• The KIPS Transactions:PartB
• /
• v.8B no.2
• /
• pp.155-163
• /
• 2001

In this paper, we propose an image registration method by optimization of mutual information to provide a significant infonnation from multimodality images. The method applies mutual infonnation to measure the statistical dependency'r information redundancy between the image intensities of corresponding pixels in both images, which is assumed to be maximal if the images are geometrically aligned. We show the registration results optimizing mutual information between brain MR image and brain CT image and the comparison results with additive gaussian noise. Since our method uses the native image rather than prior segmentation or feature extraction, no user interaction is required and the accuracy of registration is improved. In addition, it shows the robustness against the noise.

• Journal of the Korea Computer Industry Society
• /
• v.2 no.8
• /
• pp.1117-1122
• /
• 2001

Recently, issues on the internet searching of image information through various multimedia databases have drawn an tremendous attention and several researches on image information retrieval methods are on progress. By incorporating wavelet transform and correlation matrixes, we propose a novel and highly efficient feature vector extraction algorithm that has an capability of a robust similarity matching. The simulation results have yielded a faster and highly accurate candidate image retrieval performance in comparison to those of the conventional algorithms. Such an improved performance can be obtained because the used feature vectors were compressed to 256：1 while the correlation matrixes are incorporated to provide a fuel information for the better matching.

• Journal of KIISE:Software and Applications
• /
• v.28 no.11
• /
• pp.855-863
• /
• 2001

We propose an image alignment algorithm based on variable-sized blocks of cylindrical prototype model to generate a face texture for a realistic 3D face model. This is a block matching algorithm which aligns 2D images of a 3D cylindrical model using th correlation between them. While matching blocks, it does not use same sized blocks with considering a curvature of 3D model. And we make a texture of aligned images using a technique of image mosaic. For this purpose, we stitch them with assigning linear weights according to the overlapped region and using the cross-dissolve technique.

• Journal of the Institute of Electronics Engineers of Korea SC
• /
• v.49 no.1
• /
• pp.1-7
• /
• 2012

This paper presents a method for 3D image construction using the hybrid (color and depth) camera system, in which the drawbacks of each camera can be compensated for. Prior to an image generation, intrinsic parameters and extrinsic parameters of each camera are extracted through experiments. The geometry between two cameras is established with theses parameters so as to match the color and depth images. After the preprocessing step, the relation between depth information and distance is derived experimentally as a simple linear function, and 3D image is constructed by coordinate transformations of the matched images. The present scheme has been realized using the Microsoft hybrid camera system named Kinect, and experimental results of 3D image and the distance measurements are given to evaluate the method.

• Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography
• /
• v.28 no.5
• /
• pp.485-491
• /
• 2010

Acquisition of information on failure slope, which may cause apprehension of second hand damage, requires acquisition of fast and accurate topographical data and efficient expression in indirect surveying method without accessing as needed. Therefore, in this study, the images on the intended area were photographed through hovering in the air by approaching collapsed road cut slope with the use of unmanned helicopter photography system. As a result of comparing the points observed by no prism total station and the 10 coordinate points analyzed through image analysis, the averages of absolute values were shown to be 0.056m in X axis direction, 0.082m in Y axis direction and 0.066m Z axis direction. In addition, the RMSE of the error for 10 points of test points were 0.015636m in X axis direction, 0.021319m in Y axis direction and 0.018734m in Z axis direction. Therefore, this method can determine the range of slope and longitudinal and cross sections of each slope in dangerous area that cannot be approached in relational image matching method for the terrains of such collapsed cut slope.

• Proceedings of the Korean Information Science Society Conference
• /
• 2002.10d
• /
• pp.493-495
• /
• 2002

다시점 영상 부호화와 함께, 가상물체를 지능형으로 합성하기 위해서는 정합할 위치를 강인하게 추적(tracking)하여야 한다. 기존의 특징추적기법은 시점이 확장된 시스템에서, 광원의 영향과 가려진 영역의 발생으로 인하여, 추적 할 수 없게 되거나 시각적 혼동이 발생하는 등의 많은 문제점을 유발한다. 뿐만 아니라 다시점에 의한 정보의 급격한 증가는 시점간 움직임을 추적하여 정합하는 연산량의 추가적 발생을 초래한다. 이에 본 논문은 광원이나 가려진 영역(occtusion)의 영향이 적은 형상과 정보량이 많은 텍스쳐를 동시에 고려하여 영역의 변형에 따라 발생하는 최적 움직임을 예측하는 파라미터 모델을 결정함으로써 사각영역을 추적한다. 이 때 미리 연산된 카메라 파라미터와 투영관계를 사용해 시점간에 가상 객체를 지속적으로 3차원 정합한다. 결과적으로, 제안된 알고리즘은 기존의 특징추적 알고리즘에 비해 광원의 영향과 가려진 영역에 강인하며 정합의 오류가 적고 추적 할 때 시점간의 대응을 위한 연산을 줄이기 때문에 실감 혼합현실 시스템의 구현에 매우 효율적임을 실험을 통해 증명하였다.

• Journal of KIISE:Software and Applications
• /
• v.32 no.1
• /
• pp.22-33
• /
• 2005

In this paper. we propose an image registration algorithm based on variable-sized blocks of ellipsoidal prototype model which is similar in shape to human face. While matching blocks, the existing cylindrical prototype model which only takes into account left and right curvature can accomplish a correct alignment on left and right images. But, registration errors are produced from up and down images because the cylindrical prototype model not reflects characteristics of head shape and jaw structure of human. The proposed method is a block matching algorithm which uses variable-sized blocks with considering left-right and up-down curvature of ellipsoidal face model and can correctly align images by using the correlation between them. We then adapt image mosaic technique to generate a face texture from aligned images. For this purpose, we stitch them with assigning linear weights according to the overlapped region and remove ghost effects to make more realistic facial texture.

• Proceedings of the Korean Information Science Society Conference
• /
• 2003.04c
• /
• pp.346-348
• /
• 2003

본 논문에서는 얼굴 형태와 유사한 타원체형 모텔의 가변적 블록크기를 기반으로 하는 영상정렬 알고리즘을 제안한다. 실린더형 모델을 사용한 기존의 방법은 블록정합 수행 시 좌우 곡률을 고려하므로 좌우영상은 정확히 정렬하지만 상하 부분은 사람의 두상모양이나 턱 구조와 다르기 때문에 상하 영상을 정렬할 때 오류가 발생한다. 본 논문에서 제안하는 알고리즘은 얼굴 모델의 상하좌우의 곡률에 따라 가변적인 블록크기론 사용하고 영상의 상관관계를 이용하여 정확한 정렬을 할 수 있도록 한다. 그리고 얼굴 텍스쳐영상을 생성하기 위하여 영상 모자익 기법을 사용하며 영상들이 서로 잘 융합하도록 크로스디졸브(Cross dissolve) 기법을 적용하여 영상을 스티칭한다.

• Journal of Korea Multimedia Society
• /
• v.8 no.3
• /
• pp.345-354
• /
• 2005

Since motion estimation and motion compensation methods remove the redundant data to employ the temporal redundancy in images, it plays an important role in digital video compression. Because of its high computational complexity, however, it is difficult to apply to high-resolution applications in real time environments. If we have a priori knowledge about the motion of an image block before the motion estimation, the location of a better starting point for the search of an exact motion vector can be determined to expedite the searching process. In this paper presents the motion detection algorithm that can run robustly about recusive motion. The motion detection compares and analyzes two frames each other, motion of whether happened judge. Through experiments, we show significant improvements in the reduction of the computational time in terms of the number of search steps without much quality degradation in the predicted image.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.11 no.4
• /
• pp.1254-1259
• /
• 2010

In this paper, Low Computational Adaptive Expanded Block Search Motion Estimation Method is proposed. Proposed method classifies ME blocks as Average Motion Block(AMB) and Local Motion Block(LMB) according to correlation of reference frame. It could reduce the computational complexity with performing Modified Fast Search(MFS). And accuracy of MV is also increased by 4 sub-blocks on LMB and Block Expansion(BE). The experimental results show that the proposed method has better performance that increased 1.8dB than Diamond Search and 0.6dB than Full Search with 7.5 % computation of Full Search. The proposed method could be applied to video compression and Frame Rate Conversion(FRC).