Tracking algorithms-Tracking an hand-held camera by estimating it’s pose in a small Workspace environment. There are two main approaches:
Introduction
Tracking algorithms-Tracking an hand-held camera by estimating it’s pose in a small Workspace environment.
There are two main approaches
- Tracking while operate with prior knowledge of the environment (some form of a map, CAD Model, etc.)
- Extensible tracking – Tracking in scenes without any prior knowledge. Simultaneous localization and mapping
PTAM
- Tracking a calibrated camera in an unknown scene while building a map of it’s environment. Splitting tracking and mapping into two separate tasks, processed in parallel
- Require fast, accurate and robust camera tracking for Real Time purposes
- A 3D model will be assembled from point features
Problem
The robustness of these systems to rapid camera motions still lags behind that of tracking systems which use known object models.
Solution
- Add edge features to e map and exploit their resilience to motion blur to improve tracking under fast motion
- Keep a rough estimate of the motion in order to calculate each edgelet response under blur
Results
Future works
- Integrate with PTAMM project in order to reduce computation time
- Insert occlusion detecting
- Code organizing + Parallel computing and adjusting to windows/mobile phone
