• Journal of KIISE:Computing Practices and Letters
• /
• v.15 no.8
• /
• pp.611-615
• /
• 2009

Due to the popular use of digital camera, a great number of photos are taken at every usage of camera. It is essential to reveal relationship between photos to manage digital photos efficiently. We propose a method that tessellates image into unit blocks and applies 2D alignment to extend content-based similar region from seed block pair having high similarity. Through an alignment, we can get a block region scoring best matching value on whole image. The method can distinguish whether photos are sharing the same object or background. Our result is less sensitive to transition or pause change of objects. In experiment, we show how our alignment method is applied to real photo and necessities for further research like photo clustering and massive photo management.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2001.10a
• /
• pp.547-550
• /
• 2001

웹상에 공유되어진 문서의 내용을 대표하는 색인어 추출은 정보 검색 시스템의 질을 좌우한다. 한국어의 자유로운 복합명사나 띄어쓰기 규약, 사전 미등록 어휘 등으로 색인어 추출시 질의어와 색인어 사이의 형태상의 불일치(Syntactic Term Mismatch)가 발생하여 검색성능을 저하시키는 경우가 많다. 따라서 본 논문에서는 사전을 통한 형태소 해석을 통해 단위명사(Unit Noun)로 색인어를 추출하고 사전 미등륵어는 N-gram 기반 색인 방법을 이용하여 질의어와 색인어 사이의 부분 일치된 문서도 추출될 수 있는 방법을 제안하였으며, 색인어와 질의어 사이의 유사도 계산을 통해 문서의 우선순위를 정함으로써 색인기의 성능을 높이는 방법을 제안한다.

• Journal of KIISE:Computing Practices and Letters
• /
• v.10 no.6
• /
• pp.529-536
• /
• 2004

The main idea of this paper is to combine position information in sentence and query type classification to make the documents ranking to query more accessible. First, the use of conceptual graphs for the representation of document contents In information retrieval is discussed. The method is based on well-known strategies of text comparison, such as Dice Coefficient, with position-based weighted term. Second, we introduce a method for learning query type classification that improves the ability to retrieve answers to questions from Question Answering system. Proposed methods employ naive bayes classification in machine learning fields. And, we used a collection of approximately 30,000 question-answer pairs for training, obtained from Frequently Asked Question(FAQ) files on various subjects. The evaluation on a set of queries from international TREC-9 question answering track shows that the method with machine learning outperforms the underline other systems in TREC-9 (0.29 for mean reciprocal rank and 55.1% for precision).

• Proceedings of the KIEE Conference
• /
• 1999.11c
• /
• pp.751-753
• /
• 1999

본 논문에서는 영상의 텍스쳐 정보를 이용하여 일반 영상에 대한 내용기반 영상 검색을 수행할 수 있는 알고리듬을 제안한다. Gabor 웨이브렛 변환을 이용하여 Gabor 필터 뱅크 내의 각 필터에 의해 필터링된 대역의 평균과 표준편차를 영상의 특징 벡터(Gabor Texture Feature)로 추출하여 영상들간의 유사성을 계산하는데 사용한다. 논문의 목적이 영상에 가해진 외적 변형, 즉 잡음 첨가, 블러링, 샤프닝 등과 같은 변형에 강인하게 동작할 수 있는 텍스쳐 특징 기반 영상 검색 기법을 제안하는 것이므로, 기존의 Gabor 필터만을 사용하여 텍스쳐 특징을 추출하여 검색의 기준으로 삼을 경우에 발생할 수 있는 주파수 성분의 변화에 대한 민감성을 Daubechies의 웨이브렛 필터를 사용하여 낮은 해상도에서 영상을 해석함으로써, 외적 변형에 대하여도 강인하게 동작할 수 있는 알고리듬을 제시하였다. 기존의 텍스쳐를 이용한 검색이 주로 텍스쳐 영역(textured region)에 대한 해석만을 하였지만, 본 논문에서는 이를 일반 영상에 적용하였으며, 일반 영상에 대해서도 효율적인 검색을 수행할 수 있음을 보였다.

• The KIPS Transactions:PartD
• /
• v.18D no.5
• /
• pp.323-328
• /
• 2011

For content-based image retrieval, the earth mover's distance and the optimal color composition distance are proposed to measure the dissimilarity. Although providing good retrieval results, both methods are too time-consuming to be used in a large image database. To solve the problem, we propose a new distance function that calculates an approximate earth mover's distance in linear time. To calculate the dissimilarity in linear time, the proposed approach employs the space-filling curve. We have performed extensive experiments to show the effectiveness and efficiency of the proposed approach. The results reveal that our approach achieves almost the same results with the EMD in linear time.

• Journal of KIISE:Computing Practices and Letters
• /
• v.8 no.6
• /
• pp.692-704
• /
• 2002

In this paper, we present am image retrieval method based on color-spatial features using quadtree and texture information extracted from object MBRs in an image. Tile proposed method consists of creating a DC image from an original image, changing a color coordinate system, and decomposing regions using quadtree. As such, conditions are present to decompose the DC image, then the system extracts representative colors from each region. And, image segmentation is used to search for object MBRs, including object themselves, object included in the background, or certain background region, then the wavelet coefficients are calculated to provide texture information. Experiments were conducted using the proposed similarity method based on color-spatial and texture features. Our method was able to refute the amount of feature vector storage by about 53%, but was similar to the original image as regards precision and recall. Furthermore, to make up for the deficiency in using only color-spatial features, texture information was added and the results showed images that included objects from the query images.

• Journal of the Korea Society of Computer and Information
• /
• v.10 no.6 s.38
• /
• pp.177-186
• /
• 2005

In this Paper. we propose vision-based camera localization technique using 3D information which is created by mapping of DEM and mountain image. Typically, image features for localization have drawbacks, it is variable to camera viewpoint and after time information quantify increases . In this paper, we extract invariance features of geometry which is irrelevant to camera viewpoint and estimate camera extrinsic Parameter through accurate corresponding Points matching by Proposed similarity evaluation function and Graham search method we also propose 3D information creation method by using graphic theory and visual clues, The Proposed method has the three following stages; point features invariance vector extraction, 3D information creation, camera extrinsic Parameter estimation. In the experiments, we compare and analyse the proposed method with existing methods to demonstrate the superiority of the proposed methods.

• Journal of KIISE:Databases
• /
• v.34 no.4
• /
• pp.304-314
• /
• 2007

Content Based Image Retrieval (CBIR) is to store and retrieve images using the feature description of image contents. In order to support more accurate image retrieval, it has become necessary to develop features that can effectively describe image contents. The commonly used low-level features, such as color, texture, and shape features may not be directly mapped to human visual perception. In addition, such features cannot effectively describe a single image that contains multiple objects of interest. As a result, the research on feature descriptions has shifted to focus on higher-level features, which support representations more similar to human visual perception like spatial relationships between objects. Nevertheless, the prior works on the representation of spatial relations still have shortcomings, particularly with respect to supporting rotational invariance, Rotational invariance is a key requirement for a feature description to provide robust and accurate retrieval of images. This paper proposes a high-level feature named 8AB (8 Angular Bin) that effectively describes the spatial relations of objects in an image while providing rotational invariance. With this representation, a similarity calculation and a retrieval technique are also proposed. In addition, this paper proposes a search-space pruning technique, which supports efficient image retrieval using the 8AB feature. The 8AB feature is incorporated into a CBIR system, and the experiments over both real and synthetic image sets show the effectiveness of 8AB as a high-level feature and the efficiency of the pruning technique.

• Proceedings of the Korean Information Science Society Conference
• /
• 1999.10a
• /
• pp.296-298
• /
• 1999

컬러 히스토그램은 멀티미디어 이미지 데이터의 특성을 표현하기 위하여 널이 이용되어 왔다. 그러나 컬러 히스토그램을 고차원으로 설정할 경우 색인 구조에 효율적이지 못할 뿐만 아니라 유사도 계산에서도 고비용이 요구된다. 이러한 단점을 보완하기 위해 히스트그램의 차원을 줄이는 여러 방법이 제시되어 왔으나 이미지의 색상정보 손실을 피할 수 없으며, 이미지의 전체 히스토그램으로는 이미지의 레이아웃을 고려할 수 없기 때문에 필터링을 통한 후보 선정 시 상이한 이미지가 선택되어지는 문제점을 지닌다. 본 논문에서는 이미지를 일정한 크기의 타일로 분할한 이미지 타일 평균 RGB 방법을 제안하였으며, 실험을 통하여 제안한 방법의 성능을 평가하였다.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2008.11a
• /
• pp.782-784
• /
• 2008

급속하게 확산된 비즈니스 웹 사이트로 인해 웹상에 상품의 정보가 기하급수적으로 증가하여 정보 과부하 문제가 발생하였다. 이를 극복하기 위해 내용 기반 추천 시스템, 협업 필터링 추천 시스템 등의 개인화 추천 시스템이 발전했으나 사용자의 성향과 아이템의 성향을 반영하지 못하고 있다. 본 연구에서는 웹상에서 사용자의 행동을 관찰하여 상품의 구매경로와 판매의 상관관계에 따라 각 사용자의 성향과 그룹의 성향, 아이템의 성향을 측정한 뒤 벡터의 내적을 이용하여 사용자의 성향에 가장 적합한 상품의 유사도를 계산하고 추천하는 시스템을 제안한다.