Techniques for automated analysis of images and videos. Image formation, feature detection, segmentation, multiple view geometry, recognition, and video processing.
Why take this course?
Computers increasingly require the ability to "see" their surroundings in order to interact with humans and with the three-dimensional world. This course introduces theory and techniques for analyzing the content of images. Applications of computer vision include robotics, autonomous vehicle navigation, industrial automation, content-based search in image databases, face and gesture recognition, and aides for the seeing-impaired.
This material is appropriate for the 5000 level. This class will apply extensive and in-depth knowledge that builds on undergraduate learning through a conceptual understanding of the specialization.
Major Measurable Learning Objectives
contrast common image formation models
implement various ways of extracting features from images
segment an image into meaningful regions
derive the theory behind multi-view geometry
implement various approaches to recognizing objects and scenes in images
implement techniques for processing video sequences
Percentage of Course
1. Features and filters: linear filters, edge detection, binary image analysis, image pyramids, texture
2. Grouping and fitting: fitting lines and curves, robust fitting, RANSAC, Hough transform, segmentation, clustering
3. Multiple views and motion: dense motion estimation, optical flow, camera model, image formation, planar homography, image warping, Epipolar geometry, stereo and multi-view reconstruction, invariant local features
4. Recognition: instance recognition with local features, bag-of-words representations, shape matching, part-based models, face detection and recognition, sliding window detection
5. Video processing: motion descriptors, tracking, background subtraction, activity recognition