• 한국멀티미디어학회논문지
• /
• 제16권5호
• /
• pp.619-628
• /
• 2013

이 논문에서는 CAMshift 알고리즘과 칼만 필터(Kalman filter) 알고리즘을 결합하여 강건하게 개선된 추적모듈에 관해서 기술한다. 물체를 추적할 때 사용되는 CAMshift 알고리즘은 추적과정에서 탐색 윈도우를 설정할 때 물체가 이동하는 방향 및 속도를 고려하지 않는다는 단점이 있었다. 이를 해결하기 위해 칼만 필터 알고리즘을 추가한다면 현재 물체의 위치 및 속도 등의 정보를 바탕으로 다음 순간의 물체 위치를 추정할 수 있게 된다. 이 추정값을 기준으로 CAMshift 추적 시 탐색 윈도우를 재설정함으로써, 기존 CAMshift 알고리즘만으로는 추적이 불가능한 고속 이동물체에 대해서도 보다 정확한 추적이 가능하게 되었다. 또 본 연구에서는 추적 대상의 HSV와 YCrCb 두 색상정보를 동시에 고려함으로써 단일 색정보를 이용하는 검출보다 더 강인한 결과를 얻을 수 있었다.

• 대한원격탐사학회지
• /
• 제27권3호
• /
• pp.359-367
• /
• 2011

A suitable pan-sharpening method has to be chosen with respect to the used spectral characteristic of the multispectral bands and the intended application. The research on pan-sharpening algorithm in improving the accuracy of image classification has been reported. For a classification, preserving the spectral information is important. Other applications such as road detection depend on a sharp and detailed display of the scene. Various criteria applied to scenes with different characteristics should be used to compare the pan-sharpening methods. The pan-sharpening methods in our research comprise rather common techniques like Brovey, IHS(Intensity Hue Saturation) transform, and PCA(Principal Component Analysis), and more complex approaches, including wavelet transformation. The extraction of matching pairs was performed through SIFT descriptor and Canny edge detector. The experiments showed that pan-sharpening techniques for spatial enhancement were effective for extracting point and linear features. As a result of the validation it clearly emphasized that a suitable pan-sharpening method has to be chosen with respect to the used spectral characteristic of the multispectral bands and the intended application. In future it is necessary to design hybrid pan-sharpening for the updating of features and land-use class of a map.

• 대한의용생체공학회:의공학회지
• /
• 제27권3호
• /
• pp.110-116
• /
• 2006

Finite element method (FEM) provides several advantages over other numerical methods such as boundary element method, since it allows truly volumetric analysis and incorporation of realistic electrical conductivity values. Finite element mesh generation is the first requirement in such in FEM to represent the volumetric domain of interest with numerous finite elements accurately. However, conventional mesh generators and approaches offered by commercial packages do not generate meshes that are content-adaptive to the contents of given images. In this paper, we present software that has been implemented to generate content-adaptive finite element meshes (cMESHes) based on the contents of MR images. The software offers various computational tools for cMESH generation from multi-slice MR images. The software named as the Content-adaptive FE Mesh Generation Toolbox runs under the commercially available technical computation software called Matlab. The major routines in the toolbox include anisotropic filtering of MR images, feature map generation, content-adaptive node generation, Delaunay tessellation, and MRI segmentation for the head conductivity modeling. The presented tools should be useful to researchers who wish to generate efficient mesh models from a set of MR images. The toolbox is available upon request made to the Functional and Metabolic Imaging Center or Bio-imaging Laboratory at Kyung Hee University in Korea.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제16권6호
• /
• pp.2018-2043
• /
• 2022

Nowadays, COVID-19 infections are influencing our daily lives which have spread globally. The major symptoms' of COVID-19 are dry cough, sore throat, and fever which in turn to critical complications like multi organs failure, acute respiratory distress syndrome, etc. Therefore, to hinder the spread of COVID-19, a Computerized Doughty Predictor Framework (CDPF) is developed to yield benefits in monitoring the progression of disease from Chest CT images which will reduce the mortality rates significantly. The proposed framework CDPF employs Convolutional Neural Network (CNN) as a feature extractor to extract the features from CT images. Subsequently, the extracted features are fed into the Adaptive Dragonfly Algorithm (ADA) to extract the most significant features which will smoothly drive the diagnosing of the COVID and Non-COVID cases with the support of Doughty Learners (DL). This paper uses the publicly available SARS-CoV-2 and Github COVID CT dataset which contains 2482 and 812 CT images with two class labels COVID⁺ and COVI^-. The performance of CDPF is evaluated against existing state of art approaches, which shows the superiority of CDPF with the diagnosis accuracy of about 99.76%.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제16권6호
• /
• pp.1833-1848
• /
• 2022

Named entity recognition (NER) is an important basic task in the field of Natural Language Processing (NLP). Recently deep learning approaches by extracting word segmentation or character features have been proved to be effective for Chinese Named Entity Recognition (CNER). However, since this method of extracting features only focuses on extracting some of the features, it lacks textual information mining from multiple perspectives and dimensions, resulting in the model not being able to fully capture semantic features. To tackle this problem, we propose a novel Multi-view Semantic Feature Fusion Model (MSFM). The proposed model mainly consists of two core components, that is, Multi-view Semantic Feature Fusion Embedding Module (MFEM) and Multi-head Self-Attention Mechanism Module (MSAM). Specifically, the MFEM extracts character features, word boundary features, radical features, and pinyin features of Chinese characters. The acquired font shape, font sound, and font meaning features are fused to enhance the semantic information of Chinese characters with different granularities. Moreover, the MSAM is used to capture the dependencies between characters in a multi-dimensional subspace to better understand the semantic features of the context. Extensive experimental results on four benchmark datasets show that our method improves the overall performance of the CNER model.

• International Journal of Computer Science & Network Security
• /
• 제22권3호
• /
• pp.61-66
• /
• 2022

Object classification is one of the main fields in neural networks and has attracted the interest of many researchers. Although there have been vast advancements in this area, still there are many challenges that are faced even in the current era due to its inefficiency in handling large data, linguistic and dimensional complexities. Powerful hardware and software approaches in Neural Networks such as Deep Neural Networks present efficient mechanisms and contribute a lot to the field of object recognition as well as to handle time series classification. Due to the high rate of accuracy in terms of prediction rate, a neural network is often preferred in applications that require identification, segmentation, and detection based on features. Neural networks self-learning ability has revolutionized computing power and has its application in numerous fields such as powering unmanned self-driving vehicles, speech recognition, etc. In this paper, the experiment is conducted to implement a neural approach to identify numbers in different formats without human intervention. Measures are taken to improve the efficiency of the machines to classify and identify numbers. Experimental results show the importance of having training sets to achieve better recognition accuracy.

• 정보처리학회논문지D
• /
• 제13D권5호
• /
• pp.661-670
• /
• 2006

쉽게 접할 수 있는 정보의 양이 증가하고 전자상거래가 발전함에 따라, 드넓은 정보공간을 축소하기 위하여 추천과 SDI 시스템과 같은 정보 필터링 시스템이 사용되어지게 되었으며, 이에 따라 사용자들은 그들의 요구와 취향에 가장 적합한 정보들을 바로 접근할 수 있게 되었다. 지금까지 다양한 정보 필터링 방법들이 추천시스템을 지원하기 위해 제안되었다. 최근에는 새로운 정보교환 표준으로 떠오르고 있는 XML 문서를 필터링 하는 시스템들에 있어서도 다른 접근 방법을 요구하고 있다. 따라서, 본 논문에서 제안하는 시스템은 XML이 가진 구조 정보를 이용하여 여러 형태의 사용자 프로파일을 생성하는 방법을 제안한다. 시스템은 구매와 같은 트랜잭션이 발생하기 전에 사용자 구매 패턴을 분석하기 위해서 필요한 프로파일을 운영자가 직접 정의하는 운영자 프로파일과 이를 적용한 사용자 프로파일의 두 부분으로 구성된다. 운영자 프로파일은 DTD로부터 선택된 항목을 이용하여 DTD를 따르는 문서내의 특정부분을 가리킬 수 있도록 만들어진다. 제안하는 시스템은 사용자의 구매 행위에 적응력을 가질 수 있도록 보다 정확한 사용자 프로파일을 구축하고, 이와 같은 사용자 프로파일을 기반으로 사용자에게 불필요한 검색과정 없이 필요한 상품 정보를 제공할 수 있도록 한다.

• 융합정보논문지
• /
• 제11권11호
• /
• pp.57-65
• /
• 2021

본 논문은 딥 러닝(Deep Learning)을 이용하여 대기오염측정망 데이터 중 특정 증상이 나타나는 이상 데이터를 탐지하는 방법을 제시한다. 기존 방법들은 일반적으로 시계열 데이터 내에서 기존과는 다른 특이한 패턴이 나타나는 데이터를 탐지하여 이상치로 분류하며, 이는 특정 증상만을 탐지하기에는 적합하지 않다. 본 논문에서는 주로 이미지의 전경 분리(Sementic Segmentation)에 사용되는 DeepLab V3+ 모델의 2차원 합성곱 신경망 구조를 1차원 구조로 변형하여 이미지 대신 여러 센서의 시계열 측정값을 입력받고 특정 증상이 나타나는 데이터를 탐지하도록 하는 방법을 제시한다. 또한, 데이터에 '조각별 집계 근사법(Piecewise Aggregate Approximation)'을 적용하여 잡음이 많은 대기오염측정망 데이터의 복잡도를 줄임으로써 성능을 높인다. 실험 결과를 통해 준수한 성능으로 이상치 탐지를 수행할 수 있음을 확인할 수 있다.

• 한국의류학회지
• /
• 제34권12호
• /
• pp.1980-1991
• /
• 2010

This study proposes the O P E N Triad framework as a future set of tools and perspectives for individual members and institutes to further their professional and academic potential as well as prospect and vitalize the future of the Korean Clothing and Textiles discipline through a global perspective. The millennial generation desires On-demand, Personal, Engaging, and Networked (O P E N) experiences effecting cultural change for creative and influential interaction in transactions, communication, and education. O P E N Individuals offers a WebSphere model as a holistic learning system that has a synergizing value of education across academic courses, industries, and cultures. Through a digitalized and virtualized class, it complements relevant technologies already familiar to the student population. By employing environmental scanning approaches, the most influential and viable future global issues related to the clothing and textiles discipline are identified and dialogued within O P E N Institutes. For future clothing and textiles institutes, this scanning allows them to be open to new ideas, to focus on inter-engagements, to collaborate among individuals, to associate as a part of web of people, organizations, and ideas, to personalize an institutes curricula, and to dialogue generative knowledge. O P E N Industries reveals three dominant future issues that cross academia and industry, sustainability, supply chain management, and social networking. In-depth interviews with U.S. industry experts identified interdependent gaps in global consumer experience practices and suggested the following gaps as future research areas: a standardized business model to the entrepreneurial model, strategic management to a sustainable competitive advantage, standardized to differentiated products, services and operations, market segmentation to global consumer clusters, business-driven marketplaces to consumer-engaged marketspaces, and excellent services to optimal experience. This O P E N Triad framework empowers millennial students, universities, and industries to anticipate and prepare for a radically changing world.

• 대한공간정보학회지
• /
• 제20권1호
• /
• pp.83-91
• /
• 2012

다양한 산림 정보 중 수관 밀도는 단위면적 당 수관점유 면적의 비율로 정의되며, 다양한 분야에 활용되는 중요한 정보이다. 기존의 측정 방법들은 항공사진 판독 또는 현지 조사를 통해 이루어졌다. 이로 인하여 수관 밀도를 측정함에 있어 시간적/인적/경제적 자원의 소모가 크고, 판독자의 주관 및 경험이 반영되어 자료 제작의 일관성이 부족하다. 따라서 본 연구에서는 KOMPSAT-2 고해상도 위성영상을 이용하여 국내 산림 지역의 수관 밀도를 추정하였다. 고해상도 위성영상에 적합한 영역 기반의 수관 밀도를 추정하기 위해 영상 분할 기법과 임분 경계 정보를 이용하여 산림 내부를 일정 영역으로 구분하고, 판별 분석 기법과 산림 비율 기법을 통해 구분된 영역의 수관 밀도를 추정하였다. 현장 조사 및 영상 판독을 통해 구축한 참조자료와 비교해본 결과 판별 분석 기법은 약 60%, 산림비율 기법은 약 85%의 정확도를 보였다. 연구 결과와 수치 임상도의 비교를 통해 갱신이 필요한 후보지 추출 가능성을 확인하였다.