|
Modules | |
| Geometric calibration and augmentation of a set of camera. | |
| Feature points detection | |
| Photometric Calibration | |
| Viewsets | |
Detailed Description
This module depends on Starter: Basic structures and mathematical tools and provides tools to detect key points, or feature points, and to match them against a set of trained model points.
Garfeild is designed to quickly detect a planar object on an image. It is a two phases process: a training, or off-line phase, and an on-line phase. During training, yape detects feature points on a model image. Patches are extracted, their orientation corrected, and their appearance learned. The planar_object_recognizer::build_with_cache method takes care of the learning process.
During the on-line phase, planar_object_recognizer calls yape to detect feature points on the input image. It establish matches with model points. It then run RANSAC to sort out bad and good matches, and to find an homography sending relating model and input points.
