• Proceedings of the Korea Information Processing Society Conference
• /
• 2013.11a
• /
• pp.1068-1069
• /
• 2013

최근 정보 기술과 웹의 발전으로 많은 응용에서 데이터의 양이 급격이 증가하였다. MapReduce는 이러한 대용량 데이터를 처리하기 위해 구글에서 제안한 프레임워크이다. MapReduce 프레임워크는 데이터 전달 패러다임을 기반으로 한다. 이로부터, 데이터 처리 및 질의에 있어 I/O 비용이 전체 처리 비용에서 큰 부분을 차지한다. 본 논문에서는 MapReduce 프레임워크에서 I/O에 소요되는 비용을 확인하기 위해, 실제 데이터를 기반으로 실험을 수행하였다. 이를 통해, MapReduce 기반 시스템의 성능 예측이나 성능 향상을 위해 고려되어야 할 부분을 제시하고자 하였다.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2007.05a
• /
• pp.731-734
• /
• 2007

본 연구는 구조광 기반의 깊이 에지를 이용하여 조명의 변화와 복잡한 배경에 상관없이 손 제스처의 외곽선 영상을 안정적으로 획득하였고, 제스처 영상을 표현하기 위하여 Curvature Scale Space(CSS) map을 이용하였다. 기존의 CSS map은 외곽선 영상의 깊은 굴곡과 완만한 굴곡을 잘 구분하지 못하는 문제점이 있었으나, 본 연구에서는 이러한 문제점을 분석하고, 이를 개선하기 위해서 각도 좌표를 이용한 CSS map 생성 방법을 제안하였다. 실험을 통해서 제안한 방법이 기존의 CSS map보다 우수한 인식 성능이 있음을 보였다.

• Journal of KIISE
• /
• v.43 no.6
• /
• pp.700-709
• /
• 2016

Recently, with the rapid growth of LOD (Linked Open Data) existing methods based on a single machine have limitation in performance. Existing solutions use distributed framework such as Mapreduce in order to improve the performance. However, the MapReduce framework for processing SPARQL queries involves multiple MapReduce jobs and additional costs incurred. In addition, the problem of unnecessary data processing arises. In this study, we proposed a method to reduce the number of MapReduce jobs during SPARQL query processing and join indexes based on Bitmap for minimizing the costs of processing unnecessary data.

• IEMEK Journal of Embedded Systems and Applications
• /
• v.11 no.5
• /
• pp.277-285
• /
• 2016

Moving object detection system has been an emerging research field in various advanced driver assistance systems (ADAS) and surveillance system. In this paper, we propose two optical flow based moving object detection methods at dynamic scenes. Both proposed methods consist of three successive steps; pre-processing, foreground segmentation, and post-processing steps. Two proposed methods have the same pre-processing and post-processing steps, but different foreground segmentation step. Pre-processing calculates mainly optical flow map of which each pixel has the amplitude of motion vector. Dense optical flows are estimated by using Farneback technique, and the amplitude of the motion normalized into the range from 0 to 255 is assigned to each pixel of optical flow map. In the foreground segmentation step, moving object and background are classified by using the optical flow map. Here, we proposed two algorithms. One is Gaussian mixture model (GMM) based background subtraction, which is applied on optical map. Another is adaptive thresholding based foreground segmentation, which classifies each pixel into object and background by updating threshold value column by column. Through the simulations, we show that both optical flow based methods can achieve good enough object detection performances in dynamic scenes.

• KIPS Transactions on Software and Data Engineering
• /
• v.3 no.2
• /
• pp.73-80
• /
• 2014

MapReduce is a widely used programming model for analyzing and processing Big data. Aggregate queries are one of the most common types of queries used for analyzing Big data. In this paper, we propose an efficient method for processing an aggregate query stream, where many concurrent users continuously issue different aggregate queries on the same data. Instead of processing each aggregate query separately, the proposed method processes multiple aggregate queries together in a batch by a single, optimized MapReduce job. As a result, the number of queries processed per unit time increases significantly. Through various experiments, we show that the proposed method improves the performance significantly compared to a naive method.

• Proceedings of the KSRS Conference
• /
• 1999.11a
• /
• pp.304-309
• /
• 1999

In this paper, we presents a new disparity map refinement algorithm using statistical characteristics of disparity map and edge information. The proposed algorithm generate a refined disparity map using disparity maps which are obtained from area and feature-based Stereo Matching by selecting a disparity value of edge point based on the statistics of both disparity maps. Experimental results on aerial stereo image show the better results than conventional fusion algorithms in the disparity error. This algorithm can be applied to the reconstruction of building image from the high resolution remote sensing data.

• Journal of Internet of Things and Convergence
• /
• v.1 no.1
• /
• pp.11-19
• /
• 2015

The mobile and ubiquitous environment is experiencing a rapid development of information and communications technology as it provides an ever increasing flow of information. Particularly, GIS is now widely applied in daily life due to its high accuracy and functionality. GIS information is utilized through the tiling method, which divides and manages large-scale map information. The tiling method manages map information and additional information to allow overlay, so as to facilitate quick access to tiled data. Unlike past studies, this paper proposes a new architecture and algorithms for tile searching and indexing management to optimize map information and additional information for GIS mobile applications. Since this involves the processing of large-scale information and continuous information changes, information is clustered for rapid processing. In addition, data size is minimized to overcome the constrained performance associated with mobile devices. Our system has been implemented in actual services, leading to a twofold increase in performance in terms of processing speed and mobile bandwidth.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2010.11a
• /
• pp.72-75
• /
• 2010

Google을 시작으로 포털 사이트들이 Maps API를 공개하면서 맵 응용에 대한 수요가 급증하고 있다. 맵 응용의 개발은 대부분 수작업으로 이루어지고 있기 때문에 맵 응용의 생산성이 떨어질 뿐만 아니라 유지 보수에도 많은 시간과 노력이 요구되고 있다. 우리가 설계하고 구현한 MapAppGen은 ModuleGen, IndexGen, MapGen으로 구성되며 맵 응용을 자동으로 생성한다. MapAppGen은 Google Maps API를 사용하여 CMS(Content Management System) 중의 하나인 Drupal에 적용 가능한 모듈들을 생성해 맵 응용에 대한 생산성을 향상시켜 준다. MapAppGen처럼 맵 응용을 위해 Drupal 모듈을 생성하는 생성기는 현재 존재하지 않으며 Gmap이나 NodeMap과 같이 고정된 맵 인터페이스 모듈들은 존재한다. 그러나 Gmap이나 NodeMap 모듈은 Drupal의 기본 컨텐츠 타입 모듈에 의존하기 때문에 사용자는 원하는 지형/지물 컨텐츠 타입을 생성해 활용할 수 없고 지형/지물들을 유형별로 맵 상에 표시하고 연관된 컨텐츠들을 검색할 수도 없다.

• KIPS Transactions on Software and Data Engineering
• /
• v.2 no.9
• /
• pp.589-594
• /
• 2013

As the concept of web characteristics represented by openness and mind sharing grows more and more popular, device log data generated by both users and developers have become increasingly complicated. For such reasons, a log data processing mechanism that automatically produces meaningful data set from large amount of log records have become necessary for mobile device UX(User eXperience) analysis. In this paper, we define the attributes of to-be-analyzed log data that reflect the characteristics of a mobile device and collect real log data from mobile device users. Along with the MapReduce programming paradigm in Hadoop platform, we have performed a mobile device User eXperience analysis in a distributed processing environment using the collected real log data. We have then demonstrated the effectiveness of the proposed analysis mechanism by applying the various combinations of Map and Reduce steps to produce a simple data schema from the large amount of complex log records.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2006.05a
• /
• pp.177-180
• /
• 2006

체감형 게임을 하기 위한 지형 인터페이스 모델로써 스노우보드 게임에서 사용되는 지형 Map Tool을 구현한다. 게임에 쓰이게 될 3D 지형을 하이트 2D 맵 편직기만을 이용해서 만들기는 힘들며 이러한 방법은 지형의 세밀함이 떨어지고 게임상에서 계속 불러들여 확인하는 것은 비효율적이다. 하이트맵을 3D 상에서 직접 불러와 편집을 하고 저장할 수 있게 된다면 훨씬 효율적이며 퀄리티를 높일 수 있을 것이다.