#include <interest_point.h>
|
|
typedef std::vector< float > | descriptor_type |
| |
|
typedef descriptor_type::iterator | iterator |
| |
|
typedef descriptor_type::const_iterator | const_iterator |
| |
|
|
| InterestPoint (float x=0, float y=0, float scale=1.0, float interest=0.0, float ori=0.0, bool pol=false, uint32_t octave=0, uint32_t scale_lvl=0) |
| |
|
const_iterator | begin () const |
| |
|
iterator | begin () |
| |
|
const_iterator | end () const |
| |
|
iterator | end () |
| |
|
size_t | size () const |
| |
|
float | operator[] (int index) |
| |
| bool | operator< (const InterestPoint &other) const |
| |
|
std::string | to_string () const |
| | Generates a human readable string.
|
| |
|
void | setFromCvKeypoint (Eigen::Vector2d const &key, cv::Mat const &cv_descriptor) |
| | Copy IP information from an OpenCV KeyPoint object.
|
| |
|
|
float | x |
| | Subpixel (col, row) location of point.
|
| |
|
float | y |
| |
| float | scale |
| |
|
int32_t | ix |
| | Integer location (unnormalized), mainly for internal use.
|
| |
|
int32_t | iy |
| |
| float | orientation |
| |
|
float | interest |
| | The interest measure (could be Harris, LoG, etc.).
|
| |
|
bool | polarity |
| | These are some extras for SURF-like implementations.
|
| |
| uint32_t | octave |
| |
|
uint32_t | scale_lvl |
| |
| descriptor_type | descriptor |
| |
A class for storing information about an interest point in the format the NASA ASP can read it (very useful
◆ operator<()
| bool dense_map::InterestPoint::operator< |
( |
const InterestPoint & |
other | ) |
const |
|
inline |
std::sort can be used to sort InterestPoints in descending order of interest.
◆ descriptor
| descriptor_type dense_map::InterestPoint::descriptor |
And finally the descriptor for the interest point. For example, PCA descriptors would have a vector of floats or doubles...
◆ octave
| uint32_t dense_map::InterestPoint::octave |
This is the integer location in scale space (used for indexing a vector of interest images)
◆ orientation
| float dense_map::InterestPoint::orientation |
Since the orientation is not necessarily unique, we may have more than one hypothesis for the orientation of an interest point. I considered making a vector of orientations for a single point. However, it is probably better to make more than one interest point with the same (x,y,s) since the descriptor will be unique for a given orientation anyway.
◆ scale
| float dense_map::InterestPoint::scale |
Scale of point. This may come from the pyramid level, from interpolating the interest function between levels, or from some other scale detector like the Laplace scale used by Mikolajczyk & Schmid
The documentation for this struct was generated from the following file:
