Guest Editorial: Fast Energy-Minimization-Based Imaging and Vision Techniques
✍ Scribed by Emanuele Salerno
- Book ID
- 102617581
- Publisher
- Elsevier Science
- Year
- 2001
- Tongue
- English
- Weight
- 84 KB
- Volume
- 7
- Category
- Article
- ISSN
- 1077-2014
No coin nor oath required. For personal study only.
✦ Synopsis
A wide class of imaging and computer vision methods is based on the optimization of suitable functionals, which contain all the information that can be formalized on the problem. These functionals are often in the form of energies, so that all these methods are referred to as energy-minimization methods. Their attractiveness is due to their theoretical soundness, and thus to the high quality of the results expected. Unfortunately, two main drawbacks are still hampering their practical application: first, the algorithms must be ''tuned'' by means of free parameters, whose choice is difficult to be made automatically, and second, these algorithms are often very expensive computationally. For these reasons, energy-minimization methods are not used extensively in real-world. Indeed, although the velocity of an algorithm is a relative concept, depending on its application, the elapsed times for these algorithms are still prohibitive in most cases, even though real-time or nearly real-time performances are not required.