Active Contours is about the computer analysis of moving video images. Numerous applications are illustrated from computer graphics animation, user-interface design, medical imaging, automated surveillance and robotics. This book can be downloaded freely in postscript format.
In a characteristically no-nonsense, mathematically solid treatment, Blake and Isard take the subject of active contours to new heights of theoretical sophistication and practical application. The latter addresses the difficult task of visually tracking the motions of a variety of complex objects captured by a video camera feeding a frame-rate video digitizer. The impressive technical achievement of the book is a novel, probabilistic interpretation of active contours built on a geometric substrate that combines B-spline curve bases with shape spaces defined by global deformations.
This combination leads to a new class of very fast and highly robust (non-Gaussian) active contour propagation algorithms. Another noteworthy achievement is the ability of these new tracking algorithms to learn the complex motions of specific objects through observation, thereby automatically tuning the tracker with greater selectivity
to objects of interest, further enhancing its robustness.
This book defines the state-of-the-art of contour-based object tracking algorithms. It is required reading for anyone interested in computer vision theory and in the design of working computer vision systems.
Table of Contents
- Active shape models
- Spline curves
- Shape-space models
- Image processing techniques for feature location
- Fitting spline templates
- Probabilistic models of shape
- Dynamical models
- Dynamic contour tracking
- Learning motion
- Non-Gaussian models and random sampling algorithms
File size: 4.30 MB
Number of pages: 352