Principal Investigator
|
Project Title
| Estimation of Jump Curves and Surfaces using Locally-Varying Thresholds |
|---|
Brief Description for General Publications
Many conventional estimators of boundaries in images suffer from the drawback that they need to be tuned manually to yield, often only after a tedious process, an acceptable result. The exploration of statistical estimators, such as ours, is appealing on the grounds that they have theory to support them, which makes it possible to understand better their abilities and limitations. The research ventures into an area that has received relatively little attention. An important reason for this is that the incorporation of more complex boundary models, notably vertices, requires considerably higher computational efforts, yet these are more manageable with our likelihood-based method than others. |