SIFT
|
#include <interestpoint.hpp>
Public Member Functions | |
InterestPoint ()=default | |
InterestPoint (Point< u16_t, u16_t > loc, f32_t scale, u16_t octave, u16_t index) | |
Static Public Member Functions | |
static bool | cmpByFilter (const InterestPoint &a, const InterestPoint &b) |
Public Attributes | |
f32_t | scale |
u16_t | octave |
u16_t | index |
bool | filtered = false |
Point< u16_t, u16_t > | loc |
f32_t | orientation |
std::vector< f32_t > | descriptors |
A class, which holds all the necessary information about the collected interest points
|
default |
|
inlineexplicit |
|
inlinestatic |
Orders interest points by the fact if they are filtered or not. So the filtered can be deleted from the end of a vector
std::vector<f32_t> sift::InterestPoint::descriptors |
bool sift::InterestPoint::filtered = false |
A flag, which shows if the interestpoint was filtered out by sift
u16_t sift::InterestPoint::index |
This is needed because of the fact, the corresponding DoG must not be searched again. And we get a linear access to the corresponding DoG element.
u16_t sift::InterestPoint::octave |
The octave of the interestpoint, because scale alone doesn't identify clearly
f32_t sift::InterestPoint::orientation |
The orientations of the interest point
f32_t sift::InterestPoint::scale |
the scale of the interestpoint