• Journal of Korean Society for Geospatial Information Science
• /
• v.21 no.4
• /
• pp.101-107
• /
• 2013

Generally the commercial high-resolution images have their coordinates, but the locations are locally different according to the pose of sensors at the acquisition time and relief displacement of terrain. Therefore, a process of image co-registration has to be applied to use the multi-temporal images together. However, co-registration is interrupted especially when images include the cloud-covered regions because of the difficulties of extracting matching points and lots of false-matched points. This paper proposes an automatic co-registration method for the cloud-covered high-resolution images. A scale-invariant feature transform (SIFT), which is one of the representative feature-based matching method, is used, and only features of the target (cloud-covered) images within a circular buffer from each feature of reference image are used for the candidate of the matching process. Study sites composed of multi-temporal KOMPSAT-2 images including cloud-covered regions were employed to apply the proposed algorithm. The result showed that the proposed method presented a higher correct-match rate than original SIFT method and acceptable registration accuracies in all sites.

• Journal of the Korea Society of Computer and Information
• /
• v.19 no.11
• /
• pp.43-52
• /
• 2014

In this paper, we propose the target detection method using image displacement, and classification method using SURF(Speeded Up Robust Features) feature points and BAS(Beam Angle Statistics) in infrared images. The SURF method that is a typical correspondence matching method in the area of image processing has been widely used, because it is significantly faster than the SIFT(Scale Invariant Feature Transform) method, and produces a similar performance. In addition, in most SURF based object recognition method, it consists of feature point extraction and matching process. In proposed method, it detects the target area using the displacement, and target classification is performed by using the geometry of SURF feature points. The proposed method was applied to the unmanned target detection/recognition system. The experimental results in virtual images and real images, we have approximately 73~85% of the classification performance.

• Proceedings of the Korean Information Science Society Conference
• /
• 2006.06b
• /
• pp.364-366
• /
• 2006

본 논문에서는 인간의 눈 움직임이 반영된 물체 추적 기능을 모방하여 CCD 카메라를 통하여 실시간으로 입력되는 영상 데이터로부터 특징기반 정합방법을 응용하여 움직임 정보를 추출한 후, 팬-틸트(pan-tilt) 기능의 하드웨어를 제어하여 실시간으로 이동하는 물체를 효율적으로 추적하는 시스템을 제안하였다. 기존의 연구들에서는 주로 물체의 색상값을 이용하여 추적이 이루어지므로 조명이나 카메라의 변화에 따라 이동 물체를 놓치거나 유사한 색의 다른 물체를 잘못 추적하는 문제가 있었다. 이러한 문제점을 해결하기 위하여 측정기반의 정합을 응용하여 이동하는 카메라에서 이동물체를 추출하고 이 이동 물체의 좌표를 이동하여 팬-틸트 하드웨어를 제어하여 추적을 수행하였다. 실험 결과 본 시스템은 움직이는 물체를 감지해서 팬-틸트 하드웨어를 올바르게 제어하며 카메라의 움직임을 보정해가며 전체적으로 움직이는 영상 내에서 실제 움직이는 물체를 일관성 있게 추적하는 만족스러운 결과를 보인다.

• Journal of Korea Multimedia Society
• /
• v.6 no.7
• /
• pp.1178-1188
• /
• 2003

In this paper, we describe image mosaicing techniques for constructing a large high-resolution image with images taken by a video camera in hand. we propose the method which is automatically retrieving the exact matching area using color information and shape information. The proposed method extracts first candidate areas which have similar form using a Voronoi Distance Matching Method which is rapidly estimating the correspondent points between adjacent images, and calculating initial transformations of them and finds the final matching area using color information. It is a method that creates Voronoi Surface which set the distance value among feature points and other points on the basis of each feature point of a image, and extracts the correspondent points which minimize Voronoi Distance in matching area between an input image and a basic image using the binary search method. Using the Levenberg-Marquadt method we turn an initial transformation matrix to an optimal transformation matrix, and using this matrix combine a basic image with a input image.

• Journal of Korean Society for Geospatial Information Science
• /
• v.12 no.2 s.29
• /
• pp.61-66
• /
• 2004

By the advent of the high-resolution Satellite imagery, there are increasing needs in image mosaicking technology which can be applied to various application fields such as GIS(Geographic Information system). To mosaic images, various methods such as image matching and histogram modification are needed. In this study, automated image mosaicking is performed using image matching method based on the multi-resolution wavelet analysis(MWA). Specifically, both area based and feature based matching method are embedded in the multi-resolution wavelet analysis to construct seam line.; seam points are extracted then polygon clipping method are applied to define overlapped area of two adjoining images. Before mosaicking, radiometric correction is proceeded by using histogram matching method. As a result, mosaicking area is automatically extracted by using polygon clipping method. Also, seamless image is acquired using multi-resolution wavelet analysis.

• Journal of the Korea Society of Computer and Information
• /
• v.10 no.3 s.35
• /
• pp.173-182
• /
• 2005

In this paper, we propose a landmark recognition method which is irrelevant to the camera viewpoint on the navigation for localization. Features in previous research is variable to camera viewpoint, therefore due to the wealth of information, extraction of visual landmarks for positioning is not an easy task. The proposed method in this paper, has the three following stages; first, extraction of features, second, learning and recognition, third, matching. In the feature extraction stage, we set the interest areas of the image. where we extract the corner points. And then, we extract features more accurate and resistant to noise through statistical analysis of a small eigenvalue. In learning and recognition stage, we form robust feature models by testing whether the feature model consisted of five corner points is an invariant feature irrelevant to viewpoint. In the matching stage, we reduce time complexity and find correspondence accurately by matching method using similarity evaluation function and Graham search method. In the experiments, we compare and analyse the proposed method with existing methods by using various indoor images to demonstrate the superiority of the proposed methods.

• Journal of KIISE:Computer Systems and Theory
• /
• v.32 no.10
• /
• pp.520-529
• /
• 2005

A feature-based 3D modeling algorithm is presented in this paper. Since conventional methods use depth-based techniques, they need much time for the image matching to extract depth information. Even feature-based methods have less computation load than that of depth-based ones, the calculation of modeling error about whole pixels within a triangle is needed in feature-based algorithms. It also increase the computation time. Therefore, the proposed algorithm consists of three phases, which are an initial 3D model generation, model evaluation, and model refinement phases, in order to acquire an efficient 3D model. Intensity gradients and incremental Delaunay triangulation are used in the Initial model generation. In this phase, a morphological edge operator is adopted for a fast edge filtering, and the incremental Delaunay triangulation is modified to decrease the computation time by avoiding the calculation errors of whole pixels and selecting a vertex at the near of the centroid within the previous triangle. After the model generation, sparse vertices are matched, then the faces are evaluated with the size, approximation error, and disparity fluctuation of the face in evaluation stage. Thereafter, the faces which have a large error are selectively refined into smaller faces. Experimental results showed that the proposed algorithm could acquire an adaptive model with less modeling errors for both smooth and abrupt areas and could remarkably reduce the model acquisition time.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.41 no.6
• /
• pp.247-252
• /
• 2004

This paper proposes a matching algorithm using Voronoi diagram for rotation and translation invariant fingerprint identification. The proposed algorithm extracts geometrical structures that are derived from Voronoi diagram of a fingerprint image. Then two features, distances and angles are extracted from the geometrical structures and saved as indexing form for fingerprint matching. This matching algerian has a lower error rate than indexing based methods of old times.

• The KIPS Transactions:PartB
• /
• v.12B no.3 s.99
• /
• pp.265-272
• /
• 2005

In this paper, we propose a technique which is speedily searching for correspondent points of two images using Voronoi-Distance, as an image registration method for feature based image mosaics. It extracts feature points in two images by the SUSAN corner detector, and then create not only the Voronoi Surface which has distance information among the feature points in the base image using a priority based Voronoi distance algorithm but also select the model area which has the maximum variance value of coordinates of the feature points in the model image. We propose a method for searching for the correspondent points in the Voronoi surface of the base image overlapped with the model area by use of the partitive search algorithm using queues. The feature of the method is that we can rapidly search for the correspondent points between adjacent images using the new Voronoi distance algorithm which has $O(width{\times}height{\times}logN)$ time complexity and the the partitive search algerian using queues which reduces the search range by a fourth at a time.

• Journal of Biomedical Engineering Research
• /
• v.24 no.4
• /
• pp.259-266
• /
• 2003

Non-rigid registration between different modality images with shape deformation can be used to diagnosis and study for inter-patient image registration, longitudinal intra-patient registration, and registration between a patient image and an atlas image. This paper proposes a hierarchical registration method using bodily tissue based segmentation for registration between color images and CT images of the Visible Human leg areas. The cross-sectional color images and the axial CT images are segmented into three distinctive bodily tissue regions, respectively: fat, muscle, and bone. Each region is separately registered hierarchically. Bounding boxes containing bodily tissue regions in different modalities are initially registered. Then, boundaries of the regions are globally registered within range of searching space. Local boundary segments of the regions are further registered for non-rigid registration of the sampled boundary points. Non-rigid registration parameters for the un-sampled points are interpolated linearly. Such hierarchical approach enables the method to register images efficiently. Moreover, registration of visibly distinct bodily tissue regions provides accurate and robust result in region boundaries and inside the regions.