Feature Matching. Local matching, global mathing, virtual feature detectors and descriptors. More...
Classes | |
struct | vslam::Cell |
class | vslam::Grid |
Fast lookup of content by image coordinates. More... | |
class | RobustFeatureMatcher |
Handles unguided feature matching and robust estimation. More... | |
struct | RobustFeatureMatcherStats |
statistics output data structure for RobustFeatureMatcher More... | |
class | VirtualFeatureDetector |
Virtual feature detector for OpenCV feature2d framework. More... | |
Namespaces | |
namespace | vslam |
Namespace containing implementation of visual odometry. | |
Modules | |
Guided Matching | |
Matching whre approximate location of feature beeing matched is known. | |
Virtual Feature Matching | |
Implementation of virtual matching for OpenCV detector-descriptor-matcher interface. | |
Files | |
file | feature_matching.cpp |
implementation of feature_matching.h. | |
file | feature_matching.h |
Virtual feature detectors, descriptors and matchers (for simulation) and global feature matching code. | |
file | graph_builder.h |
Contains code that performs functionality of Node Builder, Node Destroyer, Edge Builder and Landmark Manager (which are components of Main Data Structure). | |
file | grid.h |
Class for fast lookup of features by their image plane coordinates. | |
Functions | |
int | vslam::col (const Cell &c) |
void | vslam::nonMaximaSuppresion (const vector< cv::KeyPoint > &keypoints, double max_num, vector< int > &supressed_keypoint_idxs) |
Reduces number of detected fetures. Causes features to be more evenly distributed across the image. | |
int | vslam::row (const Cell &c) |
Feature Matching. Local matching, global mathing, virtual feature detectors and descriptors.
int vslam::col | ( | const Cell & | c | ) | [inline] |
void vslam::nonMaximaSuppresion | ( | const vector< cv::KeyPoint > & | keypoints, |
double | max_num, | ||
vector< int > & | supressed_keypoint_idxs | ||
) | [inline] |
Reduces number of detected fetures. Causes features to be more evenly distributed across the image.
Partitions image into a grid and from each grid selects up to k points based on feature detector response strength so as to reduce number of features.
max_num | maximum number of features allowed. If some cells in the grid contain less than k features, there is that much less selected features. |
int vslam::row | ( | const Cell & | c | ) | [inline] |