• Journal of Korea Multimedia Society
• /
• v.9 no.11
• /
• pp.1421-1434
• /
• 2006

Spatial data warehouses provide analytical information for decision supports using SOLAP (Spatial On-Line Analytical Processing) operations. Many researches have been studied to reduce analysis cost of SOLAP operations using pre-aggregation methods. These methods use the index composed of fixed size nodes for supporting the concept hierarchy. Therefore, these methods have many unused entries in sparse data area. Also, it is impossible to support the concept hierarchy in dense data area. In this paper, we propose a dynamic pre-aggregation index method based on the spatial hierarchy. The proposed method uses the level of the index for supporting the concept hierarchy. In sparse data area, if sibling nodes have a few used entries, those entries are integrated in a node and the parent entries share the node. In dense data area, if a node has many objects, the node is connected with linked list of several nodes and data is stored in linked nodes. Therefore, the proposed method saves the space of unused entries by integrating nodes. Moreover it can support the concept hierarchy because a node is not divided by linked nodes. Experimental result shows that the proposed method saves both space and aggregation search cost with the similar building cost of other methods.

• Journal of Korea Multimedia Society
• /
• v.8 no.10
• /
• pp.1347-1359
• /
• 2005

We propose an efficient block classification of the document images using the second-order statistical texture features computed from spatial gray level dependence matrix (SGLDM). We studied on the techniques that will improve the block speed of the segmentation and feature extraction speed and the accuracy of the detailed classification. In order to speedup the block segmentation, we binarize the gray level image and then segmented by applying smoothing method instead of using texture features of gray level images. We extracted seven texture features from the SGLDM of the gray image blocks and we applied these normalized features to the BP (backpropagation) neural network, and classified the segmented blocks into the six detailed block categories of small font, medium font, large font, graphic, table, and photo blocks. Unlike the conventional texture classification of the gray level image in aerial terrain photos, we improve the classification speed by a single application of the texture discrimination mask, the size of which Is the same as that of each block already segmented in obtaining the SGLDM.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2013.06a
• /
• pp.14-15
• /
• 2013

본 논문에서는 공간 계층 분해를 이용한 3 차원 메쉬 시퀀스 압축 기법을 제안한다. 제안하는 기법은 우선 각 점에 대한 시간적 궤적을 공분산 행렬로 표현하고, PCA(Principal component analysis)를 적용하여 시간 궤적에 대한 고유 벡터와 PCA 계수를 획득한다. 공간적인 예측을 통해 PCA 계수에 대한 벡터 차를 추출하고, 벡터 차와 그것에 대한 고유 벡터를 전송한다. 제안하는 방법은 PCA 계수 예측의 성능을 높이기 위해 점진적 압축에서 사용하는 공간 계층 분해 기법을 적용하여, 계수 예측에 효과적인 이웃 점을 지정하도록 한다. 또한, 이웃 점 개수를 사용자가 임의로 지정할 수 있도록 하여, 성능과 복잡도간의 트레이드 오프를 제어할 수 있도록 한다. 다양한 모델에 대한 실험 결과를 통해 제안하는 방법의 성능을 확인한다.

• Proceedings of the Korean Association of Geographic Inforamtion Studies Conference
• /
• 2003.04a
• /
• pp.182-186
• /
• 2003

일반적으로 영상은 공간, 분광 및 시간 해상력을 바탕으로 고해상과 저해상 영상으로 구분된다. 최근 IKONOS 와 QuickBird 등 공간해상력이 1m 이하인 위성 영상들이 국내에 공급되어 바야흐로 고해상 위성영상을 이용한 다양한 활용분야들이 연구되고 있다. 이에 반하여 고분광해상력을 갖는 하이퍼스펙트럴 영상에 대한 연구는 미흡한 실정이다. 국제적으로는 항공기탑재 센서들을 이용한 다양하고 광범위한 조사분석 연구가 이루어지고 있으나, 국내에서는 장비와 관심의 부재로 인하여 초기적인 연구 단계에 있는 실정이다 하이퍼스펙트럴 센서는 환경, 지질, 목표물 인식 분야에 있어 많은 관심을 받고 있으며 위성탑재 초다중분광센서가 운용되기 시작하면서 연구의 활성화가 더욱 기대되고 있다. 본 연구에서는 EO-1 위성의 Hyperion 센서 데이터를 이용하여 노이즈 제거를 위한 영상 전처리 과정을 실시하고 분광특성에 따른 무감독 분류를 통한 인덱싱 기법과 널리 알려진 분광 라이브러리를 활용한 대상물, 특히 인공지물 추출 기법을 실험하였다. 이를 위하여 MNF(Maximum/Minimum Noise Filtering) 변환 및 분광 매칭(Spectral Matching) 기법, 분광 라이브러리 처리 등을 수행하였다. 결과의 비교를 위하여 동일 지역의 Landsat ETM+ 데이터를 이용하여 상호비교를 통한 검증작업으로서 그 성과를 판단하였다.

• Journal of Korean Society for Geospatial Information Science
• /
• v.22 no.1
• /
• pp.23-29
• /
• 2014

The emergence of high resolution satellite images and the evolution of spatial resolution facilitate various studies using high resolution satellite images. Above all, target detection algorithms are effective for monitoring of traffic flow and military surveillance and reconnaissance because vehicles, airplanes, and ships on broad area could be detected easily using high resolution satellite images. Recently, many satellites are launched from global countries and the diversity of satellite images are also increased. On the contrary, studies on comparison about the spatial resolution or target detection, especially, are insufficient in domestic and foreign countries. Therefore, in this study, effects of spatial resolution on target detection are analyzed using the PSO target detection algorithm. The resampling techniques such as nearest neighbor, bilinear, and cubic convolution are adopted to resize the original image into 0.5m, 1m, 2m, 4m spatial resolutions. Then, accuracy of target detection is assessed according to not only spatial resolution but also resampling method. As a result of the study, the resolution of 0.5m and nearest neighbor among the resampling methods have the best accuracy. Additionally, it is necessary to satisfy the criteria of 2m and 4m resolution for the detection of airplane and ship, respectively. The detection of airplane need more high spatial resolution than ship because of their complexity of shape. This research suggests the appropriate spatial resolution for the plane and ship target detection and contributes to the criteria of satellite sensor design.

• Journal of KIISE:Computing Practices and Letters
• /
• v.13 no.5
• /
• pp.241-251
• /
• 2007

Energy consumption in the instruction cacheaccounts for a significant portion of the total processor energy consumption. Therefore, reducing energy consumption in the instruction cache is important in designing embedded processors. This paper proposes a method for reducing dynamic energy consumption in the instruction cache by partitioning it to smaller (less energy-consuming) sub-caches. When a request comes into the proposed cache, only one sub-cache is accessed by utilizing the locality of applications. By contrast, the other sub-caches are not accessed, leading todynamic energy reduction. In addition, the proposed cache reduces dynamic energy consumption by eliminating the energy consumed in tag matching. We evaluated the energy efficiency by running cycle accurate simulator, SimpleScalar. with power parameters obtained from CACTI. Simulation results show that the proposed cache reduces dynamic energy consumption by $37%{\sim}60%$ compared to the traditional direct-mapped instruction cache.

• The KIPS Transactions:PartD
• /
• v.11D no.2
• /
• pp.281-292
• /
• 2004

Selectivity estimation for spatial query is very important process used in finding the most efficient execution plan. Many works have been performed to estimate accurate selectivity. Although they deal with some problems such as false-count, multi-count, they can not get such effects in little memory space. Therefore, we propose a new technique called MW Histogram which is able to compress summary data and get reasonable results and has a flexible structure to react dynamic update. Our method is based on two techniques : (a) MinSkew partitioning algorithm which deal with skewed spatial datasets efficiently (b) Wavelet transformation which compression effect is proven. The experimental results showed that the MW Histogram which the buckets and wavelet coefficients ratio is 0.3 is lower relative error than MinSkew Histogram about 5%-20% queries, demonstrates that MW histogram gets a good selectivity in little memory.

• Proceedings of the Korean Institute of Intelligent Systems Conference
• /
• 2007.04a
• /
• pp.109-112
• /
• 2007

온라인 음악 시장이 점차 커지고 있다. 이에 따라 사용자를 위한 다양한 서비스가 요구되고 있다. 하지만 현재 적용되는 서비스는 통계적인 수치에 기반하는 순위권 나열 혹은 테마나 장르별 음악 소개에 그치고 있다. 따라서 본 논문에서는 사용자의 성향에 가까운 음악을 분석하고 이를 추천하는 방법을 제시한다. 음악 추천 시스템을 위해 우선 사용자의 성향을 분석하기 위하여 사용자가 청취했던 음악의 음파를 분석하여 특성을 추출하여 벡터로 나타낸다. 하지만 추출된 성향과 다른 음악의 성향을 비교해야 하는데 음악의 양이 방대하기 때문에 시간이 오래 걸릴 수 있다. 따라서 이 문제를 해결하기 위해 공간 분할을 통해 검색의 범위를 축소시키고, 음악을 빠르게 추천한다. 실험 결과, 사람의 주관적인 해석이 아닌 음파의 해석을 통해 보다 객관적이고 자동화된 추천 방법을 구현할 수 있었다. 그리고 같은 성질의 음악이 추천되어짐을 확인할 수 있었다.

• Journal of Korea Spatial Information System Society
• /
• v.7 no.3 s.15
• /
• pp.55-65
• /
• 2005

This paper presents a bulk insertion technique for efficiently inserting data items. Traditional moving object database focused on efficient query processing that happens mainly after index building. Traditional index structures rarely considered disk I/O overhead for index rebuilding by inserting data items. This paper, to solve this problem, describes a new bulk insertion technique which efficiently induces the current positions of moving objects and reduces update cost greatly. This technique uses buffering technique for bulk insertion in spatial index structures such as R-tree. To analyze split or merge node, we add a secondary index for information management on leaf node of primary index. And operations are classified to reduce unnecessary insertion and deletion. This technique decides processing order of moving objects, which minimize split and merge cost as a result of update operations. Experimental results show that this technique reduces insertion cost as compared with existing insertion techniques.

• Journal of KIISE:Computing Practices and Letters
• /
• v.16 no.4
• /
• pp.427-431
• /
• 2010

Multidimensional scaling (MDS) is a widely used method for dimensionality reduction, of which purpose is to represent high-dimensional data in a low-dimensional space while preserving distances among objects as much as possible. MDS has mainly been applied to data visualization and feature selection. Among various MDS methods, the classical MDS is not readily applicable to data which has large numbers of objects, on normal desktop computers due to its computational complexity. More precisely, it needs to solve eigenpair problems on dissimilarity matrices based on Euclidean distance. Thus, running time and required memory of the classical MDS highly increase as n (the number of objects) grows up, restricting its use in large-scale domains. In this paper, we propose an efficient approximation algorithm for the classical MDS based on divide-and-conquer and CUDA. Through a set of experiments, we show that our approach is highly efficient and effective for analysis and visualization of data consisting of several thousands of objects.