1. Administrivia

• Five slip days are available for the semester
• They will be used for programming assignments and written homework

Course Overview

1. Low and mid-level vision
   1. Image processing, segmentation
2. Reconstruction
   1. Camera, geometry and physics of image formation
3. Recognition
   1. Primer on machine learning, convolution neural networks

2. Image Filtering

Images as functions

An image can be thought of a function from coordinates to values.

• Grayscale: $f(x,y)$ gives the intensity at position $(x,y)$.
  • A gray-scale image can be represented as a 2D array.
• Color: $f(x,y)$ gives the color channels at position $(x,y)$.
  • $f(x,y) = [r(x,y),g(x,y), b(x,y)]$.
  • A color image has 3 channels: RGB.

Geometric manipulations

• $f'(x,y) = f(T(x,y))$
• For example, $f'(x,y) = f(y, x)$ rotates an image 90 degrees clockwise.

Appearance transformations

• $f'(x,y) = g(f(x,y))$