과 미리 구성된 타일 정보들을 포함하는 EIS (Extraction Information Sets) SEI (Supplemental Enhancement Information), 타일 정보를 추출해내고 영상을 분할 및 추출해주는 추출기(extractor)를 구현한다. 또한 사용자 시점에 기반한 타일 추출 방법과 추출된 영상을 이용해 가상현실 기기 화면에 표현하는 방법에 대한 구현 내용을 설명한다. 따라서, 제안된 구현물을 기반으로 영상 전송을 수행하면, 사용자 시점 영역의 영상만 전송하여 불필요한 영상 전송을 하지않게 되어 화질 대비 낮은 대역폭의 향상된 영상을 표현할 수 있다.

• 복식
• /
• 제54권3호
• /
• pp.145-158
• /
• 2004

In modern art, body appears as artistic performer, material or producer. and is expressed as' social environment body' which is changed in the relationship with its society and culture. The correlation between body and clothing image appears in 'body as expression object' which directly borrows human body or sets up a section of human body, 'body as medium' in which clothing substitutes body, and 'body as image' in which image of body reappears along with clothing. The results of analysis are as follows : First, 'image of expansion and exaggeration' to expand the influence of clothing thereby disclosing illusion of material civilization prevalent in our society, and make metaphor of dwarfish human's lurking fear by transforming and exaggerating human body. Second, 'image of restriction and suppression' to express the loss of humanity, power and restriction of modern society with fixing and cruelty of body image through clothing which disregard body function. Third, 'image of open and fluidity to criticize the extinction of values of human existence and standardized figure of our society by reducing three-dimensional clothing and body to untypical form or introducing the image of absent of human body to clothing. Fourth, 'image of reversion and paradox' to express practical clothing object with unwearable material, or cause confusion of sex and identity by expressing dual aspects of body at the same time. In this study, which is focused on correlation between body and clothing and the meaning of them, I realized that, even though artistic clothing expressed image of distortion and deformation of human existence as essential subject of body, they, in most contents, were used as medium of communication to rediscover human dignity and identity, and consisted of a series of metaphoric network of meanings satirizing aspects of our society.

• Spatial Information Research
• /
• 제18권3호
• /
• pp.73-83
• /
• 2010

본 연구는 정지장면의 연속영상간 각 픽셀위치에서 일어나는 통계적 특성을 활용하여 영상을 분할하는 기법을 제안한다. 공간정보의 획득과 분석에서 디지털 영상 처리 기법의 활용은 아주 중요한 의미를 가진다. 특히 디지털 영상의 영역 구분을 위해 다양한 영상 분할(image segmentation) 기법들이 활용되고 있다. 본 연구에서는 선행 연구한 연속프레임 영상의 분광학적 특성 분석의 결과를 바탕으로 연속 프레임 간 Randomness를 활용한 이미지 분할 방법을 제안하였다. 우선 연속 프레임 간 각 화소에 통계학적인 분석 방법을 적용하여 각 화소의 평균과 표준편차 값을 구하고, 이를 통하여 대상 영상에서 가장 신뢰할 만한 화소들을 찾아 씨앗 점(seed point)을 결정하였다. 그리고 이 씨앗 점들을 시작으로 이웃 화소 간 T-test를 실시하였으며, 이를 기반으로 영역 성장(region growing)의 개념을 적용하여 영상을 분할 할 수 있는 기법을 연구하였다. 제안방식의 성능을 검증하기 위하여 실험을 통하여 기존의 방식과 비교분석을 수행하였다. 이러한 실험의 결과 영상분할에서 영상의 단일 프레임을 활용한 것보다 연속 프레임을 활용한 경우가 유리함을 확인 할 수 있었다.

• Journal of Radiation Protection and Research
• /
• 제44권4호
• /
• pp.149-155
• /
• 2019

Background: Magnetic resonance (MR) image guided radiation therapy system, enables real time MR guided radiotherapy (RT) without additional radiation exposure to patients during treatment. However, MR image lacks electron density information required for dose calculation. Image fusion algorithm with deformable registration between MR and computed tomography (CT) was developed to solve this issue. However, delivered dose may be different due to volumetric changes during image registration process. In this respect, synthetic CT generated from the MR image would provide more accurate information required for the real time RT. Materials and Methods: We analyzed 1,209 MR images from 16 patients who underwent MR guided RT. Structures were divided into five tissue types, air, lung, fat, soft tissue and bone, according to the Hounsfield unit of deformed CT. Using the deep learning model (U-NET model), synthetic CT images were generated from the MR images acquired during RT. This synthetic CT images were compared to deformed CT generated using the deformable registration. Pixel-to-pixel match was conducted to compare the synthetic and deformed CT images. Results and Discussion: In two test image sets, average pixel match rate per section was more than 70% (67.9 to 80.3% and 60.1 to 79%; synthetic CT pixel/deformed planning CT pixel) and the average pixel match rate in the entire patient image set was 69.8%. Conclusion: The synthetic CT generated from the MR images were comparable to deformed CT, suggesting possible use for real time RT. Deep learning model may further improve match rate of synthetic CT with larger MR imaging data.