An EM approach for contour tracking based on point clouds
| Autor: | Marcus Baum, Hauke Kaulbersch, Peter Willett |
|---|---|
| Rok vydání: | 2016 |
| Předmět: |
020301 aerospace & aeronautics
Noise measurement Computer science business.industry Point cloud 020206 networking & telecommunications 02 engineering and technology Kalman filter Tracking (particle physics) 0203 mechanical engineering Video tracking Expectation–maximization algorithm 0202 electrical engineering electronic engineering information engineering Batch processing Computer vision Artificial intelligence Noise (video) business |
| Zdroj: | MFI |
| DOI: | 10.1109/mfi.2016.7849542 |
| Popis: | This work considers the problem of tracking a mobile object with an unknown shape based on noisy point cloud measurements from the object contour. For this purpose, an Expectation Maximization (EM) method is developed that is capable of simultaneously estimating the location and shape parameters of an object based on a temporal sequence (i.e., batch) of point clouds. The benefits of the EM approach are evaluated with respect to a standard Kalman filter that performs a greedy association of measurement points to contour points. Especially for scenarios with high noise and sparse point clouds, the batch processing scheme of the EM approach is beneficial. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|