• 한국가시화정보학회지
• /
• 제21권1호
• /
• pp.119-126
• /
• 2023

Acquiring experimental data for PIV verification or machine learning training data is resource-demanding, leading to an increasing interest in synthetic particle images as simulation data. Conventional synthetic particle image generation algorithms do not follow physical laws, and the use of CFD is time-consuming and requires computing resources. In this study, we propose a new method for synthetic particle image generation, based on a Physics-Informed Neural Networks(PINN). The PINN is utilized to infer the flow fields, enabling the generation of synthetic particle images that follow physical laws with reduced computation time and have no constraints on spatial resolution compared to CFD. The proposed method is expected to contribute to the verification of PIV algorithms.

• ETRI Journal
• /
• 제42권1호
• /
• pp.67-77
• /
• 2020

Image captioning has received significant interest in recent years, and notable results have been achieved. Most previous approaches have focused on generating visual descriptions from images, whereas a few approaches have exploited visual descriptions for image classification. This study demonstrates that a good performance can be achieved for both description generation and image classification through an end-to-end joint learning approach with a loss function, which encourages each task to reach a consensus. When given images and visual descriptions, the proposed model learns a multimodal intermediate embedding, which can represent both the textual and visual characteristics of an object. The performance can be improved for both tasks by sharing the multimodal embedding. Through a novel loss function based on class activation mapping, which localizes the discriminative image region of a model, we achieve a higher score when the captioning and classification model reaches a consensus on the key parts of the object. Using the proposed model, we established a substantially improved performance for each task on the UCSD Birds and Oxford Flowers datasets.

• ETRI Journal
• /
• 제39권4호
• /
• pp.480-492
• /
• 2017

A holographic display based on a viewing window enables the converging of a reconstruction wave into a viewing window by means of an optical system. Accordingly, a user can observe a reconstructed hologram image, even with a small diffraction angle. It is very difficult to manufacture an optical system with no aberrations; thus, it is inevitable that a certain amount of wave aberrations will exist. A viewing-window-based holographic display, therefore, always includes distortions in an image reconstructed from a hologram pattern. Compensating the distortions of a reconstructed image is a very important technical issue because it can dramatically improve the performance when reconstructing a digital three-dimensional content image from a hologram pattern. We therefore propose a method for suppressing image distortion by measuring and compensating the wave aberration calculated from a Zernike polynomial, which can represent arbitrary wave aberrations. Through our experimental configuration using only numerical calculations, our proposed method decreased the reconstructed image distortion by more than 28%.

• 한국관광식음료학회지:관광식음료경영연구
• /
• 제15권1호
• /
• pp.19-36
• /
• 2004

This study was to investigate brand image and its influence on customers perception of fast-food restaurants. A total of 25 brand image attributes were selected on the basis of the literature review and were asked to indicate respondents' perceptions of their image targeting for fast-food restaurants such as Lotteria, McDonald, KFC, and Popeyes. A on-site survey method was applied to collect data. Out of 300 distributed survey questionnaire, a total of 271 useful sample were collected and analyzed in SPSS 10.0. From the factor analysis with 25 brand image attributes, five underline dimensions were identified and named such as interior factor, store factor, atmosphere factor, service factor, and product factor. Then, T-test and ANOVA were conducted to investigate any differences between/among the groups based on those five brand image factors. The results provided that in terms of gender, age, preference for fast food, food cluster, and four different brand, customers differently perceived restaurant brand image. It is hoped that this information may help for restaurant managers and marketers to understand customers' restaurant perception and choice behavior in order to launch appropriate marketing strategies.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제4권5호
• /
• pp.910-924
• /
• 2010

As Internet usage has increased, the risk of adolescents being exposed to adult content and harmful information on the Internet has also risen. To help prevent adolescents accessing this content, a novel detection method for adult images is proposed. The proposed method involves three steps. First, the Image Of Interest (IOI) is extracted from the image background. Second, the IOI is distinguished from the segmented image using a novel weighting mask, and it is determined to be acceptable or unacceptable. Finally, the features (color and texture) of the IOI or original image are compared to a critical value; if they exceed that value then the image is deemed to be an adult image. A Receiver Operating Characteristic (ROC) curve analysis was performed to define this optimal critical value. And, the textural features are identified using a gray level co-occurrence matrix. The proposed method increased the precision level of detection by applying a novel weighting mask and a receiver operating characteristic curve. To demonstrate the effectiveness of the proposed method, 2850 adult and non-adult images were used for experimentation.

• 한국게임학회 논문지
• /
• 제17권6호
• /
• pp.153-164
• /
• 2017

구면 파노라마 영상에서는 심한 구면 왜곡으로 인하여 기존의 평면 패턴의 탐색 방법인 SURF가 올바르게 작동되지 않는다. 평면 패턴은 구면 영상에서 왜곡되어 나타나므로 구면 파노라마 영상에서의 패턴 탐색 및 다른 텍스처로의 대치는 평면 영상의 경우와 다르게 처리되어야 한다. 본 논문에서는 구면 파노라마 영상을 큐브맵 영상으로 변환하고 구면 왜곡이 제거된 평면 영상에서 SURF를 사용하여 패턴을 탐색하고 해당 평면 텍스처를 대치한 후 다시 구면 파노라마 영상으로 변환하는 구면 파노라마 영상에서의 평면 대치 방법을 제안한다.

• 대한전기학회:학술대회논문집
• /
• 대한전기학회 2004년도 학술대회 논문집 정보 및 제어부문
• /
• pp.358-360
• /
• 2004

The digital images contain a significant amount of redundancy and require a large amount of data for their storage and transmission. Therefore, the image compression is necessary to treat digital images efficiently. The goal of image compression is to reduce the number of bits required for their representation. The image compression can reduce the size of image data using contractive mapping of original image. Among the compression methods, the mapping is affine transformation to find the block(called range block) which is the most similar to the original image. In this paper, we applied the neural network(SOM) in encoding. In order to improve the performance of image compression, we intend to reduce the similarities and unnecesaries comparing with the originals in the codebook. In standard image coding, the affine transform is performed with eight isometries that used to approximate domain blocks to range blocks.

• 한국염색가공학회지
• /
• 제18권5호
• /
• pp.61-71
• /
• 2006

This paper addresses the factors hindering the image quality of lines in inkjet printed on polyester fabric as printing media. Lines were printed onto different types of polyester fabrics in warp and filling directions. Line image quality including line width, edge blurriness, and edge raggedness was assessed. The effect of capillary wicking on line image quality of printed spun polyester fabric is discussed. The factors on the image quality include printing position(top of the yam or between the yarn), printing direction(warp or filling), yarn structures(filament or spun), thread size(yam or fiber), finishing, and ink properties(evaporation rate). More than 30% differences in image quality results were observed by changing the printing location on the spun polyester fabric. The best results of the image quality were obtained with the printed plain and spun polyester fabrics. The fiber sizes may affect capillary size; therefore, the image quality can be dissimilar. Types of finishing materials and inks greatly improve the line image quality on spun polyester fabrics.

• 경영과학
• /
• 제27권2호
• /
• pp.97-108
• /
• 2010

Previous studies have shown that country image affects consumers' valuation of products. Based on a literature review this paper identifies five dimensions (economic, political, relational, people and cultural image) and purifies them with Exploratory Factor Analysis (EFA) and Confirmatory Factor Analysis (CFA). The data was gathered by using a structured questionnaire from 252 Korean consumers. Among the five dimensions of country image, this paper derives the three factors of country image for four countries (The United States, Japan, Australia, and China) - economic, relational, and cultural image. Then it examines the impacts of the three dimensions of country image on consumers' purchase intention of two industry types : industrial products vs. agricultural products. The result shows there is no difference between both of the two types in the impacts of country image on purchase intention. This implies that for managing the country image it not necessary to develop a communication strategy which is differentiated by industry types.

• 한국콘텐츠학회:학술대회논문집
• /
• 한국콘텐츠학회 2009년도 춘계 종합학술대회 논문집
• /
• pp.79-82
• /
• 2009

본 논문에서는 화재 영상을 검출하기 위한 새로운 영상처리 기법을 제안한다. 카메라에서 획득한 영상에서 화재 영상 검출에 필요한 차영상, 이진화 영상, 윤곽선 검출, 노이즈 제거(열림, 닫힘 연산), 픽셀검사[1], 플리커링 검사[2]등 영상처리기법들을 구현하기 위하여 OpenCV라는 영상처리 라이브러리를 사용하였다.