• Proceedings of the Korea Information Processing Society Conference
• /
• 2020.05a
• /
• pp.555-558
• /
• 2020

본 논문에서는 입력 영상에 담긴 다양한 물체들과 그들 간의 관계를 효과적으로 탐지하여, 하나의 장면 그래프로 표현해내는 새로운 심층 신경망 모델을 제안한다. 제안 모델에서는 물체와 관계의 효과적인 탐지를 위해, 합성 곱 신경망 기반의 시각 맥락 특징들뿐만 아니라 언어 맥락 특징들을 포함하는 다양한 멀티 모달 맥락 정보들을 활용한다. 또한, 제안 모델에서는 관계를 맺는 두 물체 간의 상호 의존성이 그래프 노드 특징값들에 충분히 반영되도록, 그래프 신경망을 이용해 맥락 정보를 임베딩한다. 본 논문에서는 Visual Genome 벤치마크 데이터 집합을 이용한 비교 실험들을 통해, 제안 모델의 효과와 성능을 입증한다.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2021.11a
• /
• pp.775-778
• /
• 2021

악성 댓글은 언어폭력이며 사이버 범죄의 일종으로 인터넷상에서 상대방이 올린 글에 비방이나 험담을 하는 악의적인 댓글을 말한다. 악성 댓글을 단순히 차단하는 다른 프로그램들과는 달리 해당 영상의 악성 댓글의 비율을 알려주고 악플러들의 닉네임과 그 빈도를 나타내주는 것으로 차별화를 두었다. 따라서 많은 유튜버들이 겪는 악성 댓글 문제들을 탐지하여 유튜브에 달리는 악성 댓글들을 탐지하고 시각화하여 제공한다.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2023.11a
• /
• pp.1047-1048
• /
• 2023

청각장애인은 수화 언어, 즉 수어를 통해 의사소통한다. 따라서 본 논문에서는 의료 상황에서 청각 장애인이 겪는 소통의 어려움을 해결하기 위해 의료 상황 중심의 수어 데이터셋을 구축한 뒤, R(2+1)D 딥러닝 모델을 이용해 수어 동작을 영상 단위로 인식하고 분류할 수 있도록 하였다. 그리고 이를 Django를 이용한 웹 사이트로 만들어 사용할 수 있게 하였다. 이 웹 페이지는 청각장애인 개인 뿐만 아니라 의료 사회 전반적으로 긍정적인 효과를 줄 것으로 기대한다.

• Journal of Sensor Science and Technology
• /
• v.33 no.2
• /
• pp.78-85
• /
• 2024

We proposed a method for visual place recognition that represents images using objects as visual words. Visual words represent the various objects present in urban environments. To detect various objects within the images, we implemented and used a zero-shot detector based on a large-scale image language model. This zero-shot detector enables the detection of various objects in urban environments without additional training. In the process of creating histograms using the proposed method, frequency-based weighting was applied to consider the importance of each object. Through experiments with open datasets, the potential of the proposed method was demonstrated by comparing it with another method, even in situations involving environmental or viewpoint changes.

• Proceedings of the Korean Society of Computer Information Conference
• /
• 2024.01a
• /
• pp.359-361
• /
• 2024

한국의 전통탑에 대한 정보는 주로 2D 형식의 사진이나 동영상 자료를 통해 제공되고 있다. 또한, 특정 지역 및 시대에 따라 분류하여 제공하는 웹사이트도 일부 존재한다. 본 논문에서는 우리나라의 다양한 시대별 탑에 대한 서비스를 현실적으로 체험할 수 있도록 웹 기반으로 구현했다. Aframe 언어를 활용하여 웹상에서 그래픽을 표현하고, 서버는 아파치를 이용하여 서비스를 제공했다. 더불어 가상현실를 이용해 실감형서비스를 제공하기 위해 웹상에서 가상현실을 지원다. 이를 통해 사용자들은 웹상에서 가상현실과 360도 서비스를 동시에 이용할 수 있다.

• Proceedings of the Korea Society for Industrial Systems Conference
• /
• 2002.06a
• /
• pp.148-155
• /
• 2002

Multithreaded models improve the efficiency of parallel systems by combining inner parallelism, asynchronous data availability and the locality of von Neumann model. This model executes thread code which is generated by compiler and of which quality is given by the method of generation. But multithreaded models have the demerit that execution model is restricted to a specific platform. On the contrary, Java has the platform independency, so if we can translate from threads code to Java bytecode, we can use the advantages of multithreaded models in many platforms. Java executes Java bytecode which is intermediate language format for Java virtual machine. Java bytecode plays a role of an intermediate language in translator and Java virtual machine work as back-end in translator. But, Java bytecode which is translated from multithreaded models have the demerit that it is not secure. This paper, multhithread code whose feature of platform independent can execute in java virtual machine. We design and implement translator which translate from thread code of multithreaded code to Java bytecode and which check secure problems from Java bytecode.

• Cartoon and Animation Studies
• /
• s.36
• /
• pp.375-391
• /
• 2014

Animation movie is a non-photorealistic animated art that consists of formative language forming a frame based on a story and cuts describing frames that form the cuts. Therefore, in expressing an image, artistic expression methods and devices for a formative space are should be provided in a frame while cuts have the images between frames faithfully. Short animation movie is produced by various image experiments with unique image expressions rather than narration for expressing subjective discourse of a writer. Therefore, image style that forms unique images and various image directions are important factors. This study compared the experimental image directions of and , both of which showed a production method of film manipulation. First, while uses pixilation that produces images obtained from live images through painting and many optical disclosure process on a cell mat, was made with diverse collage techniques such as tearing, cutting, pasting, and folding hundreds of scenes from action movies. Second, expresses non-causal relationship of characters by their repetitive behaviors and circulatory image structure through a fixed camera angle, resisting typical scene transition. On the other hand, has an advancing structure that progresses antagonistic relationship of characters through diverse camera angles and scene transition of unique images. Third, in terms of editing, uses a long-take short cut technique in which the whole image consists of one short cut, though it seems to be many scenes with the appearance of various characters. On the other hand, maximizes visual fun and commitment by image reconstruction with hundreds of various short cuts. That is, both works have common features of an experimental work that shows expansion of animated image expressions through film manipulation that is different form general animation productions. On top of that, delivers routine life of diverse human beings without clear narration through image of conceptualized spaces. expresses it in a new image space through image reconstruction with collage technique and speedy progress, setting a binary opposition structure.

• Journal of radiological science and technology
• /
• v.36 no.4
• /
• pp.319-326
• /
• 2013

The purpose of this study is to evaluate usefulness of Hybrid image and Inverse image about detection of tumor shadow in chest radiography using ROC analysis. Original images of 60 cases are selected from Standards digital image date base issued by the Japanese Society of Radiological Technology. Through computer language of C, Inverse images of 60 cases and Hybrid image of 30 cases are made. The continues reading experiment was conducted. In the case of inverse image were observed by 5 radiographer and 2 radiologist. In the case of In case of Hybrid image were observed by 3 student radiographer and 2 experienced radiographer. ROC curve are constructed using ROCKIT Program made by Metz. In Inverse image, a Az of average ROC curve was increases from 0.742 of original image to 0.775 of inverse image. In normal cases, the effect of the detrimental is same to that of the beneficial, however In abnormal cases, the beneficial effect is greater than detrimental effect. However in Hybrid image, a Az of average ROC curve was decreases from 0.5253 of original image to 0.4868 of Hybrid image. In Normal cases, the effect of the detrimental is greater than that of the Beneficial, however In abnormal cases, the Beneficial effect is greater than detrimental effect. The inverse image can be more positively considered for the detecting of tumor than the hybrid image.

• Korean Journal of Food Science and Technology
• /
• v.28 no.4
• /
• pp.786-789
• /
• 1996

An image processing system was built to evaluate the color properties of apple and meat. The system consisted of video camera, video card, 32 bit microcomputer and an optical illuminator. The operating software was developed to carry out capturing, analyzing, displaying and storing of the 8 bit digitized images of food. The images of apples at various maturing stages were investigated to obtain the color histogram of R, G, B and Hunter value. RGB histogram showed a major difference in G value, 35.01, the minor change in R value, 6.16, and the negligible difference in B value. The image of beef cut was separated into two parts, fat and lean tissue, by applying threshold value method based on the digital value of color. The threshold value for fat was over 240 and for lean under 230 in R value, respectively. The resulting non fat image showed 2% decreased color difference value, ${\Delta}E$, than whole meat cut.

• Journal of the Korean Society for Nondestructive Testing
• /
• v.22 no.5
• /
• pp.465-473
• /
• 2002

Images are easily corrupted by noise during the data transmission, data capture and data processing. A technical method of noise analyzing and adaptive filtering for reducing of quantum noise in radiography is presented. By adjusting the characteristics of the filter according to local statistics around each pixel of the image as moving windowing, it is possible to suppress noise sufficiently while preserve edge and other significant information required in reading. We have proposed adaptive weighted median(AWM) filters based on local statistics. We show two ways of realizing the AWM filters. One is a simple type of AWM filter, whose weights are given by a simple non-linear function of three local characteristics. The other is the AWM filter which is constructed by homogeneous factor(HF). Homogeneous factor(HF) from the quantum noise models that enables the filter to recognize the local structures of the image is introduced, and an algorithm for determining the HF fitted to the detection systems with various inner statistical properties is proposed. We show by the experimented that the performances of proposed method is superior to these of other filters and models in preserving small details and suppressing the noise at homogeneous region. The proposed algorithms were implemented by visual C++ language on a IBM-PC Pentium 550 for testing purposes, the effects and results of the noise filtering were proposed by comparing with images of the other existing filtering methods.