• Journal of Broadcast Engineering
• /
• v.26 no.1
• /
• pp.70-78
• /
• 2021

Estimating 3D information from a single image is one of the essential problems in numerous applications. Since a 2D image inherently might originate from an infinite number of different 3D scenes, thus 3D reconstruction from a single image is notoriously challenging. This challenge has been overcame by the advent of recent deep convolutional neural networks (CNNs), by modeling the mapping function between 2D image and 3D information. However, to train such deep CNNs, a massive training data is demanded, but such data is difficult to achieve or even impossible to build. Recent trends thus aim to present deep learning techniques that can be trained in a weakly-supervised manner, with a meta-data without relying on the ground-truth depth data. In this article, we introduce recent developments of weakly-supervised deep learning technique, especially categorized as scene 3D reconstruction and object 3D reconstruction, and discuss limitations and further directions.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2003.05a
• /
• pp.527-530
• /
• 2003

In this paper, we implemented the voxel coloring method to reconstruct 3D object from synthetic input Images. Then compare the result between using standard voxel coloring and using coarse-to-fine method. We compared using different voxel space site to see the difference of time processing and the result of 3D object. Photorealistic 3D object reconstruction is a challenging problem in computer graphics. Vexel coloring considered the reconstruction problem as a color reconstruction problem, instead of shape reconstruction problem. This method works by discretizing scene space into yokels, then traversed and colored those in special order. Also there is an extension of voxel coloring method far decreasing the amount of processing time called coarse-to-fine method. This. method works using low resolution instead of high resolution as input and after processing finish, apply some kind of search strategy.

• KIPS Transactions on Software and Data Engineering
• /
• v.4 no.4
• /
• pp.187-194
• /
• 2015

In this paper, we present an effective visual odometry estimation system to track the real-time pose of a camera moving in 3D space. In order to meet the real-time requirement as well as to make full use of rich information from color and depth images, our system adopts a feature-based sparse odometry estimation method. After matching features extracted from across image frames, it repeats both the additional inlier set refinement and the motion refinement to get more accurate estimate of camera odometry. Moreover, even when the remaining inlier set is not sufficient, our system computes the final odometry estimate in proportion to the size of the inlier set, which improves the tracking success rate greatly. Through experiments with TUM benchmark datasets and implementation of the 3D scene reconstruction application, we confirmed the high performance of the proposed visual odometry estimation method.

• 한국HCI학회:학술대회논문집
• /
• 2006.02a
• /
• pp.24-32
• /
• 2006

본 논문에서는 다시점 카메라부터 획득된 부분적인 3D 점군을 사용하여 실내환경의 3D 복원을 위한 새로운 방법을 제안한다. 지금까지 다양한 양안차 추정 알고리즘이 제안되었으며, 이는 활용 가능한 깊이 영상이 다양함을 의미한다. 따라서, 본 논문에서는 일반화된 다시점 카메라를 이용하여 실내환경을 복원하는 방법을 다룬다. 첫 번째, 3D 점군들의 시간적 특성을 기반으로 변화량이 큰 3D 점들을 제거하고, 공간적 특성을 기반으로 주변의 3D 점을 참조하여 빈 영역을 채움으로써 깊이 영상 정제 과정을 수행한다. 두 번째, 연속된 두 시점에서의 3D 점군을 동일한 영상 평면으로 투영하고, 수정된 KLT (Kanade-Lucas-Tomasi) 특징 추적기를 사용하여 대응점을 찾는다. 그리고 대응점 간의 거리 오차를 최소화함으로써 정밀한 정합을 수행한다. 마지막으로, 여러 시점에서 획득된 3D 점군과 한 쌍의 2D 영상을 동시에 이용하여 3D 점들의 위치를 세밀하게 조절함으로써 최종적인 3D 모델을 생성한다. 제안된 방법은 대응점을 2D 영상 평면에서 찾음으로써 계산의 복잡도를 줄였으며, 3D 데이터의 정밀도가 낮은 경우에도 효과적으로 동작한다. 또한, 다시점 카메라를 이용함으로써 수 시점에서의 깊이 영상과 컬러 영상만으로도 실내환경 3D 복원이 가능하다. 제안된 방법은 네비게이션 뿐만 아니라 상호작용을 위한 3D 모델 생성에 활용될 수 있다.

• Journal of Ocean Engineering and Technology
• /
• v.30 no.3
• /
• pp.227-233
• /
• 2016

This paper presents an underwater structure 3D reconstruction method using a 2D multibeam imaging sonar. Compared with other underwater environmental recognition sensors, the 2D multibeam imaging sonar offers high resolution images in water with a high turbidity level by showing the reflection intensity data in real-time. With such advantages, almost all underwater applications, including ROVs, have applied this 2D multibeam imaging sonar. However, the elevation data are missing in sonar images, which causes difficulties with correctly understanding the underwater topography. To solve this problem, this paper concentrates on the physical relationship between the sonar image and the scene topography to find the elevation information. First, the modeling of the sonar reflection intensity data is studied using the distances and angles of the sonar beams and underwater objects. Second, the elevation data are determined based on parameters like the reflection intensity and shadow length. Then, the elevation information is applied to the 3D underwater reconstruction. This paper evaluates the presented real-time 3D reconstruction method using real underwater environments. Experimental results are shown to appraise the performance of the method. Additionally, with the utilization of ROVs, the contour and texture image mapping results from the obtained 3D reconstruction results are presented as applications.

• Journal of Institute of Control, Robotics and Systems
• /
• v.10 no.5
• /
• pp.415-420
• /
• 2004

We present a new algorithm for the calibration of a camera and the recovery of 3D scene structure up to a scale from image sequences using known angles between lines in the scene. Traditional method for calibration using scene constraints requires various scene constraints due to the stratified approach. Proposed method requires only one type of scene constraint of known angle and also it directly recovers metric structure up to an unknown scale from projective structure. Specifically, we recover the matrix that is the homography between the projective structure and the Euclidean structure using angles. Since this matrix is a unique one in the given set of image sequences, we can easily deal with the problem of varying intrinsic parameters of the camera. Experimental results on the synthetic and real images demonstrate the feasibility of the proposed algorithm.

• 제어로봇시스템학회:학술대회논문집
• /
• 2002.10a
• /
• pp.41.2-41
• /
• 2002

The stereo correspondence of two retinal images is one of the most difficult problems in stereo vision because the reconstruction of 3-D scene is a typical visual ill-posed problem. So far there still have been many unsolved problems, one of which is to reconstruct 3-D scene for a monochromatic surface because there is no clue to make a correspondence between two retinal images. We consider this problem with two layered self-organization neural network to simulate the competitive and cooperative interaction of binocular neurons. A...

• Korean Journal of Computational Design and Engineering
• /
• v.20 no.4
• /
• pp.357-366
• /
• 2015

We review recent research results on coupled line cameras (CLC) as a new geometric tool to reconstruct a scene quadrilateral from image quadrilaterals. Coupled line cameras were first developed as a camera calibration tool based on geometric insight on the perspective projection of a scene rectangle to an image plane. Since CLC comprehensively describes the relevant projective structure in a single image with a set of simple algebraic equations, it is also useful as a geometric reconstruction tool, which is an important topic in 3D computer vision. In this paper we first introduce fundamentals of CLC with reals examples. Then, we cover the related works to optimize the initial solution, to extend for the general quadrilaterals, and to apply for cuboidal reconstruction.

• Proceedings of the IEEK Conference
• /
• 2008.06a
• /
• pp.739-740
• /
• 2008

Method for 3D reconstruction from image points and geometric clues can be roughly classified as "model-based" and "constraint-based". We present a new method to reconstruct from one image a scene using depth-point. The our method is benchmarked synthetic data and its effectiveness is shown on photograph data.

• Proceedings of the IEEK Conference
• /
• 2003.11a
• /
• pp.191-194
• /
• 2003

The recovery of 3D scene structure from multiple views has been long one of the central problems in computer vision. This paper presents a new projective reconstruction method based on factorization for un-calibrated image sequences. The proposed algorithm provides an effective measure to construct frame groups by using various information between frames. The experimental results show that the proposed method can reconstruct a more precise 3D structure than the precious methods such as the merging method.