• 한국정보처리학회:학술대회논문집
• /
• 한국정보처리학회 2013년도 추계학술발표대회
• /
• pp.139-142
• /
• 2013

최근 대용량 영상데이터로부터 정보 수집, 영상 처리를 위한 클라우드 관련 연구들이 활발하다. 그러나 공개 소프트웨어를 이용한 클라우드 연구의 대부분은 라이브러리 수준이 아닌 단순히 프로그램 수준의 조합으로 작동한다. 이런 이유로 단순 조합에 따른 비효율성에 의한 성능문제는 크게 다루어지지 않는다. 본 논문에서는 이 비효율성을 해결하는데 중점을 두고 FFmpeg과 Hadoop을 라이브러리 수준으로 결합하여 기존보다 더 나은 성능의 영상클라우드 환경을 구축하였다. C기반의 영상처리 라이브러리인 FFmpeg와 JAVA기반의 클라우드 환경 Hadoop의 결합을 위해 JNI(Java Native Interface)를 이용하였다. 상세구현으로는 HDFS(Hadoop Distributed File System)을 확장하여 Hadoop MapReduce가 직접 FFmpeg을 통한 영상파일 접근이 가능하게 하였다. 이로써 FFmpeg과 Hadoop간 상이한 파일 접근 방식에서 발생하는 불필요한 작업에 의한 시스템의 성능저하를 막았다. 또한 응용의 확장성을 위해 영상작업시 작업영상을 영상처리의 최소단위인 GOP(Group of Pictures)단위로 잘라 클라우드의 노드들에게 분산시켰다. 결과적으로 기존에 존재하는 Hadoop과 FFmpeg을 프로그램적으로 결합한 영상처리 클라우드보다 총 처리시간을 앞당겼고, GOP 단위의 영상 처리는 영상기반 작업에 안정성과 응용의 확장성을 보장해주었다.

• 건축역사연구
• /
• 제3권1호
• /
• pp.57-82
• /
• 1994

This paper is the result of the researches and the field surveys of the villages and the dwellings of Korean immigrants in Yien-Pien area, north-eastern China. This study aims to persue both of the origin and the process of development of their settlements and dwelling types from late 19C to the present. Their processes are too complex to analysis by single view-point. I have eyes to interprete them from three pionts; 1)correspondences between the dwelling types and the econo-political history of their region, 2)cultural assimilation with the native dwelling types, and 3)the direction of their modernization with the economical development of modern China. Three village types have been pioneered; 1)the villages of indivisual immigration, 2)the villages of planned group immigration, and 3)the villages of socilistic reform villages of 1) were composed of organic village patterns and various shaped dwelling lots on the sloped site; villages both of 2) and 3), gird patterns and uniformed lots on open fields. Historically, villages of 1) were pioneered before 1931; villages of 2), 1936-1945; villages of 3), from 1945. Each of dwelling types had strong relations with the village types to which it belonged. Before 1931, dwellings were built up based on so called "Ham-buk dwelling type" which was dominent in north-eastern Korea. In the era of gruop-immigration, various dwelling types were flew into Yen-Pien from southern Korea. In modern China, their southern types were changed into Yen-Pien type as similar as Ham-book type. After 1945, with the Great leap Forward and the Cultural Revolution, as communization of indivisual properties and reorganization of rural communities, each of dwellings became smaller and simpler in aspects of scales as well as functions. There are two types in Yen-Pien dwellings, those are 'single-file' and 'double-file' type. Three sub-types of latter arc 'six-bays', 'eight-bays', and rarely 'ten-bays'. The most common element of all types is Chong-ju-k'an; which is large room with heated floor, openig to kitchen. Now, modern dwellings of Korean immigrants are changing their spatial compositions, materials, and structures. With cultural assimilation as well as modernization, especially in urban areas, they are compelled to accept the elements of Chinese dwellings. But the spatial element of "Chong-ju-k'an", which is the core element of Yen-Pien dwelling type, never fade away nor is changed.

• 정보처리학회논문지:컴퓨터 및 통신 시스템
• /
• 제11권10호
• /
• pp.363-372
• /
• 2022

최근 코로나 19 팬더믹 이후 원격근무의 확대와 더불어 랜섬웨어 팬더믹이 심화하고 있다. 현재 안티바이러스 백신 업체들이 랜섬웨어에 대응하고자 노력하고 있지만, 기존의 파일 시그니처 기반 정적 분석은 패킹의 다양화, 난독화, 변종 혹은 신종 랜섬웨어의 등장 앞에 무력화될 수 있다. 이러한 랜섬웨어 탐지를 위한 다양한 연구가 진행되고 있으며, 시그니처 기반 정적 분석의 탐지 방법과 행위기반의 동적 분석을 이용한 탐지 연구가 현재 주된 연구유형이라고 볼 수 있다. 본 논문에서는 단일 분석만을 이용하여 탐지모델에 적용하는 것이 아닌 ".text Section" Opcode와 실제 사용하는 Native API의 빈도수를 추출하고 K-means Clustering 알고리즘, 코사인 유사도, 피어슨 상관계수를 이용하여 선정한 특징정보들 사이의 연관성을 분석하였다. 또한, 타 악성코드 유형 중 웜과 Cerber형 랜섬웨어를 분류, 탐지하는 실험을 통해, 선정한 특징정보가 특정 랜섬웨어(Cerber)를 탐지하는 데 특화된 정보임을 검증하였다. 위와 같은 검증을 통해 최종 선정된 특징정보들을 결합하여 기계학습에 적용하여, 최적화 이후 정확도 93.3% 등의 탐지율을 나타내었다.

• 한국CDE학회논문집
• /
• 제21권4호
• /
• pp.417-425
• /
• 2016

Especially for resolving BIM data compatibility issue, as one of recently raised BIM technology issues, has also been improved by using open BIM, representatively using IFC (Industry Foundation Classes) format. As shown in many case studies, usefulness of BIM technology is increasing day by day, and the IFC-based open BIM technology is essential in recent AEC projects where the productive collaboration is of importance. One of current problems in actual projects is that there is a conflict between conventional ways and newly developed BIM ways. Using both conventional and new technologies leads construction workers to having more work loads, consequently the efficiency and productivity of on-site workers have been decreased. Thus, it is strongly necessary to facilitate 3D BIM models to extract and generate 2D precision drawings in automation, especially using open BIM technologies. Some native BIM authoring tools have limitations in there automatic generation of 2D drawings, and there is no standardized mechanism to generate 2D drawings from heterogeneous applications. For this reason, this paper aims to develop an automated stand-alone program to generate 2D drawings in automation using IFC file, totally independent from various BIM authoring tools. By using the application described in this paper, any type of general drawings such as plan, section and elevation can be extracted without additional efforts. The development approach described in this paper, based on the open BIM technologies, has a strong impact to the current process especially in the perspective of enhancing productivity when we need to find out a trade-off in-between conventional and new approaches.

• International Journal of Internet, Broadcasting and Communication
• /
• 제16권2호
• /
• pp.149-160
• /
• 2024

For examining Web3.0 video streaming (VS) platforms in terms of the decentralized technology, tokenization and decentralized autonomous organization (DAO), we look at four platforms like DLive, DTube, Livepeer, and Theta Network (Theta). As a result, DLive which firstly partnered with Medianova for CDN and with Theta for peer to peer (P2P) network and migrates to Tron blockchain (BC), receives no commission from what creators earn, gives rewards to viewers by measuring engagement, and incentivizes participation by allowing 20% of donation & fees for funding development, 5% to BitTorrent Token (BTT) stakeholders (among these 5%, 20% to partners, 80% to other BTT stakeholders). DTube on its own lower-layer BC, Avalon, offers InterPlanetary File System (IPFS), gives 90% of the created value to creators or curators, and try to empower the community. Livepeer on Ethereum BC offers decentralized CDN, P2P, gives Livepeer Token (LPT) as incentive for network participants, and delegators can stake their LPT to orchestrators doing good. Theta on its native BC pulls streams from peering caching nodes, creates P2P network, gives Theta utility token, TFUEL for caching or relay nodes contributors, and allows Theta governance token, THETA as staking token. We contribute to the categorization of Web3.0 VS platforms: DLive and DTube reduce the risk of platform censorship, promote the diverse content, and allow the community to lead to more user-friendly environments. On the other hand, Livepeer and Theta provide new methods to stream content, but they have some differences. Whereas Livepeer focuses on the transcoding layer, Theta concentrates both on the video application layer and content delivery layer. It means, Theta tries to deliver value to all participants by enhancing network quality, reducing CDN cost, and rewarding users in utility tokens for the storage and bandwidth they provide.