• Journal of KIISE:Software and Applications
• /
• v.36 no.8
• /
• pp.653-663
• /
• 2009

In this paper, we propose a blind spots elimination system using three cameras. A wide-angle camera is attached on trunk for eliminating blind spots of a rear-view mirror and two cameras are attached on each side-view mirror for eliminating blind spots of vehicle's sides. In order to eliminate blind spots efficiently, we suggest a method to build a panoramic mosaic view with two side images and one wide-angle rear image. First, we obtain an undistorted image from a wide-angle camera of rear-view and calculate the focus-of-contraction (FOC) in undistorted images of rear-view while the car is moving straight forward. Second, we compute a homography among side-view images and an undistorted image of rear-view in flat road scenes. Next, we perform an image registration process after road and background region segmentation. Finally, we generate various views such as a cylinder panorama view, a top view and an information panoramic mosaic view.

• Journal of IKEEE
• /
• v.11 no.4
• /
• pp.187-192
• /
• 2007

Since wireless sensor networks (WSN) have relatively narrow bandwidth and have limited memory space. Mosaic by inlaying images transmitted by adjacent sensors can provide wider field of view and smaller storage memory. Most WSN are used for surveillance purpose, image acquisition period should be sufficiently short, so that mosaic algorithm has to be run in real time. Proposed algorithm is derived by using the fact that position of sensor nodes are fixed and known. Transformation matrix can be calculated by using distance between sensor nodes and distance between sensor nodes and predefined object. Simulation result shows that proposed algorithm provides very short processing time whereas it preserves image quality.

• Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography
• /
• v.27 no.4
• /
• pp.453-460
• /
• 2009

For constructing 3D information using aerial photograph or video sequences, left and right stereo images having different viewing angle should be prepared in overlapping area. In video sequences, left and right stereo images would be generated by mosaicing left and right slice images extracted in consecutive video sequences. Therefore, this paper is focused on generating left and right stereo mosaic images that are able to construct 3D information and video sequences could be made for the best use. In the stereo mosaic generation, motion parameters between video sequences should be firstly determined. In this paper, to determine motion parameters, free mosaic method using geometric relationship, such as relative orientation parameters, between consecutive frame images without GPS/INS geo-data have applied. After determining the motion parameters, the mosaic image have generated by 4 step processes: image registration, image slicing, determining on stitching line, and 3D image mosaicking. As the result of experiment, generated stereo mosaic image and analyzed result of x, y-parallax have showed.

• Journal of Korean Society for Geospatial Information Science
• /
• v.17 no.3
• /
• pp.49-56
• /
• 2009

Recently, a generation of high quality mosaic images from video sequences has been attempted by a variety of investigations. Among the matter of investigation, in this paper, generation on stereo mosaic utilizing airborne-video sequence images is focused upon. The stereo mosaic is made by creating left and right mosaic which are fabricated by front and rear slices having different viewing angle in consecutive video frames. For making the stereo mosaic, motion parameters which are able to define geometric relationship between consecutive video frames are determined. For determining motion parameters, affine model which is able to explain relative motion parameters is applied by this paper. The mosaicing method using relative motion parameters is called by free mosaic. The free mosaic proposed in this paper consists of 4 step processes: image registration with reference to first frame using affine model, front and rear slicing, stitching line definition and image mosaicing. As the result of experiment, the left and right mosaic image, anaglyphic image for stereo mosaic images are showed and analyzed y-parallax for checking accuracy.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.20 no.3
• /
• pp.630-638
• /
• 2016

In this paper, a feature based panorama image generation algorithm using FAST(Features from Accelerated Segment Test) method that is faster than SIFT(Scale Invariant Feature Transform) and SURF(Speeded Up Robust Features) is proposed. Cylindrical projection is performed to generate natural panorama images with numerous images as input. The occurred error can be minimized by applying RANSAC(Random Sample Consensus) for the matching process. When we synthesize numerous images acquired from different camera angles, we use blending techniques to compensate the distortions by the heterogeneity of border line. In that way, we could get more natural synthesized panorama image. The proposed algorithm can generate natural panorama images regardless the order of input images and tilted images. In addition, the image matching can be faster than the conventional method. As a result of the experiments, distortion was corrected and natural panorama image was generated.

• Journal of Broadcast Engineering
• /
• v.24 no.1
• /
• pp.3-24
• /
• 2019

Techniques related to 360 VR service have been developed to improve the quality of the stitched image and video, where illumination compensation scheme is one of the important tools. Among the conventional illumination compensation algorithms, Gain-based compensation and Block Gain-based compensation algorithms have shown the outstanding performances in the process of making panorama picture. However, those are ineffective in the 360 VR service, because the disparity between illuminations of the multiple pictures in 360 VR is much more than that in making the panorama picture. In addition, the number of the pictures to be stitched in 360 VR system is more than that in the conventional panorama image system. Thus, we propose a preprocessing tool to enhance the illumination compensation algorithm so that the method reduces the degradation in the stitched picture of 360 VR systems. The proposed algorithm consists of 'color compensation' and 'illumination compensation'. The simulation results show that the proposed technique improve the conventional techniques without additional complexity.

• Smart Structures and Systems
• /
• v.25 no.5
• /
• pp.631-641
• /
• 2020

To meet the urgent requirements of safety surveillance from civil engineering management authorities, this study proposes a refined and efficient approach to generate full-field high-resolution panorama of construction sites using camera-amounted UAV (Unmanned Aerial Vehicle). GPS (Global Position System) information extraction for pre-registration, feature points filtering for efficient registration and optimal seaming line seeking for fusion are performed in sequence to form the full-field panorama generation framework. Advantages of the proposed method are as follows. First, GPS information can sort images for pre-registration, avoiding inefficient repeated pairwise calculations and matching. Second, the feature points are filtered according to the characteristics of the construction site images to reduce the amount of calculation. The proposed framework is validated on a road construction site and results demonstrate that it can generate an accurate and high-quality full-site panorama for the safety supervision in a much efficient manner.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2005.05a
• /
• pp.185-188
• /
• 2005

영상 기반 렌더링에서 평면에 투영된 사영 영상만을 가지고 3 차원 영상을 재구성 하는 여러 가지 모델링 기법들이 연구되어 왔다. 4D Plenoptic Function 을 사용하는 Light Field Rendering 이나 Lumigraph 방법은 여러 개의 영상으로 새로운 시점의 영상을 생성하는 기법이다. 이러한 방법은 사용자가 가상 세계에서의 항해를 가능하게 하고 2 차원의 정보만으로 3 차원 환경을 구성 할 수 있다. Concentric Mosaic, Plenoptic Stitching, Sea of Image 등은 Light Field 를 이용하여 사용자가 여러 가지 환경에서 항해할 수 있게 하는 기법이다. 특히 Takahashi 는 도시의 거리와 같은 대규모 환경에서의 항해에 관한 연구를 발표하였다. 단일 경로를 따라 파노라마 영상을 획득한 다음 Light Field 방법을 사용해서 새로운 시점의 영상을 생성한다. 하지만 대규모 환경에서 사용자가 이동할 수 있는 경로의 범위는 매우 넓고 경로를 따라 조밀하게 파노라마 영상을 획득해야 하기 때문에 데이터의 양이 많아지고 영상획득의 어려움이 있다. 이러한 단점으로 인하여 참조 영상의 네트워크 전송 시에 네트워크의 부하가 증가될 수 있다. 본 논문에서는 Takahashi 의 방법을 기본으로 파노라마 영상 모핑 방법을 이용하여 임의 시점 (Arbitrary View)의 영상을 렌더링 하는 방법을 제안한다. 파노라마 영상의 획득 간격을 비교적 크게 하면서 파노라마 영상 모핑 기법을 이용하여 중간 영상을 생성한 후 Takahashi 의 방법을 사용하여 임의 영상을 생성하는 방법이다. 적은 수의 파노라마 영상으로 비교적 좋은 임의 시점의 영상을 재구성 할 수 있었다.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2018.06a
• /
• pp.309-312
• /
• 2018

지금까지 영상 콘텐츠 제작 기술의 발전은 SD(Standard Definition)에서 시작하여 HD(High Definition)와 FHD(Full High Definition)를 거쳐, UHD(Ultra High Definition)에 이르기까지 화질을 중심으로 이루어져 왔다. UHD 에 이르며 육안으로는 그 이상의 해상도로 제작된 콘텐츠와 구분하는 것이 힘들어졌으며, 이에 영상 콘텐츠 제작은 화질이 아닌 제한된 촬영 장비들로부터 촬영 방법, 영상 화각의 개선 작업 등으로 그 방향을 전환하고 있다. 이의 연장선 상에서 360 도 영상에 대한 기술개발이 활발히 이루어 지고 있다. 방송 분야에서는 360 도 영상의 실시간 스트리밍 적용 가능성이 모색되고 있는데, 이것이 가능 하려면 대량의 동영상 데이터를 실시간으로 스티칭하여 전달하는 기술이 필요하다. 따라서 고속 이미지 스티칭이 가능해질 경우 실시간 동영상 스티칭을 통해 방송 통신 분야에서의 서비스 향상에 기여할 것으로 보인다. 본 논문은 이미지의 edge 정보를 방향성을 가진 데이터로 분할하여 특징점을 추출하고, 이후 가중치를 통한 특징점 매칭으로 기존의 이미지 스티칭 방법 보다 빠른 속도의 알고리즘을 제안한다.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2016.06a
• /
• pp.349-351
• /
• 2016

본 논문에서는 VR 영상의 스티칭을 위한 특징점 추출 방식의 하나인 SIFT 알고리즘의 고속화 방법을 제안한다. 이 방법은 SIFT 의 각 단계 모두에 최적화 방법을 적용하여 CPU 에 최적화된 알고리즘을 구축하였다. 그리고 비독립적인 과정들로 이루어진 SIFT 특징점 추출 연산을 병렬화하기 위한 방법으로, 영상 분할 방법을 제시하며 SIFT 의 새로운 병렬화 방법을 제안한다. 특히 최적화 과정을 통해 Scale-space Extrema Detection 과 Orientation Assignment 과정에서 큰 시간 단축 효과를 보여 총 75.5%의 시간을 단축하였다. 이를 OpenMP 와 영상 분할 방법을 활용한 CPU 병렬화로 FullHD($1920{\times}1080$)해상도 영상에서 약 4000 개의 특징점을 추출하는 데 평균 91ms 의 성능을 보이며 기존 GPU 고속화 논문 대비 약 30%의 성능 개선 효과를 보였다.