CE6451: Introduction to Photogrammetry
Lecture 1
Introduction
Acknowledgements: part of the materials of the all the lecture notes are
from Cyrill Stachniss, Ping Tan, Marc Pollefey, Wolfgang Foerstner,
Bernhard Wrobel, James Hays, A. Dermanis, Armin Gruen, Alper
Yilmaz.
Instructor: Rongjun Qin, Ph.D. Page 1
Introduction – Photogrammetry
“Photogram” --- photograph
“metry” --- Geometry
Mensuration with light!
-Photogrammetry is the science of making
measurements from photographs,
especially for recovering the exact
positions of surface points.
Early user - Surveyors
Instructor: Rongjun Qin, Ph.D. Page 7
Photogrammetry – Brief History
Albrecht Dürer, in 1525,
Perspective Drawing
1840 - Dominique François Jean Arago, daguerreotype
1855 - Nadar(Gaspard Felix Tournachon) Air balloon,
first aerial photograph
1882 – photograph with kites E. D. Archibald
1909 – First aerial photograph Wrights
Instructor: Rongjun Qin, Ph.D. Page 8
Photogrammetry – Brief History
New
LIDAR, X-Ray, electro-magnetic imaging !
millennium
Digital
1974 Invention of CCD cameras
Analytical
1942 Invention of computer
Analog
1913 Invention of plane
First generation
Plane Table
Photogrammetry
1840 Invention of photography
Instructor: Rongjun Qin, Ph.D. Page 9
Plane Table Photogrammetry
1850 – 1900
Instructor: Rongjun Qin, Ph.D. Page 10
Analog Photogrammetry
1900-1960
Orientation being mechanical movement
Instructor: Rongjun Qin, Ph.D. Page 11
Analytical Photogrammetry
1960-2010
Orientation being
computed by
computers, as well as
the intersection of rays
Instructor: Rongjun Qin, Ph.D. Page 12
Digital Photogrammetry
1993 - now
All replaced
by digital
devices
Automated
topography
generation
Imag courtesy: http://www.gpsi-corp.com
Instructor: Rongjun Qin, Ph.D. Page 13
Digital Photogrammetry
Fully Automated Digital Surface Model Generation
Google Earth
Instructor: Rongjun Qin, Ph.D. Page 14
Computer Vision
A science-driven discipline that interpret the
physical environment from images, and
simulating human vision systems and
intelligence
Does not focus on particular applications
Instructor: Rongjun Qin, Ph.D. Page 15
Computer Vision – Examples of Topic
Instructor: Rongjun Qin, Ph.D. Page 16
Computer Vision – Examples of Topic
Instructor: Rongjun Qin, Ph.D. Page 17
Computer Vision – Examples of Topic
Instructor: Rongjun Qin, Ph.D. Page 18
Computer Vision – Examples of Topic
Instructor: Rongjun Qin, Ph.D. Page 19
Computer Vision – Examples of Topic
Instructor: Rongjun Qin, Ph.D. Page 20
Computer Vision – Examples of Topic
Instructor: Rongjun Qin, Ph.D. Page 21
Computer Vision – Remarks
T. S. Huang
Practically successful computer Vision algorithm
may :
1) Allow Human Interaction
2) Forgiving
Instructor: Rongjun Qin, Ph.D. Page 22
Computer Vision – A Brief Intro
1960s, Larry Roberts Recover 3D from synthetic 2D views
(MIT)
Extract primal sketches of real-world images, David Marr,
1978 (MIT). Tasks such as segmentation, edge detection,
low-level vision
1980s, Artificial Neural networks. 3D recovery from 2D
images, mathematical model development. Low-level
vision
1990s, face recognition, object recognition, tracking of
moving objects, high-level vision
2000s, large annotated datasets, video data processing,
structure from motion.
2012s, deep convolutional neural network, applied to
both low-level and high-level vision problem
Snavely 06,07
Russakovsky et
al 2015
Instructor: Rongjun Qin, Ph.D. Page 23
Photogrammetry & Computer Vision
We aim to merge the mensuration nature of
photogrammetry methods and intelligent
image understanding techniques with a major
focus on perceiving, understanding and
measuring the Physical environment
Instructor: Rongjun Qin, Ph.D. Page 24
Synergy and Difference
Photogrammetry Computer Vision
Mensuration, Topographic Mapping Achieving human vision capability
Mature, developed with the need of Not particularly developed
surveying industry
Application mainly cover large scale, e.g. Application mainly cover intelligent systems,
aerial surveillance, earthquake, flood, GIS, drive-less car, speech recognition, human face
manufacturing detection, movie, robotic navigation.
Rarely deal with top-view data
Engineering driven Largely science-driven
All kinds: expensive metric camera, satellite dealing with consumer grade cameras
camera, mobile. Including laser scanning
3D vision and photogrammetry share very similar theoretical basis, though formulation might
be different
Mapping requires intelligence and computer vision demand for accuracy
Instructor: Rongjun Qin, Ph.D. Page 25
Connections
Image courtesy: Foerstner and Wrobel
Instructor: Rongjun Qin, Ph.D. Page 26
Common Applications
Large-scale semantic 3D reconstruction of the
earth object.
Precision navigation using images and multi-
sensory data
Simultaneous localization and mapping
3D object tracking
Virtual / Augmented Reality
…
Instructor: Rongjun Qin, Ph.D. Page 27
Civil Applications
Disaster Response
Instructor: Rongjun Qin, Ph.D. Page 28
Civil Applications – Flood Simulation
Instructor: Rongjun Qin, Ph.D. Page 29
Civil Applications – Car tracking
Instructor: Rongjun Qin, Ph.D. Page 30
Environmental Applications
Tree type identification
Instructor: Rongjun Qin, Ph.D. Page 31
Machine Learning on Satellite Images
Instructor: Rongjun Qin, Ph.D. Page 32
Machine Learning on Satellite Images
Instructor: Rongjun Qin, Ph.D. Page 33
Heritage Preservation
Instructor: Rongjun Qin, Ph.D. Page 34
Autonomous Driving
Google Car, Courtesy of Google
Instructor: Rongjun Qin, Ph.D. Page 35
Visual Navigation
From Cyrill Stachniss
Instructor: Rongjun Qin, Ph.D. Page 36
3D Modeling for Virtual Reality
Instructor: Rongjun Qin, Ph.D. Page 37
Geometry-Aware Satellite-to-Ground Image Synthesis for Urban Areas ,CVPR 2020
Instructor: Rongjun Qin, Ph.D. Page 38
Instructor: Rongjun Qin, Ph.D. Page 39
Instructor: Rongjun Qin, Ph.D. Page 40
Instructor: Rongjun Qin, Ph.D. Page 41
Cross-view SLAM
Instructor: Rongjun Qin, Ph.D. Page 42
Registration between point clouds
Instructor: Rongjun Qin, Ph.D. Page 43
Course Content
Introduction
Sensors and Radiometry
Features
Camera Geometry
Two-view Geometry
Bundle Adjustment
A brief intro to linear array cameras
Surface Reconstruction
Instructor: Rongjun Qin, Ph.D. Page 44
Radiometry and Features
Photometric Stereo
Feature point matching
Credits: wiki & opencv
Instructor: Rongjun Qin, Ph.D. Page 45
Perspective Geometry
How to measure the height of a person with a single image?
Credits: nzdl.com
Instructor: Rongjun Qin, Ph.D. Page 46
Image Orientation
Aero-triangulation
Image courtesy: Ackermann
Pose estimation in structure from motion
Instructor: Rongjun Qin, Ph.D. Page 47
Surface Reconstruction from Oriented Images
Credit: CMU
Instructor: Rongjun Qin, Ph.D. Page 48
Course logistics
Text Book: Lecture notes; Reference list;
Wolgang Foestner and Bernhard Wrobel,
Photogrammetric Computer Vision, Springer;
Richard Harley and Andrew Zisserman; Multi-
view Geometry in Computer Vision,
Cambridge University Press.
Instructor: Rongjun Qin, Ph.D. Page 49
Course logistics
Phone: (614) 292-4356
Office: Bolz Hall 218B – Currently Virtual
Office Hours/ Availability Outside of Class Time: Tu 2 -3 pm or by appointment
outside these hours.
Class Meeting Times: Tu- Th 12:45 am -2:05 pm
Classroom: Journalism Bldg 387
GRADING
Homework assignments 60%
Attendance 10%
Midterm 10%
Final 20%
Instructor: Rongjun Qin, Ph.D. Page 50
Assignments
Tri-weekly/monthly, including programming
assignment and/or reading notes
It should be written in C/C++, MATLAB is
discouraged but allowable. Refers to the slide of
how to set up OpenCV
Assignment Release Time and Submission
Deadlines – refer to syllabus (Always Thursday)
Report format -
Instructor: Rongjun Qin, Ph.D. Page 51
Questions?
Instructor: Rongjun Qin, Ph.D. Page 52
