By the end of this course Students will be able to explain the concepts behind computer based recognition and the extraction of features from raster images. Students will also be able to illustrate some successful applications of vision systems and will be able to identify the vision systems limitations.
Things you will cover in this course:
- Concepts behind computer-based recognition and extraction of features from raster images.
- Applications of vision systems and their limitations.
- Overview of early, intermediate and high level vision,
- Segmentation:
- region splitting and merging;
- quad tree structures for segmentation;
- mean and variance pyramids;
- computing the first and second derivatives of images using the isotropic,
- Sobel and Laplacian operators;
- grouping edge points into straight lines by means of the Hough transform;
- limitations of the Hough transform;
- parameterization of conic sections.
- Perceptual grouping:
- failure of the Hough transform;
- perceptual criteria;
- improved Hough transform with perceptual features;
- grouping line segments into curves.
- Overview of mammalian vision:
- experimental results of Hubel and Weisel;
- analogy to edge point detection and Hough transform;
- Relaxation labeling of images:
- detection of image features;
- Grouping of contours and straight lines into higher order features such as vertices and facets.
- Depth measurement in images.
Reference material:
1. “Computer Vision: A Modern Approach”, David Forsyth, Jean Ponce, Prentice Hall, 2003.2. “Computer Vision”, Linda G. Shapiro, George C. Stockman, Prentice Hall, 2001.
3. “Handbook of Mathematical Models in Computer Vision”, Nikos Paragios, Yunmei Chen, Olivier Faugeras, Birkhäuser, 2006.
0 Comments