Automatic Dense Reconstruction from Uncalibrated Video Sequences. Front Cover. David Nistér. KTH, – pages. Automatic Dense Reconstruction from Uncalibrated Video Sequences by David Nister; 1 edition; First published in aimed at completely automatic Euclidean reconstruction from uncalibrated handheld amateur video system on a number of sequences grabbed directly from a low-end video camera The views are now calibrated and a dense graphical.

Author: Tebar Gushakar
Country: Algeria
Language: English (Spanish)
Genre: Career
Published (Last): 28 April 2009
Pages: 226
PDF File Size: 17.20 Mb
ePub File Size: 5.77 Mb
ISBN: 305-9-88569-853-8
Downloads: 1672
Price: Free* [*Free Regsitration Required]
Uploader: Shaktizragore

This thesis describes a system that completely automaticallybuilds a three-dimensional model of a scene given a sequence ofimages of the scene. Then, after depth—map refinement and depth—map fusion, a dense 3D point data cloud can be obtained. Because of the rapid development of the unmanned aerial vehicle UAV industry in recent years, civil UAVs have been used in agriculture, energy, environment, public safety, infrastructure, and other fields.

The weight is w j after an experimental comparison, vidoe value of 20 is suitable for w j.

The running times of the algorithm are recorded in Table 2and the precision is 1 s. The flight height is about 15 m from the ground and is kept unchanged. The first step involves recovering the 3D structure of the scene and the camera motion from the images. The process is illustrated in Figure 3. Different color means different value of distance. There are several improved SfM methods such as the method proposed by Wu [ 814 ].

The flight blocks are integrated for many parallel strips. The calculation of distance is performed only on the common part of the two point clouds. SLAM mainly consists in the simultaneous estimation of the localization of the robot and the map of the environment.

Maxime Lhuillier’s home page

The main contribution of the thesis is in building acomplete system and applying it to full-scale real worldproblems, thereby facing the practical difficulties of far fromideal imagery.

Then, the structure of the images in the queue is computed, and the queue is updated with new images. The fundamental uncapibrated of the two images is obtained by the random sample consensus RANSAC method [ uncalibeated ], and the essential matrix between the two images is then calculated when the intrinsic matrix obtained by the calibration method proposed in [ 23 ] is known. Distinctive image features from scale-invariant keypoints. Discrete-continuous optimization for large-scale structure from motion.


In contrast, P c can be used in the later weighted bundle adjustment to ensure the continuity of the structure. This problem can be addressed by using control points, which are the points connecting two sets of adjacent feature points of the image, as shown in Figure 5. Finally, dense 3D point cloud data of the scene are obtained by using depth—map fusion. The accuracy of our result is almost the same as result of openMVG and MicMac, but the speed of our algorithm is faster than them.

In order to reconstruct the 3D structure of scenes using image sequences, we propose a rapid and accurate 3D reconstruction method based on an image queue. The general 3D reconstruction algorithm without a priori positions and orientation information can be roughly divided into two steps.

With the continuous development of computer hardware, multicore technologies, and GPU technologies, the SfM algorithm can now be used in several areas.

Adaptive structure from motion with a contrario model estimation; Proceedings of the Asian Conference on Computer Vision; Daejeon, Korea.

Automatic dense reconstruction from uncalibrated video sequences | BibSonomy

The following matrix is formed by the image coordinates of the feature points:. As is shown in Figure 18 c, the 3D point cloud is generated by depth—map fusion. First, a principal component analysis method of the feature points is zequences to select the esquences images suitable for 3D reconstruction, which ensures that the algorithm improves the calculation speed with almost no loss of accuracy.

The algorithm first obtains the feature points in the structure calculated by the SfM.

Then, the mesh is used as an outline of the object, which is projected onto the plane of the images to obtain the estimated depth maps. And Figure 7 d is standard point cloud provided by roboimagedata.

This article has been cited by other articles in PMC. Equation 9 is the reprojection error formula of the weighted bundle adjustment. Next, the image queue is updated, several images are deleted from the front of the queue, and the same number of images is placed at the end of the queue.


In order to reduce the computational complexity of feature point matching, we propose a method of reconstructoon the feature points based on principal component analysis PCA. One motivation is to make it possible forany amateur photographer to produce graphical models of theworld with the use of a computer. This step is usually completed by generating a dense point data cloud or mesh data cloud from multiple images. Speed Evaluation In order to test the speed of the proposed algorithm, we compared the time consumed by our method with those consumed by openMVG and MicMac.

National Center for Biotechnology InformationU. Incremental smoothing and mapping using the Bayes tree. By carrying a digital camera on a UAV, two-dimensional 2D images can be obtained. In this case, the UAV flight is over a factory land. A paradigm for model fitting with applications to image analysis and automated cartography.

Figure 4 illustrates the process of the sequeences. This task is frequently carried outin movie making but is then performed with a great deal ofexpensive manual work. The main text gives a detailed coherent account of thetheoretical foundation for the system and its components. Finally, unxalibrated structure of all of the images can be calculated by repeating the following two procedures alternately: Second, these key images are inserted into a fixed-length image queue.

After the first update of the image queue, the formula for the projection error of the bundle adjustment used in step 6 will be altered. The calculation of the bundle adjustment is a nonlinear least-squares problem. Author information Article notes Copyright reconstfuction License information Disclaimer. The UAV is launched from the ground and flies over the house.

In addition, the research into Real-time simultaneous localization and mapping SLAM and 3D reconstruction of the environment have become popular over the past few years. The study of the methods in which 3D structures are generated by 2D images is an important branch of computer vision.