Communication Systems Group Technische Universität Berlin S. Knorr A Geometric Segmentation...
-
Upload
erik-lawrence -
Category
Documents
-
view
214 -
download
0
Transcript of Communication Systems Group Technische Universität Berlin S. Knorr A Geometric Segmentation...
![Page 1: Communication Systems Group Technische Universität Berlin S. Knorr A Geometric Segmentation Approach for the 3D Reconstruction of Dynamic Scenes in 2D.](https://reader035.fdocuments.in/reader035/viewer/2022070404/56649f345503460f94c51435/html5/thumbnails/1.jpg)
Communication Systems Group
Technische Universität Berlin
S. Knorr
A Geometric Segmentation Approach for the 3D Reconstruction of Dynamic Scenes in 2D Video
Sequences
Technical University of BerlinCommunication Systems Group
Director: Prof. Thomas Sikora
Sebastian Knorr, Evren Imre*, A. Aydin Alatan*, and Thomas Sikora
* Middle East Technical UniversityEEE Department
Assoc. Prof. Aydin Alatan
![Page 2: Communication Systems Group Technische Universität Berlin S. Knorr A Geometric Segmentation Approach for the 3D Reconstruction of Dynamic Scenes in 2D.](https://reader035.fdocuments.in/reader035/viewer/2022070404/56649f345503460f94c51435/html5/thumbnails/2.jpg)
EUSIPCO 2006 2Communication Systems GroupS. Knorr
Technische Universität Berlin
Outline
• Motivation• Overview of proposed solution• Feature tracking and motion segmentation• Multiview 3D reconstruction• Simulation results • Future work
![Page 3: Communication Systems Group Technische Universität Berlin S. Knorr A Geometric Segmentation Approach for the 3D Reconstruction of Dynamic Scenes in 2D.](https://reader035.fdocuments.in/reader035/viewer/2022070404/56649f345503460f94c51435/html5/thumbnails/3.jpg)
EUSIPCO 2006 3Communication Systems GroupS. Knorr
Technische Universität Berlin
Motivation
S. Knorr, E. Imre, B. Özkalayci, A. A. Alatan, and T. Sikora, A Modular Scheme for 2D/3D Conversion of TV Broadcast, 3DPVT'06
E. Imre, S. Knorr, A. A. Alatan, and T. Sikora, Prioritized Sequential 3D Reconstruction in Video Sequences of Dynamic Scenes, ICIP'06
![Page 4: Communication Systems Group Technische Universität Berlin S. Knorr A Geometric Segmentation Approach for the 3D Reconstruction of Dynamic Scenes in 2D.](https://reader035.fdocuments.in/reader035/viewer/2022070404/56649f345503460f94c51435/html5/thumbnails/4.jpg)
EUSIPCO 2006 4Communication Systems GroupS. Knorr
Technische Universität Berlin
Motivation
This work:
• 3D reconstruction of dynamic scenes with independently moving objects (IMOs) using Structure from Motion (SfM) techniques
• Scenes are captured with a single camera
• Static background and IMOs are reconstructed independently
Main goal: 3D reconstruction of TV broadcast video
![Page 5: Communication Systems Group Technische Universität Berlin S. Knorr A Geometric Segmentation Approach for the 3D Reconstruction of Dynamic Scenes in 2D.](https://reader035.fdocuments.in/reader035/viewer/2022070404/56649f345503460f94c51435/html5/thumbnails/5.jpg)
EUSIPCO 2006 5Communication Systems GroupS. Knorr
Technische Universität Berlin
Examples
![Page 6: Communication Systems Group Technische Universität Berlin S. Knorr A Geometric Segmentation Approach for the 3D Reconstruction of Dynamic Scenes in 2D.](https://reader035.fdocuments.in/reader035/viewer/2022070404/56649f345503460f94c51435/html5/thumbnails/6.jpg)
EUSIPCO 2006 6Communication Systems GroupS. Knorr
Technische Universität Berlin
Overview of the Proposed Solution
Feature Detection & Tracking
Iterative F-Matrix Estimation &
Motion Segmentation
Initial Structure Computation
Prioritized Sequential Structure
Estimation
![Page 7: Communication Systems Group Technische Universität Berlin S. Knorr A Geometric Segmentation Approach for the 3D Reconstruction of Dynamic Scenes in 2D.](https://reader035.fdocuments.in/reader035/viewer/2022070404/56649f345503460f94c51435/html5/thumbnails/7.jpg)
EUSIPCO 2006 7Communication Systems GroupS. Knorr
Technische Universität Berlin
Outline
• Motivation• Overview of proposed solution• Feature tracking and motion segmentation• Multiview 3D reconstruction• Simulation results • Future work
![Page 8: Communication Systems Group Technische Universität Berlin S. Knorr A Geometric Segmentation Approach for the 3D Reconstruction of Dynamic Scenes in 2D.](https://reader035.fdocuments.in/reader035/viewer/2022070404/56649f345503460f94c51435/html5/thumbnails/8.jpg)
EUSIPCO 2006 8Communication Systems GroupS. Knorr
Technische Universität Berlin
Feature tracking and motion segmentation
• Harris-Corner-Detector for feature selection
• pyramidal Lucas-Kanade tracker to track features along the whole sequence
• Geometric Robust Information Criterion (GRIC) for keyframe selection
kdneGRIC i 212
2D motion model, H (homography), for small baselines vs. 3D motion model, F (epipolar geometry), for large baselines
[P.H.S. Torr, A.W. Fitzgibbon and A. Zisserman (ICCV'98)]
![Page 9: Communication Systems Group Technische Universität Berlin S. Knorr A Geometric Segmentation Approach for the 3D Reconstruction of Dynamic Scenes in 2D.](https://reader035.fdocuments.in/reader035/viewer/2022070404/56649f345503460f94c51435/html5/thumbnails/9.jpg)
EUSIPCO 2006 9Communication Systems GroupS. Knorr
Technische Universität Berlin
Keyframe Selection
240-frame sequence (TUB-room) with 14 keyframes
![Page 10: Communication Systems Group Technische Universität Berlin S. Knorr A Geometric Segmentation Approach for the 3D Reconstruction of Dynamic Scenes in 2D.](https://reader035.fdocuments.in/reader035/viewer/2022070404/56649f345503460f94c51435/html5/thumbnails/10.jpg)
EUSIPCO 2006 10Communication Systems GroupS. Knorr
Technische Universität Berlin
Motion segmentation
• F-matrix estimation for consecutive keyframes (RANSAC and re-RANSAC) labeling of background and IMO trajectories
• Increasing the number of features on the IMO (guided-matching)
021 xFx iT
For each independent motion in the sequence, there exists a corresponding F-matrix, Fi, which
fulfills the epipolar constraint
![Page 11: Communication Systems Group Technische Universität Berlin S. Knorr A Geometric Segmentation Approach for the 3D Reconstruction of Dynamic Scenes in 2D.](https://reader035.fdocuments.in/reader035/viewer/2022070404/56649f345503460f94c51435/html5/thumbnails/11.jpg)
EUSIPCO 2006 11Communication Systems GroupS. Knorr
Technische Universität Berlin
Outline
• Motivation• Overview of proposed solution• Feature tracking and motion segmentation• Multiview 3D reconstruction• Simulation results • Future work
![Page 12: Communication Systems Group Technische Universität Berlin S. Knorr A Geometric Segmentation Approach for the 3D Reconstruction of Dynamic Scenes in 2D.](https://reader035.fdocuments.in/reader035/viewer/2022070404/56649f345503460f94c51435/html5/thumbnails/12.jpg)
EUSIPCO 2006 12Communication Systems GroupS. Knorr
Technische Universität Berlin
Multiview 3D reconstruction
• Fast convergence to a reliable estimate:Since the quality of subsequent reconstructions depend on the current (intermediate) structure estimate, errors in the first few frame pairs may cause the entire estimation procedure to collapse.
• Fast recovery of the scene structure: The number of reconstructed 3-D points should be maximized, while processing a minimum number of frame pairs.
Main goals:
![Page 13: Communication Systems Group Technische Universität Berlin S. Knorr A Geometric Segmentation Approach for the 3D Reconstruction of Dynamic Scenes in 2D.](https://reader035.fdocuments.in/reader035/viewer/2022070404/56649f345503460f94c51435/html5/thumbnails/13.jpg)
EUSIPCO 2006 13Communication Systems GroupS. Knorr
Technische Universität Berlin
Prioritized Sequential Reconstruction
1. Compute the initial reconstruction and camera path (e.g. Pollefeys et al.).
2. Compute the priority metric and order the frame pairs.
Given the internal calibration parameters and the feature trajectories:
2-view rec.1:6
2-view rec.2:4
2-view rec.7:9
Add 5, 2-viewrec. 1:5
Add 3, 2-viewrec. 3:4
Merge sub-reconstruction
3:6
Add 8, 2-viewrec. 3:8
Merge sub-reconstruction
7:8
Two-view reconstruction
Sub-estimate fusion using 3D-2D correspondencesSub-reconstruction fusion using 3D-3D correspondences
Ordered pair list: 1:6, 2:4, 1:5, 3:4, 3:6, 7:9, 3:8, 7:8
![Page 14: Communication Systems Group Technische Universität Berlin S. Knorr A Geometric Segmentation Approach for the 3D Reconstruction of Dynamic Scenes in 2D.](https://reader035.fdocuments.in/reader035/viewer/2022070404/56649f345503460f94c51435/html5/thumbnails/14.jpg)
EUSIPCO 2006 14Communication Systems GroupS. Knorr
Technische Universität Berlin
Outline
• Motivation• Overview of proposed solution• Feature tracking and motion segmentation• Multiview 3D reconstruction• Simulation results • Future work
![Page 15: Communication Systems Group Technische Universität Berlin S. Knorr A Geometric Segmentation Approach for the 3D Reconstruction of Dynamic Scenes in 2D.](https://reader035.fdocuments.in/reader035/viewer/2022070404/56649f345503460f94c51435/html5/thumbnails/15.jpg)
EUSIPCO 2006 15Communication Systems GroupS. Knorr
Technische Universität Berlin
Simulation Results (1)
frame 1
frame 170
![Page 16: Communication Systems Group Technische Universität Berlin S. Knorr A Geometric Segmentation Approach for the 3D Reconstruction of Dynamic Scenes in 2D.](https://reader035.fdocuments.in/reader035/viewer/2022070404/56649f345503460f94c51435/html5/thumbnails/16.jpg)
EUSIPCO 2006 16Communication Systems GroupS. Knorr
Technische Universität Berlin
Simulation Results (2)
frame 1
frame 200
![Page 17: Communication Systems Group Technische Universität Berlin S. Knorr A Geometric Segmentation Approach for the 3D Reconstruction of Dynamic Scenes in 2D.](https://reader035.fdocuments.in/reader035/viewer/2022070404/56649f345503460f94c51435/html5/thumbnails/17.jpg)
EUSIPCO 2006 17Communication Systems GroupS. Knorr
Technische Universität Berlin
Future Work
2D TV-Display
Autostereoscopic Display
Shutter-Glases
Anaglyph
![Page 18: Communication Systems Group Technische Universität Berlin S. Knorr A Geometric Segmentation Approach for the 3D Reconstruction of Dynamic Scenes in 2D.](https://reader035.fdocuments.in/reader035/viewer/2022070404/56649f345503460f94c51435/html5/thumbnails/18.jpg)
EUSIPCO 2006 18Communication Systems GroupS. Knorr
Technische Universität Berlin
Future Work
initial dense depth map refined depth map virtual right stereo view
original left stereo view
![Page 19: Communication Systems Group Technische Universität Berlin S. Knorr A Geometric Segmentation Approach for the 3D Reconstruction of Dynamic Scenes in 2D.](https://reader035.fdocuments.in/reader035/viewer/2022070404/56649f345503460f94c51435/html5/thumbnails/19.jpg)
EUSIPCO 2006 19Communication Systems GroupS. Knorr
Technische Universität Berlin
Future Work
original left stereo view virtual right stereo view
![Page 20: Communication Systems Group Technische Universität Berlin S. Knorr A Geometric Segmentation Approach for the 3D Reconstruction of Dynamic Scenes in 2D.](https://reader035.fdocuments.in/reader035/viewer/2022070404/56649f345503460f94c51435/html5/thumbnails/20.jpg)
EUSIPCO 2006 20Communication Systems GroupS. Knorr
Technische Universität Berlin
Thank you for your attention!
Questions?
?
![Page 21: Communication Systems Group Technische Universität Berlin S. Knorr A Geometric Segmentation Approach for the 3D Reconstruction of Dynamic Scenes in 2D.](https://reader035.fdocuments.in/reader035/viewer/2022070404/56649f345503460f94c51435/html5/thumbnails/21.jpg)
EUSIPCO 2006 21Communication Systems GroupS. Knorr
Technische Universität Berlin
Motion segmentation
• For each independent motion in the sequence, there exists a corresponding F-matrix, Fi, which fulfills the epipolar constraint
where x1 and x2 are corresponding points in two views.
• A RANSAC-based F-matrix estimation algorithm identifies the feature pairs belonging to the dominant motion and labels the rest as outliers.
• Some of the outliers should satisfy the epipolar constraint according to a new F-matrix, which corresponds to the motion of an independent moving object (IMO)
021 xFx iT
![Page 22: Communication Systems Group Technische Universität Berlin S. Knorr A Geometric Segmentation Approach for the 3D Reconstruction of Dynamic Scenes in 2D.](https://reader035.fdocuments.in/reader035/viewer/2022070404/56649f345503460f94c51435/html5/thumbnails/22.jpg)
EUSIPCO 2006 22Communication Systems GroupS. Knorr
Technische Universität Berlin
Segmentation Algorithm
1. Compute the F-matrix corresponding to the first and the second key-frame by using a RANSAC-based procedure and label the inliers as background trajectories.
2. Compute the F-matrix on the outliers of step 1 by using again RANSAC and label the inliers as IMO trajectories.
3. Compute the centroid of the inliers of step 2 and check their distances. If the distance is higher than a threshold, reject the feature.
4. Increase the number of features on the IMO in consecutive key-frames with guided-matching.
5. Repeat step 2 to 4 as long as the F-matrix estimation is still reliable and most of the remaining features are spatially close.
6. Proceed to the next key-frame. Estimate the F-matrix between the last and the current key-frame for each motion using the labeled trajectories and classify new trajectories using step 1 to 5.
7. Repeat step 6 for all key-frames.
![Page 23: Communication Systems Group Technische Universität Berlin S. Knorr A Geometric Segmentation Approach for the 3D Reconstruction of Dynamic Scenes in 2D.](https://reader035.fdocuments.in/reader035/viewer/2022070404/56649f345503460f94c51435/html5/thumbnails/23.jpg)
EUSIPCO 2006 23Communication Systems GroupS. Knorr
Technische Universität Berlin
Priority Metric
cnb
adp
exp1
The priority metric, p, utilized in the algorithm to evaluate the feasibility of a frame pair for reconstruction is defined as
where d is the baseline distance between the cameras, n the number of feature matches, a, b and c are the design parameters of the sigmoid function appearing in the second term. The non-linear (sigmoidal) weighting keeps the contribution of the second term within a bound, when there is a relatively small or large number of matching features.
![Page 24: Communication Systems Group Technische Universität Berlin S. Knorr A Geometric Segmentation Approach for the 3D Reconstruction of Dynamic Scenes in 2D.](https://reader035.fdocuments.in/reader035/viewer/2022070404/56649f345503460f94c51435/html5/thumbnails/24.jpg)
EUSIPCO 2006 24Communication Systems GroupS. Knorr
Technische Universität Berlin
Simulation Results (3)
# frame pairs
(used/total)
# 3D points
(recov./total)
reproj. error
Cliff 45 / 5778 5890 / 8212 0.95
Palace 25 / 19900 2771 / 3546 0.90
TUB-Room
17 / 28680 4716 / 6095 0.27
![Page 25: Communication Systems Group Technische Universität Berlin S. Knorr A Geometric Segmentation Approach for the 3D Reconstruction of Dynamic Scenes in 2D.](https://reader035.fdocuments.in/reader035/viewer/2022070404/56649f345503460f94c51435/html5/thumbnails/25.jpg)
EUSIPCO 2006 25Communication Systems GroupS. Knorr
Technische Universität Berlin
Segmentation Results
frame 83frame 1
frame 201frame 170