• The KIPS Transactions:PartB
• /
• v.8B no.5
• /
• pp.515-522
• /
• 2001

Feature-based similarity retrieval become an important research issue in image database systems. The features of image data are useful to discrimination of images. In this paper, we propose the highspeed k-Nearest Neighbor search algorithm based on Self-Organizing Maps. Self-Organizing Map(SOM) provides a mapping from high dimensional feature vectors onto a two-dimensional space. A topological feature map preserves the mutual relations (similarity) in feature spaces of input data, and clusters mutually similar feature vectors in a neighboring nodes. Each node of the topological feature map holds a node vector and similar images that is closest to each node vector. We implemented about k-NN search for similar image classification as to (1) access to topological feature map, and (2) apply to pruning strategy of high speed search. We experiment on the performance of our algorithm using color feature vectors extracted from images. Promising results have been obtained in experiments.

• Korean Journal of Environment and Ecology
• /
• v.29 no.3
• /
• pp.410-420
• /
• 2015

In order to estimate the influence of environmental factors on zooplankton communities in lentic freshwater ecosystems, 20 reservoirs and wetlands were monitored by season in 2013. A total of 109 species of zooplankton were identified during the study period. Zooplankton assemblage showed a different distribution in its density and diversity in accordance with the seasons. In particular, the density of zooplankton (98 species and 603ind. L-1) was the most in autumn when compared to the other seasons. In order to effectively analyze zooplankton distribution that are affected by various environmental factors, a Self-Organizing Map (SOM) was used, which extracts information through competitive and adaptive properties. A total of 11 variables (8 environment factors and 3 groups of zooplankton) were patterned on to the SOM. Based on a U-matrix, four clusters were identified from the model. Among zooplankton communities, rotifer displayed a positive relationship with water temperature, and cladocerans and copepod were positively related to conductivity, chlorophyll a, and nutrient factor (i. e. TN and TP). In contrast, high dissolved oxygen appeared to have a negative effect on zooplankton distribution. Consequently, the SOM results depicted a clear pattern of zooplankton density clusters partitioned by environmental factors, which play a key role in determining the seasonal distribution of zooplankton groups in lentic freshwater ecosystem.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.13 no.10
• /
• pp.1981-1986
• /
• 2009

This paper presents new matting techniques. The matting is an alpha estimation technique of object in an image. We can extract the object in an image naturally using the matting technique. The proposed algorithms begin by segmenting an image into three regions: definitely foreground, definitely background, and unknown. Then we estimate foreground, background, and alpha for all pixels in the unknown region. The proposed algorithms learn the definitely foreground and definitely background using self-organizing map(SOM), and estimate an alpha value of each pixel in the unknown region using SOM learning result. SOM matting is distinguished between global SOM matting and local SOM matting by learning method. Experiment results show the proposed algorithms can extract the object in an image.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.5 no.1
• /
• pp.102-109
• /
• 2001

In this paper, in order to reduce the computational complexity of codebook, we propose a fast search algorithm which takes advantage of the information generated in the process of the self-organizing map (SOM). In an attempt to demonstrate the influence of the ordering of codebook on the performance of the partial distance search (PDS), we present the results of computation savings for three cases of ordering of codebooks.

• International Journal of Ocean System Engineering
• /
• v.2 no.1
• /
• pp.57-61
• /
• 2012

In this paper we use the Kohonen neural network based Self Organizing Map (SOM) algorithm for Urdu Character Recognition. Kohenen NN have more efficient in terms of performance as compare to other approaches. Classification is used to recognize hand written Urdu character. The number of possible unknown character is reducing by pre-classification with respect to subset of the total character set. So the proposed algorithm is attempt to group similar character. Members of pre-classified group are further analyzed using a statistical classifier for final recognition. A recognition rate of around 79.9% was achieved for the first choice and more than 98.5% for the top three choices. The result of this paper shows that the proposed Kohonen SOM algorithm yields promising output and feasible with other existing techniques.

• Proceedings of the Korean Information Science Society Conference
• /
• 2005.07b
• /
• pp.622-624
• /
• 2005

대규모 지문인식 시스템에서 비교해야할 지문의 수를 줄이기 위해서 지문분류는 필수적인 과정이다. 최근 이진분류기인 지지 벡터 기계(Support Vector Machine: SVM)를 이용한 지문분류 기법이 많이 연구되고 있다. 본 논문에서는 다중부류 지문분류에 적합하도록 자기 구성 지도(Self-Organizing Map:SOM)를 이용하여 OVA(One-Vs-All) SVM들을 결합하는 지문분류 기법을 제안한다. SOM을 이용하여 OVA SVM들을 동적으로 결합하기 위한 결합 지도를 생성하여 지문분류 성능을 높인다. 지문분류에 있어 대표적인 NIST-4 지문 데이터베이스를 대상으로 Jain이 구축한 FingerCode 데이터베이스에 제안하는 방법을 적용하여 $1.8\%$의 거부율에서 $90.5\%$의 분류율을 획득하였으며, 기존의 결합 방법인 승자독식(Winner-takes-all)과 다수결 투표(Majority vote)보다 높은 성능을 확인하였다.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2014.05a
• /
• pp.73-75
• /
• 2014

본 논문에서는 원본 초음파 영상에서 스케일을 측정한 후, 영상의 확대 비율을 분석하여 충수염 객체의 크기에 대한 범위를 설정한다. 제안된 방법은 초음파 영상에서 ROI 영역을 추출한 후, 사다리꼴 타입의 소속 함수를 이용한 Fuzzy 이진화와 8방향 윤곽선 추적 기법을 적용하여 잡음을 제거한 후에 근막을 추출한다. 추출된 복부 근육의 근막 하단 경계선을 Cubic Spline 보간법을 이용하여 근막의 하단 영역을 추출한다. 초음파 영상의 근막을 기준으로 근막 영역을 제거한 후, SOM(Self-Organizing Map) 알고리즘을 이용하여 충수염의 후보 영역을 추출한다. 추출된 충수염의 후보 영역에 8방향 윤곽선 추적기법을 적용하여 충수염을 추출한다. 제안된 방법을 초음파 영상에 적용하여 실험한 결과, 기존의 충수염 추출 방법보다 충수염 영역이 비교적 정확히 추출되고 충수염의 크기를 측정할 수 있는 것을 실험을 통하여 확인하였다.

• KSCE Journal of Civil and Environmental Engineering Research
• /
• v.29 no.3D
• /
• pp.347-356
• /
• 2009

This paper is described clustering analysis of traffic characteristics-based highway classification in order to deviate from methodologies of existing highway functional classification. This research focuses on comparing the clustering techniques performance based on the total within-group errors and deriving the optimal number of cluster. This research analyzed statistical clustering method (Hierarchical Ward's minimum-variance method, Nonhierarchical K-means method) and Kohonen self-organizing maps clustering method for highway characteristic classification. The outcomes of cluster techniques compared for the number of samples and traffic characteristics from subsets derived by the optimal number of cluster. As a comprehensive result, the k-means method is superior result to other methods less than 12. For a cluster of more than 20, Kohonen self-organizing maps is the best result in the cluster method. The main contribution of this research is expected to use important the basic road attribution information that produced the highway characteristic classification.

• Korean Journal of Fisheries and Aquatic Sciences
• /
• v.41 no.4
• /
• pp.272-281
• /
• 2008

Seasonal and annual variation in the species composition of bag-net catch in the coastal waters of Incheon, Korea were examined from April 2000 to November 2004. To analyze seasonal variation of the fisheries data, we implemented a self-organizing map(SOM), an unsupervised artificial neural network, with the catch amount of 97 species. Over 5 years, we caught 68 species of fish, 23 species of crustaceans and six species of cephalopods. The total number of fish species were gradually increased during the study period. The number of species was higher during the spring than the autumn. The SOM identified four groups of the sampling months based on seasonal changes in communities. In the spring, the dominant species were Leptochela gracilis and Pholis fangi; whereas, in the autumn, Engraulis japonicus and Portunus trituberculatus were dominant species in bag-net catch. Our results will be used to estimate seasonal and annual variation in fisheries resources of Korean coastal waters.

• Journal of Ecology and Environment
• /
• v.38 no.2
• /
• pp.133-143
• /
• 2015

Freshwater macrophytes improve the structural heterogeneity of microhabitats in water, often providing an important habitat for zooplankton. Some studies have focused on the overall influence of macrophytes on zooplankton, but the effects of macrophyte in relation to different habitat characteristics of zooplankton (e.g., epiphytic and pelagic) have not been intensively studied. We hypothesized that different habitat structures (i.e., macrophyte habitat) would strongly affect zooplankton distribution. We investigated zooplankton density and diversity, macrophyte characteristics (dry weight and species number), and environmental parameters in 40 shallow wetlands in South Korea. Patterns in the data were analyzed using a self-organizing map (SOM), which extracts information through competitive and adaptive properties. A total of 20 variables (11 environmental parameters and 9 zooplankton groups) were patterned onto the SOM. Based on a U-matrix, 3 clusters were identified from the model. Zooplankton assemblages were positively related to macrophyte characteristics (i.e., dry weight and species number). In particular, epiphytic species (i.e., epiphytic rotifers and cladocerans) exhibited a clear relationship with macrophyte characteristics, while large biomass and greater numbers of macrophyte species supported high zooplankton assemblages. Consequently, habitat heterogeneity in the macrophyte bed was recognized as an important factor to determine zooplankton distribution, particularly in epiphytic species. The results indicate that macrophytes are critical for heterogeneity in lentic freshwater ecosystems, and the inclusion of diverse plant species in wetland construction or restoration schemes is expected to generate ecologically healthy food webs.