ENGN 1610 Image Understanding

Pedro Felzenszwalb
Email: pff (at) brown.edu

Lectures: MWF 2:00-2:50 PM Barus & Holley 157

Office hours: Wednesday 3-4pm or by appointment

Course description
Image processing is a technology experiencing explosive growth; it is central to medical image analysis and transmission, industrial inspection, image enhancement, indexing into pictorial and video databases, e.g., WWW, and to robotic vision, face recognition, and image compression. This senior-level undergraduate course covers theoretical underpinnings of this field and includes a series of practical MATLAB image processing projects. ENGN 1570 is recommended but not required.

Image formation
Low-level image processing
3D reconstruction
Motion estimation
Image segmentation
Object recognition

Computer Vision: Algorithms and Applications. Szeliski. Springer.
A draft PDF is available here.


Topic 1: Image Formation
Topic 2: Image Filtering and Edge detection
Topic 3: Multiview geometry and stereo matching
Topic 4: Dynamic Programming
Topic 5: Image Segmentation
Topic 6: Graph algorithms
Topic 7: Template matching
Topic 8: Distance transform and Hausdorff matching
Topic 9: Convolutional Neural Networks
Topic 10: Geometric Methods for Recognition
Topic 11: Motion and Optical flow

Readings Assignments

1) Szeliski chapter 2
2) Edge detection Handout
3) Szeliski chapter 11
4) Section 4 of this survey
5) Section 3 of this survey
6) Slides on distance transforms and Hausdorff matching
7) Paper on learning templates for Hausdorff matching
8) LeNet paper
9) ImageNet CNN
10) Recognition by Linear Combination of Models
11) Determining Optical Flow


Using images in MATLAB: example.m

Assignment 1 and test images
Due: Friday September 30

Assignment 2 and test images
Due: Friday October 14

Assignment 3 and files
Due: Tuesday November 1

Assignment 4 and files
Due: Thursday November 17

Assignment 5 and test images
Due: Friday December 16