Introduction

This course serves as an introduction to Computational Photography, a study of the perceptual and technical aspects of the capture and depiction of reality by computational means.

Topics include basic image operations, filters, transformations and feature point detection. Applications such as creating panoramas, high-dynamic range images, image morphing, inpainting and retargeting will be used to illustrate applicability of the basic operations. There are programming assignments in this class, and a familiarity with Python is required. A working knowledge of linear algebra is helpful, and we will brush up against topics from calculus and probability as well.

Summer 2024 - Jay Summet

Weekly Schedule & Resources

Week 1: Welcome, OpenCV, Digital Images

Course Textbook - Computer Vision: Algorithms and Applications, 2nd ed. Richard Szeliski.

Textbook Readings:
- The Bidirectional Reflectance Distribution Function (BRDF) section on pages 68-69 - Optionally you may want to skim the diffuse & specular sections that follow.
- Section 2.3 The digital Camera (on pages 79-84)
- Section 10.0 (pg 609) - Introduction to the Computational Photography Chapter

Installing Python (3.10) and OpenCV (4.5.4):
- cs4475_env_setup.pdf - Enviornmental setup instructions
- cp_env.yml - Enviromental setup yml file.
- test_env_setup.py - Test your setup post-install with this file.
- We recommend miniconda
- Main OpenCV Website
- SciPy.org
- On Ubuntu: sudo apt-get install python3-opencv python3-scipy python3-numpy
OpenCV Code Examples
- blueSky.jpg
- holes.jpg
- slice-Dice.py - Python slicing operators with images.
- a.jpg
- b.jpg
- blending.py - Full image point operations using numpy array math.
- pixel-addressing.py - Accessing individual pixels with python looping constructs, making per-pixel decisions with conditionals.

OpenCV Documentation & Tutorials
- OpenCV-Python Tutorial
- API Documentation (For both C & Python, not all functions will have Python documentation...)
- OpenV - Full Tutorial in Python

Week 2: Pinhole Cameras, Rays

Suggested Readings: Szelisky - Point Operators, Pixel & Color Transforms & Compositing and Matting (3.1.1-3.1.3) (pgs 109-115)
Turning a Paris Apartement into a Giant Camera!
Pinhole calculator
Dual Photography [YouTube Video]
Accidental Cameras
The Exposure Triangle
Effects of Aperture & Shutter speed on moving subjects

Week 3: Linear Filters: Correlation & Convolution

Suggested Readings: Szelinski Section 3.2 - Linear Filtering, 3.21 - Separable Filtering, 3.22 - Examples of linear filtering (pgs 119-127)
Hugin - Open source panorama software.
Wikipedia - Convolution
Image Kernels - Explained Visually
Separable filters and Is a filter separable?
Open CV Code Examples
- oscar.jpg
- shiftLeft.py - using an offset identity kernel to shift all pixels 4 to the left.
- sharpen.py - Enhance high frequency elements by subtracting a smoothed version from the original image. Read more about the unsharp filter math

Week 4: Gradients & Edges

Suggested readings: Szeliski 7.2 Edges & Contours, 7.2.1 Edge Detection
Image Differentation (math)
Image Derivative (Wikipedia)
OpenCV Code Examples:

Week 5: Optics / Lenses / Sensors

Suggested Readings: Szelinski Section 2.1.5 Lens Distortions (pgs 63-66), 2.2.3 Optics (pgs 74-79).
The "Vertigo" Effect - Dolly zoom - camera motion combined with focal length change (zoom) to keep the subject at the same size while compressing/expanding the background. [Jaws | GOTG2 | Zoom vs physical dolly movement | Simulated zoom in software]
CD: Shutter Speed - Shutter speed, CCD vs CMOS, Mechanical vs Digital & Rolling shuters

Week 6: Frequency & Laplacian Pyramids

Suggested readings: Szeliski Section 3.4 Fourier Transforms, 3.5.3 Multi-Resolution representations (Sections 3.5.1 Interpolation and 3.5.2 Decimation are also germane to image pyramids)
Examples of filtering images in the Frequency (Fourier) domain
The Laplacian Pyramid as a Compact Image Code
OpenCV Python code example:
- original_minus_blurred.py
- oscar512.png