Personal tools
You are here: Home Teaching Computational Photography

Computational Photography

8652, Vorlesung mit √úbungen, Master, 5.0 ECTS Punkte

Dozent: Prof. Dr. Matthias Zwicker
Ort: Engehaldenstrasse 8, Room 002
Zeit: Thursdays, 14:15-17:00

Poisson blend

 Gradient based image manipulation: the user provides the input on the left, and the result on the right is computed automatically (from Perez et al., "Poisson image editing").


The combination of digital cameras and computer processing provides fascinating opportunities to enhance and manipulate digital images. This class explores the field of computational photography, which is at the intersection of image processing, computer vision, and computer graphics. The research objective of computational photography is to allow everyday users to capture better pictures, and to give them the ability to easily and creatively manipulate and enhance their images.

In this class we will cover the fundamentals of image formation in digital cameras and discuss limitations and deficiencies in this process. We will discuss image restoration techniques to remove low-level artifacts of the imaging process such as noise, blur, and contrast limitations. We investigate the representation of color and how to manipulate it. We discuss algorithms that allow users to easily and creatively manipulate images with minimum effort, as shown in the example image below. We also explore devices and representations that capture light beyond two-dimensional images. For an overview of the class, see also the introductory slides of the first lecture.

The class includes practical exercises. Students will implement and experiment with algorithms that are discussed in class. All class material is available on Ilias.

Learning Outcomes

On successful completion of the this class, you will be able to:

  • Identify the potential of computer algorithms to enhance digital photography. 
  • Describe a number of automatic and user assisted algorithms to manipulate digital photographs, including: image denoising, deblurring, warping, segmentation, and compositing; high dynamic range imaging; automatic panorama stitching; reconstruction of 3D information. 
  • Explain mathematical techniques that build the background for these algorithms, including: Fourier analysis; gradient domain techniques; linear least squares problems; optimization algorithms based on graph cuts. 
  • Apply the mathematical techniques to implement these algorithms in Matlab or other programming environments. 


This schedule is preliminary and subject to change. Material for each lecture is available on Ilias.

22.09. Introduction, image formation
29.09. Color: color imaging, demosaicing, white balancing
06.10. Tone mapping: high dynamic range imaging, style transfer
13.10. Sampling, reconstruction, & the frequency domain
20.10. Image restoration: denoising & deblurring
27.10. Gradient domain image manipulation
03.11. Image manipulation using optimization
Hallucinating image content: patch-based and data-driven methods
17.11. Decomposing images: intrinsic images, alpha matting, non-local editing
24.11. Warping & morphing
01.12. Panoramas
08.12. Automatic alignment
15.12. Cameras that do not capture images: coded aperture, light fields
22.12. Reserve


The schedule for the exercises is listed below. You will find all material for the exercises on the Ilias platform.

Due Date
02.10. Color
16.10. High dynamic range imaging
30.10. Fourier transforms, deblurring
13.11. Gradient domain & graph cuts
27.11. Warping, panoramas
18.12. Light fields

Class Materials

All class materials will be available on the Ilias platform.


There will be a written exam. It will take place on February 12, 2015, from 10:00-12:00, in room 003 at Engehaldenstrasse 8. You may bring two sheets (four pages) of handwritten notes to the exam. No other tools or materials are allowed.

Suggested Reading

These books go far beyond the material covered in class. The list is intended as a suggestion for independent further reading if you are interested in a broader overview of the field or an in-depth understanding of specific topics. Some of the books are available in the CGG library for viewing (but not for home loans).

  • Computer Vision: Algorithms and Applications, by Richard Szeliski, to be published.A draft is available online. This text provides a broad overview of computer vision, image processing, and computational photography. It discusses several topics covered in the class, and many more.
  • Photography, by Barbara London, John Upton, and Jim Stone, published by Prentice Hall, 10th edition, 2010. A classic on the craft of photography. 
  • Color: An Introduction to Practice and Principles, by Rolf Kuehni, published by Wiley, 2nd Edition 2004. A short and concise introduction to color.
  • High Dynamic Range Imaging, by Erik Reinhard et al., published by Morgan Kaufmann, 2nd edition 2010. A comprehensive and up-to-date text on HDR imaging.
  • Digital Image Processing, by Rafael Gonzalez and Richard Woods, published by Prentice Hall, 3rd edition 2007. A classic book on image processing, covers standard topics such as image representation, color images, filtering, Fourier transforms, wavelets, and image compression.
  • Image Processing: The Fundamentals, by Maria Petrou and Costas Petrou, published by Wiley, 2nd Edition 2010. This book focuses on statistical modeling of images using random fields and its applications to image restoration and segmentation.
  • Multiple View Geometry in Computer Vision, by Richard Hartley and Andrew Zisserman, published by Cambridge University Press, 2nd edition 2004. A standard text on 3D reconstruction from mutliple views. It provides an in-depth explanation of the mathematics behind multiview 3D reconstruction, going far beyond the class material.
Document Actions