• Journal of Korea Multimedia Society
• /
• v.15 no.6
• /
• pp.770-783
• /
• 2012

In this paper, we propose a novel structure recovery algorithm in the projective space using image feature points. We use normalized image feature coordinates for the numerical stability. To acquire an initial value of the structure and motion, we decompose the scaled measurement matrix using the singular value decomposition. When recovering structure and motion in projective space, we introduce matrix decomposition constraints. In the reconstruction procedure, a nonlinear iterative optimization technique is used. Experimental results showed that the proposed method provides proper accuracy and the error deviation is small.

• Proceedings of the IEEK Conference
• /
• 2001.09a
• /
• pp.903-906
• /
• 2001

본 논문에서는 조명의 변화에 의해 컬러 영상의 컬러 성분이 달라지더라도 영상 내 컬러간의 편차값을 나타내는 공분산 행렬(covariance matrix)의 고유벡터(eigenvector)와 영상 내 화소들의 컬러 성분과의 상관관계는 거의 변화하지 않는 특징을 이용한 조명 변화에 강인한 영상 검색 방법을 제안한다. 제안된 방법은 영상에서 컬러 성분들의 공분산 행렬과 공분산 행렬의 고유치(eigenvalue), 고유벡터를 계산한 후, 가장 큰 고유치에 관계된 고유벡터로 화소를 투영시키고, 투영된 벡터의 크기 성분으로 영상을 재구성한다. 재구성된 영상으로부터 7개의 불변 모멘트(moment)를 계산하고, 공분산의 가장 큰 고유치를 가중치로 부과하여 특징벡터를 추출한다. 7개의 불변 모멘트로부터 구한 특징벡터는 영상 내 물체의 이동, 영상의 회전, 크기 변화뿐만 아니라, 조명의 변화에 의해 컬러가 변화할 경우에도 유사한 영상을 잘 검색한다. 제안된 방법의 성능 확인을 위하여 5가지 조명에서 얻은 영상 데이터베이스를 이용하여 실험하였으며, 실험 결과 히스토그램 인터섹션에 비해 적은 특징량으로 검색이 가능하면서 조명 변화에도 대응할 수 있는 검색 결과를 얻을 수 있었다.

• Journal of Advanced Navigation Technology
• /
• v.12 no.4
• /
• pp.372-377
• /
• 2008

This paper proposes a new query-based document summarization method using the semantic feature matrix and the semantic variable matrix. The proposed method doesn't need the training phase using training data comprising queries and query specific documents. And it exactly summarizes documents for the given query by using semantic features and semantic variables that is better at identifying sub-topics of document. Because the NMF have a great power to naturally extract semantic features representing the inherent structure of a document. The experimental results show that the proposed method achieves better performance than other methods.

• Proceedings of the Korea Multimedia Society Conference
• /
• 2000.04a
• /
• pp.157-160
• /
• 2000

본 논문은 제약되지않은 카메라에서 얻은 영상중 회전과 크기 변화를 가진 두 장의 디지털 영상을 자동적으로 하나의 통합된 영상으로 모자이킹 하는 방법에 대해 기술한다. 기존에 제시되었던 영상 모자이킹 방법은 영상의 중첩 영역이 많이 존재하거나 회전이 거의 없는 경우만을 고려하고, 카메라 제약이 많이 존재하였다. 우선, 한 쌍의 영상으로부터 각각 특징점을 찾고, 각 특징 점에 대하여 위상을 측정하여 계층적으로 매칭을 하는 방법을 제안한다. 다음으로 비선형 이승오차 최적화 알고리즘을 이용해 최적의 변환행렬을 구한후 , 변환 행렬에 대해 하나의 영상을 만들어내는 과정을 기술한다.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2012.04a
• /
• pp.362-364
• /
• 2012

컬러 정보는 이미지 처리 시스템에서 이미지에 대한 중요한 특징값을 제공한다. 그러나 조명이나 객체의 물리적 특성으로 인하여 녹취된 이미지의 컬러값을 활용하기에는 많은 문제점이 따르게 된다. 이러한 문제점을 해결하기 위하여 컬러불변치에 대한 많은 연구가 있어 왔다. 이 연구에서는 컬러불변치와 병발행렬 특징값과의 관계에 대한 기초 데이터를 제공함으로써 위치기반 어플리케이션에서 이미지 유사도를 측정하는데 활용하고자 한다.

• Korean Journal of Remote Sensing
• /
• v.30 no.6
• /
• pp.769-776
• /
• 2014

It is necessary to detect the feature points existing simultaneously in both images and then find the corresponding relationship between the detected feature points. We propose a new feature detector based on geometric mean of two eigenvalues of gradient matrix which is able to measure the change of pixel intensities. The corner response of the proposed detector is proportional to the geometric mean and also the difference of two eigenvalues in the case of same geometric mean. We analyzed the localization error of the feature detection using aerial image and artificial image with various types of corners. The localization error of the proposed detector was smaller than that of the typical corner detector, Harris detector.

• 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.

• Journal of the Korean Society of Radiology
• /
• v.6 no.3
• /
• pp.167-171
• /
• 2012

In this paper, we present an algorithm to extract features about disease region in digital stomach images. For feature extraction, DCT coefficients of gastrointestinal imaging matrix was obtained. DCT coefficent matrix is concentrated energy in low frequency region, we were extracted 128 feature parameters in low frequency region. Extracted feature parameters can using for differential compression of PACS and, can using for input parameter in CAD.

• Journal of the Institute of Electronics Engineers of Korea CI
• /
• v.48 no.4
• /
• pp.27-34
• /
• 2011

In this paper, it was presented a design on the dynamic multimedia fingerprinting code for anti-collusion code(ACC) in the protection of multimedia content. Multimedia fingerprinting code for the conventional ACC, is designed with a mathematical method to increase k to k+1 by transform from BIBD's an incidence matrix to a complement matrix. A codevector of the complement matrix is allowanced fingerprinting code to a user' authority and embedded into a content. In the proposed algorithm, the feature points were drawing from a content which user bought, with based on these to design the dynamical multimedia fingerprinting code. The candidate codes of ACC which satisfied BIBD's v and k+1 condition is registered in the codebook, and then a matrix is generated(Below that it calls "Rhee matrix") with ${\lambda}+1$ condition. In the experimental results, the codevector of Rhee matrix based on a feature point of the content is generated to exist k in the confidence interval at the significance level ($1-{\alpha}$). Euclidean distances between row and row and column and column each other of Rhee matrix is working out same k value as like the compliment matrices based on BIBD and Graph. Moreover, first row and column of Rhee matrix are an initial firing vector and to be a forensic mark of content protection. Because of the connection of the rest codevectors is reported in the codebook, when trace a colluded code, it isn't necessity to solve a correlation coefficient between original fingerprinting code and the colluded code but only search the codebook then a trace of the colluder is easy. Thus, the generated Rhee matrix in this paper has an excellent robustness and fidelity more than the mathematically generated matrix based on BIBD as ACC.

• Journal of the Institute of Electronics and Information Engineers
• /
• v.54 no.3
• /
• pp.55-62
• /
• 2017

Non-negative matrix factorization (NMF) is one of the typical parts-based representation in which images are expressed as a linear combination of basis vectors that show the lcoal features or objects in the images. In this paper, we represent face images using various NMF methods and recognize their face identities based on extracted features using a learning vector quantization. We analyzed the various NMF methods by comparing extracted basis vectors. Also we confirmed the availability of NMF to the face recognition by verification of recognition rate of the various NMF methods.