Emerging Topics in Computer Vision

4.5. Conclusion

Our goal in this chapter was to approach robust estimation from the point of view of a practitioner. We used a common statistical framework with solid theoretical foundations to discuss the different types and classes of robust estimators. Therefore, we did not dwell on techniques that have an excellent robust behavior but are of a somewhat ad hoc nature. These techniques, such as tensor voting (see Chapter 5), can provide valuable tools for solving difficult computer vision problems.

Another disregarded topic was the issue of diagnosis. Should an algorithm be able to determine its own failure, one can already talk about robust behavior. When in the late 1980s robust methods became popular in the vision community, the paper [ ...