• Journal of Korea Spatial Information System Society
• /
• v.6 no.1 s.11
• /
• pp.19-29
• /
• 2004

Recently, studies have been performed to improve the cache performance of the R-Tree in main memory. A general mothed to improve the cache performance of the R-Tree is to reduce size of an entry so that a node can store more entries and fanout of it can increase. However, this method generally requites additional process to reduce information of entries and do not support incremental updates. In addition, the cache miss always occurs on moving between a parent node and a child node. To solve these problems efficiently, this paper proposes and evaluates the PR-Tree that is an extended R-Tree indexing method using prefetching in main memory. The PR-Tree can produce a wider node to optimize prefetching without additional modifications on the R-Tree. Moreover, the PR-Tree reduces cache miss rates that occur in moving between a parent node and a child node. In our simulation, the search performance, the update performance, and the node split performance of the PR-Tree improve up to 38%. 30%, and 67% respectively, compared with the original R-Tree.

• The KIPS Transactions:PartD
• /
• v.10D no.6
• /
• pp.927-934
• /
• 2003

Database query optimates the selectivety of a query to find the most efficient access plan. Multi-dimensional selectivity estimation technique is required for a query with multiple attributes because the attributes are not independent each other. Histogram is practically used in most commercial database products because it approximates data distributions with small overhead and small error rates. However, histogram is inadequate for a query with multiple attributes because it incurs high storage overhead and high error rates. In this paper, we propose a novel method for multi-dimentional selectivity estimation. Compressed information from a large number of small-sized histogram buckets is maintained using the discrete cosine transform. This enables low error rates and low storage overheads even in high dimensions. Extensive experimental results show adventages of the proposed approach.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.45 no.3
• /
• pp.30-36
• /
• 2008

The multimedia that is characterized by multi-media, multi-features, multi-representations, huge volume, and varieties, is rapidly spreading out due to the increasing of application domains. Thus, it is urgently needed to develop a multimedia information system that can retrieve the needed information rapidly and accurately from the huge amount of multimedia data. For the content-based retrieval of moving picture, picture information is generally used. It is generally used when video is segmented. Through that, it can be a structural video browsing. The tasking that divides video to shot is called video segmentation, and detecting the cut for video segmentation is called cut detection. The goal of this paper is to divide moving picture using HMMD(Hue-Mar-Min-Diff) color model and edge histogram descriptor among the MPEG-7 visual descriptors. HMMD color model is more familiar to human's perception than the other color spaces. Finally, the proposed retrieval system is implemented as hardware.

• Korean Journal of Remote Sensing
• /
• v.35 no.2
• /
• pp.279-288
• /
• 2019

We evaluated the land cover classification performance of SegNet, which features semantic segmentation of aerial imagery. We selected four semantic classes, i.e., urban, farmland, forest, and water areas, and created 2,000 datasets using aerial images and land cover maps. The datasets were divided at a 8:2 ratio into training (1,600) and validation datasets (400); we evaluated validation accuracy after tuning the hyperparameters. SegNet performance was optimal at a batch size of five with 100,000 iterations. When 200 test datasets were subjected to semantic segmentation using the trained SegNet model, the accuracies were farmland 87.89%, forest 87.18%, water 83.66%, and urban regions 82.67%; the overall accuracy was 85.48%. Thus, deep learning-based semantic segmentation can be used to classify land cover.

• Spatial Information Research
• /
• v.18 no.4
• /
• pp.89-98
• /
• 2010

Recently, the target of navigation system is moving from the cars to pedestrians. Many researches are in progress regarding pedestrian navigation, However, in most cases, the path-finding is based on the existing node/link network model. which is widely used for the car navigation, and thus showing its limitation. The reasons arc that a) unlike with a car, the paths that pedestrians take arc not limited to the roads, b) pedestrians an~ not restricted in rotation or direction, and c) they can freely move within the walkable space. No alternatives have been offered yet, especially for openspaces such as a park or square. Therefore, in this research, we suggested appropriate methods to create paths that can be used in pedestrian navigation service, by using motion-planning technology, which is used in the field of robotics for planning the motion of an object, and conducted tests for their applicability.

• Proceedings of the Korean Information Science Society Conference
• /
• 2008.06b
• /
• pp.292-297
• /
• 2008

본 논문은 FPGA 기반 설계에서 주변보다 급격한 온도 변화를 보이는 hotspot들을 탐지하기 위한 열 감지 센서 수를 정하고, 센서의 놓여야 할 배치 장소를 결정하는 알고리즘을 제안한다. 열 감지 센서로는 동적으로 설계가 가능한 ring oscillator 센서 기술을 사용한다는 가정 하에, 센서의 사용 개수를 최소화함과 동시에 최적의 센서 배치 위치 찾는다. 기존의 연구의 단점은 센서가 감지하는 영역 범위를 적당한 크기의 정사각형으로 간주하였기에, 실제 원형의 관측 범위를 보이는 센서 감지 영역의 현실을 올바로 반영하지 못하였으며, 또한 잘 알려진 회로 분할(partition) 기법에 의존한 휴리스틱으로 최적의 결과를 보장하지는 못하였다. 이와는 달리 본 연구에서는 센서의 관측 범위를 원형으로 할 수도 있게 함과 동시에 최적의 해를 보장하는 센서 할당 및 배치 알고리즘을 제안한다. 구체적으로 본 제안 알고즘에서는 소위 “Candidate Coloring 기법”을 통해 센서가 놓여야 할 모든 후보 영역을 표시하며, “Candidate Filtering 기법”을 통해 불필요한 후보 영역들을 완전히 삭제하여 탐색 공간을 줄이게 되며 (해의 최적 해는 항상 유지 되도록 하면서), 마지막으로 Branch-and-Bound 알고리즘을 적용해 최적의 센서 할당 및 배치 결과를 찾아내었다.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.10 no.5
• /
• pp.417-422
• /
• 2000

In this paper, an efficient fuzzy rule generation scheme for adaptive neuro-fuzzy system modeling using the Independent Component Analysis(ICA) as a preprocessing is proposed. Correlation between inputs was not considered in the conventional neuro- fuzzy modeling schemes, such that enormous number of rules and large amount of error were unavoidable. The correlation between inputs is weakened by employing ICA so that the number of rules and the amount of error are reduced. In simulation, the Box-Jenkins furnace data is used to verify the effectiveness of the proposed method.

• Annual Conference of KIPS
• /
• 2014.04a
• /
• pp.196-198
• /
• 2014

최근 차세대 네트워크 패러다임으로 주목받는 소프트웨어 정의 네트워킹 (SDN)에서는 네트워크를 컨트롤 플레인과 데이터 플레인으로 나누고 중앙집중형 제어를 통해 효과적이고 유연한 네트워크 관리를 가능하게 한다. 하지만 잦은 컨트롤 이벤트 발생으로 인한 컨트롤러 및 컨트롤 채널의 부하와 거대한 플로우 엔트리 크기로 인한 스위치 내 TCAM(Temary Content Addressable Memory) 메모리 부족문제 등의 본질적인 문제로 실제 네트워크 적용 시 확장성 문제가 야기된다. 이러한 문제를 해결하기 위해 기존의 연구들은 컨트롤러의 연산능력을 향상시키거나, 컨트롤 이벤트의 발생을 줄이는데 초점이 맞춰져 왔으며, 한정적인 TCAM 공간의 효율적인 사용에 대한 연구는 부족한 상황이다. 따라서 본 논문에서는 효율적인 TCAM 자원 활용을 위한 플로우테이블 관리 기법을 제안한다. 제안 기법은 플로우 엔트리의 클러스터링을 통해 플로우 엔트리를 특성에 따라 그룹화하고 사용빈도를 기준으로 분할 및 병합을 수행함으로써 스위치 내의 가용한 플로우 수를 최대화한다.

• Journal of KIISE:Databases
• /
• v.36 no.2
• /
• pp.99-111
• /
• 2009

In wireless sensor networks, various schemes have been proposed to store and process sensed data efficiently. A Data-Centric Storage(DCS) scheme assigns distributed data regions to sensors and stores sensed data to the sensor which is responsible for the data region overlapping the data. The DCS schemes have been proposed to reduce the communication cost for transmitting data and process exact queries and range queries efficiently. Recently, KDDCS that readjusts the distributed data regions dynamically to sensors based on K-D tree was proposed to overcome the storage hot-spots. However, the existing DCS schemes including KDDCS suffer from Query Hot-Spots that are formed if the query regions are not uniformly distributed. As a result, it causes reducing the life time of the sensor network. In this paper, we propose a new DCS scheme, called TPDCS(Time-Parameterized DCS), that avoids the problems of storage hot-spots and query hot-spots. To decentralize the skewed. data and queries, the data regions are assigned by a time dimension as well as data dimensions in our proposed scheme. Therefore, TPDCS extends the life time of sensor networks. It is shown through various experiments that our scheme outperform the existing schemes.

• Journal of Korean Society of Steel Construction
• /
• v.16 no.1 s.68
• /
• pp.123-134
• /
• 2004

In this research, a multilevel decomposition technique to enhance the efficiency of the configuration optimization of truss structures was proposed. On the first level, the nonlinear programming problem was formulated considering cross-sectional areas as design variables, weight, or volume as objective function and behavior under multiloading condition as design constraint. Said nonlinear programming problem was transformed into a sequential linear programming problem. which was effective in calculation through the approximation of member forces using behavior space approach. Such approach has proven to be efficient in sensitivity analysis and different form existing shape optimization studies. The modified method of feasible direction (MMFD) was used for the optimization process. On the second level, by treating only shape design variables, the optimum problem was transformed into and unconstrained optimal design problem. A unidirectional search technique was used. As numerical examples, some truss structures were applied to illustrate the applicability. and validity of the formulated algorithm.