Subject description - AE4M33TDV

Summary of Study | Summary of Branches | All Subject Groups | All Subjects | List of Roles | Explanatory Notes               Instructions
AE4M33TDV 3D Computer Vision Extent of teaching:2+2c
Guarantors:  Roles:PO,V Language of
Teachers:  Completion:Z,ZK
Responsible Department:13133 Credits:6 Semester:Z


This course introduces methods and algorithms for 3D geometric scene reconstruction from images. The student will understand these methods and their essence well enough to be able to build variants of simple systems for reconstruction of 3D objects from a set of images or video, for inserting virtual objects to video-signal source, or for computing ego-motion trajectory from a sequence of images. The labs will be hands-on, the student will be gradually building a small functional 3D scene reconstruction system.

Study targets:

To master conceptual and practical knowledge of the basic methods in 3D computer vision.

Course outlines:

1. 3D computer vision, goals and applications, the course overview
2. Real perspective camera
3. Calibration of real perspective camera
4. Epipolar geometry
5. Computing camera matrices and 3D points from sparse correspondences
6. Autocalibration
7. Consistent multi-camera reconstruction
8. Optimal scene reconstruction
9. Epipolar image rectification
10. Stereoscopic vision
11. Algorithms for binocular stereoscopic matching, multi-camera
12. Shape from shading and contour
13. Shape from texture, defocus, and color
14. Surface reconstruction

Exercises outline:

1. Labs introduction and overview, experimental data, entrance test
2. Camera calibration without radial distortion from a known scene
3. Camera calibration with radial distortion from a known scene
4. Computing epipolar geometry from 8 points
5. Computing epipolar geometry from 7 points, RANSAC
6. Constructing projection matrices from epipolar geometry, computing camera motion and scene structure
7. Autocalibration of intrinsic camera parameters
8. Consistent reconstruction of a many-camera system
9. Accuracy improvement by bundle adjustment
10. Time slot to finish all pending assignments
11. Epipolar rectification for stereoscopic vision
12. Stereoscopic matching by dynamic programming
13. 3D point cloud reconstruction
14. 3D sketch reconstruction


R. Hartley and A. Zisserman. Multiple View Geometry. 2nd ed. Cambridge
University Press 2003.
Y. Ma, S. Soatto, J. Kosecka, S.S. Sastry. An Invitation to 3D
Vision. Springer 2004.


Knowledge equivalent to Geometry for Computer Vision and Graphics and Computer Vision Methods. Detailed up-to-date information on the course at



computer vision, digital image and video processing

Subject is included into these academic programs:

Program Branch Role Recommended semester
MEKME1 Wireless Communication V 3
MEKME5 Systems of Communication V 3
MEKME4 Networks of Electronic Communication V 3
MEKME3 Electronics V 3
MEKME2 Multimedia Technology V 3
MEEEM1 Technological Systems V 3
MEEEM5 Economy and Management of Electrical Engineering V 3
MEEEM4 Economy and Management of Power Engineering V 3
MEEEM3 Electrical Power Engineering V 3
MEEEM2 Electrical Machines, Apparatus and Drives V 3
MEOI3 Computer Vision and Image Processing PO 3
MEKYR4 Aerospace Systems V 3
MEKYR1 Robotics V 3
MEKYR3 Systems and Control V 3
MEKYR2 Sensors and Instrumentation V 3

Page updated 18.6.2019 17:53:02, semester: Z,L/2020-1, L/2018-9, Z,L/2019-20, Send comments about the content to the Administrators of the Academic Programs Proposal and Realization: I. Halaška (K336), J. Novák (K336)